university of groningen on the mechanism of cationic lipid ... · antisense molecules bind to...
TRANSCRIPT
University of Groningen
On the mechanism of cationic lipid-mediated delivery of oligonucleotidesShi, Fuxin
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite fromit. Please check the document version below.
Document VersionPublisher's PDF, also known as Version of record
Publication date:2004
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):Shi, F. (2004). On the mechanism of cationic lipid-mediated delivery of oligonucleotides: towards anantisense therapy. s.n.
CopyrightOther than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of theauthor(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
Take-down policyIf you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediatelyand investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons thenumber of authors shown on this cover page is limited to 10 maximum.
Download date: 27-05-2020
On the mechanism of cationiclipid-mediated delivery of oligonucleotides:
towards an antisense therapy
Fuxin Shi2004
This research project was supported by a grant from The Netherlands
Organization for Scientific Research (NWO)/NDRF Innovative Drug Research
(940-70-001).
The studies described in this thesis were performed at the Department of
Membrane Cell Biology, Faculty of Medical Sciences, University of Groningen,
the Netherlands.
Antisense oligos described in this thesis were designed and manufactured by
Biognostik, Germany.
This thesis is available in the Library of University of Groningen
© 2004 by F. ShiAll the rights reserved. No part of this book may be reproduced or transmittedin any form or by any means without written permission of the author and thepublisher holding the copyright of the published articles.
Printed by: Facilitair bedrijf, University of Groningen
International Standard Book Number: 90-367-1985-2
RIJKSUNIVERSITEIT GRONINGEN
On the mechanism of cationiclipid-mediated delivery of oligonucleotides:
towards an antisense therapy
Proefschrift
ter verkrijging van het doctoraat in de
Medische Wetenschappen
aan de Rijksuniversiteit Groningen
op gezag van de
Rector Magnificus, dr. F. Zwarts,
in het openbaar te verdedigen op
maandag 29 maart 2004
om 14.45 uur
door
Fuxin Shi
geboren op 27 augustus 1972
te Harbin, China
Promotor: Prof. Dr. D. Hoekstra
Beoordelingscommissie: Prof. Dr. C. G. Kruse Prof. Dr. H. J. Haisma Prof. Dr. H. W. G. M. Boddeke
ISBN: 90-367-1985-2
Paranimfen: Xuedong Yan Anita Nomden
Financial support for printing:Faculty of Medical SciencesNWOBCNSolvay Pharmaceuticals, B.V.Biognostik
Front cover: cells incubated with fluorescently labelled oligos (green) complexed
with cationic lipids (red) (upper picture); a brain slice incubated with
fluorescently labelled oligos(green) and nuclei (red) counterstained with
propidium iodide (lower picture).
Back cover: isolated nuclei incubated with fluorescently labelled oligos (green)
complexed with cationic lipids (red) (upper image); a brain slice incubated with
fluorescently labelled oligos (green) complexed with cationic lipids (red).
Contents Page
Chapter 1: Introduction and scope of the thesis 9
Chapter 2: Make sense of antisense oligonucleotides 15
Chapter 3: Efficient cationic lipid-mediated delivery of antisense 45
oligonucleotides into eukaryotic cells: down-regulation
of the corticotrophin-releasing factor receptor
Chapter 4: Antisense oligonucleotides reach mRNA targets via 65
the RNA matrix; downregulation of the 5-HT1A receptor
Chapter 5: Interference of polyethylene glycol-lipid analogues 89
with cationic lipid-mediated delivery oligonucleotides;
role of lipid exchangeability and non-lamellar transitions.
Chapter 6: Oligonucleotides enter cells mainly through a nucleic 115
acid channel on the explant of rat brain.
Chapter 7: Cationic liposome-mediated delivery of proteins into 133
eukaryotic cells: entry along the pathway of
caveolae-mediated endocytosis.
Chapter 8: Summary and Perspectives 151
Samenvatting 157
Acknowledgements 163
Publication list 165
9
Chapter1
General introduction and scope of the thesis
Chapter 1
10
Genetic information
Literally, biology means “study of life”. Life originates from the correct organization of one or
more cells, governed by a unique arrangement of DNA information. DNA is defined as a nucleic
acid that harbors the genetic information of the cell, is capable of self-replication, and enables
the synthesis of RNA. The sequence of the nucleotides determines individual hereditary
characteristics. The central dogma of genetics is that information flows from DNA to functional
proteins via a specific template mRNA molecule. During the information flow, amplification
occurs---from a single DNA molecule, multiple RNA transcripts are produced, from which even
more protein molecules are translated. It is vital that each conversion is carefully regulated.
Obviously, defects in transcription and translation may be detrimental to cells, and accordingly,
harmful if not lethal to individuals. At pathological conditions, like inflammation and tumor
development, the expression of numerous genes may be perturbed. Hence, an appropriate
adaptation and/or modulation of gene expression could be beneficial to individuals in order to
efficiently protect the cellular system against tumors, viral infection and inflammation alike.
What we can learn from nature.
The majority of current treatments of illness usually rely on the application of low molecular
weight chemical compounds to interact with the target proteins and alter their function. Although
such approaches are aimed to correct the outcome and alleviate the syndrome, the
malfunctioning gene that causes illness is not altered. Each gene consists of two long chains of
nucleotides, the sense chain and antisense chain, respectively. The copy of the sense chain
represents the basic structure of the mRNA, which is the blueprint for construction of a protein.
Accordingly, to alter the functioning of a potentially harmful gene, it would make ‘sense’ to
target to the mRNA. In prokaryotes, there is abundant evidence which shows that natural
antisense transcripts, i.e., the copies of the antisense chain on the genes that regulate gene
expression, can control translation, mRNA turnover or transport. In fact, by a systematic search
of mRNA sequences in human, also a large number of antisense transcripts have been identified.
Thus, the natural regulation of gene expression with antisense sequences and double-stranded
RNAs may well be a common phenomenon in cellular life in general. These findings have been
(co-)instrumental in the initiation of the antisense technology, which relies on the concept that
antisense molecules bind to complementary mRNA through Watson-Crick base pairing, thereby
interfering with gene expression.
Antisense technology.
Antisense oligonucleotides are synthetic DNA or RNA analogues. The mechanism of
antisense action is generally believed to occur through Watson-Crick base pairing, causing
Introduction
11
antisense oligonucleotides to bind/hybridize to complementary mRNA. As a result, translation is
arrested or mRNA degradation is induced via subsequent activation of RNase H. mRNA often
consists of thousands of nucleotides and displays a three-dimensional structure. The selection of
an appropriate antisense sequence as antisense oligomer target is a crucial step and remains a
major challenge in the successful application of antisense technology. In principle, antisense
oligonucleotides are targeted to the “open” region on the mRNA. The accessible mRNA regions
can be picked by spacing oligonucleotides along the mRNA, identifying the RNase H cleavage
sites on mRNA upon binding of antisense, or by determining the potential “open sites” by
predicting mRNA structure. Another crucial step toward the success of antisense technology is
the need to improve the stability of antisense molecules and enhance the affinity to the targeted
mRNA. Chemical modification of either the nucleotides or their backbones is aimed at fulfilling
such requirements.
After a “golden” antisense molecule is chosen with high stability and good affinity to mRNA,
the next step is how to direct the antisense molecule to the target mRNA and interfere with its
functions in the cells.
Overcoming extracellular and intracellular barriers
Systemically applied oligonucleotides have to travel a long journey to reach the target mRNA
in a cell. They are first facing a cleaning machinery in the body—the reticuloendothelial system,
like liver, spleen and lung. After bypassing the reticuloendothelial system, oligonucleotides may
arrive at a given organ, where they have to travel through the extracellular matrix to reach the
cell surface. Once within the cells, there are numerous intracellular organelles, which the
oligonucleotides may encounter. The mRNAs transcribed from the DNA are edited in the
nucleus, and then transported to the endoplamic reticulum (ER) in the cytosol as a template for
protein synthesis. Thus, to meet the target mRNA, oligonucleotides need to be released into the
cytoplasm so that they can reach the nucleus or interact with the mRNA on its way to the
ribosome, free in the cytosol or attached to the ER. In tissue culture, oligonucleotides are poorly
internalized by cells. Moreover, the small fraction that is taken up by the cells usually resides in
endosomes and lysosomes, without being deposited into the cytosol. Accordingly, delivery
vectors have been developed to enhance the cellular uptake and to promote release into the
cytosol, or to target delivery of oligonucleotides in vivo to desired organs. In general, delivery
vectors are classified as viral and non-viral vectors. The viral vectors have a high efficacy but
they usually also suffer from a high immunogenicity. The non-viral vectors show no
immunogenicity but often display a relatively low efficacy.
Chapter 1
12
Scope of the thesis
This thesis focuses on improving our understanding of antisense action. To do so, an
investigation was carried out of the entry pathway and intracellular processing of cationic lipids
as delivery agents for ODNs and nucleic acids. In addition, the delivery potential of such a
carrier for proteins was investigated.
First, the recent developments and obstacles in antisense application are reviewed (Chapter 2).
Antisense oligonucleotides are poorly taken up by cells. We therefore utilized the synthetic
cationic amphiphile, SAINT-2, in conjunction with helper lipid
dioleoylphosphatidylethanolamine (DOPE) to facilitate the translocation of antisense
oligonucleotides across the plasma membrane. SAINT-2/DOPE not only enhanced the cellular
uptake of oligonucleotides over more than a thousand fold, but also enabled the oligonucleotides
to escape from endosomal compartments in tissue culture. In this manner, an antisense effect was
elicited in the nanomolar concentration range, causing the specific down-regulation of both
targeted mRNA and protein in a non-toxic manner. Also the effect of serum on the cationic lipid-
mediated delivery of oligonucleotides was mimicked in tissue culture (Chapter3). The
subcellular trafficking of oligonucleotides and antisense action was studied in detail in
subsequent work. We observed that oligonucleotides quickly moved into the nucleus after
escaping from endosomes. In the nucleus they became extensively bound to the nuclear matrix.
The potency of antisense oligonucleotides was revealed by demonstrating specific
downregulation of both endogenously and exogenously expressed protein targets, and by
showing the functional elimination of the targeted proteins (Chapter 4).
Cationic lipoplexes usually display a relatively poor stability in an in vivo environment, which
is why a protocol was developed to stabilize cationic lipoplexes with polyethylene-glycol lipid
derivatives. We observed that the polyethylene glycol analogues could stabilize the cationic
lipoplexes by preventing hexagonal phase formation of the complex. Although complex
internalization was not significantly affected, the presence of polyethylene glycol interfered with
the interaction of cationic lipoplexes with the endosomal membranes, and consequently impaired
the release of oligonucleotides from endosomes, thereby abolishing the antisense effect. We
therefore developed a programmed and functional delivery of antisense oligonucleotides from
stabilized cationic lipoplexes by incorporating exchangeable polyethylene-glycol lipids (Chapter
5).
Although a successful application of an antisense technology in vitro relies on the use of
appropriate carriers, this may not always be needed for in vivo delivery of antisense
oligonucleotides. Quite remarkably, we found that oligonucleotides locally applied in rat brain
Introduction
13
were able to enter brain cells without the need of any carrier, indicating that in vivo entry may be
accomplished via a non-endocytic mechanism (Chapter 6).
Finally, since the delivery of therapeutic proteins is also of interest, we investigated whether
the SAINT-2/DOPE system, capable of efficiently carrying plasmids and oligonucleotides into
cells, could also be applied for protein delivery. Indeed, we observed that cationic lipids could
facilitate translocation of proteins across the cell membrane in an efficient and functional
manner. In tissue culture, the protein complexes are likely internalized via caveolae-mediated
endocytosis (Chapter 7).
A summary and perspectives for future developments concludes this thesis (Chapter 8).
15
Chapter 2
Make sense of antisense oligonucleotides
Fuxin Shi and Dick Hoekstra
Submitted
Chapter 2
16
Abstract
For more than two decades antisense oligonucleotides (ODNs) have been used to modulate
gene expression for the purpose of applications in cell biology, and for development of novel
sophisticated medical therapeutics. Conceptually, the antisense approach represents an elegant
strategy, involving the targeting to and association of an ODN sequence with a specific mRNA
via base-pairing. As a result, the translation of the gene into protein will be impaired, thereby
potentially revealing its functional and/or harmful effect in normal and diseased cells/tissue,
respectively. It is still poorly resolved how to design the most efficient antisense
oligonucleotides, but chemical modification of their structure can improve their efficiency of
action, in part due to an enhanced intracellular stability. Appropriate carriers have been
developed to allow efficient entry of ODNs into cells in vitro, and the mechanisms of delivery,
both in terms of biophysical requirements for the carrier and cell biological features of uptake,
are gradually becoming apparent. Remarkably, for some tissues such as liver and brain, it has
been shown that ODNs may acquire intracellular access and delivery into the nucleus, a step
needed for accomplishing the antisense effect, without the need of packaging into a delivery
vehicle. This suggests differences in entry mechanisms, knowledge that is imperative to further
appreciate and bolster antisense technology in its broadest sense. In this context, proper
consideration will be given here to antisense design and chemistry, and the challenge of extra-
and intracellular barriers to be overcome.
Make sense of antisense oligos
17
Introduction
In 1978, synthetic oligonucleotides (ODNs) complementary to Rous sarcoma virus mRNA
were shown to inhibit virus replication (1). This important observation represents one of the
hallmarks in the initiation of the development of antisense technology for therapeutic and cell
biological purposes. In fact, the principle of the approach reflects a naturally occurring
physiological event since endogenously expressed antisense molecules exist which are able to
regulate endogenous gene expression and to defend both prokaryotes and eukaryotes from viral
invasion (2, 3).
Over the last two decades, numerous studies have been carried out to obtain insight into the
mechanism of the action of antisense ODNs, including issues concerning optimal design,
chemical modification, specificity and pharmacology (4, 5). In principle, antisense technology
has a wide perspective of potential applications. At present, prompted by the rapid developments
in genomics, most of these applications are in fundamental research and are largely focussed on
an inhibition of gene expression, which provides crucial insight into the function and potentential
regulation of novel genes (6-8). However, there is a growing interest in developing drugs based
upon an antisense protocol, mostly aimed at interfering with viral infections and cancer (9, 10).
In addition, the antisense approach may be a good alternative to traditional agents for diagnosis
and treatment in nuclear medicine, although relatively few attempts have been made thus far to
evaluate this option (11-13).
Irrespective of the goal of application, a major challenge in antisense technolgy represents the
design of an antisense sequence, which is target-specific, effective and nontoxic (14-19). In spite
of drawbacks, significant progress has also been made in recent years in this area, which has
resulted in the commercial production of antisense therapeutics such as Vitravene, used in the
treatment of CMV retinitis in HIV infections (20, 21) and the undertaking of several promising
phase III trails (22-24).
As with drug development in numerous areas, a detailed understanding of antisense
mechanism and pharmacology are essential for successful application and further improvement
of its effectiveness. Here, we will discuss therefore the understanding thus far of molecular
obstacles and intracellular and extracellular barriers in antisense action and application.
Principle features of the mechanism of antisense action
Crucial toward a prosperous development and widely applied antisense technology is the need
for a detailed understanding of the mechanism of antisense action per se (Fig. 1). Conceptually,
the ability of antisense molecules to bind to complementary mRNA through Watson-Crick base
Chapter 2
18
pairing may sound simple. However, insight into issues as where and how antisense reaches its
target(s) are still largely obscure, especially in case of in vivo studies. Moreover, no evidence is
available which demonstrates the direct binding of ODNs within the cell to the targeted
sequence(s). It is generally believed that antisense either sterically blocks mRNA’s function (Fig.
1A) or promotes enzyme-mediated mRNA degradation (Fig.1B). As an example of the former
possibility, in a cell-free translation system it has been shown that antisense ODNs, such as
peptide nuclei acids (PNAs), which are unable to activate RNase H, can be targeted to mRNA
coding sites. As a result, an interference with polypeptide chain elongation occurs, resulting in
the biosynthesis of truncated protein fragments (Fig.1A; 4, 25, 26). Similarly, steric occupancy
of the 5’cap region with 2’-MOE ODNs interferes with the assembly of the 80S translation
initiation complex, and with the elevation of target mRNA (27). Steric interference of antisense
molecules on distinct RNA regions may also cause an inhibition of pre-mRA splicing (Fig. 1C)
or polyadenylation editing (Fig. 1D), and/or induce mutations of the encoding gene (28, 29).
Antisense-mediated down-regulation of mRNA as a result of the latter’s degradation by
RNase H action, may occur following binding of phosphodiester- or thioate-linked ODNs, as
revealed in cell lysates (30-33). Consequently, the expression of the encoded protein will become
reduced, an effect that will be apparent only after significant turnover of the pre-existing
endogenous protein pool. When the targeted ODN sequences are within the 5’or 3’ non-encoding
regions(Fig. 1 D), the antisense induced cleavage at these non-translation sites accelerates the
degradation of the entire target mRNA. Nevertheless, progress in carefully defining the
intracellular mechanism(s) of processing and action of antisense ODN has been particularly
frustrated by the technical inability to verify the localization of the relatively small amounts of
antisense that gain access within the cells, and the limited recovery of targeted mRNA or protein
fragments (Fig.1).
As noted, in the antisense field, RNase H is thought to play a pivotal role. However, it is
essential to take into account that only phosphodiester, phosphothioate and chimeric ODNs (cf.
Fig. 3) can activate RNase H. Any (chemical) modification of the nucleotide and/or changes in
the backbone will reduce or eliminate RNaseH recognition of the DNA-RNA duplex(31, 32).
Given the potentially effective role of RNases in antisense technology, further insight into the
presence of other intracellular RNases may prove worthwhile in fully exploiting the potency of
this technology (34). For example, the 2’, 5’oligoadenylates bind to and activate the latent RNase
L, cleaving viral and cellular RNAs at the 3’side of UpNp sequences, thus leading to an
inhibition of protein synthesis (35). In fact any RNase, which recognizes the double stranded
RNAs or RNA/DNA duplexes could play a key role in the overall action of antisense ODN. In
Make sense of antisense oligos
19
both plants and animals, RNAi is present which consists of about 22 nucleotides in length that is
homologous to the gene that is being suppressed. These 22-nucleotide sequences serve as a guide
sequence that instructs a multicomponent nuclease, RNA-induced silencing complex (RISC), to
destroy specific messenger RNAs (36-38). The RISC contains a helicase activity that unwinds
the double RNA strand, thereby allowing the antisense to bind to the targeted RNA sequence and
causing the ensuing activation of an endonuclease activity that hydrolyzes the targeted RNA at
the site where the antisense strand is bound. Recently, RNase III, which is involved in the
maturation of prokaryotic and eukaryotic RNA, has emerged as a key player in this new and
exciting biological field of RNA silencing or RNA interference (39). In fact RNase III appears
to contain very high affinity RNA binding sites, which readily interact with the double-stranded
RNA binding domain (dsRBD), thus initiating rapid and efficient cleavage.
Figure 1. Major sites of the actions of antisense oligonucleotides. A. Translationalarrest. Antisense oligonucleotides bound to complementary mRNA inhibit polypeptideselongation. B. RNase H activation. Hybridized mRNA is degraded by RNase H that isactivated upon the binding of antisense oligonucleotide on the mRNA. C. Inhibition ofthe splicing of pre-mRNA. Antisense oligonucleotides bound on the splicing site of pre-mRNA interfere with the maturation of mRNA by preventing the binding of spliceosomeon pre-mRNA. D. Pre-mRNA destabilization. Antisense oligonucleotides bound onencoding or non-encoding regions on pre-mRNA can accelerate the RNase-mediateddegradation of the mRNA, or interfere with polyadenylation or cap formation.
DNA
Pre-RNA
AAACAP
AAACAP
spliceosomes
AAACA
AAACA
CAP AAA
CAP AAA
CAP AAA
RNase
CAPAAA
proteinB. RNase H activationA. Translational arrest
C. Inhition of splicingD. Pre-mRNA destabilization
nucleus
cytosol
RNase
Nuclear pore
Chapter 2
20
Protocols for designing proper antisense sequences
Obviously, the proper selection of an appropriate antisense sequence is a crucial step and
remains a major challenge in the successful application of antisense technology (Fig. 2).
Knowledge of the RNA secondary structure, antisense affinity and antisense chemistry are key
factors in antisense design. mRNA is not a single stranded random coil but displays a secondary
or tertiary structure, which plays a significant role in determining the efficacy of ODN antisense
activity (40-47). Usually, a stretch of 10-30 nucleotides on the mRNA is selected as a potential
target for the antisense. This preselected mRNA fragment may engage in intramolecular base
pairing, thus constituting stable secondary and tertiary structures, which may render large parts
of the mRNA inaccessible towards interaction with the matching antisense ODNs. The number
of potential conformational states grows exponentially with the chain length. For example, the
number of hairpin conformations increases from 138 for a 10-nucleotide chain to 24,666 for a
16-nucleotide chain. RNA hairpins are stabilized predominantly by base-stacking interactions
(48).
mRNA folding and stem loops can be predicted by certain computer programs (49, 50), such
as mfold (51-52). However, precise prediction of these structures remains difficult. In addition,
cellular factors may influence the mRNA structures as well. For example, the association of
AUF1, one of A + U-rich binding factors, with RNA substrates induces the formation of
condensed RNA structures (53). The spliceosome (cf. Fig.1C) removes introns from pre-
messenger RNAs by a mechanism that entails extensive remodeling of the RNA structure. The
most conspicuous rearrangement involves disruption of 24 base pairs between U4 and U6 small
nuclear RNAs (snRNAs). In this case, the yeast RNA binding protein Prp24 has been shown to
re-anneal these snRNAs (54). Thus ODN ‘walks’, i.e., spacing ODNs of a given length at
intervals along the RNA and choosing the one with the highest activity, still play an important
role in selecting appropriate antisense molecules. In fact, currently a variety of approaches are
applied to design the most appropriate antisense sequence (Fig.2). Also sequences that are
targeted to initial coding regions are frequently applied since these regions lack secondary
structure. However, although an attractive target, such an approach has been shown to be of little
value in general, since ODNs targeted to these regions have been shown to often generate a poor
downregulation of mRNA content (14). Furthermore, translation initiation sites may display a
shared homology in both related and non-related genes, since ODNs targeted to these sites
displayed both antisense specific effects toward the targeted gene, and non-specific effects on
cell proliferation.
Make sense of antisense oligos
21
Figure 2. Schematic representation of antisense design. A. mRNA walking.Oligonucleotides of a given length, complementary to sequences along the RNAsequence, are synthesized and screened for antisense activity. B. Computer folding ofmRNA. By prediction of mRNA structure, oligonucleotides are designed to accessiblesites on the mRNA. C. Oligonucleotide array. In the scanning array, alloligonucleotides complementary to the target mRNA sites are synthesized andhybridized with mRNA transcripts. Antisense sequences are chosen by selecting theones with high affinity to the mRNA on the array. D. RNase H mapping. RNase H isused to cleave mRNA that hybridizes to a random oligonucleotide library. Appropriateantisense oligonucleotides are selected based on identifying RNase H cleavage sites oftarget mRNA.
Overall, four distinct approaches have been developed which are currently actively pursued in
the design of antisense molecules, though selection of active antisense molecules still remains
largely a matter of trial and error (Fig. 2). First, mRNA walking or sequence walking (Fig.2A),
which relies on synthesizing a variety of sequences which are targeted to distinct regions on a
given sequence. Usually some 100 different sequences will be tested in such an approach. The
outcome is that often only a few sequences turn out to be effective (55, 56). Obviously, the
advantage of this approach is that given the number of sequences applied an appropriate and
effective sequence may be recovered, but the procedure is costly, time-consuming and laborious.
However, such an approach may suit large-scale pharmaceutical purposes for screening of
optimal antisense drugs. A second procedure relies on computer facilitated screening (Fig.2B),
based on the predicted folding of mRNA in order to identify accessible sites (57-58). As
A: mRNA walking
B: Computer folding of mRNA
C: Oligonucleotide array
D: RNase H mapping
Prediction of mRNA folding
Oligonucleotide
RNase H
Sequencing RNA
Oligonucleotide
mRNA hybridisation
mRNA hybridisation
mRNA
oligonucleotides
mRNA
Scanning the array
Designing oligonucleotidesto the mRNA accessible sites
Analysing the activity of sequences
Chapter 2
22
discussed above, the precise prediction still suffers from structural uncertainties. On the other
hand, this is an inexpensive, fast and easy approach involving the screening of a few sequences.
With this approach usually an effective sequence can be obtained, even though it may not be the
most effective one. On the other hand, it should also be realized that it is not always necessary to
completely inhibit gene expression, particularly in case of functional studies.
In recent years, ODN arrays (Fig.2C) and RNase H susceptibility (Fig.2D) emerged as
important approaches to select antisense molecules. In the scanning array approach, all ODNs
complementary to the target mRNA sites are synthesized, and hybridization is performed with
mRNA transcripts in vitro, i.e., under non-physiological conditions (45, 59). Since these studies
are done in vitro, relevant cellular factors are not present in the system. Nevertheless, with this
approach, a good correlation has been obtained between the eventual antisense effect in cells,
and the binding affinity of antisense ODNs to the mRNA in vitro. With the progress in
automation and miniaturisation of DNA chips, this approach is quite promising and can be
readily adapted to a routine screening in common research laboratories.
Finally, appropriate antisense ODNs can be selected based upon the principle of RNase H-
mediated cleavage of target mRNA after binding of a random ODN library, followed by
analyzing accessible ODN binding sites on the mRNA by gel electrophoresis (60-62). The
advantage of this method is that the random ODN library is easy to synthesize and suitable to
any target mRNA, although the challenge remains to identify the precise cleavage sites, given
that a multifold of such sites may exist whereas the resolution, as obtained by gel
electrophoresis, is usually limited. Accordingly, further improvement is needed to adapt this
procedure as a generally applicable approach.
On improving Antisense Chemistry
Chemical modification of antisense ODNs (Fig. 3) is aimed at (i) improving stability and
affinity of ODNs to target mRNA, (ii) facilitating recruitment of intracellular enzymes (mostly
RNase H) to efficiently cleave targeted mRNA, and (iii) reducing/eliminating the toxicity of
ODNs as such. Early during development, antisense ODNs contained phosphorodiester (PO-
ODNs) linkages. PO-ODNs are capable to recruit and activate RNase H to ODN/mRNA hybrids.
The PO- ODNs per se are sensitive to nucleases and show relatively short half-lives, both
intracellularly and in circulation. Although still in use, currently most applications rely on the
employment of chemically modified ODNs. The most frequently used chemically-modified
antisense compounds are phosphorothioate-linked ODNs (PS-ODNs), in which unbridged
oxygen is replaced by sulfur. In fact PS-ODNs harbor more advantages than only their ability to
Make sense of antisense oligos
23
resist nuclease-mediated degradation. PS-ODNs not only exhibit a long half-life in vitro and in
vivo, but the modified compound retains the ability to recruit RNase H in order to degrade PS-
ODN/mRNA hybrids (63, 64). Importantly, when applied systemically, PS-ODNs may
spontaneously enter tissues, mostly liver, kidney, spleen, intestine and lung. A disadvantage of
the PS linkage is its affinity for proteins, which on the one hand prolongs the circulation time of
ODNs in vivo by protecting them from removal by filtration, but on the other hand frustrates the
interpretation of the antisense effect since non-sequence specific inhihition of cell proliferation
may occur (65, 66). For example, PS-ODNs may bind in a length and to some extent in a
sequence-dependent manner to heparin-binding proteins, such as fibroblast growth factors,
platelet-derived growth factor (PDGF), vascular endothlial growth factor (VEGF) and its
receptor, as well as to the epidermal growth factor receptor (EGFR) (17, 67, 68).
Figure 3. Oligonucleotide analogues and chemical modifications.
Chapter 2
24
Morpholino ODNs are nonionic DNA analogues, which, compared to PO-ODNs, show less
affinity to proteins, but similar affinity to mRNA. However, they do not recruit RNase H, and
whether these compounds are active as appropriate antisense molecules in vivo, remains to be
determined (69-71). In PNA, the uncharged N-(2-aminoethyl)-glycine linkage replaces the
phosphate deoxyribose backbone. Nonionic PNA binds with a relatively enhanced affinity to
target mRNA, reduces the nonspecific binding to protein, but like morpholino ODNs, does not
recruit RNase H (72-75). It is rapidly cleared from the circulation with an elimination half-life of
about 17 min, in contrast to PS-ODNs, which show a half life of approx. 1 h (76-79). Locked
nucleic acid (LNA) bases are RNA analogues that contain a methylene bridge connecting the 2’-
oxygen of the ribose with the 4’-carbon. This modification renders outstanding affinity of this
nucleotide to complementary mRNA, but reduces RNase H cleavage on the mRNA due to
modification of the sugar. To improve the binding affinity and activation of RNase H, chimeras
of LNA and PO-ODNs have been developed (80, 81). The in vivo activity of such chimeras has
been demonstrated recently (82). 2’-O-methoxyethyl ODNs(2’-MOE) modification conveys
nuclease resistance, high affinity to the target RNA, similar pharmacokinetics as PS linked
ODNs, and importantly, causes activation of RNaseH (83-88).
Although some comparative work on the biostability, the antisense effeciency and in vivo
pharmacokinetics of the various chemically modified ODNs has been done, there is no clear-cut
picture as to a single best design for an antisense structure. For example, it has been reported
that the antisense efficacy of neutral morpholino derivatives and cationic PNA were much higher
than that of negatively charged 2'-O-Me and 2'-O-MOE congeners in a cell culture model (89).
However, the same laboratory also reported that 2'-O-methoxyethyl (2'-O-MOE)-
phosphorothioate and PNA-4K oligomers (peptide nucleic acid with four lysines linked at the C
terminus) exhibited sequence-specific antisense activity in a number of mouse organs.
Morpholino oligomers were less effective, whereas PNA oligomers with only one lysine (PNA-
1K) were completely inactive (90). LNA-DNA ODNs were found to be the most efficient single-
stranded antisense ODN, when compared to PS-ODNs and 2’-ME ODNs, the former in turn
being more active than the latter (91). Claims have also been made that the efficiency of an ODN
in supporting RNase H cleavage correlates with its affinity for the target RNA. One such a
comparative study established an order of efficiency of LNA > 2'-O-methyl > DNA >
phosphorothioate (81). In summary, no conclusive data have been presented which would
support the general superiority and versatility of a specifically preferred chemically-modified
antisense construct. Rather, data currently available suggest that apart from some obvious
parameters such as biostability, the preferred antisense structure for efficient down-regulation of
Make sense of antisense oligos
25
a given mRNA target can only be revealed by direct comparison in a given experimental model
system, and no prior prediction can be made.
Overcoming cellular barriers in ODN delivery
Crossing the plasma membrane.
When naked ODNs are added to cells, they usually do not permeate across plasma membrane.
A small fraction can be endocytosised via adsorption (Fig.4A). Yet, this fraction is usually very
low since adsorption of the negatively charged ODNs to the net negatively charged plasma
membranes is rather poor. Hence, attempts have been made to chemically modify ODNs to
eliminate the negative charge in order to promote adsorption. However, an enhanced adsorption
does not necessarily improve the antisense effect, because endosomal escape (Fig.4B) poses as a
next barrier in ODN delivery into the cytosol, a minimal requirement for accomplishing an
antisense effect. In some studies, some antisense activity has been reported upon addition of free
antisense ODNs, provided a relatively high concentration was applied, i.e., 10-20 µM. Often,
significant escape of oliogonucleotides from the endosomal compartments was not observed in
these studies, implying that arrival of ODNs at the nucleus was not detectable. Yet, substantial
accumulation of administered naked ODNs into the nucleus has been reported to occur in
cultured bovine adrenal cells (92). Whether a cell type-dependent ODN translocation mechanism
may exist is at present unclear. One such a possibility could be a cell-type dependent expression
of a nuclei acid transporting channel, as identified on rat renal brush border membranes (Fig. 4
A2, 93, 94).
Bypassing the plasma membrane can be accomplished experimentally by microinjection,
electroporation or membrane permeabilization with chemical agents(Fig. 4 A2). Microinjection
has generated valuable insight into the understanding of intracellular antisense processing and
the potential correlation with efficiency. For example, both microinjection and permeabilization
lead to nuclear localization of antisense ODNs (95-98), a localization now thought to be crucial
in order to acquire a potential antisense effect. For practical purposes however, these procedures
are too laborious (microinjection) or harmful to cells (permeabilization and electroporation),
requiring careful adjustment for each cell type employed.
Vector-facilitated delivery of ODNs, relying on the use of liposomes, polymers and peptides,
appears an appropriate means to overcome the plasma membrane barrier in an efficient manner.
In essence, liposomes and polymers are capable of efficiently complexing ODNs and enhancing
the adsorption of complexed ODNs to the plasma membrane. Together, both features promote a
strong enhancement in the intracellular concentration of ODNs as accomplished by endocytosis
Chapter 2
26
of such complexes (Fig.4. A1). Some plasma membrane proteins have been claimed to be
involved in liposome-mediated delivery of oliogonucleotides, which trigger endocytosis (99,
100). The exact function of these membrane proteins still needs to be clarified, but the evidence
to support a specific role of distinct proteins is scanty. More likely, these complexes enter cells
by non-specifically exploiting the endocytic mechanism, presumably mainly involving clathrin-
mediated endocytosis, as has been well-defined for complexes of plasmids with cationic lipids
(lipoplexes; 101, 102) and polymers (polyplexes; 102, 103).
Figure 4. Cellular entry and subcellular trafficking of oligonucleotids. A.Oligonucleotides associate with the cell membrane via absorption or charge attraction.Folowing their binding, the ODN complexes are internalized by endocytosis (A1);alternatively, oligonucleotides may become translocated across cell membranes vianucleic acid channels, fusion or pore formation on the plasma membrane (A2); viacoupling of specific ligands to the complex, targeted internalization may beaccomplished by receptor-mediated endocytosis (A3). B. Following endocytosis,oligonucleotides are either released from endosomes and/or are transported tolysosomes. C. After endosomal release, oligonucleotides are transferred into thecytosol, and acquire access to the nucleus and eventually to mRNA. See text fordetails.
Distinct peptides, which have the ability to penetrate into the cell membrane, thereby forming
membrane-localized channels, may mediate ODN delivery via a non-endocytic pathway (Fig. 4
A2). Thus the uptake of ODNs complexed with the antennapedia homeodomain peptide and Tat
protein showed a temperature-, energy- and receptor-independent pathway of internalization,
characteristics which are typical of a non-endocytic mode of uptake (104-106). Also, proteins
DNA Pre-mRNA
mRNA
B. Endosomal release
C. Transportation to active sites
Targettedcomplexes
complex
A. Cell association and entry
protein
A1 A2A1
A2
receptor
lysosome
A3
Make sense of antisense oligos
27
and ligands, which are known to be processed by receptor-mediated endocytosis, such as
transferrin or certain growth factors, may serve the purpose of ODN delivery (Fig.4 A3). Indeed,
intracellular uptake of antisense ODNs linked to transferrin and folic acid was more effective
than addition of unmodified antisense (107, 108). Similarly, antisense ODNs targeted to cancer
cells via the epidermal growth factor receptor or to macrophages via the mannose receptor were
found to be taken up more efficiently than naked ODNs (109, 110). However, as noted, once the
plasma membrane barrier is overcome by exploiting the endocytic entry path, the next
intracellular barrier constitutes the endosomal membrane (Fig.4B), which can be considered the
crucial limiting step in the overall pathway that eventually elicites the antisense effect. This
conclusion may be inferred from the notion that cytosol-localized ODNs, as readily
accomplished by microinjection, rapidly accumulate in the nucleus.
Release from endosomes.
Naked ODNs are not able to permeate endosomal membranes. Accordingly, a perturbation of
the endosomal membrane stability is a necessary step in the translocation process, and it has
been concluded that cationic liposomes (but also polymers) are able to accomplish such a
destabilization. Cationic liposomes readily accomodate DNA via electrostatic interactions in a
complex structure, known as lipoplexes (111, 112). Cationic liposomes not only enhance the
uptake of ODNs over 1000-fold, but also promote their translocation across the endosomal
membrane. As a result, the antisense activity as mediated by a cationic lipid vector can already
be revealed in the nanomolar concentration range of the ODN (113-115). However, the exact
mechanism of this translocation is not clear, although some key factors have been revealed,
which interfere with this process. For plasmid release, following lipoplex entry, it has been
proposed that after the cationic lipid/DNA complexes are internalized into cells by endocytosis,
lipid flip-flop occurs at the level of the endosomal membrane. Thus anionic phospholipids, in
particular phosphatidylserine (PS), are thought to translocate from the cytoplasm-facing
endosomal monolayer to the inner monolayer, from where the lipid may laterally diffuse into the
complex and form a charge neutral ion pair with the cationic lipids. The potential ability of the
negatively charged lipid to transfer into the lipoplex may be triggered and/or facilitated by the
non-lamellar phase properties of (part of) the membrane phase in the lipoplex. In addition, the
translocation and mixing with PS may further promote such a non-lamellar structure of the
lipoplex, factors that are likely further potentiating the process of membrane destabilization
(116-118). Moreover, the presence of PS will also cause displacement of the DNA from the
cationic lipids and release of the DNA into cytoplasm (119-120), a phenomenon that can be
Chapter 2
28
readily simulated in vitro (121). An unresolved question in this context is the issue why the
release of DNA (largely, if not exclusively) occurs at the endosomal membrane, and not at the
plasma membrane, particularly since the composition of its outer monolayer is thought to be
reminiscent of that of the inner endosomal membrane.
From observations that several parameters interfere with ODN transport from the endosome
into the nucleus, lipoplex-mediated transfection or release of DNA or ODN from cationic lipid
complexes in in vitro model systems, insight into a potential mechanism of the release of both
(plasmid-) DNA and ODNs from the endosomal compartment has been generated. Endosomal
release of DNA is relatively inhibited when the lipoplexes are consisting of cationic lipids and
dioleoylphosphatidylcholine (DOPC), rather than DOPE. While DOPE displays polymorphic
properties and readily adopts a nonlamellar, i.e. hexagonal HII organization in isolation, DOPC is
a lamellar membrane bilayer-stabilizing lipid, which moreover is much more strongly hydrated
than DOPE (117, 122, 123). The latter parameter will preclude opposing membranes from tight
intermembrane interactions, which are needed in ODN (or plasmid translocation). Indeed, when
including a membrane spacer such as PEGylated lipids, the bulky and extended polyethylene
glycol headgroup separating opposed membranes, ODN delivery from endosome to nucleus is
strongly inhibited and only becomes apparent when the PEG-lipid has diffused from the
complex, the latter being accomplished by incorporating exchangeable PEG-lipids in the
complex (124). Furthermore, structural studies of ODN cationic lipid complexes as well as
lipoplexes, relying on small angle X-ray scattering and cryo-electronmicroscopy have elucidated
that the complex’ ability to adopt a non-lamellar, i.e., hexagonal phase, is crucial for obtaining
efficient plasmid or ODN delivery, respectively. Close intermembrane interactions between
complex and target membrane are instrumental in this event, which is promoted by parameters
such as the nature of helper lipids, as noted above, and the ionic environment. The latter may
modulate the state of hydration of the interphase and thereby govern head group repulsion within
the lateral plane of the bilayer, close head group- head group interactions facilitating nonlamellar
transitions in bilayer organization. PEGylated lipids and strongly hydrated lipids, such DOPC,
may oppose such transitions by maintaining a distinct distance between and within (laterally)
membranes due to steric interference or by means of hydration repulsion and bilayer
stabilization, respectively (117, 123-126). Indeed, protruding PEG-spacers strongly inhibit
endosomal release of ODN (124, 127). Remarkably, PEGylated lipids, incorporated at mole
ratios as low as 1mol% of the total lipid completely eliminated endosomal release of ODNs,
without significantly interfering with the internalization of the complexes. As noted, upon
Make sense of antisense oligos
29
diffusion-mediated removal of polyethelene glycol analogues from the complexes, a time-
dependent release of ODNs was observed in the cells, as reflected by their nuclear accumulation.
Apart from the overall structural and physical considerations that apply to accomplishing
optimal ODN delivery, endosomal release of ODNs can also be affected by the presence of
serum proteins. When such proteins absorb and/or penetrate into the complexes, the ODNs also
remain trapped in the endosomal compartments, and significant ODN release does not occur
(115, 128). In such cases, serum proteins might stabilize the complex membrane by penetration
or via surface adsorption, and/or may cause a steric interference which precludes the intimate
interaction between complex and endosomal membranes in order to induce the release of ODNs
into the cytosol, as described above.
Passage across the nuclear membrane.
Upon microinjection into the cytosol or following their escape from endosomes, ODNs rapidly
accumulate into the nucleus by passive diffusion through the nuclear membranes, whereas other
organellar membranes pose as an effective barrier. A recent study suggested that the diffusion
rate of ODNs is dependent on their length (129). A fragment consisting of 100 bp appears fully
mobile in the cytoplasm, displaying a diffusion rate compatible to that of similarly sized FITC
dextran, and only 5 times slower than that obtained in water. The diffusion of larger fragments is
remarkably slowed down, with little or no diffusion for nucleotides with a length beyond
2000bp. Small ODNs, up to several hundreds of base pairs, readily acquire access into the
nucleus. However, although ODNs of 500 bp diffuse relatively rapidly through the cytoplasm,
their avid crossing of the nuclear membrane was not observed. In passing, such observations are
also highly relevant in elucidating the mechanism by which (much larger sized) plasmids are
supposed to acquire access towards the nuclear machinery.
Within the nucleus, nucleic acid fragments of all sizes were nearly immobile on a distance
scale of 1 micron over a time interval of several minutes. Interestingly, similarly sized FITC
dextrans i.e., up to 580 kDa, diffused freely in the nucleus. Accordingly, it was speculated that
once entering the nucleus, the ODNs are rapidly recruited and immobilized upon their interaction
with nuclear components, including the positively charged histones(Lukacs,-G-L,2000). Other
studies have shown that intranuclear ODNs may also extensively bind to the nuclear RNA matrix
(130, 131). In deed, some controversy may exist concerning the (fractional) nuclear mobility of
ODNs. In fact, it has been shown, that at least a substantial fraction of the ODNs may shuttle
between the cytoplasm and the nucleus (132). Thus, when ODNs were microinjected into one
nucleus of a binuclear cell, they were found to appear readily in the other nucleus. Presumably,
Chapter 2
30
the extent of shuttling may very much depend on the nature of the ODN (length) and whether or
not appropriate target sites are reached. However, at steady state, fluorescently tagged ODNs are
usually predominantly found in the nucleus, and a significant pool in the cytosol is not observed.
When cationic lipids are used to deliver ODNs, the overall distribution of the ODNs is similar
to that obtained upon microinjection of ODNs in the cytosol, except for a minor fraction that
may become entrapped in endocytic compartments. The lipids of the vector itself remain
associated with endocytic compartments, and in contrast to polymers (103), cationic lipids have
not been observed in association with the nuclear membrane (115, 133). However, when the
complexes are released from artificially ruptured endosomes, the dissociation of cationic lipids
and ODNs can be seen to occur at the nuclear membranes, the ODNs becoming detectable in the
nucleus, while the cationic lipids diffuse into the nuclear membrane (130). As noted, at least
some cationic polymers may directly mediate the delivery of ODNs at the nuclear membrane,
likely as a result of endosomal rupture. A typical example is polyethylenimine (PEI), which is
capable of mediating a cell cycle independent nuclear entry of plasmid DNA (103). Also with
ODNs as cargo, PEI arrives at the nuclear membrane, in contrast to cationic liposomes (134).
Nuclear transport of ODNs, following their deposition into the cytosol, occurs by diffusion since
it is not affected by depletion of the intracellular ATP pool or temperature; neither is the effect of
nuclear association affected by the presence of excess unlabeled oligomers. It has been suggested
that nuclear entry can be accomplished by translocation across the nuclear pores, a conclusion
that was derived from the observation that entry could be inhibited by wheat germ agglutinin, a
nucleoporin inhibitor (135, 136). Although the size-dependent ODN transport as noted above,
may also reflects a size-limiting pore-mediated mechanism of nuclear ODN translocation, more
specific experiments, for example. by antibody-specific inhibition, would be helpful to firmly
establish the role of nuclear pores in entry. Indeed, further work will be needed, as no consensus
has been reached yet (99).
Within the nucleus, accumulation of the oligomers involves at least in part their association
with a distinct set of proteins, as revealed by crosslinking of photosensitive oligomers (95).
These proteins can be extracted in the presence of high salt (0.2 M-0.4 M NaCl). Importantly, all
experimental evidence available thus far supports the view that nuclear accumulation of ODNs is
important for eventually obtaining an antisense activity. In the following section, we will
discuss the potential mechanisms by which antisense activity is acquired.
Make sense of antisense oligos
31
Intranuclear distribution of ODNs; a correlation with antisense activity.
Once ODNs cross the nuclear membrane, they distribute throughout the nuclear lumen.
Examination by light microscopy revealed that ODNs usually do not reach the nucleoli, where
the ribosomal RNAs are transcribed from a number of chromosomes and assembled with
ribosomal proteins into ribosomal subunits. In fact, ODNs can form condensed spherical foci
(150-300nm) in a concentration dependent manner, which are not colocalized with the known
nuclear bodies. The function of these oliogonucleotide bodies is not known. It has been
speculated that the accumulation of ODNs as nuclear bodies is a response of the cell to sequester
excess ODNs in order to reduce their toxic effect (137). The exact localization of the
predominantly nucleoplasma-localized distribution of ODNs is still largely unresolved. We (130)
and others (137) have shown that the diffusely distributed ODNs within the nucleoplasm are
resistant to mild removal of loosely bound nuclear proteins and to removal of chromatin with
DNase. However, they are susceptible to the removal of RNA, following Rnase treatment (137),
and following treatment with high salt (95). These data thus indicate that ODNs are extensively
bound with non-chromatin structures in the nucleus, which is usually designated as the nuclear
matrix. Among others, these structures are composed of heteronuclear RNA (hnRNA), the
fraction of nuclear RNA which contains primary transcripts of the DNA, representing a nuclear
precursor fraction that is processed to messenger RNA, and hnRNPs, a large family of proteins,
implicating in the processing of nascent mRNA transcripts.
In permeabilized cells, ODNs readily migrate into the nucleus at ambient temperature, at
which conditions PS-ODNs, rather than PO-ODNs, give rise to a multitude of large, irregular
aggregates (138). High-affinity PS-ODN, but not PO-ODN, presumably reacts with the nuclear
lamina. Simultaneously, ODNs cause decompaction of chromatin, the PS-ODN aggregates
appearing as compact inclusions in homogeneously dispersed chromatin. After microinjection of
S-ODN into intact cells, these effects were not observed, although the nucleic acids rapidly
moved into the nucleus and condensed into a large number of well-defined, spherical speckles or
longitudinal rodlets (135, 136, 139). A high degree of complex formation between ODNs and
nuclear proteins has also been reported, as demonstrated in a gel shift assay (65). However, the
nature and number of these proteins is unknown, and whether such interaction may also occur
intracellularly, remains to be determined. Even though ODNs presumably bind to nuclear matrix
proteins, this will shorten the spatial and temporal distance to target mRNA. Nuclear matrix
binding of ODNs is not sequence-dependent, implying that nuclear matrix binding per se does
not reflect the efficiency of an antisense effect. Thus it remains to be determined how such
Chapter 2
32
interactions provide insight as to how antisense, but not mismatched sequences, finds their
targets.
Meeting intranuclear targets and identifying the fate of targets.
Obviously, the target of antisense molecules is the complementary mRNA. In spite of the fact
that antisense ODNs can hybridize to in vitro transcribed mRNA or to synthesized
complementary RNA or DNA fragments, the evidence that such an event also occurs within cells
is scanty. In an effort to gain such evidence, we have delivered radio-labeled or fluorescently-
labeled antisense ODNs and mismatched sequences into target cells. Following delivery, both
ODNs become extensively bound to the nuclear matrix, similarly as described above. When total
RNA was isolated from cells treated with the labeled antisense or mismatched ODNs, the
antisense sequence showed the expected high affinity to target mRNA, but not to other mRNAs
whereas the mismatched sequence did not display affinity to target mRNA. These data were
taken to indicate that antisense ODNs hybridize to target mRNA with high affinity over
mismatched control via their initial binding to the nuclear matrix (130). Accordingly, this
scenario would be consistent with the notion that upon hybridization of antisense ODNs to target
mRNA, its induced degradation occurs, and as a consequence protein synthesis becomes
impaired. Importantly, intracellular cleavage of target mRNAs by antisense binding could be
revealed in these studies (30, 32, 130).
In order to appreciate the biological effect of antisense treatment, it should be taken into
account that the existing pool of a target protein will not be affected other than by its natural
turnover. Thus, the antisense can only inhibit or eliminate de novo synthesis of target proteins.
Hence, the time course of the appearance of an antisense effect will vary according to the half-
life of target protein degradation, implying that relatively shortly following treatment, a large
preexisting pool of target proteins may obscure a potential biological effect of the antisense in
causing effective reduction of newly synthesized targets.
In vitro versus in vivo; role of extracellular barriers.
It is becoming apparent that many fundamental issues on the mechanism of nucleic acid
delivery can be conveniently and properly investigated in vitro, including intracellular
processing and intracellular ODN stability. However, it has become equally clear that a direct
extrapolation from in vitro to in vivo should be done but with caution. Apart from factors related
to the circulation, the role of the extracellular matrix is still puzzling, and it seems that thus far
its role has been inadequately mimicked in vitro. One such an example involves studies on the
delivery of ODNs to brain tissue. Thus, ODNs can be readily introduced into a variety of
Make sense of antisense oligos
33
neuronal cell lines, provided that delivery is mediated via an appropriate vector, such as cationic
lipids. By contrast, when injected intracerebrally in rats, the ODNs also acquire ready access to
neuronal cells, accumulating in cytoplasm and nuclei without the need of such a carrier (Shi et
al., unpublished observ.;140). Up to some 4 hr after injection of the free ODNs into whole brain,
the ODNs show a diffuse distribution in the extracellular matrix and in the cells. After 24 hr, the
associated pool of ODNs with the extracellular matrix has strongly diminished while
concomitantly, the pool of intracellular ODNs greatly increases. Another example represents the
uptake of ODNs by liver tissue. Following intravenous administration of naked ODNs, they
become associated with the extracellular matrix, which is followed by internalization in all cell
types of the liver, including appearance of ODNs, albeit in relatively minor amounts, in the
nucleus. Although the mechanism of uptake remains to be determined, it is again apparent that
also in this case nuclear access of ODNs in vivo can be accomplished in a vector- independent
manner (141).
How these remarkable differences between in vivo and in vitro application in terms of cellular
access of ODNs should be accounted for, has not been addressed thus far. Whether the
extracellular matrix harbors means to effectuate and/or to facilitate the mechanism of entry,
remains to be determined. In the context of in vitro observations it is difficult to envision that if
endocytosis would represent a major means of entry in vivo as well how endosome- localized
ODNs would readily acquire access to the cytosol in vivo and not in vitro. This raises the issue of
the existence of alternative (facilitated) mechanisms of entry in vivo, which should be capable of
mediating direct delivery of ODNs into the cytosol, thus allowing rapid nuclear delivery. ODN
transport via a plasma membrane-localized nucleic acid channel as recently described to be
present in rat renal brush board plasma membranes could fulfill such an alternative (93, 94). It is
possible that an enhanced endocytic capacity in vitro and/or a less efficient translocation
mechanism at such conditions, possibly depending on extracellular modulators for optimal
activity, may obscure the seemingly low delivery to nucleus of added free ODNs in vitro.
Even though naked ODNs can enter cells in certain tissues, they permeated very poorly into
others such as brain or tumor after systematic administration. This necessitates the use of
appropriate vectors in vivo, particularly in targeted delivery.
Overcoming the reticuloendothelial system (RES).
Dictated by pharmacokinetic distribution studies in vivo on drug delivery, organs of the RES,
such as liver, spleen, kidney and lungs represent most likely effective clearance sites, following
systemic administration of ODN complexes. Even though a wide tissue distribution of ODNs has
Chapter 2
34
been reported, except for the brain and testes, a major fraction, irrespective of the route of
administration, appears to end up in RES organs. Accordingly, directing ODNs to non-RES
organs remains a challenge in antisense technology.
Free ODNs are cleared from plasma according to a kinetics with two exponentials, showing
half-lives of about 20 min and approx. 24 h, respectively, although it may vary with the chemical
nature of administered ODNs. Provided that the majority of naked ODNs are cleared in the first
few hours, liposomes and other delivery vectors are used to protect and prolong the half-life of
ODNs in the circulation in order to improve the interaction frequency with non-RES organs.
However, to avoid the RES, additional modification of the delivery vehicle per se is needed, like
the inclusion of PEGylated lipids, modulation the surface charge and inclusion of appropriate
targeting devices for tissue/cell-specific delivery. Such systems have proved useful, for example,
in targeted delivery of DNA into the brain. Thus a PEGylated (DSPE-PEG) DNA lipoplex,
consisting of phosphatidylcholine and the cationic lipid didodecylammonium bromide, was
conjugated with transferrin receptor antibodies, OX-26. Following systemic administration, such
complexes were shown to cross the blood brain barrier by transcytosis via transferrin receptors
on brain endothelial cells. The complexes subsequently distributed over a large area in the
central nervous system, including neurons, choroid plexus epithelium, and the brain
microvasculature (12, 142, 143). A priori, in this manner it should thus be possible to specifically
target to defined brain cell populations, following systemic administration. Importantly however,
successful application of such an approach will require the dissociation of the PEGylated lipid
from the complex, as discussed above. Hence rather than the essentially non-exchangeable
DSPE-PEG, this kind of approach requires a controllable release of PEGylated lipid analogues
(124, 144, 145)
Reducing the toxicity of ODNs.
In spite of the fact that successful clinical trials have been reported (22-24), scepticism
remains as to unambiguous proof for accomplishing a genuine antisense effect. Obviously this
will frustrate a more general acceptance and effectuation of antisense therapy. In fact, for any
kind of drug application, including the application of antisense technology, potential toxic side
effects are a primary concern. In cell culture, the main toxicity is due the inhibition of cell
growth as a result of the binding of ODNs to membrane or other intracellular proteins. In vitro
cytotoxicity can be usually controlled by adjusting the proper dosage of ODNs or by their
packaging in delivery vectors. The in vivo toxicity often correlates with the capture and long-
term deposit of ODNs in RES organs, causing harmful side effects as renal tubule
Make sense of antisense oligos
35
degeneration/necrosis, splenomegaly, thrombocytopenia and elevation of liver transaminases
(146, 147). Some of these side effects are similar to those associated with delivery of other non-
related polyanions such as dextran sulfate (148). Improved ODN chemistry, for example by
reducing the negative charge, may at least in part relieve these problems. The potential toxicity
of ODNs with chemically modified linkages has not been addressed thus far. Thus future work
should also be aimed at clarifying the metabolic fate of ODNs. However, in general, from the
clinical toxicity profile of ODNs directed against viruses and tumors, this new generation of
drugs is still more tolerable than conventional chemotherapy and radiotherapy (149-151).
Conclusion and remarks In the past 5 years, considerable progress has been made in
understanding the mechanism(s) of antisense activity, in developing ODN chemistry and in
setting up antisense-related clinical trails, with as landmark the production of the first
commercial antisense drug against CMV retinitis. Evidently, the activities leading to these
prosperous developments have raised a wealth of insight into many fundamental cellular
processes ranging from events related to the regulation of gene function to signal transduction
pathways. Thus therapeutic progress is closely related to that of fundamental development and
the spin off for the latter is far beyond the actual purpose of ODN research as such. In terms of
therapeutic application of antisense technology, the window could be extended beyond treatment
of primarily cancers and viral infections, which may be accomplished by improving therapeutic
index, better specifying the antisense effect and lowering the costs.
AcknowledgmentWork carried out in the authors laboratory was in part supported by a grant from TheNetherlands Organization for Scientific Research (NWO)/NDRF Innovative DrugResearch (940-70-001).
References:1. Zamecnik, P. C. and Stephenson, M. L. (1978) Inhibition of Rous sarcoma virus
replication and cell transformation by a specific oligodeoxynucleotide. Proc. Natl.Acad. Sci. U. S. A., 75: 280-284.
2. Vanhee-Brossollet, C. and Vaquero, C. (1998) Do natural antisense transcriptsmake sense in eukaryotes? Gene, 211: 1-9.
3. Lehner, B., Williams, G., Campbell, R. D. and Sanderson, C. M. (2002) Antisensetranscripts in the human genome. Trends Genet., 18: 63-65.
4. Dias, N., Dheur, S., Nielsen, P. E., Gryaznov, S., Van-Aerschot, A., Herdewijn, P.,Helene, C. and Saison-Behmoaras, T. E. (1999) Antisense PNA tridecamers
Chapter 2
36
targeted to the coding region of Ha-ras mRNA arrest polypeptide chainelongation. J. Mol. Biol., 294: 403-416.
5. Sazani, P., Gemignani, F., Kang, S. H., Maier, M. A., Manoharan, M., Persmark,M., Bortner, D. and Kole, R. (2002) Systemically delivered antisense oligomersupregulate gene expression in mouse tissues. Nat. Biotechnol., 20: 1228-1233.
6. De-Backer, M.D., Nelissen, B., Logghe, M., Viaene, J., Loonen, I., Vandoninck, S.,de-Hoogt, R., Dewaele, S., Simons, F. A., Verhasselt, P., Vanhoof, G., Contreras,R.and Luyten, W. H. (2001) An antisense-based functional genomics approachfor identification of genes critical for growth of Candida albicans. Nat.Biotechnol., 19: 235-241.
7. Dean, N. M. (2001) Functional genomics and target validation approaches usingantisense oligonucleotide technology. Curr. Opin. Biotechnol., 12: 622-625.
8. Huber, M., Mundlein, A., Dornstauder, E., Schneeberger, C., Tempfer, C. B.,Mueller, M. W. and Schmidt, W. M. (2002) Accessing single nucleotidepolymorphisms in genomic DNA by direct multiplex polymerase chain reactionamplification on oligonucleotide microarrays. Anal. Biochem., 303: 25-33.
9. Wagner, R. W. and Flanagan, W. M. (1997) Antisense technology and prospectsfor therapy of viral infections and cancer. Mol. Med. Today, 3: 31-38.
10. Holmlund, J. T., Monia, B. P., Kwoh, T. J. and Dorr, F. A. (1999) Towardantisense oligonucleotide therapy for cancer: ISIS compounds in clinicaldevelopment. Curr. Opin. Mol. Ther., 1: 372-385.
11. Hjelstuen, O. K., Tonnesen, H. H., Bremer, P. O. and Verbruggen, A. M. (1998)Hybridization of a 99Tcm-labelled oligodeoxynucleotide to CAPL RNA. Nucl. Med.Biol., 25: 651-657.
12. Shi, N., Boado, R. J. and Pardridge, W. M. (2000) Brain-specific expression of anexogenous gene after i.v. administration. Proc. Natl. Acad. Sci. U. S. A., 97:14709-14714.
13. Lee, H. J., Boado, R. J., Braasch, D. A., Corey, D. R. and Pardridge, W. M. (2002)Imaging gene expression in the brain in vivo in a transgenic mouse model ofHuntington's disease with an antisense radiopharmaceutical and drug-targetingtechnology. J. Nucl. Med., 43: 948-956.
14. Storey, A., Oates, D., Banks, L., Crawford, L. and Crook, T. (1991) Anti-sensephosphorothioate oligonucleotides have both specific and non-specific effects oncells containing human papillomavirus type 16. Nucleic. Acids Res., 19: 4109-4114.
15. Wang, W., Chen, H. J., Schwartz, A., Cannon, P. J., Stein, C. A. and Rabbani, L.E. (1996) Sequence-independent inhibition of in vitro vascular smooth musclecell proliferation, migration, and in vivo neointimal formation byphosphorothioate oligodeoxynucleotides. J. Clin. Invest., 98: 443-450.
16. Anselmet, A., Mayat, E., Wietek, S., Layer, P. G., Payrastre, B. and Massoulie, J.(2002) Non- antisense cellular responses to oligonucleotides. FEBS Lett., 510:175-180.
17. Rockwell, P., O'Connor, W. J., King, K., Goldstein, N. I., Zhang, L. M. and Stein,C. A. (1997) Cell-surface perturbations of the epidermal growth factor andvascular endothelial growth factor receptors by phosphorothioateoligodeoxynucleotides. Proc. Natl. Acad. Sci. U. S. A., 94: 6523-6528.
18. Coulson, J. M., Poyner, D. R., Chantry, A., Irwin, W. J. and Akhtar, S. (1996) Anonantisense sequence-selective effect of a phosphorothioateoligodeoxynucleotide directed against the epidermal growth factor receptor inA431 cells. Mol. Pharmacol., 50: 314-325.
19. Papucci, L., Schiavone, N., Donnini, M., Lapucci, A., Luzi, E., Tempestini, A.,Witort, E and Capaccioli, S. (2002) Phosphodiester oligonucleotides inhibitmitosis and trigger apoptosis by a non-antisense, p53-mediated mechanism.Antisense Nucleic Acid Drug Dev., 12: 21-31.
Make sense of antisense oligos
37
20. Anonymous (1998) Advances in CMV management: fomivirsen (Vitravene)approved. PI. Perspect., 26: 7.
21. Anonymous (1998) AIDS-treatment-news Fomivirsen approved for CMV retinitis:first antisense drug. AIDS Treat News,302: 7.
22. Li, K. and Zhang, J. (2001) ISIS-3521. Isis Pharmaceuticals. Curr. Opin. Investig.Drugs, 2: 1454-1461.
23. Gewirtz, A. T. and Sitaraman, S. (2001) Alicaforsen. Isis Pharmaceuticals. Curr.Opin. Investig. Drugs, 2: 1401-1406.
24. Banerjee, D. (2001) Genasense (Genta Inc). Curr. Opin. Investig. Drugs., 2: 574-580.
25. Arzumanov, A., Walsh, A. P., Rajwanshi, V. K., Kumar, R., Wengel, J. and Gait,M. J. (2001) Inhibition of HIV-1 Tat-dependent trans activation by steric blockchimeric 2'-O-methyl/LNA oligoribonucleotides. Biochemistry, 40: 14645-14654.
26. Dias, N., Senamaud-Beaufort, C., Forestier-El, E., Auvin, C., Helene, C. andEster-Saison-Behmoaras, T. (2002) RNA hairpin invasion and ribosomeelongation arrest by mixed base PNA oligomer. J. Mol. Biol., 320: 489-501.
27. Baker, B. F., Lot, S. S., Condon, T. P., Cheng-Flournoy, S., Lesnik, E. A., Sasmor,H. M. and Bennett, C. F. (1997) 2'-O-(2-Methoxy)ethyl-modified anti-intercellularadhesion molecule 1 (ICAM-1) oligonucleotides selectively increase the ICAM-1mRNA level and inhibit formation of the ICAM-1 translation initiation complex inhuman umbilical vein endothelial cells. J. Biol. Chem., 272: 11994-2000.
28. Kole, R., and Mercatante, D. (2001) in Antisense Technology: Principles,Strategies and Applications (Crooke, S. T., ed) pp. 517-539, Marcel Dekker, Inc.,New York
29. Vickers, T. A., Wyatt, J. R., Burckin, T., Bennett, C. F. and Freier, S. M. (2001)Fully modified 2' MOE oligonucleotides redirect polyadenylation. Nucleic AcidsRes., 29: 1293-1299
30. Boiziau, C., Thuong, N. T. and Toulme, J. J. (1992) Mechanisms of the inhibitionof reverse transcription by antisense oligonucleotides. Proc. Natl. Acad. Sci. U. S.A., 89: 768-772.
31. Crooke, S. T., Lemonidis, K. M., Neilson, L., Griffey, R., Lesnik, E. A. and Monia,B. P. (1995) Kinetic characteristics of Escherichia coli RNase H1: cleavage ofvarious antisense oligonucleotide-RNA duplexes. Biochem. J., 312: 599-608.
32. Veal, G. J., Agrawal, S. and Byrn, R. A. (1998) Sequence-specific RNase Hcleavage of gag mRNA from HIV-1 infected cells by an antisense oligonucleotidein vitro. Nucleic Acids Res., 26: 5670-5675.
33. Furrer, P., Billeci, T. M., Donati, A., Kojima, C., Karwowski, B., Sierzchala, A.,Stec, W. and James, T. L. (1999) Structural effect of complete [Rp]-phosphorothioate and phosphorodithioate substitutions in the DNA strand of amodel antisense inhibitor-target RNA complex. J. Mol. Biol., 285: 1609-1622.
34. Wu, H., MacLeod, A. R., Lima, W. F. and Crooke, S. T. (1998) Identification andpartial purification of human double strand RNase activity. A novel terminatingmechanism for oligoribonucleotide antisense drugs. J. Biol. Chem., 273: 2532-2542.
35. Carroll, S. S., Chen, E., Viscount, T., Geib, J., Sardana, M. K., Gehman, J. andKuo, L. C. (1996) Cleavage of oligoribonucleotides by the 2',5'-oligoadenylate-dependent ribonuclease L. J. Biol. Chem., 271: 4988-4992.
36. Zamore, P. D., Tuschl, T., Sharp, P. A. and Bartel, D. P. (2000) RNAi: double-stranded RNA directs the ATP-dependent cleavage of mRNA at 21 to 23nucleotide intervals. Cell, 101: 25-33.
37. Bernstein, E., Caudy, A. A., Hammond, S. M., and Hannon, G. J. (2001) Role fora bidentate ribonuclease in the initiation step of RNA interference. Nature, 409:363-366.
Chapter 2
38
38. Zamore, P. D. (2002) Ancient pathways programmed by small RNAs. Science,296: 1265-1269.
39. Conrad, C. and Rauhut, R. (2002) Ribonuclease III: new sense from nuisance.Int. J. Biochem. Cell Biol., 34: 116-129.
40. Ecker, D. J., Vickers, T. A., Bruice, T. W., Freier, S. M., Jenison, R. D.,Manoharan, M. and Zounes, M. (1992) Pseudo--half-knot formation with RNA.Science, 257: 958-961.
41. Stull,R.A., Taylor,L.A. and Szoka,F.C.,Jr (1992) Predicting antisenseoligonucleotide inhibitory efficacy: a computational approach using histogramsand thermodynamic indices. Nucleic Acids Res., 20, 3501–3508.
42. Hamilton,S.E., Pitts,A.E., Katipally,R.R., Jia,X., Rutter,J.P., Davies,B.A.,Shay,J.W., Wright,W.E. and Corey,D.R. (1997) Identification of determinants forinhibitor binding within the RNA active site of human telomerase using PNAscanning. Biochemistry, 36: 11873-11880.
43. Petryshyn,R.A., Ferrenz,A.G. and Li,J. (1997) Characterization and mapping ofthe double-stranded regions involved in activation of PKR within a cellular RNAfrom 3T3-F442A cells. Nucleic Acids Res., 25: 2672-2678.
44. Lima,W.F., Brown-Driver,V., Fox,M., Hanecak,R. and Bruice,T.W. (1997)Combinatorial screening and rational optimization for hybridization to foldedhepatitis C virus RNA of oligonucleotides with biological antisense activity. J.Biol. Chem., 272, 626–638.
45. Milner, N., Mir,K.U. and Southern,E.M. (1997) Selecting effective antisensereagents on combinatorial oligonucleotide arrays. Nature Biotechnol., 15, 537–541.
46. Eckardt, S., Romby, P. and Sczakiel, G. (1997) Implications of RNA structure onthe annealing of a potent antisense RNA directed against the humanimmunodeficiency virus type 1. Biochemistry., 36: 12711-12721.
47. Vickers, T. A., Wyatt, J. R. and Freier, S. M. (2000) Effects of RNA secondarystructure on cellular antisense activity. Nucleic Acids Res., 28: 1340-1347.
48. Zhang, W. and Chen, S. J. (2002) RNA hairpin-folding kinetics. Proc. Natl. Acad.Sci. U.S.A., 99: 1931-1936.
49. Zuker, M. (1989) On finding all suboptimal foldings of an RNA molecule. Science,244: 48-52.
50. Chalk, A. M. and Sonnhammer, E. L. (2002) Computational antisense oligoprediction with a neural network model. Bioinformatics, 18: 1567-1575.
51. Butler,-B-A (1998) Sequence analysis using GCG. Methods Biochem. Anal., 39:74-97.
52. Zuker,-M (2003) Mfold web server for nucleic acid folding and hybridizationprediction. Nucleic-Acids-Res., 31: 3406-3415.
53. Wilson, G. M., Sutphen, K., Moutafis, M., Sinha, S. and Brewer, G. (2001)Structural remodeling of an A + U-rich RNA element by cation or AUF1 binding.J. Biol. Chem., 276: 38400-38409.
54. Raghunathan, P. L. and Guthrie, C. (1998) A spliceosomal recycling factor thatreanneals U4 and U6 small nuclear ribonucleoprotein particles. Science, 279:857-860.
55. Peyman, A., Helsberg, M., Kretzschmar, G., Mag, M., Grabley, S. and Uhlmann,E. (1995) Inhibition of viral growth by antisense oligonucleotides directed againstthe IE110 and the UL30 mRNA of herpes simplex virus type-1. Biol-Chem-Hoppe-Seyler., 376: 195-198.
56. Monia, B. P., Johnston, J. F., Geiger, T., Muller, M. and Fabbro, D. (1996)Antitumor activity of a phosphorothioate antisense oligodeoxynucleotide targetedagainst C-raf kinase. Nat. Med., 2: 668-675.
Make sense of antisense oligos
39
57. Ding, Y. and Lawrence, C. E. (2001) Statistical prediction of single-strandedregions in RNA secondary structure and application to predicting effectiveantisense target sites and beyond. Nucleic Acids Res., 29: 1034-1046.
58. Patzel, V. and Sczakiel, G. (1998) Theoretical design of antisense RNA structuressubstantially improves annealing kinetics and efficacy in human cells. Nat.Biotechnol., 16: 64-68.
59. Sohail, M., Hochegger, H., Klotzbucher, A., Guellec, R. L., Hunt, T. and Southern,E. M. (2001) Antisense oligonucleotides selected by hybridisation to scanningarrays are effective reagents in vivo. Nucleic-Acids-Res., 29: 2041-2051.
60. Ho, S. P., Britton, D. H., Stone, B. A., Behrens, D. L., Leffet, L. M., Hobbs, F. W.,Miller, J. A. and Trainor, G. L. (1996) Potent antisense oligonucleotides to thehuman multidrug resistance-1 mRNA are rationally selected by mapping RNA-accessible sites with oligonucleotide libraries. Nucleic Acids Res., 24: 1901-1907.
61. Matveeva, O., Felden, B., Audlin, S., Gesteland, R. F. and Atkins, J. F. (1997) Arapid in vitro method for obtaining RNA accessibility patterns for complementaryDNA probes: correlation with an intracellular pattern and known RNA structures,Nucleic Acid Res. 25: 5010-5016.
62. Ho, S. P., Bao, Y., Lesher, T., Malhotra, R., Ma, L. Y., Fluharty, S. J. and Sakai,R. R. (1998) Mapping of RNA accessible sites for antisense experiments witholigonucleotide libraries. Nat. Biotechnol., 16: 59-63.
63. Eckstein, F. (2000) Phosphorothioate oligodeoxynucleotides: what is their originand what is unique about them? Antisense Nucleic Acid Drug Dev., 10, 117-121.
64. Geary, R.S., Yu, R. Z. and Levin A. A. (2001) Pharmacokinetics ofphosphorothioate antisense oligodeoxynucleotides. Curr. Opin. Invest. NewDrugs, 2: 562-573.
65. Brown, D. A., Kang, S. H., Gryaznov, S. M., DeDionisio, L., Heidenreich, O.,Sullivan, S., Xu, X. and Nerenberg, M. I. (1994) Effect of phosphorothioatemodification of oligodeoxynucleotides on specific protein binding. J. Biol. Chem.,269: 26801-26805.
66. Stein, C. A. (1999) Two problems in antisense biotechnology: in vitro delivery andthe design of antisense experiments. Biochim. Biophys. Acta, 1489, 45-52.
67. Guvakova, M. A., Yakubov, L. A., Vlodavsky, I., Tonkinson, J. L. and Stein, C. A.(1995) Inhibition of high affinity basic fibroblast growth factor binding byoligonucleotides. J. Biol. Chem., 270: 2620-2627.
68. Fennewald, S. M. and Rando, R. F. Cell-surface perturbations of the epidermalgrowth factor and vascular endothelial growth factor receptors byphosphorothioate oligodeoxynucleotides. J. Biol. Chem., 270: 21718-21721.
69. Summerton,J. (1999) Morpholino antisense oligomers: the case for an RNase H-independent structural type. Biochim. Biophys. Acta, 1489: 141-158.
70. Giles, R. V., Spiller, D. G., Clark, R. E. and Tidd, D. M. (1999) Antisensemorpholino oligonucleotide analog induces missplicing of C-myc mRNA.Antisense Nucleic Acid Drug Dev., 9: 213-220.
71. Schmajuk, G., Sierakowska, H. and Kole, R. (1999) Antisense oligonucleotideswith different backbones. Modification of splicing pathways and efficacy ofuptake. J. Biol. Chem., 274: 21783-21789.
72. Koppelhus, U., Zachar, V., Nielsen, P. E., Liu, X., Eugen-Olsen, J. and Ebbesen,P. (1997) Efficient in vitro inhibition of HIV-1 gag reverse transcription by peptidenucleic acid (PNA) at minimal ratios of PNA/RNA . Nucleic-Acids-Res., 25: 2167-2173.
73. Lee, R., Kaushik, N., Modak, M. J., Vinayak, R. and Pandey, V. N. (1998)Polyamide nucleic acid targeted to the primer binding site of the HIV-1 RNAgenome blocks in vitro HIV-1 reverse transcription. Biochemistry, 37: 900-910.
Chapter 2
40
74. Lomakin, A. and Frank-Kamenetskii, M. D. (1998) A theoretical analysis ofspecificity of nucleic acid interactions with oligonucleotides and peptide nucleicacids (PNAs). J. Mol. Biol., 276: 57-70.
75. Pitts, A. E. and Corey, D. R. (1998) Inhibition of human telomerase by 2'-O-methyl-RNA. Proc. Natl. Acad. Sci. U.S.A., 95: 11549-11554.
76. Boado, R. J., Kang, Y. S., Wu, D and Pardridge, W. M. (1995) Phase I safety andpharmacokinetic profile of an intercellular adhesion molecule-1 antisenseoligodeoxynucleotide (ISIS 2302). Drug Metab. Dispos., 23: 1297-1300.
77. Glover, J. M., Leeds, J. M., Mant, T. G., Amin, D., Kisner, D. L., Zuckerman, J.E., Geary, R. S., Levin, A. A. and; Shanahan, W. R. Jr (1997) Phase I safety andpharmacokinetic profile of an intercellular adhesion molecule-1 antisenseoligodeoxynucleotide (ISIS 2302). J. Pharmacol. Exp. Ther., 282: 1173-1180.
78. Leeds, J. M., Henry, S. P., Geary, R., Burckin, T. and Levin, A. A. (2000)Comparison of the pharmacokinetics of subcutaneous and intravenousadministration of a phosphorothioate oligodeoxynucleotide in cynomolgusmonkeys. Antisense Nucleic Acid Drug Dev., 10: 435-441.
79. McMahon, B. M., Mays, D., Lipsky, J., Stewart, J. A., Fauq, A. and Richelson, E.(2002) Pharmacokinetics and tissue distribution of a peptide nucleic acid afterintravenous administration. Antisense Nucleic Acid Drug Dev., 12: 65-70.
80. Wahlestedt, C., Salmi, P., Good, L., Kela, J., Johnsson, T., Hokfelt, T., Broberger,C., Porreca, F., Lai, J., Ren, K., Ossipov, M., Koshkin, A., Jakobsen, N., Skouv,J., Oerum, H., Jacobsen, M. H. and Wengel, J. (2000) Potent and nontoxicantisense oligonucleotides containing locked nucleic acids. Proc. Natl. Acad. Sci.U. S. A., 97: 5633-5638.
81. Kurreck, J., Wyszko, E., Gillen, C. and Erdmann, V. A. (2002) Design ofantisense oligonucleotides stabilized by locked nucleic acids. Nucleic Acids Res.,30: 1911-1918.
82. Fluiter, K., ten-Asbroek, A. L., de-Wissel, M. B., Jakobs, M. E., Wissenbach, M.,Olsson, H., Olsen, O., Oerum, H. and Baas, F. (2003) In vivo tumor growthinhibition and biodistribution studies of locked nucleic acid (LNA) antisenseoligonucleotides. Nucleic Acids Res., 31: 953-962.
83. Brown-Driver, V., Eto, T., Lesnik, E., Anderson, K. P. and Hanecak, R. C. (1999)Inhibition of translation of hepatitis C virus RNA by 2-modified antisenseoligonucleotides. Antisense Nucleic Acid Drug Dev., 9: 145-154.
84. Teplova, M., Minasov, G., Tereshko, V., Inamati, G. B., Cook, P. D. Manoharan,M. and Egli, M. (1999) Crystal structure and improved antisense properties of 2'-O-(2-methoxyethyl)-RNA. Nat. Struct. Biol., 6: 535-539.
85. Geary, R. S., Watanabe, T. A., Truong, L., Freier, S., Lesnik, E. A., Sioufi, N. B.,Sasmor, H., Manoharan, M. and Levin, A. A. (2001) Pharmacokinetic propertiesof 2'-O-(2-methoxyethyl)-modified oligonucleotide analogs in rats. J. Pharmacol.Exp. Ther., 296: 890-897.
86. Geary, R. S., Khatsenko, O., Bunker, K., Crooke, R., Moore, M., Burckin, T.,Truong, L., Sasmor, H. and Levin, A. A. (2001) Absolute bioavailability of 2'-O-(2-methoxyethyl)-modified antisense oligonucleotides following intraduodenalinstillation in rats. J. Pharmacol. Exp. Ther., 296: 898-904.
87. Elayadi, A. N., Demieville, A., Wancewicz, E. V., Monia, B. P. and Corey, D. R.(2001) Inhibition of telomerase by 2'-O-(2-methoxyethyl) RNA oligomers: effect oflength, phosphorothioate substitution and time inside cells. Nucleic Acids Res.,29: 1683-1689.
88. Zellweger, T., Miyake, H., Cooper, S., Chi, K., Conklin, B. S., Monia, B. P. andGleave, M. E. (2001) Antitumor activity of antisense clusterin oligonucleotides isimproved in vitro and in vivo by incorporation of 2'-O-(2-methoxy)ethylchemistry. J. Pharmacol. Exp. Ther. 298: 934-940.
Make sense of antisense oligos
41
89. Sazani, P., Kang, S. H., Maier, M. A., Wei, C., Dillman, J., Summerton, J.,Manoharan, M. and Kole, R. (2001) Nuclear antisense effects of neutral, anionicand cationic oligonucleotide analogs. Nucleic Acids Res., 29: 3965-3974.
90. Sazani, P., Gemignani, F., Kang, S. H., Maier, M. A., Manoharan, M., Persmark,M., Bortner, D. and Kole, R. (2002) Systemically delivered antisense oligomersupregulate gene expression in mouse tissues. Nat. Biotechnol., 20: 1228-1233.
91. Grunweller, A., Wyszko, E., Bieber, B., Jahnel, R., Erdmann, V. A. and Kurreck,J. (2003) Comparison of different antisense strategies in mammalian cells usinglocked nucleic acids, 2'-O-methyl RNA, phosphorothioates and small interferingRNA Nucleic Acids Res., 31: 3185-3193.
92. Li, B., Hughes, J. A. and Phillips, M. I. (1997) Uptake and efflux of intactantisense phosphorothioate deoxyoligonucleotide directed against angiotensinreceptors in bovine adrenal cells. Neurochem. Int., 31: 393-403.
93. Hanss, B., Leal-Pinto, E., Bruggeman, L. A., Copeland, T. D. and Klotman, P. E.(1998) Identification and characterization of a cell membrane nucleic acidchannel. Proc. Natl. Acad. Sci. U. S. A., 95: 1921-1926.
94. Hanss, B., Leal-Pinto, E., Teixeira, A., Christian, R. E., Shabanowitz, J., Hunt, D.F. and Klotman, P. E. (2002) Cytosolic malate dehydrogenase confers selectivityof the nucleic acid-conducting channel. Proc. Natl. Acad. Sci. U. S. A., 99: 1707-1712.
95. Leonetti, J. P., Mechti, N., Degols, G., Gagnor, C. and Lebleu, B. (1991)Intracellular distribution of microinjected antisense oligonucleotides.Proc-Natl-Acad-Sci-U-S-A., 88: 2702-2706.
96. Fisher, T. L., Terhorst, T., Cao, X. and Wagner, R. W. (1993) Intracellulardisposition and metabolism of fluorescently-labeled unmodified and modifiedoligonucleotides microinjected into mammalian cells. Nucleic Acids Res., 21:3857-3865.
97. Giles, R. V., Spiller, D. G. and Tidd, D. M. (1995) Detection of ribonuclease H-generated mRNA fragments in human leukemia cells following reversiblemembrane permeabilization in the presence of antisense oligodeoxynucleotides.Antisense Res. Dev., 5: 23-31.
98. Spiller, D.G. and Tidd, D. M. (1995) Nuclear delivery of antisenseoligodeoxynucleotides through reversible permeabilization of human leukemiacells with streptolysin O. Antisense Res. Dev., 5: 13-21.
99. Beltinger, C., Saragovi, H. U., Smith, R. M., LeSauteur, L., Shah, N., DeDionisio,L., Christensen, L., Raible, A., Jarett, L. and Gewirtz, A. M. (1995) Binding,uptake, and intracellular trafficking of phosphorothioate-modifiedoligodeoxynucleotides. J. Clin. Invest., 95: 1814-1823.
100. Hawley, P. and Gibson, I. (1996) Interaction of oligodeoxynucleotides withmammalian cells. Antisense Nucleic Acid Drug Dev., 6: 185-195.
101. Zuhorn, I. S., Kalicharan, R. and Hoekstra, D. (2002) Lipoplex-mediatedtransfection of mammalian cells occurs through the cholesterol-dependentclathrin-mediated pathway of endocytosis. J. Biol. Chem., 277: 18021-18028.
102. de-Semir, D., Petriz, J., Avinyo, A., Larriba, S., Nunes, V., Casals, T., Estivill, X.and Aran, J. M. (2002) Non-viral vector-mediated uptake, distribution, andstability of chimeraplasts in human airway epithelial cells. J. Gene Med., 4: 308-322.
103. Godbey, W. T., Wu, K. K. and Mikos, A. G. (1999) Tracking the intracellular pathof poly(ethylenimine)/DNA complexes for gene delivery. Proc. Natl. Acad. Sci. U.S. A., 96: 5177-5181.
104. Fawell, S., Seery, J., Daikh, Y., Moore, C., Chen, L. L., Pepinsky, B. andBarsoum, J. (1994) Tat-mediated delivery of heterologous proteins into cells.Proc. Natl. Acad. Sci. U. S. A., 91: 664-668.
Chapter 2
42
105. Allinquant, B., Hantraye, P., Mailleux, P., Moya, K., Bouillot, C. and Prochiantz,A. (1995) Downregulation of amyloid precursor protein inhibits neurite outgrowthin vitro. J. Cell Biol., 128: 919-927.
106. Troy, C. M., Derossi, D., Prochiantz, A., Greene, L. A. and Shelanski, M. L. (1996)Downregulation of Cu/Zn superoxide dismutase leads to cell death via the nitricoxide-peroxynitrite pathway. J. Neurosci., 16: 253-261.
107. Citro, G., Perrotti, D., Cucco, C., D'Agnano, I., Sacchi, A., Zupi, G. andCalabretta, B. (1992) Inhibition of leukemia cell proliferation by receptor-mediated uptake of c-myb antisense oligodeoxynucleotides. Proc. Natl. Acad. Sci.U. S. A., 89: 7031-7035.
108. Citro, G., Szczylik, C., Ginobbi, P., Zupi, G. and Calabretta, B. (1994) Inhibitionof leukaemia cell proliferation by folic acid-polylysine-mediated introduction of c-myb antisense oligodeoxynucleotides into HL-60 cells. Br. J. Cancer, 69: 463-467.
109. Bonfils, E., Depierreux, C., Midoux, P., Thuong, N. T., Monsigny, M. and Roche,A. C. (1992) Drug targeting: synthesis and endocytosis of oligonucleotide-neoglycoprotein conjugates. Nucleic Acids Res., 20: 4621-4629.
110. Liang, W. W., Shi, X., Deshpande, D., Malanga, C. J. and Rojanasakul, Y. (1996)Oligonucleotide targeting to alveolar macrophages by mannose receptor-mediatedendocytosis. Biochim. Biophys. Acta, 1279: 227-234.
111. Oberle, V., Bakowsky, U., Zuhorn, I. S. and Hoekstra, D (2000) Lipoplexformation under equilibrium conditions reveals a three-step mechanism.Biophys. J., 79: 1447-1454.
112. Maurer, N., Wong, K. F., Stark, H., Louie, L., McIntosh, D., Wong, T., Scherrer,P., Semple, S. C. and Cullis, P. R. (2001) Spontaneous entrapment ofpolynucleotides upon electrostatic interaction with ethanol-destabilized cationicliposomes. Biophys. J., 80: 2310-2326.
113. Bennett, C. F., Chiang, M. Y., Chan, H., Shoemaker, J. E. and Mirabelli, C. K.(1992) Cationic lipids enhance cellular uptake and activity of phosphorothioateantisense oligonucleotides. Mol.Pharmacol., 41: 1023-1033.
114. DeLong, R. K., Yoo, H., Alahari, S. K., Fisher, M., Short, S. M., Kang, S. H., Kole,R., Janout, V., Regan, S. L. and Juliano, R. L. (1999) Novel cationic amphiphilesas delivery agents for antisense oligonucleotides. Nucleic Acids Res., 27:3334-3341.
115. Shi, F., Nomden, A., Oberle, V., Engberts, J. B. and Hoekstra, D. (2001) Efficientcationic lipid-mediated delivery of antisense oligonucleotides into eukaryoticcells: down-regulation of the corticotropin-releasing factor receptor. Nucleic AcidsRes., 29: 2079-2087.
116. Radler, J. O., Koltover, I., Salditt, T., and Safinya, C. R. (1997) Structure of DNA-cationic liposome complexes: DNA intercalation in multilamellar membranes indistinct interhelical packing regimes. Science, 275: 810-814.
117. Koltover, I., Salditt, T., Radler, J. O. and Safinya, C. R. (1998) An invertedhexagonal phase of cationic liposome-DNA complexes related to DNA release anddelivery. Science, 281: 78-81.
118. Smisterova, J., Wagenaar, A., Stuart, M. C., Polushkin, E., ten-Brinke, G., Hulst,R., Engberts, J. B. and Hoekstra, D. (2001) Molecular shape of the cationic lipidcontrols the structure of cationic lipid/dioleylphosphatidylethanolamine-DNAcomplexes and the efficiency of gene delivery. J. Biol. Chem., 276: 47615-47622.
119. Xu, Y. and Szoka, F. C. Jr. (1996) Mechanism of DNA Release from CationicLiposome/DNA Complexes Used in Cell Transfection. Biochemistry, 35:5616 –5623.
120. Zelphati, O. and Szoka, F. C. Jr (1996) Mechanism of oligonucleotide releasefrom cationic liposomes. Proc. Natl. Acad. Sci. U. S. A., 93: 11493-11498.
Make sense of antisense oligos
43
121. Zuhorn, I. S., Oberle, V., Visser, W. H., Engberts, J. B., Bakowsky, U., Polushkin,E. and Hoekstra, D. Phase behavior of cationic amphiphiles and theirmixtures with helper lipid influences lipoplex shape, DNA translocation, andtransfection efficiency. Biophys. J. 83: 2096-2108.
122. Radler, J. O., Koltover, I., Salditt, T. and Safinya, C. R. (1997) Structure of DNA-cationic liposome complexes: DNA intercalation in multilamellar membranes indistinct interhelical packing regimes. Science, 275: 810-814.
123. Koltover, I., Salditt, T., Safinya, C. R. (1999) Phase diagram, stability, andovercharging of lamellar cationic lipid-DNA self-assembled complexes. Biophys-J., 77: 915-924.
124. Shi, F., Wasungu, L., Nomden, A., Stuart, M. C., Polushkin, E., Engberts, J. B.and Hoekstra, D. (2002) Interference of poly(ethylene glycol)-lipid analogues withcationic-lipid-mediated delivery of oligonucleotides; role of lipid exchangeabilityand non-lamellar transitions. Biochem. J. 366: 333-341.
125. Holland, J. W., Cullis, P. R. and Madden, T. D. (1996) Poly(ethylene glycol)-lipidconjugates promote bilayer formation in mixtures of non-bilayer-forming lipids.Biochemistry, 35: 2610-2617.
126. Johnsson, M. and Edwards, K. (2001) Phase behavior and aggregate structure inmixtures of dioleoylphosphatidylethanolamine and poly(ethylene glycol)-lipids.Biophys. J., 80: 313-323.
127. Song, L. Y., Ahkong, Q. F., Rong, Q., Wang, Z., Ansell, S., Hope, M. J. and Mui,B. (2002) Characterization of the inhibitory effect of PEG-lipid conjugates onthe intracellular delivery of plasmid and antisense DNA mediated by cationiclipid liposomes. Biochim. Biophys. Acta., 1558: 1-13.
128. Zelphati, O., Uyechi, L. S., Barron, L. G. and Szoka, F. C. Jr (1998) Effect ofserum components on the physico-chemical properties of cationiclipid/oligonucleotide complexes and on their interactions with cells. Biochim.Biophys. Acta. 1390: 119-133.
129. Lukacs, G. L., Haggie, P., Seksek, O., Lechardeur, D., Freedman, N. andVerkman, A. S. (2000) Size-dependent DNA mobility in cytoplasm and nucleus. J.Biol. Chem., 275: 1625-1629.
130. Shi, F., Visser, W. H., de Jong, N. M. J., Liem, R. S. B., Ronken, E. and Hoekstra,D (2003) Antisense oligonucleotides reach mRNA targets via the RNA matrix;downregulation of the 5-HT1A receptor. Exp. Cell Res. in press.
131. Lorenz, P., Baker, B. F., Bennett, C. F. and Spector, D. L. (1998)Phosphorothioate antisense oligonucleotides induce the formation of nuclearbodies. Mol. Biol. Cell. 9: 1007-1023.
132. Lorenz, P., Misteli, T., Baker, B. F., Bennett, C. F. and Spector, D. L. (2000)Nucleocytoplasmic shuttling: a novel in vivo property of antisensephosphorothioate oligodeoxynucleotides. Nucleic Acids Res. 28: 582-592.
133. Marcusson, E. G., Bhat, B., Manoharan, M., Bennett, C. F. and Dean, N. M.(1998) Phosphorothioate oligodeoxyribonucleotides dissociate from cationic lipidsbefore entering the nucleus. Nucleic Acids Res. 26: 2016-2023.
134. Brunner, S., Furtbauer, E., Sauer, T., Kursa, M. and Wagner, E. (2002)Overcoming the nuclear barrier: cell cycle independent nonviral gene transferwith linear polyethylenimine or electroporation. Mol. Ther., 5: 80-86.
135. Shoeman, R. L., Hartig, R., Huang, Y., Grub, S. and Traub, P. (1997)Fluorescence microscopic comparison of the binding of phosphodiester andphosphorothioate (antisense) oligodeoxyribonucleotides to subcellular structures,including intermediate filaments, the endoplasmic reticulum, and the nuclearinterior. Antisense Nucleic Acid Drug Dev., 7: 291-308.
136. Hartig, R., Shoeman, R. L., Janetzko, A., Grub, S. and Traub, P. (1998) Activenuclear import of single-stranded oligonucleotides and their complexes with non-karyophilic macromolecules. Biol-Cell, 90: 407-426.
Chapter 2
44
137. Lorenz, P., Baker, B. F., Bennett, C. F. and Spector, D. L. (1998)Phosphorothioate antisense oligonucleotides induce the formation of nuclearbodies. Mol. Biol. Cell, 9: 1007-1023.
138. Zupan, J. R., Citovsky, V. and Zambryski, P. (1996) Agrobacterium VirE2 proteinmediates nuclear uptake of single-stranded DNA in plant cells. Proc. Natl. Acad.Sci. U. S. A., 93: 2392-2397.
139. Hartig, R., Huang, Y., Janetzko, A., Shoeman, R., Grub, S. and Traub, P (1997)Binding of fluorescence- and gold-labeled oligodeoxyribonucleotides tocytoplasmic intermediate filaments in epithelial and fibroblast cells. Exp. CellRes. 233: 169-186.
140. Chauhan, N. B. (2002) Trafficking of intracerebroventricularly injected antisenseoligonucleotides in the mouse brain. Antisense Nucleic Acid Drug Dev., 12: 353-357.
141. Graham, M. J., Crooke, S. T., Monteith, D. K., Cooper, S. R., Lemonidis, K. M.,Stecker, K. K., Martin, M. J. and Crooke, R. M. (1998) In vivo distribution andmetabolism of a phosphorothioate oligonucleotide within rat liver afterintravenous administration. J. Pharmacol. Exp. Ther., 286: 447-458.
142. Pardridge, W. M., Boado, R. J. and Kang, Y. S. (1995) Vector-mediated delivery ofa polyamide ("peptide") nucleic acid analogue through the blood-brain barrier invivo. Proc. Natl. Acad. Sci. U. S. A., 92: 5592-5596.
143. Penichet, M. L., Kang, Y. S., Pardridge, W. M., Morrison, S. L. and Shin, S. U.(1999) An antibody-avidin fusion protein specific for the transferrin receptorserves as a delivery vehicle for effective brain targeting: initial applications inanti-HIV antisense drug delivery to the brain. J. Immunol., 163: 4421-4426.
144. Wheeler, J. J., Palmer, L., Ossanlou, M., MacLachlan, I., Graham, R. W., Zhang,Y. P., Hope, M. J., Scherrer, P. and Cullis, P. R. (1999) Stabilized plasmid-lipidparticles: construction and characterization, Gene Ther., 6: 271-281.
145. Hu, Q., Shew, C. R., Bally, M. B.and Madden, T. D. (2001) Programmablefusogenic vesicles for intracellular delivery of antisenseoligodeoxynucleotides: enhanced cellular uptake and biological effects. Biochim.Biophys. Acta. 1514: 1-13.
146. Sarmiento, U. M., Perez, J. R., Becker, J. M. and Narayanan, R. (1994) In vivotoxicological effects of rel A antisense phosphorothioates in CD-1 mice. AntisenseRes. Dev., 4: 99-107.
147. Agrawal, S., Rustagi, P. K. and Shaw, D. R. (1995) Novel enzymatic andimmunological responses to oligonucleotides. Toxicol. Lett., 82-83: 431-434.
148. Bitter-Suermann, D., Burger, R. and Hadding, U. (1981) Activation of thealternative pathway of complement: efficient fluid-phase amplification byblockade of the regulatory complement protein beta1H through sulfatedpolyanions. Eur. J. Immunol., 11: 291-295.
149. Wallace, T. L., Bazemore, S. A., Kornbrust, D. J. and Cossum, P. A. (1996)Single-dose hemodynamic toxicity and pharmacokinetics of a partialphosphorothioate anti-HIV oligonucleotide (AR177) after intravenous infusion tocynomolgus monkeys. J. Pharmacol. Exp. Ther., 278: 1306-1312.
150. Henry, S. P., Novotny, W., Leeds, J., Auletta, C. and Kornbrust, D. J. (1997)Inhibition of coagulation by a phosphorothioate oligonucleotide. AntisenseNucleic Acid Drug Dev., 7: 503-510.
151. Flaherty, K. T., Stevenson, J. P. and O'Dwyer, P. J. (2001) Antisensetherapeutics: lessons from early clinical trials. Curr. Opin. Oncol., 13: 499-505.
45
Chapter 3
Efficient cationic lipid-mediated delivery of antisenseoligonucleotides into eukaryotic cells: down-regulation
of the corticotropin-releasing factor receptor
Fuxin Shi, Anita Nomden, Volker Oberle, Jan. B. F. N. Engberts1 and Dick Hoekstra
Department of Membrane Cell Biology, University of Groningen, Faculty of Medical Sciences,Antonius Deusinglaan 1, 9713 AV Groningen, The Netherlands and 1Physical OrganicChemistry Unit, Stratingh Institute, University of Groningen, Nijenborgh 4, 9747 AG
Groningen, The Netherlands
Nucleic Acids Research, 29(2001): 2079-2087
Chapter 3
46
Abstract
Oligonucleotides (ODNs) can be employed as effective gene-specific regulators. However,
before ODNs can reach their targets, several physical barriers have to be overcome, as although
ODNs may pass cell membranes, most become sequestered in endocytic compartments.
Accordingly, sophisticated strategies are required for efficient delivery. Here we have employed
a pyridinium-based synthetic amphiphile, called SAINT-2, which carries ODNs into cells in a
highly efficient, essentially non-toxic and serum-insensitive manner. Intracellular delivery was
examined by monitoring the trafficking of fluorescent ODNs and lipid, and by measuring the
effect of specific antisense ODNs on target mRNA and protein levels of the receptor for the
neuropeptide corticotropin-releasing factor (CRF-R), expressed in Chinese hamster ovary cells.
ODN delivery is independent of lipoplex size, and fluorescently tagged ODNs readily acquire
access to the nucleus, whereas the carrier itself remains sequestered in the endosomal–lysosomal
pathway. While the release is independent of the presence of serum, it is not observed when
serum proteins gain access within the lipoplex, and which likely stabilizes the lipoplex
membrane. We propose that the amphiphile-dependent aggregate structure governs complex
dissociation, and hence, the biological efficiency of ODNs. We demonstrate an essentially non-
toxic and effective antisense-specific down-regulation of the CRF-R, both at the mRNA and
protein level.
Cationic-lipid-mediated delivery of oligos
47
Introduction
Cell biological analyses or therapeutic treatment may often benefit from an ability to down-
regulate or modify the expression of proteins. The application of antisense technology, as has
been described for a variety of receptors (1,2), provides such a possibility. This technology relies
on the use of antisense oligonucleotides (ODNs), which are short DNA or modified DNA
fragments (7–30 nt in length) that contain a complementary base sequence to their target RNAs.
Upon selective hybridization, a specific interference with gene transcription, RNA transport,
splicing or translation of the desired target protein can, thus, be accomplished, thereby allowing a
selective regulation of gene expression. For example, the corticotropin-releasing factor (CRF) is
a neuropeptide present in the central nervous system that is intimately involved in the expression
of stress in mammals (3). Its effects, including behavioral, immunological and endocrine
responses, are propagated via the CRF-receptor (CRF-R), which belongs to the G-protein
coupled receptor family. Thus far, the biochemical and physiological characteristics of this
receptor have been poorly characterized. Therefore, from both a biochemical (2) and a
therapeutic point of view, down-regulation of receptor expression and, thereby, its function
would be of interest.
The specificity of the antisense technology has been under debate, as often high ODN
concentrations and long term incubations are required (2,4), conditions that may readily lead to
cell cytotoxic effects of the ODN per se. Also, non-specific hybridization with intracellular
proteins rather than mRNAs has been reported (5), thus challenging the occurrence of a genuine
antisense effect. Other limiting factors include ODN stability, and selective cellular uptake and
processing. To improve the latter features, antisense phosphorothioate ODNs are preferred over
phosphodiester derivatives as the latter are readily degraded by nucleases (6). In contrast, thioate
ODNs display a capacity to induce degradation of the target sequence by RNase H (7,8).
Delivery of ODNs in vitro has been accomplished by simple exogenous addition, microinjection
and electroporation (9,10). Also, liposomes have been employed for this purpose (11). However,
such vehicles may display their own inherent drawback of a relatively poor encapsulation and
delivery. Some cationic lipids may in part overcome these disadvantages, although such systems
often suffer from high cytotoxic side effects and a high sensitivity towards serum (12), whereas
selective uptake by only a restricted number of cells within a population may also occur.
Moreover, the efficiency of ODN delivery may strongly depend on the chemical nature of the
cationic surfactant, but the underlying cause for such differences remains entirely obscure (1).
In the present work, we have employed a synthetic amphiphile, 1-methyl-4(dioleoyl)methyl-
pyrimidinium chloride (SAINT-2), and examined the mechanism of uptake and cellular
Chapter 3
48
processing of CRF-R-specific, unmodified phosphodiester (D-)ODNs and phosphorothioate (S-)
ODNs. We demonstrate that the intracellular delivery of antisense ODNs improved by more than
two orders of magnitude when complexed with SAINT-2–dioleoylphosphatidyl ethanolamine
(DOPE) vesicles, in a serum-insensitive and virtually non-cytotoxic manner. As a result, an
effective and specific down-regulation of the CRF-R, cloned in Chinese hamster ovary (CHO)
cells, was accomplished, using only nanomolar amounts of S-ODNs.
Materials and methods
Synthesis and formulation of amphiphile
SAINT-2 was synthesized as described in detail elsewhere (13,14). For vesicle preparation,
equimolar amounts of SAINT-2 and DOPE (Avanti Polar Lipids Inc., USA) were mixed and the
solvent was removed by evaporation under a stream of nitrogen, followed by placing the vial
under vacuum overnight. The lipids were then resuspended in water (1 ml) and sonicated to
clarity in a bath sonicator in a closed vial. Where indicated, 1 mol % N-(lissamine Rhodamine
sulfonyl)-PE (N-Rh-PE; Avanti Polar Lipids, Inc.) was included in the lipid mixture to monitor
the fate of the lipid complex by confocal fluorescence microscopy.
Cell culture
CHO cells, stably expressing the CRF receptor under the control of CMV promoter were
kindly provided by Solvey Pharmaceuticals (Weesp, The Netherlands). The cells were grown in
CHO-S-SFM medium (Gibco, Breda, The Netherlands) supplemented with 10% heat-inactivated
fetal calf serum, 2 mM L-glutamine and penicillin (50 U/ml)/streptomycin (50 µg/ml) under the
selection of 0.5 mg/ml geneticin in 5% CO2/95% air at 37°C.
ODNs
Two antisense S-ODNs, complementary to CRF-R mRNA (sequence see Gene Bank,
accession L25438), targeted to bp 474–490 (antisense 1) and bp 788–801 (antisense 2) with the
sequence: 5'-GGA TGA AAG CCG AGA TG-3' and 5'-GTA TAC CCC AGG AC-3',
respectively, as well as the GC-mismatched randomized-sequence ODNs with the sequence: 5'-
ACT ACG ACC TAC GTG AC-3' and 5'-GAA CCA AGA GCA CC-3', and a fluorescein-
labeled randomized-sequence ODN with the sequence: 5'-ACT ACG ACC TAC GTG AC-3'
were designed and manufactured by Biognostik (Göttingen, Germany). All ODNs were thioated
and purified by high-performance liquid chromatography, cross-flow dialysis and ultrafiltration.
A 25mer Cy5-labeled phosphodiester ODN was custom-synthesized by Pharmacia Biotech (NJ,
USA).
Cationic-lipid-mediated delivery of oligos
49
Preparation of SAINT-2–DOPE–antisense complexes
The complexes were prepared as follows. Twenty nanomoles of vesicles, prepared as
described above, were diluted in 100 µl serum-free CHO medium. Then, 0.1 nmol ODN in TE
buffer (pH 7.4) was added, mixed gently, and the complex was allowed to assemble by
incubating the mixture at room temperature. After 20 min, 900 µl pre-warmed (37°C) 10%
serum-containing or serum-free CHO medium was added. Alternatively, complexes were diluted
in 100 µl 0.9% NaCl/10 mM HEPES buffer (instead of serum-free CHO medium). Then, 0.1
nmol ODN in TE (pH 7.4) was added, mixed gently, and the complex was allowed to assemble
by incubating the mixture at room temperature. After 20 min, 900 µl pre-warmed (37°C) serum-
free CHO medium was added. Finally, complexes were prepared by diluting vesicles in 100 µl
10% serum-containing medium. After 20 min, 900 µl pre-warmed (37°C) 10% serum-containing
medium was added. The experiments were carried out in 6-well plates by adding the complexes,
as prepared, to the washed cells and, unless indicated otherwise, the experiments were carried
with ODN lipoplexes that had been assembled in serum-free CHO medium. The sizes of the
lipoplexes were determined by particle size analysis, using a Nicomp 370 submicron particle
sizer (CA, USA). The efficiency of ODN association with the complex was determined by using
the Oligreen® ssDNA Quantitation kit (Molecular Probes, OR, USA).
Cellular binding and uptake studies
500 000 cells/well were seeded in 6-well plates. After 24 h, when the cells had reached 70–
80% confluency, the complexes were prepared as described above, added to cells and incubated
at 37°C during a time interval as indicated. Cells were then rinsed with PBS, trypsinized and
resuspended in medium prior to quantifying fluorescence by fluorescence-activated cell sorting
(FACS) measurements.
For characterization of intracellular uptake and distribution, FITC-labeled ODNs and N-Rh-
PE-containing SAINT-2–DOPE complexes were visualized by TCS Leica confocal laser
scanning microscopy (Heidelberg, Germany). To this end, CHO cells were grown on coverslips
in 6-well plates and treated with SAINT-2–DOPE–ODNs complex as described above. To verify the endosomal–lysosomal localization of the SAINT-2–DOPE complex, the cells
were incubated with 2 mg/ml FITC-dextran (mol. wt 71 600 Da; Sigma) for 12 h. After removal
by washing, the cells were incubated for another 4 h with the lipoplexes. Finally, the cells were
rinsed three times with PBS, and analyzed directly or fixed for 10 min in 2.5% paraformaldehyde
in PBS, washed and mounted on microscope slides for examination.
Chapter 3
50
Cytotoxicity studies
The cytotoxicity of SAINT-2–DOPE, S-ODNs and their complex was determined in CHO
cells, using the MTT assay. In brief, following the incubation with lipid, antisense ODN or the
complex, the cells were incubated for 24 h. The surviving fraction was determined by the MTT
dye assay, measuring the absorbance at 520 nm with an automated microplate reader, as
described (15).
Antisense assay
The ability of SAINT-2–DOPE to deliver antisense ODNs in pharmacologically active form
was evaluated by examining the level of CRF-R mRNA in transfected CHO cells, treated with
SAINT-2–DOPE–antisense ODN complex. CHO cells, stably expressing CRF-R under the
control of CMV promoter, were plated in 10 cm dishes and grown for 24 h. Cells were then
treated with the complex of S-ODNs and SAINT-2–DOPE, prepared as described above. After a
24 h incubation at 37°C, total RNA was extracted from 6 x 106 cells using the Qiagen RNeasy kit
(Hilden, Germany). The purity and quantity of the RNA preparation was determined by
recording the absorbance at 260 and 280 nm. Equal amounts of RNA were resolved on a 1.2%
agarose gel, containing 6.7% formaldehyde, and transferred to a nylon membrane by vacuum
transfer. Total RNAs were fixed onto the membrane by heating for 1 h at 80°C. The RNA was
intact without any degradation as indicated by 18S and 28S bands on the membrane, as
visualized by methanol blue staining. CRF-R and GAPDH probes were prepared by amplifying
CRF-R and GAPDH cDNA by RT–PCR from total RNA in CHO cells. The CRF-R and GAPDH
primers were ATTATGGGACGGCGCCCG/ TCACACTGCTGTGGACTG and
CCACCCATGCAAATTCCATGGCA/ TCTAGACGGCAGGTCAGGTCCACC, respectively.
PCR was performed using oligo(dT)-primed cDNAs (synthesized with reverse transcriptase;
Boehringer Mannheim, Germany) as a template under the following conditions: 30 cycles of
94°C for 30 s, 58°C for 30 s, 72°C for 45 s, and a final 72°C extension of 5 min). The probes
from the PCR products were gel purified (QiaexII Gel Extraction kit, Germany) and labeled with
[32P]ATP using a random priming kit (Gibco BRL, Breda, The Netherlands). The level of CRF-R
mRNA was probed by the cDNA probe for the CRF-R gene, visualized by autoradiography and
quantified by densitometry, by scanning an area that contained the band of interest, and which
was of the same size for each band examined. To normalize the level of RNA loading, the CRF-
R probe was removed by stripping the membrane in boiling SDS solution (0.5%) for 10 min. The
membrane was then probed with the control GAPDH probe.
Levels of CRF receptors were evaluated by western immunoblot. An aliquot of 106 CRF-R-
expressing and control CHO cells were seeded in 10 cm dishes and grown for 24 h. The cells
Cationic-lipid-mediated delivery of oligos
51
were then treated with the S-ODN–SAINT-2–DOPE complex, and after a 24 h incubation at
37°C, the complex was removed and fresh medium was added. The cells were harvested after 48
h and lyzed, and cell membranes were isolated as described (16). A 100 µg sample of proteins
was run on 12.5% SDS–PAGE (Bio-Rad, Hercules, CA), electrotransferred to ImmobilonTM-P
transfer membranes (Millipore Corp., MA), and probed with goat anti-rat CRF-R (1:100, Santa
Cruz), followed by alkaline phosphatase-conjugated rabbit anti-goat antibody (1:3000, Sigma).
The blot was color processed by nitro blue tetrazolium and 5-bromo-4-chloro-3-indolyl (Sigma),
and analyzed by the software of image tool IT2.00.
Results
SAINT-2–DOPE (1:1) efficiently mediates antisense S-ODN uptake in an almost non-toxic
manner
In previous work, we have demonstrated that SAINT-2–DOPE (molar ratio 1:1), when
complexed with plasmid DNA, effectively transfects eukaryotic cells (13). Indeed, transfection
efficiencies up to 90% are obtained at a charge ratio (+/–) of approximately 2.5 (10–20 µM
lipid/1 µg DNA). The next goal was, therefore, to examine whether SAINT-2–DOPE could
similarly tranfer antisense ODNs into cells to regulate gene expression for therapeutic or cell
biological purposes. Using fluorescently tagged ODNs, we established that >95% of the added
ODNs, at all conditions described in the present study, associated with the SAINT-2–DOPE
vesicles prepared as described in the Materials and Methods, thus giving rise to efficient lipoplex
assembly (not shown). In order to determine the optimal ratio between SAINT-2–DOPE and S-
ODNs in terms of ODN delivery efficiency, CHO cells were treated either with 20 µM SAINT-
2–DOPE, complexed with various concentrations of FITC-labeled S-ODNs, or with a complex
consisting of 100 nM S-ODNs and various concentrations of SAINT-2–DOPE. Treated cells
were harvested after a 4 h incubation period, and the S-ODN delivery in terms of the efficiency
and number of cells that contained internalized complex was then measured by FACS. As shown
in Figure 1A at a fixed lipid concentration of 20 µM, both the absolute delivery and number of
cells that had taken up the complex increased with increasing S-ODN concentration. Thus, at an
ODN concentration of 80–100 nM, essentially all cells displayed the presence of the complex.
Conversely, when adding complexes to the cells at identical conditions, assembled from various
amounts of SAINT-2–DOPE and a fixed concentration of S-ODN of 100 nM, no further increase
in absolute delivery was obtained relative to that obtained for 100 nM ODN/20 µM SAINT-2–
DOPE (Fig. 1B). Therefore, the latter composition was used in this study to further examine the
mechanism and to define the efficiency of antisense delivery. To appreciate the efficiency of the
Chapter 3
52
delivery capacity of the present system, we first investigated the cellular uptake of non-
complexed S-ODN and the effect of serum. To this end, the cells were incubated with S-ODNs
alone or the S-ODN–SAINT-2–DOPE complex for 24 h at various conditions, and cellular
uptake was determined by measuring total cellular-associated fluorescence, using fluorescently
tagged S-ODN. As shown in Figure 1C, SAINT-2–DOPE enhanced S-ODN cellular uptake by
100–250-fold compared with the uptake of S-ODN alone. Interestingly, the absolute amount of
uptake, as reflected by the total cell-associated fluorescence, increased by 1.5-fold in the
presence of 10% FCS. Note, however, that in either case, essentially all cells displayed the
presence of cell-associated fluorescence.
The application of both synthetic amphiphiles and treatment with antisense ODNs has often
suffered from serious cytotoxic side effects. Although our previous work (13,17,18) clearly
implied the relative non-toxic nature of the SAINT amphiphile compared with other cationic
lipids, we next verified its potential toxicity in conjunction with the application of ODNs. CHO
cells were incubated with 20 µM SAINT-2–DOPE alone, 100 nM antisense S-ODNs alone, or
the ODN–amphiphile complex in the presence or absence of 10% serum. After 24 h, the cells
were harvested and the cytotoxicity was determined by the MTT assay. As shown in Figure 2,
the cells treated with antisense S-ODNs did not show any cytotoxicity. Cells treated with the
SAINT-2–DOPE alone, and the complex of SAINT-2–DOPE and S-ODNs showed only modest
cytotoxicity, which was <10% in the presence of serum. We conclude, therefore, that at an
appropriate molar ratio SAINT-2–DOPE (20 µM) effectively delivers ODNs (100 nM) to cells,
leading to complex uptake by essentially all cells in the culture system in an almost non-toxic
manner, which is promoted rather than inhibited by the presence of (10%) serum. To further
define the parameters of uptake, the mechanism of internalization and ensuing biological effects,
the next experiments were undertaken.
SAINT-2-mediated uptake of S-ODN is time-dependent and promoted by serum
To determine the kinetics of cellular uptake of the amphiphile–ODN complex, the cells were
incubated with FITC-labeled antisense S-ODN complex, prepared in CHO medium (see
Materials and Methods), in the presence or absence of serum. After various time intervals, the
cells were harvested and the cell-associated fluorescence was determined by FACS
measurements. As shown in Figure 3, already after 1 h of incubation, a significant association of
fluorescently tagged S-ODNs with the cells was apparent, the amount increasing over a time
interval of 12 h. Following this time interval, the subsequent increase over the next 12 h was only
minor, amounting to an additional 10–15% of the total uptake. Interestingly, as already noted
Cationic-lipid-mediated delivery of oligos
53
above, the efficiency of uptake could be enhanced when the cells were incubated with the
complex in the presence of 10% serum, the difference between both conditions becoming most
pronounced after 4 h of incubation. It is further relevant to note that over the entire incubation
period, all cells display a capacity in taking up the complex, even after 1 h, all cells of the
population show cell-associated fluorescence and, moreover, that this uptake is not affected by
the presence of serum.
Figure 1. Optimization of parameters for SAINT-2–DOPE-mediated delivery ofantisense S-ODNs into cells. Lipoplexes, containing 5' fluorescein-labeled S-ODN, ornon-complexed ODN were incubated with the cells as described in Materials andMethods, and the amount of cell-associated antisense was determined by FACSmeasurements. (A) The cell-associated fluorescence intensity (bars, left y-axis) and thepercentage of fluorescently labeled cells (line, right y-axis), as a function of theconcentration of antisense ODN complexed with 20 µM SAINT-2–DOPE (1:1). (B) Theassociation of the ODNs (100 nM) as a function of SAINT-2–DOPE concentration. Thenumber of cells that showed cell-associated lipoplexes is indicated by the line (right y-axis). (C) CHO cells were treated with S-ODNs alone (AS), or when complexed withSAINT-2–DOPE (AS + SD), in the presence (shaded bars) or absence of serum (filledbars). The line indicates the number of cells that were labeled in each population atthe various conditions indicated. Note that when complexed with SAINT-2–DOPE (100nM/20 µM of lipid) S-ODN delivery is enhanced by 100–250-fold, while complexuptake in the presence of serum is enhanced. Data are the mean values (±SD) of threedeterminations.
Figure 2. Antisense treatment and cell toxicity.CHO cells were incubated for 24 h in either thepresence (shaded bars) or absence of 10%serum (filled bars). The toxicity was determinedusing the MTT assay, and the data areexpressed as the percentage of surviving cells,relative to untreated CHO cells (100%). 1. S-ODNs 2. SAINT-2/DOPE 3. S-ODNs+SAINT-2/DOPE Mean values ± SD were obtained fromtwo to three experiments, carried out induplicate.
0
20
40
60
80
20 40 80 100
S-ODNs (nM)
cell-
asso
ciat
ed
fluor
esce
nce(
a.u.
)
40
60
80
100S-
OD
Ns
cont
aini
ng c
ells
(%)
A.
0
20
40
60
80
25 50 75 100
SAINT/DOPE (uM)
cell-
asso
ciat
ed
flure
nsce
nce(
a.u.
)
0
20
40
60
80
100
S-O
DN
s co
ntai
ning
cel
ls(%
)
B.
0
100
200
300
400
AS+SD AS
cell-
asso
ciat
ed fl
uore
scen
ce
0
20
40
60
80
100
S-O
DN
s co
ntai
ning
cel
ls(%
)
C
0
2 0
4 0
6 0
8 0
1 0 0
1 2 0
1 2 3
cell
viab
ility
(%)
Chapter 3
54
Thus far, we have carried out our studies using S-ODN in anticipation of potential nuclease-
mediated degradation of the ODN. To rationalize its use, it was of interest to examine whether D-
ODN delivery was processed differently by the cells, using the current carrier system. As shown
in Figure 4, this indeed appeared to be the case. Whereas the cell-associated fraction of thiolated
ODN continuously increased as a function of the incubation time, the cell-associated D-ODN
fraction, although increasing with fairly similar kinetics over the first 4 h of incubation as those
observed for S-ODN, rapidly decreased over the next 8 h of incubation. These data thus imply
that a major fraction of the ODNs becomes exposed to a mechanism that must involve
degradation prior to cellular expulsion within 4 and 12 h after the onset of the incubation. To
further examine the intracellular processing, we subsequently determined the (intra-)cellular
localization of cell-associated complexes by confocal laser scanning microscopy, employing
fluorescently labeled S-ODN, while the carrier itself was labeled by the non-exchangeable lipid
probe N-Rh-PE (19).
The presence of serum does not affect effective nuclear localization of S-ODNs
CHO cells were incubated with the fluorescently labeled (S-ODN and lipid) complex,
prepared in CHO-medium, for 3 h at 37°C in the presence or absence of serum. After this time
interval, serum-containing medium was added, and the cells were further incubated at 37°C.
After 21 h, the cells were extensively washed, and either examined directly or after fixation with
2.5% paraformaldehyde by confocal laser scanning microscopy (Fig. 5). The data reveal that
irrespective of the presence of serum in the incubation medium (initial 3 h), essentially all cells
contain fluorescence, which is largely localized in the nuclei. In the presence of serum very little
if any fluorescence is accumulating at the cell periphery (which would have been apparent as
patchy fluorescence), implying that most of the cell-associated fluorescence has been internalized
by the cells. The internalized red-labeled N-Rh-PE, marking the lipid carrier, can be discerned as
a fine-punctate fluorescence, which primarily localizes at perinuclear regions. Some
colocalization (as reflected by the yellow colour) of lipid and ODN is occasionally observed.
Note that laterally diffused membrane staining of N-Rh-PE is not detectable. The perinuclear
localization reflects the presence of the carrier in the endosomal–lysosomal pathway, as
demonstrated by a prominent colocalization (yellow) of the lysosomal marker FITC-dextran
(green), internalized over a 12 h period, and N-Rh-labeled (red) complexes, internalized for a
subsequent time interval of 4 h (Fig. 5C). Relative to its distribution in the presence of serum, a
more irregular appearance of N-Rh-PE fluorescence is observed in the absence of serum (Fig.
5A). Thus, in contrast to a fine-punctate appearance in its presence, in the absence of serum,
Cationic-lipid-mediated delivery of oligos
55
large clusters are apparent, which are often localized to the cell periphery, presumably reflecting
the incapacity of the cells to internalize such complexes. Consistent with this notion, the absolute
uptake of complexes in the absence of serum is less than in its presence (Fig. 3). Clearly, this
decrease does not affect the ultimate fate of the internalized S-ODNs as its nuclear localization
is, at least in a qualitative sense, as prominent as observed in the presence of serum (Fig. 5A
versus B). Finally, when non-complexed fluorescently labeled S-ODN was incubated at similar
conditions, an occasional fluorescent dot in the cytoplasm in <5% of the cells was observed. At
none of these conditions was any fluorescence localized in the nucleus (not shown).
Figure 3. Kinetics of S-ODN uptake and theeffect of serum. The cells were incubated withFITC-labeled S-ODN–SAINT-2–DOPEcomplexes for the indicated time intervals,either in the presence or absence of serum.After extensive washing, the cells wereharvested and the cell-associated fluorescence(filled bars, without serum; shaded bars, inthe presence of serum) and the number offluorescently labeled cells in the population(triangles and crosses, presence and absenceof serum, respectively) were measured byFACS. Results are the mean values ±SD ofthree different experiments, carried out induplicate.
Figure 4. Comparison of the kinetics ofprocessing of D-ODNs versus S-ODNs inCHO cells. CHO cells were incubated witheither FITC-labeled S-ODNs or Cy5-labeledD-ODNs, which were complexed withSAINT-2–DOPE. The cells were incubatedwith these complexes at 37°C for varioustimes, and the cell-associated fluorescencewas measured by FACS. The amount ofcell-associated S-ODNs continuouslyincreased over the 24 h incubation (shadedbars). However, the cell-associated D-ODNsonly transiently increased, the cell-associated fraction decreasing again after 4h of incubation. Results are the meanvalues ±SD of three determinations.
0
100
200
300
400
1h 2h 4h 12h24hincubate time
cell-
asso
ciat
ed
fluor
esce
nce(
a. u
.)
0
20
40
60
80
100
S-O
DN
s co
ntai
ning
ce
lls(%
)
0
20
40
60
80
100
120
1h 2h 4h 12h 24hincubation time
cell-
asso
ciat
ed
fluor
ensc
ence
(%)
Chapter 3
56
Only lipoplex-penetrating serum proteins affect ODN localization
The data presented above indicated that the presence of serum in the extracellular medium
may have a profound influence on the size of the complex, and accordingly, on the efficiency of
complex internalization. In addition, it is possible that serum proteins might penetrate to different
degrees into the complex, thereby affecting complex stability and, hence, the release of the ODN.
Both complex internalization and release of ODN from the complex are likely important
parameters in governing the eventual antisense effect. During these studies, we noted that
different sizes of lipoplexes could be obtained, depending on the methodology of preparation, as
indicated in the Materials and Methods (see section entitled Preparation of SAINT-2–DOPE–
antisense complexes). Thus, when prepared in serum-free CHO medium, complexes with a
diameter of up to 1000 nm were obtained, as determined by particle size analysis. When such
complexes were mixed in serum-containing medium within 20 min after preparation, the size of
the complexes was 750 nm. Interestingly, when ODN–lipid complex assembly was carried out in
buffer (0.9 % NaCl/10 mM HEPES), particles were obtained with a diameter of only 80 nm. In
contrast, when for complex assembly the buffer was replaced by serum-containing CHO
medium, thereby simulating the potential penetration of serum proteins into the lipid core, the
size of the complexes was 200–300 nm. Importantly, at all conditions, lipoplexes effectively
assembled, involving >95% of the added ODN (not shown). When added to cells and examining
the cellular localization of the complexes, i.e. as prepared in either CHO-medium, NaCl–HEPES
buffer, or in serum-containing CHO medium, it was observed that in all but one case, nuclear
fluorescence was apparent, and essentially indistinguishable in appearance from the data shown
in Figure 5A and B. The exception concerned complexes that had been assembled in serum-
containing CHO medium. In this case, the level of cell-associated fluorescence was similar to
that seen for complexes prepared in the absence of serum, but intriguingly, no fluorescence was
apparent in the nucleus. Rather, only punctate fluorescence, distributed throughout the cytoplasm
was present, indicating that neither vesicle clustering nor growth occurred during the incubation,
consistent with incubations of complexes assembled in serum-free media, but incubated with the
cells in the presence of serum. The typical cytoplasmic distribution and perinuclear localization,
in conjunction with the colocalization of nucleotide and lipid as reflected by the yellow colour,
likely represents entrapment of the complex in the endosomal–lysosomal pathway (Fig. 5C
versus D), implying that dissociation of ODN from the complex did not occur. Accordingly, as
an incubation with the 80 nm particles, as obtained by preparing the complex in NaCl–HEPES,
result in efficient delivery, the data indicate that dissociation rather than particle size represents a
rate limiting step in ODN delivery.
Cationic-lipid-mediated delivery of oligos
57
Figure 5. Confocal microscope images of intracellular uptake and distributionpatterns of S-ODNs, mediated by SAINT-2–DOPE. CHO cells were incubated withcomplexes, prepared from FITC-labeled S-ODN, and SAINT-2–DOPE, labeled with 1mol% N-Rh-DOPE, at 37°C for 24 h in the presence or absence of 10% serum. Imagesshow fields of live cells in which the S-ODNs are visualized in green, while the SAINT-2–DOPE vehicles are visualized in red. (A) Distribution of ODNs and carrier afterincubation in the absence of serum. Note the presence of S-ODNs in the cell nuclei(green), whereas SAINT-2–DOPE was present as large, heterogenous dots (red) in thecytoplasm and at the cell surface. (B) Cells were incubated with complexes, preparedas in (A), in the presence of serum. The S-ODNs were localized in the cell nuclei(green), while in this case, the SAINT-2–DOPE carrier was seen as relatively small andhomogenous dots (red) in the cytoplasm, giving rise to a fine-punctate distribution.Note that in both (A) and (B), only occasionally yellow dots are seen, reflecting thecolocalization of the fluorescent ODN and lipid. The perinuclear localization of the red-labeled carrier lipid (in A and B) most likely originated from the presence of the carrierin the endosomal–lysosomal pathway. When cells were incubated with the lysosomalmarker FITC-dextran (green) for 12 h, followed by a 4 h incubation with complexes ofS-ODNs (non-labeled) and N-Rh-PE–SAINT-2–DOPE (red; in the absence of serum), aprominent colocalization of lysosomal marker and SAINT-2–DOPE was apparent, asreflected by the yellow fluorescence (C). (D) When the complexes had been prepared inthe presence of 10% serum, only non-dissociated complexes were seen within thecells, as reflected by the almost exclusive appearance of yellow fluorescence.
Chapter 3
58
Following dissociation and arrival at the nucleus, we finally examined whether the ODN thus
delivered proved to be effective in conveying a biological response.
SAINT-2-mediated delivery of antisense S-ODNs efficiently down-regulates CRF-R mRNA
and protein levels
To determine the capacity of SAINT-2–DOPE to deliver ODNs in pharmacologically active
quantities, the ability of antisense ODNs targeted to rat CRF-R to down-regulate the levels of
mRNA was examined by Northern blot analysis. CHO cells were incubated for 3 h with the
complex, consisting of 100 nM antisense S-ODNs and 20 µM SAINT-2–DOPE in the absence of
serum, followed by the addition of 10% serum-containing medium. After an incubation of 24 h,
the cells were harvested and total RNA was extracted and transferred onto nylon membranes.
The blots were first probed by 32P-labeled CRF-R probes, and then stripped and reprobed by 32P-
labeled GAPDH probes. GAPDH was used as an internal control for the RNA loading. The
relative amounts of mRNA were determined by densitometric measurement of the
autoradiographs by scanning a defined spot-encompassing area, which was the same for all spots,
and the amount of CRF-R mRNA was normalized to that of GAPDH mRNA. The use of the
antisense S-ODN as applied led to a reduction in CRF-R mRNA by 50%. Either antisense
sequence (1 and 2, see Materials and Methods) displayed approximately the same efficiency. No
effect was seen when the cells had been treated with antisense S-ODNs only, or a complex
consisting of mismatched ODNs and SAINT-2–DOPE, emphasizing the specificity of the
observed effect. The latter control also indicates that the cationic lipid mixture itself does not
significantly affect CRF-R expression in a non-specific manner, such as by affecting cell
cytotoxicity. This is entirely consistent with the data presented in Figure 2, in which toxicity was
monitored directly. The seemingly slight decrease in activity seen in lane 9 (Fig. 6), in which free
vesicles rather than complexes were added, may therefore have been caused by fluctuations seen
in GAPDH expression, rather than by a toxic effect of the amphiphile. At any rate, the data in
conjunction with those shown in Figure 2 support the notion that the ODN–lipid complexes
display little if any toxicity towards the cells. Thus, SAINT-2–DOPE was able to effectively
deliver antisense ODNs into the CRF-R expressing CHO cells and consequently elicit a potent
and selective inhibition of gene expression.
To verify the consequences of mRNA down-regulation in terms of down-regulation in receptor
expression per se, we subsequently determined the level of CRF-R expression by Western
immunoblot, following antisense ODN treatment. The cells were incubated with complexes as
described above, i.e. 100 nM ODN and 20 µM SAINT-2–DOPE in the absence of serum for 3 h,
Cationic-lipid-mediated delivery of oligos
59
after which serum-containing medium was added, and the cells were left for 24 h. The ODN
complexes were then removed by washing the cells, and fresh medium was added. After another
48 h, the cell membranes were isolated, and the presence of CRF-R was determined as described
in the Materials and Methods. As shown in Figure 7, a prominent reduction of >50% in CRF-R
was only seen when the cells had been treated with either antisense 1 or antisense 2 ODN,
complexed with SAINT-2–DOPE. Accordingly, the data demonstrate that SAINT-2–DOPE-
mediated delivery of CRF-R antisense effectively down-regulated both mRNA and protein
levels.
Figure 6. Specific down-regulation of CRF-R messenger RNA by antisense S-ODNs.CRF-R expressing CHO cells were treated with lipoplexes prepared from 100 nM S-ODNs and 20 µM SAINT-2–DOPE for 3 h (in the absence of serum), and weresubsequently incubated for 24 h in 10% serum-containing medium. Total RNA wasthen isolated, fractionated on agarose formaldehyde gels, and blotted on nylonmembrane as described in the Materials and Methods. This membrane was probedwith 32P-radiolabeled CRF cDNA (CRF-R) and then stripped and reprobed with the 32P-radiolabeled GAPDH cDNA (GAPDH). GAPDH is used as the internal control for RNAloading. Note that a decrease of CRF-R mRNA was only seen when antisense ODNshad been delivered by SAINT-2–DOPE(SD). Each of the above experiments wasrepeated in duplicate, and similar results were obtained. Two antisense sequenceswere employed, antisense 1 and antisense 2 (for sequences see Materials andMethods). MAS1(2) + SD indicates results obtained with lipoplexes containingmismatched antisense (AS) 1 or 2. As a negative control, CHO cells without CRF-Rexpression were treated and analyzed similarly (CHO-cells, no CRF-R).
Figure 7. Effect of antisense S-ODNs on CRF-R expression. CRF-R was analyzed byWestern immunoblot. CHO cells expressing CRF-R under the control of CMV promoterwere treated with 100 nM S-ODNs and 20 µM SAINT-2–DOPE for 3 h in the absence ofserum, followed by an incubation of 24 h in 10% serum-containing medium. Thelipoplexes were removed by washing and fresh medium with serum was added. Afteranother 48 h, the cell membranes were isolated, the proteins were separated on PAGE,probed with goat anti-rat CRF-R and then with alkaline phosphatase-conjugatedrabbit anti-goat antibody. The reduction of CRF-R was only observed when antisenseODNs had been delivered by SAINT-2–DOPE (antisense 1 and 2). MAS representstreatment with mismatched antisense 1 and 2. Each of the above experiments wasrepeated in duplicate, and similar results were obtained.
Chapter 3
60
Discussion
To design a successful antisense-ODN-based approach for either biochemical or therapeutic
purposes, it is crucial to develop a convenient carrier system that transfers ODNs efficiently into
cells and across membranes, followed by an equally efficient delivery into the nucleus, high
enough to promote effective RNA binding. The present work reveals that the cationic lipid
SAINT-2, which has been previously demonstrated to act as an efficient, virtually non-toxic
carrier for gene delivery (13), also effectively translocates small ODNs into cells, in a non-toxic
and serum-insensitive manner. The relative insensitivity of the present system towards serum is
highly relevant as many previous studies revealed a more or less perturbing interference with
ODN or gene delivery (1,20,21). Recently, some novel umbrella amphiphile systems have been
developed that showed only a moderate capacity in overcoming serum-inhibited delivery of
antisense ODN (22). Relative to free ODNs, the delivery capacity of the SAINT-2–DOPE carrier
system is enhanced by at least two orders of magnitude. In passing, the efficient uptake of ODN–
SAINT-2–DOPE complexes is not restricted to their uptake by CHO cells. Cellular
internalization of such complexes with a similar efficiency as in CHO cells has also been
obtained in COS7 cells, mouse pituitary AtT20 cells and in a mouse L cell line, implying a
potential for a wide application of the present ODN carrier system (F.Shi and D.Hoekstra,
unpublished observations). Here evidence has been presented that CRF-R antisense ODNs
introduced in this manner effectively and specifically down-regulate receptor expression at both
the mRNA and protein level. Within 24 h, the mRNA level was down-regulated by 50%. The
level of protein expression diminished even to a higher extent after the same period of antisense
treatment, but because of the half time of the receptor (60 h) quantified after another 48 h,
thereby indicating the stability of the introduced ODNs. Moreover, the significance of the
efficiency of this down-regulation is emphasized when taking into account that exposure of a
pituitary gland primary cell culture to 10 µM free CRF-R antisense ODN for 40–67 h resulted in
decrease in CRF binding to the receptor of 17–36% (2). Similarly, others (1,2) reported in vitro
studies in which antisense effects and down-regulation (of among others the CRF-R) could only
be accomplished in the micromolar range, whereas SAINT-2–DOPE elicits such effects in the
nanomolar range. It should also be noted that the drastic reduction of CRF-R expression was
obtained while being under control of the CMV promoter, which is much stronger than the
promoter (23), effective at physiological conditions, i.e. during stress. Our studies also provide insight into the mechanism of ODN delivery. Thus, the data suggest
that particle size is not a crucial factor in ODN delivery, as even particles of a size of only 80 nm,
as obtained by complex assembly in NaCl–HEPES, effectively carry ODNs into cells. Rather,
Cationic-lipid-mediated delivery of oligos
61
stabilization of the carrier system by serum proteins, as accomplished when complex assembly is
carried out in the presence of serum-containing medium, and which presumably leads to a partial
insertion of serum proteins into the carrier’s membrane, leads to endocytic internalization of the
complex, without the release of ODNs. In this manner, the presence of protein may preclude the
complex from destabilizing the endosomal or plasma membrane, necessary for ODN
translocation. Note that these conditions differ from those in which complexes, assembled in
serum-free medium, are added to cells, cultured in the presence of serum. Thus, when serum
proteins present in the medium are recruited onto the surface of SAINT-2–DOPE–ODN
complexes, prepared in serum-free medium, such an inhibitory effect is not observed. In contrast,
in the presence of serum, the delivery appears to be enhanced. Therefore, the inhibitory activity
of serum proteins seen after complex preparation in the presence of serum must be related to a
requirement for the cationic lipid of being able to (locally) perturb the structural integrity of its
target membrane in order to accomplish efficient delivery. Clearly, such a perturbation does not
involve a fusion step. At none of the incubation conditions at which ODN translocation had taken
place was the presence of a mobile fraction of N-Rh-PE in cellular membranes apparent. It can
be argued whether the resolution suffices for discerning such a distinction at the level of the
endocytic membrane. However, as there is a continuous recycling from such membranes to the
cell surface, recycling should have caused the reappearance of N-Rh-PE at the plasma membrane,
resulting in laterally diffused membrane-associated fluorescence, as previously demonstrated for
recycling of fluorescently tagged sphingolipids between endosomes and plasma membrane
(24,25). Accordingly, we exclude that SAINT-2–DOPE-mediated ODN translocation includes a
fusion-mediated step, but rather may facilitate (transient) pore formation as argued previously
(13,26), presumably involving the polymorphic features of SAINT-2 which, in conjunction with
the presence of DOPE, may involve the ability of the cationic lipid to convert into non-lamellar,
hexagonal phases (27,28 and our unpublished observations).
Our data also indicate that the structure of the ODN is not the rate-determining step in actual
delivery. Both D-ODNs and S-ODNs are efficiently translocated during the early phase of
complex–cell interaction, i.e. between 0 and 4 h. After 4 h, however, the D-ODN is rapidly
expelled from the cells, presumably because of its degradation by nucleases, consistent with
observations of others (29). However, S-ODNs are resistant to such degradation for at least 24 h
(30,31), i.e. a time that suffices for the ODNs to reach their mRNA targets, consistent with the
efficient down-regulation observed in the present work.
Chapter 3
62
Although some colocalization of fluorescently tagged ODNs and lipid was occasionally seen
in the endosomal pathway, suggesting that dissociation had not occurred, free ODNs (green
fluorescence) were almost exclusively present in the nucleus. In contrast, when antisense ODNs
were added to cells directly, little uptake was seen, but once internalized, they were sequestered
in compartments that presumably represented endocytic compartments (not shown), consistent
with results presented by others (32–34). These data thus imply that for reaching the cytosol,
ODN translocation and passage of the endosomal membrane barrier requires the presence of the
cationic lipid. For subsequent arrival at the nucleus, no cationic lipid appears to be needed as,
based upon the localization of the lipid marker N-Rh-PE, the carrier system itself is largely
retained in the endosomal–lysosomal track, as revealed by its colocalization with the lysosomally
processed marker dextran (Fig. 5). Indeed, it has long been documented that when microinjected
into the cytosol, ODNs readily diffuse into the nucleus (35,36), a localization that appears to be
closely related to the capacity of antisense ODN to down-regulate target mRNA (37). In this
context it is finally interesting to note that polyplex-mediated delivery of ODN may be
accomplished by co-entry of the polymer (PEI) into the nucleus (38). The fact that the cationic
lipid does not reach/enter the nucleus may imply a different mechanism, but it excludes potential
interference of the cationic lipid with antisense effectiveness.
In this regard, an ideal lipoplex-based ODN delivery system requires low cytotoxicity, little or
no serum sensitivity, high efficiency of delivery into the nucleus, and a simple, economically
advantageous large-scale production of ODN–amphiphile assemblies. SAINT-2–DOPE appears
to comply with such requirements. Considering also its ability to efficiently deliver genes into
cells for biochemical and therapeutic purposes, the present carrier can, thus, be regarded as a
versatile and generally applicable delivery system.
AcknowledgementsWe gratefully acknowledge Mr Rense Veenstra (Biomedical Technology Center,University of Groningen) for expert help in carrying out the mRNA analyses, Dr JanJansen (Solvay Pharmaceutics, The Netherlands) for providing the CRF-R CHO cells,and Anno Wagenaar for synthesizing SAINT-2. This work was supported by a grantfrom The Netherlands Organization for Scientific Research (NWO)/NDRF InnovativeDrug Research (940-70-001).
References:1. Tao,M., Miyano-Kurosaki,N., Takai,K. and Takaku,H. (1999) Specific inhibition of
human telomerase activity by transfection reagent, FuGENE6-antisensephosphorothioate oligonucleotide complex in HeLa cells. FEBS Lett., 454, 312–316.
2. Owens,M.J., Mulchahey,J.J., Kasckow,J.W., Plotsky,P.M. and Nemeroff,C.B. (1995)Exposure to an antisense oligonucleotide decreases corticotropin-releasing factorreceptor binding in rat pituitary cultures. J. Neurochem., 64, 2358–2361.
Cationic-lipid-mediated delivery of oligos
63
3. Owens,M.J. and Nemeroff,C.B. (1991) Physiology and pharmacology ofcorticotropin-releasing factor. Pharmacol. Rev., 43, 425–473.
4. Gonzalez,M.A., Serrano,F., Llorente,M., Abad,J.L., Garcia-Ortiz,M.J. and Bernad,A.(1998) A hammerhead ribozyme targeted to the human chemokine receptor CCR5.Biochem. Biophys. Res. Commun., 251, 592–596.
5. Krieg,A.M. and Stein,C.A. (1995) Phosphorothioate oligodeoxynucleotides:antisense or anti-protein? Antisense Res. Dev., 5, 241.
6. Eckstein,F. and Thomson,J.B. (1995) Phosphate analogs for study of DNApolymerases. Methods Enzymol., 262, 189–202.
7. Stein,C.A. and Cohen,J.S. (1988) Oligodeoxynucleotides as inhibitors of geneexpression: a review. Cancer Res., 48, 2659–2668.
8. Uhlmann,E. and Peyman,A. (1990) Antisense oligonucleotides: a new therapeuticprinciple. Chem. Rev. 90, 544–584.
9. Bergan,R., Hakim,F., Schwartz,G.N., Kyle,E., Cepada,R., Szabo,J.M., Fowler,D.,Gress,R. and Neckers,L. (1996) Electroporation of synthetic oligodeoxynucleotides:a novel technique for ex vivo bone marrow purging. Blood, 88, 731–741.
10. Mathiesen,I. (1999) Electropermeabilization of skeletal muscle enhances genetransfer in vivo. Gene Ther., 6, 508–514.
11. Zelphati,O., Zon,G. and Leserman,L. (1993) Inhibition of HIV-1 replication incultured cells with antisense oligonucleotides encapsulated in immunoliposomes.Antisense Res. Dev., 3, 323–338.
12. Delong,R., Stephenson,K., Loftus,T., Fisher,M., Alahari,S., Nolting,A. andJuliano,R.L. (1997) Characterization of complexes of oligonucleotides withpolyamidoamine starburst dendrimers and effects on intracellular delivery. J.Pharm. Sci., 86, 762–764.
13. Van der Woude,I., Wagenaar,A., Meekel,A.A., ter Beest,M.B., Ruiters,M.H.,Engberts,J.B.F.N. and Hoekstra,D. (1997) Novel pyridinium surfactants forefficient, nontoxic in vitro gene delivery. Proc. Natl Acad. Sci. USA, 94, 1160–1165.
14. Meekel,A.A.P., Wagenaar,A., Smisterova,J., Kroeze,J., Haadsma,P., Bosgraaf,B.,Stuart,M.C.A., Brisson,A., Ruiters,M.H.J., Hoekstra,D. et al. (2000) Synthesis ofpyrimidinium amphiphiles used for transfection and some characteristics ofamphiphile/DNA complex formation. Eur. J. Org. Chem., 2000, 665–673.
15. Carmichael,J., DeGraff,W.G., Gazdar,A.F., Minna,J.D. and Mitchell,J.B. (1987)Evaluation of a tetrazolium-based semiautomated colorimetric assay: assessmentof chemosensitivity testing. Cancer Res., 47, 936–942.
16. Kramer,E.M., Koch,T., Niehaus,A. and Trotter,J. (1997) Oligodendrocytes directglycosyl phosphatidylinositol-anchored proteins to the myelin sheath inglycosphingolipid-rich complexes. J. Biol. Chem., 272, 8937–8945.
17. Oberle,V., Zuhorn,I.S., Audouy,S., Bakowsky,U., Smisterova,J., Engberts,J.B.F.N.and Hoekstra,D. (2000) Cationic amphiphiles as delivery system for genes intoeukaryotic cells. NATO ASI Series, 323, 146–155.
18. Audouy,S., Molema,G., de Leij,L.M.F.H. and Hoekstra,D. (2000) Serum asmodulator of lipoplex-mediated gene transfection: Dependence of amphiphile, celltype and complex stability. J. Gene Med., 2, 465–476.
19. Struck,D.K., Hoekstra,D. and Pagano,R.E. (1981) Use of resonance energy transferto monitor membrane fusion. Biochemistry, 20, 4093–4099.
20. Tomlinson,E. and Rolland,A.P. (1996) Controllable gene therapy. Pharmaceutics ofnon-viral gene delivery systems. J. Control Release, 39, 357–372.
21. Litzinger,D.C., Brown,J.M., Wala,I., Kaufman,S.A., Van,G.Y., Farrell,C.L. andCollins,D. (1996) Fate of cationic liposomes and their complex with oligonucleotidein vivo. Biochim. Biophys. Acta, 1281, 139–149.
22. DeLong,R.K., Yoo,H., Alahari,S.K., Fisher,M., Short,S.M., Kang,S.H., Kole,R.,Janout,V., Regan,S.L. and Juliano,R.L. (1999) Novel cationic amphiphiles asdelivery agents for antisense oligonucleotides. Nucleic Acids Res., 27, 3334–3341.
Chapter 3
64
23. Boshart,M., Weber,F., Jahn,G., Dorsch-Hasler,K., Fleckenstein,B. andSchaffner,W. (1985) A very strong enhancer is located upstream of an immediateearly gene of human cytomegalovirus. Cell, 41, 521–530.
24. Kok,J.W., Eskelinen,S., Hoekstra,K. and Hoekstra,D. (1989) Salvage ofglucosylceramide by recycling after internalization along the pathway of receptor-mediated endocytosis. Proc. Natl Acad. Sci. USA, 86, 9896–9900.
25. Kok,J.W., Hoekstra,K., Eskelinen,S. and Hoekstra,D. (1992) Recycling pathways ofglucosylceramide in BHK cells: distinct involvement of early and late endosomes. J.Cell Sci., 103, 1139–1152.
26. Van der Woude,I., Visser,H.W., ter Beest,M.B., Wagenaar,A., Ruiters,M.H.,Engberts,J.B.F.N. and Hoekstra,D. (1995) Parameters influencing the introductionof plasmid DNA into cells by the use of synthetic amphiphiles as a carrier system.Biochim. Biophys. Acta, 1240, 34–40.
27. Hirsch-Lerner,D. and Barenholz,Y. (1999) Hydration of lipoplexes commonly usedin gene delivery: follow-up by laurdan fluorescence changes and quantification bydifferential scanning calorimetry. Biochim. Biophys. Acta, 1461, 47–57.
28. Oberle,V., Bakowsky,U., Zuhorn,I.S. and Hoekstra,D. (2000) Lipoplex formationunder equilibrium conditions reveals a three-step mechanism. Biophys. J., 79,1447–1454.
29. Sixou,S., Szoka,F.C.,Jr, Green,G.A., Giusti,B., Zon,G. and Chin,D.J. (1994)Intracellular oligonucleotide hybridization detected by fluorescence resonanceenergy transfer (FRET). Nucleic Acids Res., 22, 662–668.
30. Leonetti,J.P., Mechti,N., Degols,G., Gagnor,C. and Lebleu,B. (1991) Intracellulardistribution of microinjected antisense oligonucleotides. Proc. Natl Acad. Sci. USA,88, 2702–2706.
31. Fisher,T.L., Terhorst,T., Cao,X. and Wagner,R.W. (1993) Intracellular dispositionand metabolism of fluorescently-labeled unmodified and modified oligonucleotidesmicroinjected into mammalian cells. Nucleic Acids Res., 21, 3857–3865.
32. Wagner,R.W., Matteucci,M.D., Lewis,J.G., Gutierrez,A.J., Moulds,C. andFroehler,B.C. (1993) Antisense gene inhibition by oligonucleotides containing C-5propyne pyrimidines. Science, 260, 1510–1513.
33. Loke,S.L., Stein,C.A., Zhang,X.H., Mori,K., Nakanishi,M., Subasinghe,C.,Cohen,J.S. and Neckers,L.M. (1989) Characterization of oligonucleotide transportinto living cells. Proc. Natl Acad. Sci. USA, 86, 3474–3478.
34. Tonkinson,J.L. and Stein,C.A. (1994) Patterns of intracellularcompartmentalization, trafficking and acidification of 5'-fluorescein labeledphosphodiester and phosphorothioate oligodeoxynucleotides in HL60 cells. NucleicAcids Res., 22, 4268–4275.
35. Chin,D.J., Green,G.A., Zon,G., Szoka,F.C.,Jr and Straubinger,R.M. (1990) Rapidnuclear accumulation of injected oligodeoxyribonucleotides. New Biol., 2, 1091–1100.
36. Leonetti,J.P., Mechti,N., Degols,G., Gagnor,C. and Lebleu,B. (1991) Intracellulardistribution of microinjected antisense oligonucleotides. Proc. Natl Acad. Sci. USA,88, 2702–2706.
37. Marcusson,E.G., Bhat,B., Manoharan,M., Bennett,C.F. and Dean,N.M. (1998)Phosphorothioate oligodeoxyribonucleotides dissociate from cationic lipids beforeentering the nucleus. Nucleic Acids Res., 26, 2016–2023
38. Godbey,W.T., Wu,K.K. and Mikos,A.G. (1999) Tracking the intracellular path ofpoly(ethylenimine)/DNA complexes for gene delivery. Proc. Natl Acad. Sci. USA, 96,5177–5181.
65
Chapter 4
Antisense oligonucleotides reach mRNA targets via theRNA matrix: downregulation of the 5-HT1A receptor
Fuxin Shia, Willy H. Vissera, Natasja M. J. de Jongb, Robert S. B. Liemc, Eric Ronkenb andDick Hoekstraa
a Department of Membrane Cell Biology, Faculty of Medical Sciences, University of Groningen,Antonius Deusinglaan 1,9713 AV, Groningen, The Netherlands
b Solvay Pharmaceuticals, Research Laboratories, Weesp, The Netherlandsc Department of Electron Microscopy, Faculty of Medical Sciences, Groningen, The Netherlands
Published in Experimental Cell Research, 291(2003): 313-325
Chapter 4
66
Abstract
Successful application of antisense oligonucleotides (ODNs) in cell biology and therapy will
depend on the ease of design, efficiency of (intra)cellular delivery, ODN stability, and target
specificity. Equally essential is a detailed understanding of the mechanism of antisense action.
To address these issues, we employed phosphorothioate ODNs directed against specific regions
of the mRNA of the serotonin 5HT1A receptor, governed by sequence and structure. We
demonstrate that rather than various intracellular factors, the gene sequence per se primarily
determines the antisense effect, since 5HT1a autoreceptors expressed in RN46A cells,
postsynaptic receptors expressed in SN48 cells, and receptors overexpressed in LLP-K1 cells are
all efficiently downregulated following ODN delivery via a cationic lipid delivery system. The
data also reveal that the delivery system as such is a relevant parameter in ODN delivery.
Antisense ODNs bound extensively to the RNA matrix in the cell nuclei, thereby interacting with
target mRNA and causing its subsequent degradation. Antisense delivery effectively diminished
the mRNA pool, thus resulting in downregulation of newly synthesized 5HT1A proteins, without
the appearance of truncated protein fragments. In conjunction with the selected mRNA target
sequences of the ODNs, the latter data indicated that effective degradation rather than a steric
blockage of the mRNA impedes protein expression. The specificity of the antisense approach, as
described in this study, is reflected by the effective functional downregulation of the 5-HT1A
receptor.
Author Keywords: Antisense oligoribonucleotides; Gene therapy; Nuclear matrix; Drug
carriers; Lipids-Liposomes-; Receptors, -serotonin; RNA messenger metabolism
Oligos reach mRNA via RNA matrix
67
Introduction
Antisense oligonucleotides (ODNs) offer a great potential as sequence-specific modulators of
gene expression, which may be exploited for therapeutic and cell biological purposes. In fact,
recent evidence suggests that mammalian cells do use natural antisense transcripts to regulate
gene expression [1 and 2].
Work carried out thus far indicates that ODNs per se poorly permeate across cellular
membranes. As transport vehicles, cationic lipids strongly promote cellular entry, which appears
to proceed along the pathway of endocytosis [3]. Physical parameters of the complex, in
particular a lamellar-to hexagonal phase transition of the lipid phase [4], subsequently determine
the ability of the ODNs to translocate across the endosomal membrane and to acquire access to
the nucleus, reflected by the specific accummulation of fluorescently tagged ODNs [5 and 6].
Once in the nucleus, antisense ODNs may form compact nuclear bodies or longitudinal rodlets in
a concentration-dependent manner [7], but the implication to the antisense effect remains
unclear.
Several mechanisms have been proposed as to how gene expression is inhibited [8], including
antisense ODN hybridization with target mRNA, thereby providing a steric block for the
translation machinery [9 and 10] or an effective target for RNase H, resulting in mRNA
degradation [11 and 12]. Furthermore, to reliably attribute the effect of ODNs to a specific
antisense effect, it is equally essential to demonstrate a direct link between target specificity and
functional consequences, resulting from the downregulation of protein expression [13, 14, 15, 16
and 17].
To obtain further insight into the mechanism by which ODNs cause their antisense effect, we
have investigated the effect of ODNs directed against the serotonin receptor 5-HT1A as a target
gene. Since this receptor is involved in diseases like depression and anxiety, the receptor is of
obvious neurological interest as a potential therapeutic target. In addition, these receptors are
widely distributed in the CNS, being expressed on different cell types, yet genetically and
structurally indistinguishable. This provided an opportunity to investigate cell type-dependent
antisense effects involving the same protein receptor, i.e., the 5-HT1A receptor as naturally
expressed in raphe neuronal cells (RN46A) and septal neuronal cells (SN48) and overexpressed
in mouse LLP-K1 cells. Here we show a direct correlation between mRNA downregulation and
an inhibition of both expression and functioning of a membrane receptor, 5-HT1A. Our data
support a mechanism involving antisense ODN-induced degradation of the target mRNA,
triggered via binding to the RNA matrix in the nucleus.
Chapter 4
68
Materials and methods
Materials and ODNs
SAINT-2 was synthesized as described in detail elsewhere [18]. Lipofectamine 2000 was
purchased from Invitrogen. All other chemicals were from Sigma (St. Louis, MO, USA), unless
stated otherwise. Two antisense PS-ODNs, complementary to rat 5-HT1A mRNA (for sequence
see GeneBank Accession No. J05276), targeted to bp 115–128 (antisense 1, AS1) and bp 885–
902 (antisense 2, AS2) with the sequences ATC CAT GCC TGC CT and ACT ACC TGG CTG
TCC GTT, respectively, as well as two mismatched randomized ODNs with the sequences TCC
TCT TCG ACT GCT CTC (MAS1) and ACC TAC GTG ACT ACG T (MAS2), and a
fluorescein-labeled randomized-sequence, ACT ACT ACA CTA GAC TAG, were designed and
manufactured by Biognostik (Göttingen, Germany). All ODNs were thioated and purified by
HPLC, cross-flow dialysis, and ultrafiltration. Biotin-labeled phosphorothiate ODNs were
synthesized by Invitrogen Ltd. (Breda, The Netherlands).
Cell culture
RN46A and SN48 cells were kindly provided by Dr. Paul Albert, University of Ottawa,
Ottawa, Canada [19 and 20]. RN46A cells were maintained in Dulbecco's modified Eagle's
medium (DMEM, Life technologies, Paisley, UK) supplemented with 10% fetal calf serum, 2
mM L-glutamine, and penicillin (50 U/ml)/ streptomycin (50 U/ml) at a permissive temperature
of 33°C in 5% CO2. RN46A cells were differentiated by substituting serum for a mixture of 1 %
bovine serum albumin in DMEM, containing 1 g/ml transferrin, 5 g/ml insulin, 6.3 g/ml
progesterone, and 16.1 g/ml putrescine at 39°C [21]. SN48 cells were maintained in DMEM
supplemented with 10% fetal calf serum, 2 mM L-glutamine, and penicillin (50 U/ml)/
streptomycin(50 U/ml) at 37°C in 5% CO2. After transfection, SN48 cells were differentiated by
a reduction of the fetal calf serum concentration to 1% and the addition of 1 M retinoic acid [20].
LLP-K1 cells stably transfected with the 5-HT1A receptors (LLP-5-HT1A) were kindly provided
by Dr. Michèle Darmon, Boulevard de I'Hôpital, Paris, France [22]. LLP-5-HT1A cells were
maintained in DMEM supplemented with 10% fetal calf serum, 2 mM L-glutamine, and
penicillin (50 U/ml)/streptomycin (50 U/ml) at 37°C in 5% CO2 under the selection of 1 mg/ml
geneticin.
Preparation of antisense complexes and their incubation with cells
Antisense ODNs were complexed with sonicated cationic lipid vesicles consisting of SAINT-2
and DOPE (1:1) as described in detail elsewhere [6]. In preliminary work, Lipofectamine 2000,
Oligos reach mRNA via RNA matrix
69
purchased from Invitrogen, was also employed, and antisense ODN complexes were prepared
essentially as instructed by the manufacturer. Where indicated, 0.5 mol% N-(lissamine
rhodamine sulfonyl)-PE (N-Rh-PE, Avanti Polar Lipids Inc) was included in the lipid mixture to
monitor the fate of the lipid complex by confocal fluorescence microscopy. Antisense/cationic
lipid complexes were prepared as follows: 15 nmol SAINT-2/DOPE vesicles or 1–3 µl
Lipofectamine 2000, diluted in 500 µl DMEM, was gently mixed (4–5 min) with 0.1 nmol
ODNs, also diluted in 500 µl DMEM. The complexes were allowed to equilibrate for 20 min at
room temperature, after which time they were immediately added to the cell cultures. After 4 h
incubation with antisense ODN complexes, the cells were washed with Hanks' balanced salt
solution (HBSS, Life Technologies), and analyzed directly with a TCS Leica SP2 confocal laser
scanning microscope (Wetzlar, Germany). To quantify cellular binding and uptake of FITC-
ODNs, cells were rinsed with PBS, trypsinized, and resuspended in medium prior to quantifying
fluorescence by FACS. FITC-labeled beads were used as a standard.
Analysis of newly synthesized 5-HT1A after antisense treatment
Complexes (100 nM ODNs), prepared as described above, were incubated with 70–80%
confluent cultures of LLP-5-HT1A cells, or SN48 cells and RN46A cells for 6 h, respectively.
After the cells were washed, methionine-free DMEM containing 5% fetal calf serum was added,
and the incubation was continued for 1 h. Metabolic labeling of newly synthesized protein was
then carried out with 100 Ci/ml [35S]methionine[35S]cysteine (PRO-MIX -[35S] in vitro Cell
Labelling Mix, Amersham Pharmacia Biotech, Buckinghamshire, UK) in methionine-free
medium containing 5% fetal calf serum or differentiation supplement. During metabolic labeling,
RN46A and SN48 cells were incubated under differentiation culture conditions to enhance 5-
HT1A expression. After labeling for 16 h, medium was removed and RIPA buffer (9.1 mM
dibasic sodium phosphate, 1.7 mM monobasic sodium phosphate, 150 mM NaCl, pH 7.4, 1%
Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS. 10 l/ml RIPA) was added at 4°C. After 10
min the cells were disrupted by repeated aspiration through a 25-gauge needle, and the
supernatant was collected by centrifugation at 10,000g for 10 min. 5-HT-1A antibodies (Santa
Cruz) were conjugated to Sepharose (protein A Sepharose CL-4B, Amersham Pharmacia
Biotech, Uppsala, Sweden) for 8 h at 4°C. Equal amounts of the supernatant fraction and
antibody–Sepharose conjugate were then incubated for 16 h at 4°C. Immunoprecipitates were
washed, collected by centrifugation, and analyzed by SDS–PAGE (Bio-Rad, Hercules, CA,
USA). Proteins were autoradiographed for 16 h at -80°C. The relative protein amount was
analyzed by the software Scion Image on the film.
Chapter 4
70
mRNA analysis by Northern blot after antisense treatment
The pharmacological effect of antisense ODNs was evaluated by examining the level of 5-
HT1A mRNA in LLP-5-HT1A cells treated with SAINT-2/DOPE/100 nM antisense ODN
complex. LLP-5-HT1A cells were grown to 70–80% confluency and treated with PS-
ODN/SAINT-2/DOPE complexes prepared as described above. After 16 h, total RNA was
extracted from 6 × 106 cells using the Qiagen RNeasy kit (Hilden, Germany). The quantity and
purity of the RNA preparation were determined by recording the absorbance at 260 and 280 nm.
Equal amounts of RNA were run on a 1.2% agarose gel containing 6.7% formaldehyde, and
transferred to a nylon membrane by vacuum transfer. Total RNAs were fixed onto the membrane
by heating for 1 h at 80°C. The RNA was intact without any degradation, as indicated by 18S
and 28S bands on the membrane, visualized by methanol blue staining. 5-HT1A and β-actin
probes were prepared by amplifying 5-HT1A cDNA by RT-PCR from total RNA in LLP-5-
HT1A cells. The 5-HT1A and β-actin primers were CCA AAG AGC ACC TTC CTC TG/TAC
CAC CAC CAT CAT CAT CA and AAC ACC CCA GCC ATG TAC/ATG TCA CGC ACG
ATT TCC, respectively. PCR was performed using oligo(dT)-primed cDNAs (synthesized with
reverse transcriptase, Boehringer Mannheim, Germany) as a template under the following
conditions: 30 cycles, 94°C 30 s, 55°C 30 s, 72°C 45 s, and a final 72°C extension of 5 min). The
probes from the PCR products were gel purified (QIAEXII Gel Extraction Kit, Germany) and
labeled with [32P] ATP using a random priming kit (Gibco-BRL, Breda, the Netherlands). The
level of 5-HT1A mRNA was probed by the cDNA probe for the 5-HT1A gene, and visualized by
autoradiography. To normalize the level of RNA loading, the 5-HT1A probe was removed by
stripping the membrane in boiling SDS solution (0.5%) for 10 min. The membrane was then
probed with β-actin probe. The relative amount of mRNA was analyzed by the software of Scion
Image on the film.
Removal of nuclear protein, chromatin, and RNA matrix
RN46 cells were grown on coverslips in a 12-well plate and treated with FITC–ODN–SAINT–
2/DOPE complexes for 6 h. Cells were either fixed directly or subjected to RNase digestions.
Nuclear RNA was removed by digesting cells with 1 mg/ml RNase A (Roche, Indianapolis, IN,
USA, boiled for 10 min to destroy residual DNase) and 0.1% Triton X-100 in HBSS for 10 min
at room temperature. After digestion, cells were fixed in 3% PFA or directly examined by
confocal laser microscopy.
Oligos reach mRNA via RNA matrix
71
Nuclear localization of ODNs by electron microscopy
RN46 cells were grown to 70% confluency and treated with 100 nM biotin-labeled
ODN/cationic lipid complexes as described above. After 6 h, the complexes were removed by
washing with PBS. The cells were fixed with 4% paraformaldehyde in PBS for 15 min, and
permeabilized with 0.1% Triton X-100 for 30 min. Subsequently, the cells were incubated with
FluoroNanogold–streptavidin conjugate (1.4-nm gold particles, attached to streptavidin;
Nanoprobes Inc., Stony Brook, NY, USA) for 2 h, washed three times with PBS, and
immediately examined by epifluorescence microscopy. After examination, the cells were washed
with 0.2 M sodium citrate, and the silver enhancement procedure was performed according to the
manufacturer's instructions (Nanoprobes Inc.). The cells were rinsed with PBS and postfixed
with 1% glutardialdehyde for 10 min. Finally, the samples were embedded in Epon, serially
sectioned (60-nm thin sections), and examined at 60 kV with a Philips EM-201 or CM-100
transmission electron microscope (FEI Electron Optics, Eindhoven, The Netherlands).
Binding of antisense to target mRNA
To determine the intracellular binding of antisense ODNs to target mRNA, 1 nmol ODN was
labeled with 10 µl [-32P] ATP (3000 Ci/mmol, 10 µCi/l) with Ready-To-Go T4 polynucleotide
kinase. Nonincorporated nucleotide was removed, using MicroSpin G-25 columns (Amersham
Pharmacia Biotech, Buckinghamshire, UK). LLP-5-HT1A cells were seeded on 75-cm2 flasks,
and after reaching approx 70% confluency, the cells were incubated for 16 h with 5 ml
lipoplexes containing either 100 nM 32P-labeled antisense or mismatched sequences. The cells
were lysed and total RNA was isolated with the Qiagen RNeasy kit (Hilden, Germany). To
exclude that antisense hybrization to target mRNA had occurred during or after RNA extraction,
the same amount of labeled antisense was added to untreated cells during RNA isolation. Then
RNAs were separated on a 1.2% agarose gel containing 6.7% formaldehyde, and transferred to a
nylon membrane by vacuum transfer. A film was exposed to the membrane for 3 days. To
determine the position of the antisense-bound mRNA on the developed film, experiments were
done in parallel with unlabeled ODNs directed against 5-HT1A mRNA. 5-HT1A mRNA was
probed with 32P-labeled 5-HT1A probes by Northern blot, as described above.
Analysis of the total 5-HT1A pool after antisense treatment
Protein levels of 5-HT1A were evaluated by Western immunoblot. An aliquot of 106 SN48
cells were seeded in 10-cm dishes and grown to 50% confluency. Cells were treated with 100
nM antisense ODN complexes for 6 h. After this interval the complexes were removed and fresh
Chapter 4
72
medium was added. Cells (the medium was refreshed daily) were harvested 5 days after the
initial treatment and lysed. Samples (40 µg of protein) were then analyzed by 12.5% SDS–
PAGE (Bio-Rad), blotted on a pure nitrocellulose membrane (Trans-Blot transfer medium, Bio-
Rad), and probed with rabbit anti-rat SR1A antibodies (1:500, Santa Cruz), followed by
horseradish peroxidase conjugate anti-rabbit antibody (1:5000, Nenox, Japan). The blot was
processed with ECL (Amersham Pharmacia Biotech, Buckingham, UK) according to the
manufacturer's protocol. For the second blotting, the blot was washed with PBST (9.1 mM
dibasic sodium phosphate, 1.7 mM monobasic sodium phosphate, and 150 mM NaCl, pH 7.4,
0.3% Tween 20), and then stripped with stripping buffer (Restore Western Blot Stripping Buffer,
Pierce, Rockford, IL, USA). The membrane was probed with mouse anti-β-actin antibody
(1:1000, Sigma), followed by alkaline phosphatase-conjugated sheep anti-mouse antibody
(1:3000, Chemicon, Temecula, CA, USA). The blot was color processed with nitroblue
tetrazolium and 5-bromo-4-chloro-3-indolyl. The relative amounts of protein were analyzed by
Scion Image software.
Functional assay for 5-HT1A receptor activity in SN48 cells
The function of 5-HT1A in SN48 cells was validated using a cytosensor, which measures the
metabolic change of the cell as reflected by pH changes on a challenge with an agonist. The cells
were plated on cytosensor membranes and grown for 1 day before the measurements. Cells were
either challenged with the agonist flesinoxan for 2 min at indicated concentrations from 0.003 to
0.1 mM or blocked with the antagonist WAY100635 at 0.1 mM for 10 min, before challenge
with flesinoxan (0.1 mM for 2 min). To determine the effect of antisense treatment on 5-HT1A
receptor activity, SN48 cells were treated with antisense ODNs as described above, after which
they were plated on cytosensor membranes, and further grown for 16 h. To normalize the amount
of cells on each membrane, the cells were first challenged with acetylcholine, which was
followed by a challenge with flesinoxan for 2 min at the concentrations indicated. The cells were
brought back to base level for 5 min between challenges.
Results
Parameters affecting uptake of antisense ODNs
When phophorothioate ODNs were incubated with LLP-5-HT1A, RN46A, or SN48 cells at a
relatively high concentration of 2 µM, they were endocytosed and observed as punctuate
structures within the cells. However, in either cells type, accumulation of ODNs in the nucleus
Oligos reach mRNA via RNA matrix
73
was not apparent (Figs. 1A–C). By contrast, when complexed with cationic liposomes, not only
was uptake of ODNs by either cell type significantly enhanced, but also their translocation across
the endosomal compartments was evident, as reflected by their nuclear accumulation ( Figs. 1D–
F, G–I). Since the eventual effect of the antisense ODN is likely dose-dependent, and
conceivably depends on nuclear uptake, we compared two types of cationic liposome
formulations as delivery vehicles to accomplish the most optimal delivery efficiency. As shown
in Fig. 1 (G–I vs D–F), in all three cell lines, the formulation based on SAINT-2/DOPE was
substantially more potent in facilitating the uptake and translocation of ODNs than
Lipofectamine 2000. Indeed, FACS analysis revealed that net uptake of the Lipofectamine 2000
lipoplexes was two- to sixfold lower than that observed for the SAINT lipoplexes. In fact, cell
type-dependent differences in uptake efficiency were also seen when employing Lipofectamine
2000, whereas in the case of SAINT-2-containing complexes, the delivery was approximately
equally effective in either cell type ( Fig. 1, numbers in upper left corner in G–I). Elsewhere, we
[5 and 6] and others [7] have reported that the antisense efficiency correlates well with the
cellular uptake of ODNs. Indeed, in line with the distinctions in uptake, SAINT-2/DOPE is
approximately two to three times more potent than Lipofectamine 2000 in eliciting an antisense
effect and in transfecting cells, when monitored by a GFP reporter plamid (transfection
efficiency 25–30% (LF2000) vs 50–70% (SAINT-2/DOPE); data not shown). Accordingly, in
the following experiments, SAINT-2/DOPE was applied as ODN vector. The antisense uptake
was not dependent on the sequence, since all sequences used in the present work, including
fluorescent and biotinylated ODNs, showed essentially the same amount of uptake (not shown).
In addition, none of the sequences applied caused significant cytotoxicity, as verified by
morphological criteria and the MTT assay (not shown). We next examined whether the
internalized antisense ODN in either cell type effectively interfered with expression of the target
5-HT1A receptor.
Downregulation of newly synthesized 5-HT1A receptor
To reveal an antisense effect, we monitored the effect on the pool of newly synthesized
proteins, since the existing pool is not affected other than by metabolic turnover. Two antisense
sequences (AS1 and AS2, see Materials and Methods), selected by computer-facilitated analysis
of 5-HT1A mRNA, were examined. Throughout this work, the cells were treated with 100 nM
antisense or mismatched ODN complexed with SAINT-2/DOPE (1:1). In preliminary,
concentration-dependent experiments, carried out as described in detail previously [5], it was
Chapter 4
74
established that at this concentration the antisense effect is optimal, without significant cell
cytotoxic effects. After antisense treatment for 6 h, metabolic labeling was carried out for 18 h in
Fig. 1. Intracellular uptake of FITC-labeled PS-ODNs in free form and whencomplexed with cationic lipid complexes. PS-ODNs alone, PS-ODN/Lipofectamine 2000 complexes, and PS-ODN/SAINT-2/DOPE complexeswere incubated with LLP-5-HT1A (A, D, G), RN46A (B, E, H), and SN48 cells (C,F, I). The concentration of free ODNs was 2 µM, and the incubation was carriedout for 24 h. Note the punctate appearance in intracellular compartments inthe cytosol (A–C). 100 nM PS-ODN complexed with Lipofectamine 2000 (D–F)and SAINT-2/DOPE (G–I) was incubated with the cells for 4 h. Note thedifferences in efficiency of uptake and the nuclear accumulation of PS-ODNs inthis case. The efficiency of ODN uptake in each case is indicated by the relativecell-associated fluorescence, indicated in the left upper corner.
LLP-5-HT1A RN46 SN48
20 6 12
73 42 25
148 146 142
a b c
d e f
g h i
Oligos reach mRNA via RNA matrix
75
differentiation medium to enhance 5-HT1A expression. Subsequently, the cells were lysed, and
5-HT1A was immunoprecipitated and analyzed by PAGE, as described under Materials and
Methods. The data, shown in Fig. 2, reveal that the mismatched sequence had no significant
effect on the target (lane 3 vs lane 4, 99, 113, and 110% relative to untreated cells in LLP-5-
HT1A, SN48, and RN46A, respectively). However, the two antisense sequences display
differences in potency, the AS1 sequence being the most effective one (lanes 1 and 2 vs lane 4).
Thus, following AS1 antisense treatment the expression of newly synthesized receptor in LLP-5-
HT1A, SN48, and RN46A was inhibited by 64, 34, and 32%, respectively. In the 5-HT1A
overexpressed LLP-K1 cells, AS2 inhibited receptor expression by approx 45%. It is of interest
to note that AS1 was directed to a region in the mRNA at the beginning of the translation site (bp
115–128), whereas AS2 was directed to a region localized in the middle of the translation
sequence (bp 885–902). Accordingly, in the latter case, truncated protein fragments of the
receptor should have been produced if the antisense effect relies on steric interference with
mRNA translation. However, such truncated protein fragments were never detected in the gel.
Remarkably, in contrast to AS1, AS2 was essentially without effect on the newly synthesized
protein fraction in neuronal SN48 and RN46A cells. However, in line with the strongly
diminished activity of AS1 in LLP-5-HT1A cells, these observations might suggest a cell- type-
dependent effect of antisense efficiency, discriminating a natural target (as in SN48 and RN46A)
from one that is overexpressed (LLP-5HT1A).
Antisense ODN treatment reduces the expression of target mRNA
As noted above, steric interference of the antisense ODN with translation is less likely the
cause of the decrease in protein expression. Rather, the data may imply mRNA degradation. At
steady state the mRNA levels of 5-HT1A differ markedly in a cell-dependent manner, being
most pronounced in the overexpressing cells (data not shown). For quantitative reasons, we
therefore employed LLP-5HT1A cells. In line with the diminished expression at the protein level
of the receptor in LLP-5-HT1A cells (Fig. 2, LLP-5-HT1A), a similar decrease in mRNA levels
was observed following treatment with AS1 (Fig. 3A, 75% inhibition; lane 1). Evidently, AS2
was less effective in decreasing the protein level (Fig. 2, lane 2), and consistently, the decrease in
mRNA expression was less pronounced (approx 60% inhibition). Note that the effect of the
mismatched sequences (MAS1 and MAS2) was essentially negligible (lanes 3 and 4 vs lane 5,
98 and 106% compared with untreated cells). Given the close correlation between the inhibition
of protein expression and the reduction of mRNA levels, and the absence of detectable truncated
Chapter 4
76
protein fragments, the data suggest that the consequences of antisense treatment might be due to
the rapid and efficient degradation of mRNA.
Fig. 2. Antisense treatment efficientlydownregulates newly synthesized 5-HT1Atarget receptors. LLP-5-HT1A cells, SN48cells, and RN46A cells were treated with 100nM antisense ODN sequence 1 (AS1),sequence 2 (AS2), and a mismatchedsequence (MAS), complexed with SAINT-2/DOPE. The newly synthesized 5-HT1A wasmonitored by metabolic labeling,immunoprecipitated and separated by PAGE,as described. Numbers under the bandsindicate the amount of protein (%) relative tothe amount in untreated cells (UN), asobtained in two independent experiments.
Fig. 3. Downregulation of antisense ODN to 5-HT1A mRNA. (A) LLP-5-HT1A cells weretreated with 100 nM antisense ODN sequence1 (AS1), sequence 2 (AS2), and mismatchedsequences (MAS1 and MAS2), complexed withSAINT-2/DOPE. Total RNA was isolated and5-HT1A was probed with 32P-radiolabeled 5-HT1A cDNA. (B). The same blot was probedwith 32P-radiolabeled β-actin cDNA probe.Numbers under the bands indicate theamount of mRNA (%) relative to the amount inuntreated cells (UN), as obtained in twoindependent experiments.
Fig. 4. Accumulation of PS-ODNs in the nuclei. When FITC labeled PS-ODNs (100 nM)were incubated with isolated nuclei (30 min), they readily accumulated in the nuclei(A). Note that for intranuclear accumulation neither cytosolic proteins nor cationiclipid complexes are needed. When cationic lipids/PS-ODNs were incubated with intactcells. PS-ODNs (FITC-labeled, green) also accumulated in cell nuclei, while the cationiclipids (Rh-label, red) remained in the endosomal compartments (B). When cationiclipids/PS-ODNs were incubated with isolated nuclei. PS-ODNs (FITC-label, green) alsoaccumulated in cell nuclei, and cationic lipids (with Rh-label, red) partly fused withnuclear membranes The intense red staining represents clustered lipoplexes (C).
SN48
RN46A
LLP-5-HT1A
AS1 AS2 MAS UN
36±5 55 ±10 99 ±5 100 ±7
66 ±7 121 ±11 113 ±9 100 ±17
68 ±11 125 ±20 110 ±5 100 ±2
AS1 AS2 MAS1 MAS2 UN
25�3 40�6 98�9 106�12 100�6
A5-HT1A
Bß-actin
110�15 75�20 80�9 110�2 100�10
A B C
Oligos reach mRNA via RNA matrix
77
To obtain further insight into the pathway by which the ODNs gain nuclear access and the
potential correlation of this localization with the eventual cytosolic degradation of mRNA, the
next experiments were carried out.
Nuclear delivery and interaction of antisense ODNs with the RNA nuclear matrix
It is generally assumed that once in the cytosol, ODNs may gain rapid access to the nucleus
via diffusion through the nuclear pores. Some studies indicated that antisense ODN may bind to
intracellular proteins [17], but whether such proteins are instrumental in nuclear homing remains
to be determined. It is also unclear whether cationic lipids are needed or may facilitate this event,
other than their functioning in effectively translocating ODNs across endosomal membranes
[23]. To investigate these issues, we isolated nuclei, which were then incubated with free and
complexed FITC-labeled ODNs. As shown in Fig. 4, free ODNs rapidly and efficiently acquired
nuclear access (Fig. 4A), irrespective of the presence of cytosol, implying that proteins were not
needed to accomplish efficient nuclear accumulation of the ODNs. Similarly, when whole cells
or nuclei were incubated with ODN complexes, which also contained the fluorescent lipid
analogue N-Rh-PE as a marker of the cationic lipid phase, the net extent of nuclear accumulation
of the ODNs in either case was virtually indistinguishable. Note that when added to whole cells
(Fig. 4B), lipid-derived fluorescence (red) remains localized in the endosomal/lysosomal
compartments [5], without any (detectable) appearance at or in the nucleus. The ODNs show a
largely random and diffuse distribution throughout the nucleus (green fluorescence) except for
the nucleolus (dark round spheres), and occasionally form concentrated ODN bodies (bright
green dots) ( Fig. 4B). When incubating ODN/lipid complexes with isolated nuclei, delivery of
ODNs within the nucleus occurs (green fluorescence). Concomitant entry of the lipids is not
seen, as they associate largely with the nuclear membrane as attached clustered complexes
(intense red dots, Fig. 4C) or become laterally diffused within the plane of the membrane (red
ring, Fig. 4C). Hence, these data indicate that neither (cytosolic) proteins nor entry of the
cationic lipid (complex) into the cytosol is required for effective nuclear delivery of ODNs.
To obtain further insight into a potential correlation between intranuclear access and antisense
mechanism, we subsequently determined the intranuclear fate of the ODNs, following the
treatment of intact cells with FITC-labeled ODN complexes. First, nuclear proteins were
extracted and nuclear chromatins were digested with RNase-free DNase. Following this
treatment, no significant alteration in intranuclear ODN distribution was apparent, when
compared with the distribution in control cells (Fig. 5A). Interestingly, when nuclear RNA was
digested with RNase A to remove the RNA matrix in the nucleus, most of the ODNs disappeared
and only ODNs assembled into so-called phosphorothioate ODN (PS [7]) bodies remained
Chapter 4
78
(Fig.5B). As a control, when the cells were treated with the same buffer without RNase, no effect
on ODN distribution was seen (data not shown). Furthermore, ODNs are not seen in the
nucleolus, where rRNAs are synthesized and processed and ribosome subunits are assembled.
Identical results were seen for LLP-5HT1A and SN48 cells (data not shown). From these
experiments, we conclude that following nuclear entry, ODNs associate largely with the nuclear
RNA matrix. To obtain further support, the intranuclear localization of ODNs was further
examined by EM. RN46 cells were incubated with biotin-labeled ODN cationic lipid complexes,
and then the ODNs were visualized by subsequent labeling with a streptavidin–gold conjugate,
as described under Materials and Methods. Consistent with the fluorescence distribution, the
gold labeling was seen throughout nucleoplasm, with little if any labeling in the nucleolus ( Fig.
6A, Nu.). Dense gold labeling was seen on the ODN bodies, identified as bright dots by
fluorescence microscopy, especially at the periphery (as indicated by arrows in Figs. 6A and B).
At higher magnification, gold labeling of ODNs in the nucleoplasm is apparent along fibers of
the internal nuclear matrix (arrowheads in Fig. 6B [24]), visible as a diffuse distribution by
fluorescence microscopy. As a control, labeling was also carried out with unlabeled ODNs, and
was followed by the addition of streptavidin–gold particles. Labeling was not detectable under
those conditions (data not shown).
In conjunction with the observed mRNA degradation as shown in Fig. 3B, these data would be
consistent with the notion that the antisense ODNs might act on mRNA or hnRNA in the
nucleus. The next experiments were aimed at elucidating how sequence selectivity is expressed
in terms of antisense activity following its binding to the RNA matrix.
Fig. 5. Intranuclear PS-ODNs associate with the nuclear matrix. When cationiclipids/PS-ODNs were incubated with cells, PS-ODNs (FITC labeled) ccumulated in thecell nuclei (A). Following digestion with RNase A, the PS-ODNs were largely removedfrom the nuclei, and only PS-ODN nuclear bodies remained in the nuclei. Also, somediffused fluorescence was now apparent in the cytoplasm (B).
A B
Oligos reach mRNA via RNA matrix
79
Fig. 6. EM visualization of ODNs, localized in the nucleus. One hundrednanomolar biotin-labeled ODNs, complexed with SAINT-2/DOPE, wereincubated with RN46 cells. The ODNs were visualized with FluoroNanogold–streptavidin conjugate. Note that gold labeling is seen throughout thenucleoplasm (black dots), with little if any labeling in the nucleolus (A, Nu).Dense gold labeling is apparent on the ODN bodies, especially at the peripheryof the bodies (A and B, arrows). At higher magnification (B), gold labeling ofODNs in the nucleoplasm can be seen associated along the fibers of the nuclearmatrix (arrowheads) Bars = 344 nm and (A), 100 nm (B).
Intracellular affinity of antisense ODNs for target mRNA
Although antisense has been shown to bind to isolated target mRNA or mRNA fragments in a
test tube, direct evidence of the association of antisense with mRNA within cells has not been
provided thus far. The very small amounts of target mRNA, as observed here in the RN46 and
SN48 cells, and the potential rapid degradation of mRNA in the cells may constitute a limitation
in resolution in that regard. However, the abundance of 5-HT1A mRNA in LLP-5-HT1A cells
evidently provided an opportunity for further investigation of this issue. Lipoplexes containing
100 nM 32P-labeled antisense (Fig. 7A. AS) or mismatched (Fig. 7A, MAS) sequences were
incubated with LLP-5-HT1A cells for 16 h. To exclude that antisense hybrization to target
mRNA might have occurred during or after extraction, the same amount of 32P-labeled antisense
was added to untreated cells during RNA isolation (Fig. 7, UN/AS+). Following total RNA
extraction and its concentration by ethanol precipitation, RNA was separated on a Northern gel
and blotted on a nylon membrane. The potential binding of antisense to mRNA was then
visualized by autoradiography. As shown in Fig. 7, antisense binding to mRNA in LLP-5-HT1A
cells could be readily revealed even though the signal was rather low (Fig. 7A, AS). The
relatively low signal may well be related to mRNA degradation, triggered by antisense binding,
B
Nu.
A
Chapter 4
80
as seen in Fig. 7B (AS vs MAS and UN/AS+). The mismatched (Fig. 7A, MAS) or antisense
(Fig. 7, US/AS+) sequences added during RNA isolation failed to bind to the mRNA. These data
are consistent with the notion that the antisense effect is elicited via binding of ODNs to the
nuclear matrix in the nucleus, the antisense specificity being conveyed through high binding
affinity to target mRNA. Having established the efficiency of the antisense ODNs in the
reduction of newly synthesized 5-HT1A and target mRNA, we finally examined the ultimate
goal of antisense treatment, i.e., whether and under what conditions the total pool of the receptor
could be modulated to accomplish a biological effect.
Fig. 7. Antisense ODNs bind to intracellular mRNA targets. One hundred nanomolar32P-labeled antisense ODNs (AS) or mismatched sequences (MAS), complexed withcationic lipid vectors, were incubated with LLP-5-HT1A cells, and total RNA wasisolated. Alternatively, 100 nM 32P-labeled antisense ODNs were added to untreatedcells during the RNA isolation (UN/AS+). The RNAs were separated on agarose gel andODN–mRNA binding was determined by autoradiography (A). The location of 5-HT1AmRNA on the gel in (A) was determined from that in (B). LLP-5-HT1A cells were treatedas in (A), except that the ODNs were not labeled. RNA was isolated and separated onthe gel. The 5-HT1A mRNA was then probed with 32P-labeled 5-HT1A cDNA probe (B).The specific antisense binding to 5-HT1A mRNA could be identified (A, AS), as itlocates at the same position as the 5-HT1A mRNA (B, AS). The binding to 5-HT1AmRNA was not seen with mismatched sequences (A, MAS) or on addition of antisenseODNs during RNA isolation (A, UN/AS+).
Functional downregulation of 5-HT1A receptor activity following antisense treatment
The three cell types were treated over a period of 5 days with antisense complexes or control,
mismatched complexes. Each treatment was carried out by incubating the ODN complexes with
the cells for 6 h, after which they were washed and grown in medium for another 48 h. This
protocol was repeated twice and on the sixth day the cells were harvested and analyzed by
A. 32P-ODNs B. 32P-5-HT1A
6,5834,981
1,908
1,383
955
623
281
AS MAS UN/MAS AS MAS UN/MAS
Oligos reach mRNA via RNA matrix
81
A. B.
Fig. 8. (A). Downregulation of the total 5-HT1A receptor pool in SN48 cells. The cellswere treated according to a protocol as described for 5 days with 100 nM antisenseODN sequence 1 (AS1), sequence 2 (AS2), and mismatched sequences (MAS1 andMAS2), complexed with SAINT-2/DOPE. The presence of 5-HT1A was analyzed byWestern immunoblot. β-Actin was used as an internal control. The numbers under thebands indicate the amount of 5-HT1A (%) relative to untreated cells (UN) from twoexperiments. (B) Functional downregulation of 5-HT1A receptor activity after antisensetreatment. Control activity and receptor activity in 100 nM antisense (AS1), andmismatch ODN complex-treated SN48 was determined using a cytosensor approach,as described under Materials and methods. The cells were challenged withacetylcholine and the 5-HT1A agonist flesinoxan. Flesinoxan activation was plotted asa percentage relative to acetylcholine activation. Flesinoxan activated 5-HT1A controlcells in a concentration-dependent manner (hatched bars). Mismatched sequence didnot show a significant change (open bars) relative to control. Note that flesinoxanactivation of the 5-HT1A receptor was abolished in cells treated with antisense AS1ODN under the same conditions (black bars). Also, no change was observed, relative tocontrol, when cells were treated with AS 2 ODNs (data not shown). The data areexpressed relative to the acetylcholine effect, which was taken for calibration (100%),and flesinoxan activation was plotted relative to acetylcholine activation.
Western immunoblot. As shown in Fig. 8A for SN48 cells, over the period of treatment, the total
pool of 5-HT1A can be effectively downregulated by approx 67%, following treatment with
AS1. Similar results were obtained for the treatment of RN46A cells, resulting in a
downregulation of 64% with AS1. In both cell lines, AS2 was less effective. In LLP-5-HT1A
cells, AS1 and AS2 showed 68 and 51% downregulation of 5-HT1A. Mismatched sequences
(MAS1 and MAS2) did not alter 5-HT1A expression, and expression of the internal control, β-
actin, was not affected by any treatment (Fig. 8A). In addition, no alterations were seen in total
protein expression when analyzed by PAGE for control and ODN-treated cells, further
emphasizing the specificity on the 5-HT1A receptor. The functional consequence of the
downregulation of the receptor was examined with a cytosensor, which measures the metabolic
change of the cell on challenge with an agonist. The procedure was validated by challenging
AS1 AS2 MAS1 MAS2 UN UN5-HT1A
b-actin 33±1 105 ±5 95 ±10 98 ±12 103 ±15 100 ±5
SR1a functional assay
-50
50
150
250
3E-06 1E-05 3E-05 1E-04 3E-04flexinoxan(M)
Effe
ct(%
vs
acet
ylch
olin
e)
controlmismatchantisense
Chapter 4
82
SN48 cells with flesinoxan, an agonist to 5-HT1A. The cells responded to this treatment in a
concentration-dependent manner, and the activation could be blocked by preincubation with
WAY 100635, an antagonist to 5-HT1A (data not shown). Next, the cells were treated for 5 days
with 100 nM antisense AS1 complexes, as described above. The treated cells were plated onto
the cytosensor membrane 16 h before the measurements. To normalize the activity for the
number of cells on each membrane, the cells were first challenged with acetylcholine, which is a
ligand for choline receptors expressed on SN48 cells. Then, the cells were challenged with
increasing concentrations of flesinoxan. To further support specificity, the 5-HT1A was
subsequently blocked with WAY 1000635, and the cells were challenged once more with
flesinoxan. As shown in Fig. 8B, flesinoxan activated control cells and cell treated with
mismatched ODNs in a concentration-dependent manner, which could be blocked by WAY
100635. In contrast, activation of the antisense-treated (AS1) cells was completely abolished.
Note that the negative value is due to acidification on addition of flesinoxan, an effect that is
normally overcome on activation of the 5-HT1A receptor. Antisense sequence 2 (AS2) did not
alter the function of 5-HT1A receptor (data not shown). This is consistent with the observation
that a change was not seen in protein level following AS 2 treatment ( Fig. 8A). Hence the data
indicate that following antisense treatment, the function of the 5-HT1A receptors can be
inhibited in a highly efficient manner.
Discussion
For an antisense treatment to be effective, it is essential to design a sequence that effectively
matches the secondary and/or tertiary structure of target mRNA. Since studies on mRNA
secondary and tertiary structure are still scarce, these structures are usually predicted by
computer-facilitated analysis. Thus, antisense sequences can be selected by prediction of the
mRNA structure [25], although empirical design and experimental mRNA mapping are also used
[26, 27 and 28]. Here, sequences were applied that relied on computer-aided antisense sequence
design and selection technology, called RADAR (Rational Algorithmic Design of ANTISENSE
Reagents), which avoids costly, time-consuming, and laborious screening for appropriate
antisense sequences [29, 30, 31, 32, 33 and 34].
The two antisense sequences produced in this manner and applied in the present work showed
different potencies but, interestingly, did show a cell type-dependent effect when comparing
naturally expressed and overexpressed receptors. Whether this distinction is related to structural
differences of the overexpressed mRNA in LLP-5HT1A cells remains to be determined. Yet, one
Oligos reach mRNA via RNA matrix
83
of the sequences (AS1) effectively downregulated the target expression in all three cell types
studied, suggesting an antisense sequence-dependent effect as well.
To successfully apply antisense ODNs, insight into their optimal design, mechanism of
internalization, and mechanism of action is crucial. As shown here, for effective nuclear
delivery, a vector is needed to facilitate effective translocation of the ODNs from the endosome
into the cytosol [5], rather than a delivery vehicle required for entry into the nucleus per se (Fig.
4). Indeed, elsewhere we have provided evidence that SAINT-2/DOPE-mediated delivery of
plasmids relies on entry of these complexes via clathrin-mediated endocytosis [3]. Similarly,
ODN internalization via the same vector localizes to endosomal/lysosomal compartments [5 and
6] and its uptake is inhibited on energy or potassium depletion, which inhibits clathrin-mediated
endocytosis (data not shown).
In permeabilized cells, it has been reported that ODNs bind extensively to intermediate
filaments in the cytosol, with a preference for cytokeratin intermediate filaments [35]. In the
same study, ODNs localized in the nucleus interacted with nuclear lamina and caused the
decompaction of chromatin [35]. Such phenomena were not seen on microinjection of ODNs
into the cytosol [7, 36 and 37] or following cationic lipid-mediated ODN delivery [5, 7 and 38].
Possibly, in permeabilized cells, a rapid and high degree of saturation of potential intracellular
binding sites might be accomplished on exogenous addition of ODNs. Obviously, such
conditions are not readily achieved on vector-mediated delivery, either in vitro or in vivo.
Furthermore, actin filaments are less likely encountered when ODNs are released from the
endosomal compartment, and the data as shown here ( Fig. 4) and elsewhere [5, 7 and 38]
indicate a rapid transfer of ODNs into the nucleus under such conditions. Nuclear access requires
neither cationic lipids nor cytosolic proteins ( Fig. 4).
The exact subnuclear localization of ODNs is still not well defined. The eukaryotic nucleus
consists of chromatin, RNA, and proteins. Interestingly, we observed that the ODNs almost
exclusively bind to the nuclear matrix, which consists largely of hnRNA and a large family of
proteins. This is consistent with previous findings reported in an elegant study by Lorenz et al.
[7]. Others have found a high degree of complex formation between oligonucleotides and
nuclear proteins, as revealed by a gel shift assay [39]. However, no protein has been identified
thus far.
Interference with a biological function is evidently the final goal of antisense technology, a
misleading factor for proper interpretation often being the specificity of the antisense effect.
Clearly, in all three cell types we observed a substantial decrease in newly synthesized protein,
the first effect to become apparent, following ODN treatment. Also, a specific diminishment in
Chapter 4
84
mRNA level of the target protein was observed, supporting the specificity of the observed
antisense effect in the present work. Since truncated protein fragments were not observed, our
data support an efficient degradation of target mRNA by RNase H as a leading mechanism in
directing ODN antisense efficiency, and in the present work we provided direct evidence for the
intracellular association of antisense ODN and target mRNA. Given the intranuclear localization
of RNase H [40 and 41], the need for a nuclear localization of antisense to accomplish its effect
can thus be readily rationalized.
Taken together, here we have shown that with appropriate sequences, and in a cell type-
dependent manner a functional antisense effect can be accomplished, which originated from a
nuclear association of the ODN with target mRNA via binding to the nuclear matrix. Our data
indicate that such an interaction causes degradation of the target RNA and, consequently,
downregulation of the newly synthesized serotonin receptor 5-HT1A. In conjunction with
metabolic turnover, the pool of the receptor can then be specifically diminished to a level that
results in complete abolition of its biological activity as well (Fig. 8B). Both cell biological
studies and therapeutic ex vivo treatments may benefit from such an approach.
Oligos reach mRNA via RNA matrix
85
AcknowledgementsThis work was supported by a grant from The Netherlands Organization for ScientificResearch (NWO)/NDRF Innovative Drug Research (940-70-001). The technicalassistance of Anita Nomden and Greet Kas in some of the experiments is gratefullyacknowledged. Anno Wagenaar and Professor Jan Engberts are thanked for helpfuldiscussions and for providing us with SAINT-2.
References:1. C. Vanhee-Brossollet and C. Vaquero, Do natural antisense transcripts make sense
in eukaryotes?. Gene 211 (1998), pp. 1–9.2. B. Lehner, G. Williams, R.D. Campbell and C.M. Sanderson, Antisense transcripts
in the human genome. Trends Genet. 18 (2002), pp. 63–65.3. I.S. Zuhorn, R. Kalicharan and D. Hoekstra, Lipoplex-mediated transfection of
mammalian cells occurs through the cholesterol-dependent clathrin-mediatedpathway of endocytosis. J. Biol. Chem. 277 (2002), pp. 18021–18028.
4. J. Smisterova, A. Wagenaar, M.C. Stuart, E. Polushkin, G.T. Brinke, R. Hulst, J.B.Engberts and D. Hoekstra, Molecular shape of the cationic lipid controls thestructure of cationic lipid/dioleylphosphatidylethanolamine–DNA complexes andthe efficiency of gene delivery. J. Biol. Chem. 276 (2001), pp. 47615–47622.
5. F. Shi, A. Nomden, V. Oberle, J.B. Engberts and D. Hoekstra, Efficient cationiclipid-mediated delivery of antisense oligonucleotides into eukaryotic cells: down-regulation of the corticotropin-releasing factor receptor. Nucleic Acids Res. 29(2001), pp. 2079–2087.
6. F. Shi, L. Wasungu, A. Nomden, M.C.A. Stuart, E. Polushkin, J.B. Engberts, F.N.D.Hoekstra et al., Interference of poly(ethylene glycol)–lipid analogues with cationic-lipid-mediated delivery of oligonucleotides: role of lipid exchangeability and non-lamellar transitions. Biochem. J. 366 (2002), pp. 333–341.
7. P. Lorenz, B.F. Baker, C.F. Bennett and D.L. Spector, Phosphorothioate antisenseoligonucleotides induce the formation of nuclear bodies. Mol. Biol. Cell 9 (1998), pp.1007–1023.
8. S.T. Crooke, Molecular mechanisms of action of antisense drugs. Biochim. Biophys.Acta 1489 (1999), pp. 31–44.
Chapter 4
86
9. N. Dias, S. Dheur, P.E. Nielsen, S. Gryaznov, A.V. Aerschot, P. Herdewijn, C.Helene and S.T.E. Behmoaras, Antisense PNA tridecamers targeted to the codingregion of Ha-ras mRNA arrest polypeptide chain elongation. J. Mol. Biol. 294(1999), pp. 403–416.
10. I. Robbins and B. Lebleu, Vesicular stomatitis virus as model system for studies ofantisense oligonucleotide translation arrest. Methods Enzymol. 313 (2000), pp.189–203.
11. G.J. Veal, S. Agrawal and R.A. Byrn, Sequence-specific RNase H cleavage of gagmRNA from HIV-1 infected cells by an antisense oligonucleotide in vitro. Nucleic.Acids Res. 26 (1998), pp. 5670–5675.
12. C. Boiziau, N.T. Thuong and J.J. Toulme, Mechanisms of the inhibition of reversetranscription by antisense oligonucleotides. Proc. Natl. Acad. Sci. USA 89 (1992),pp. 768–772.
13. I. Lebedeva, L. Benimetskaya, C.A. Stein and M. Vilenchik, Cellular delivery ofantisense oligonucleotides. Eur. J. Pharm. Biopharm. 50 (2000), pp. 101–119.
14. P. Rockwell, W.J. O'Connor, K. King, N.I. Goldstein, L.M. Zhang, C.A. Stein, W.J.O'Connor, K. King, N.I. Goldstein, L.M. Zhang and C.A. Stein, Cell-surfaceperturbations of the epidermal growth factor and vascular endothelial growthfactor receptors by phosphorothioate oligodeoxynucleotides. Proc. Natl. Acad. Sci.USA 94 (1997), pp. 6523–6528.
15. W. Wang, H.J. Chen, A. Schwartz, P.J. Cannon, C.A. Stein and L.E. Rabbani,Sequence independent inhibition of in vitro vascular smooth muscle cellproliferation, migration, and in vivo neointimal formation by phosphorothioateoligodeoxynucleotides. J. Clin. Invest. 98 (1996), pp. 443–450.
16. A. Anselmet, E. Mayat, S. Wietek, P.G. Layer, B. Payrastre and J. Massoulie, Non-antisense cellular responses to oligonucleotides. FEBS Lett. 510 (2002), pp. 175–180.
17. I. Brukner and G.A. Tremblay, Cellular proteins prevent antisensephosphorothioate oligonucleotide (SdT18) to target sense RNA (rA18): developmentof a new in vitro assay. Biochemistry 39 (2000), pp. 11463–11466.
18. A.A.P. Meekel, A. Wagenaar, J. Smisterova, J. Kroeze, P. Haadsma, B. Bosgraaf,M.C.A. Stuart, A. Brisson, M.H.J. Ruiters and D. Hoekstra, Synthesis ofpyrimidinium amphiphiles used for transfection and some characteristics ofamphiphile/DNA complex formation. Eur. J. Org. Chem. 2000 (2000), pp. 665–673.
19. X.M. Ou, H.J. Nejad, J.M. Storring, J.H. Meng, S. Lemonde and P.R. Albert, Noveldual repressor elements for neuronal cell-specific transcription of the rat 5-HT1Areceptor gene. J. Biol. Chem. 275 (2000), pp. 8161–8168.
20. X.M. Ou, J.M. Storring, N. Kushwaha and P.R. Albert, Heterodimerization ofmineralocorticoid and glucocorticoid receptors at a novel negative response elementof the 5-HT1A receptor gene. J. Biol. Chem. 276 (2001), pp. 14299–14307.
21. P.L. Andersen, C.A. Webber, S.R. Whittemore and D.J. Schreyer, Divergentregulation of GAP-43 expression and CNS neurite outgrowth by cyclic AMP. J.Neurosci. Res. 61 (2000), pp. 626–635.
22. X. Langlois, S.E. Mestikawy, M. Arpin, A. Triller, M. Hamon and M. Darmon,Differential addressing of 5-HT1A and 5-HT1B receptors in transfected LLC-PK1epithelial cells: a model of receptor targeting in neurons. Neuroscience 74 (1996),pp. 297–302.
23. I.S. Zuhorn, V. Oberle, W.H. Visser, J.B. Engberts, U. Bakowsky, E. Polushkin andD. Hoekstra, Phase behavior of cationic amphiphiles and their mixtures with helperlipid influences lipoplex shape, DNA translocation, and transfection efficiency.Biophys. J. 83 (2002), pp. 2096–2108.
24. J.A. Nickerson, G. Krockmalnic, K.M. Wan and S. Penman, The nuclear matrixrevealed by eluting chromatin from a cross-linked nucleus. Proc. Natl. Acad. Sci.USA 94 (1997), pp. 4446–4450.
Oligos reach mRNA via RNA matrix
87
25. V. Patzel and G. Sczakiel, Theoretical design of antisense RNA structuressubstantially improves annealing kinetics and efficacy in human cells. Nat.Biotechnol. 16 (1998), pp. 64–68.
26. S.P. Ho, D.H. Britton, B.A. Stone, D.L. Behrens, L.M. Leffet, F.W. Hobbs, J.A. Millerand G.L. Trainor, Potent antisense oligonucleotides to the human multidrugresistance-1 mRNA are rationally selected by mapping RNA-accessible sites witholigonucleotide libraries. Nucleic Acids Res. 24 (1996), pp. 1901–1907.
27. M. Sohail, H. Hochegger, A. Klotzbucher, R.L. Guellec, T. Hunt and E.M. Southern,Antisense oligonucleotides selected by hybridisation to scanning arrays areeffective reagents in vivo. Nucleic Acids Res. 29 (2001), pp. 2041–2051.
28. N. Milner, K.U. Mir and E.M. Southern, Selecting effective antisense reagents oncombinatorial oligonucleotide arrays. Nat. Biotechnol. 15 (1997), pp. 537–541.
29. H.C. Welch, W.J. Coadwell, C.D. Ellson, G.J. Ferguson, S.R. Andrews, H.Erdjument-Bromage, P. Tempst, P.T. Hawkins and L.R. Stephens, P-Rex1, a PtdIns(3,4,5) P3- and G-beta-gamma-regulated guanine-nucleotide exchange factor forRac. Cell 108 (2002), pp. 809–821
30. V.L. Raghavendra Rao, K.K. Bowen, V.K. Dhodda, G. Song, J.L. Franklin, N.R.Gavva and R.J. Dempsey, Gene expression analysis of spontaneously hypertensiverat cerebral cortex following transient focal cerebral ischemia. J. Neurochem. 83(2002), pp. 1072–1086.
31. Y. Kubota, T. Tanaka, A. Kitanaka, H. Ohnishi, Y. Okutani, M. Waki, T. Ishida andH. Kamano, Src transduces erythropoietin-induced differentiation signals throughphosphatidylinositol 3-kinase. EMBO J. 20 (2001), pp. 5666–5677.
32. Y. Kashima, T. Miki, T. Shibasaki, N. Ozaki, M. Miyazaki, H. Yano and S. Seino,Critical role of cAMP-GEF11–Rim2 complex in incretin-potentiated insulinsecretion. J. Biol. Chem. 276 (2001), pp. 46046–46053.
33. A. Von Eckardstein, C. Langer, T. Engel, I. Schaukal, A. Cignarella, J. Reinhardt,S. Lorkowski, Z. Li, X. Zhou, P. Cullen and G. Assmann, ATP binding cassettetransporter ABCA1 modulates the secretion of apolipoprotein E from humanmonocyte-derived macrophages. FASEB J. 15 (2001), pp. 1555–1561.
34. J. Fang, Y. Shing, D. Wiederschain, L. Yan, C. Butterfield, G. Jackson, J. Harper,G. Tamvakopoulos and M.A. Moses, Matrix metalloproteinase-2 is required for theswitch to the angiogenic phenotype in a tumor model. Proc. Natl. Acad. Sci. USA 97(2000), pp. 3884–3889.
35. R.L. Shoeman, R. Hartig, Y. Huang, S. Grub and P. Traub, Fluorescencemicroscopic comparison of the binding of phosphodiester and phosphorothioate(antisense) oligodeoxyribonucleotides to subcellular structures, includingintermediate filaments, the endoplasmic reticulum, and the nuclear interior.Antisense Nucleic Acid Drug Dev. 7 (1997), pp. 291–308.
36. J.P. Leonetti, N. Mechti, G. Degols, C. Gagnor and B. Lebleu, Intracellulardistribution of microinjected antisense oligonucleotides. Proc. Natl. Acad. Sci. USA88 (1991), pp. 2702–2706.
37. T.L. Fisher, T. Terhorst, X. Cao and R.W. Wagner, Intracellular disposition andmetabolism of fluorescently-labeled unmodified and modified oligonucleotidesmicroinjected into mammalian cells. Nucleic Acids Res. 21 (1993), pp. 3857–3865.
38. Y. Xu and F.C. Szoka, Jr., Mechanism of DNA release from cationic liposome/DNAcomplexes used in cell transfection. Biochemistry 35 (1996), pp. 5616–5623.
39. D.A. Brown, S.H. Kang, S.M. Gryaznov, L. DeDionisio, O. Heidenreich, S. Sullivan,X. Xu and M.I. Nerenberg, Effect of phosphorothioate modification ofoligodeoxynucleotides on specific protein binding. J. Biol. Chem. 269 (1994), pp.26801–26805.
40. C. Crum, J.D. Johnson, A. Nelson and D. Roth, Complementaryoligodeoxynucleotide mediated inhibition of tobacco mosaic virus RNA translationin vitro. Nucleic Acids Res. 16 (1988), pp. 4569–4581.
Chapter 4
88
41. R.J. Crouch and M.L. Dirksen, Ribonucleases H. In: S.M. Linn and R.J. Roberts,Editors, Nucleases, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY(1985), pp. 211–241.
89
Chapter 5
Interference of PEGylated lipids with cationic lipid-
mediated delivery of oligonucleotides; role of PEG-lipid
exchangeability and non-lamellar transitions.
Fuxin Shi1, Luc Wasungu1, Anita Nomden1, Marc C.A. Stuart2, Evgeny Polushkin3, Jan
B.F.N.Engberts4 and Dick Hoekstra1,*
Department of Membrane Cell Biology1, University of Groningen, Faculty of Medical Sciences,Antonius Deusinglaan 1, 9713 AV Groningen, Departments of Biophysical Chemistry2 and
Polymer Chemistry3, and Physical Organic Chemistry Unit, Stratingh Institute4, University ofGroningen, Groningen, The Netherlands
Part published in Biochemical Journal, 366(2002), 333-341
Chapter 5
90
ABSTRACT
Cationic liposomes are applied to transfer oligonucleotides (ODNs) into cells to regulate gene
expression for gene therapeutic or cell biological purposes. In vivo, polyethylene glycol (PEG)-
lipid derivatives are employed to stabilize and prolong the circulation lifetime of nucleic acid-
containing particles, and to improve targeting strategies. In this study, we have employed PEG-
lipid analogues, i.e., PEG coupled to either phosphatidylethanolamine (PE-PEG) and ceramide,
to investigate their effect on the mechanism of cationic lipid-mediated delivery of ODNs in vitro.
Inclusion of 10mol% PEG-PE in ODN lipoplexes inhibited cellular internalization by more than
70 %. The internalized fraction remained entrapped in the endosomal-lysosomal pathway and no
intracellular release of ODNs was seen. Similar observations were made for complexes prepared
from liposomes that contained ceramide-PEGs. Interestingly, delivery resumed when lipoplexes
had been externally coated with ceramide-PEGs. In this case, the kinetics of delivery were
dependent on the length of the ceramide acyl chain, consistent with a requirement for PEG-lipid
to dissociate from the complex. Moreover, although the chemical nature of the PEG-ceramides
distinctly affected the net internalization of the complexes, the impediment of delivery was
largely related to an inhibitory effect of the PEG-lipid on the release of ODNs from the
endosomal compartment. Cryo electronmicroscopy and small angle X-ray scattering revealed
that the PEG-lipids stabilize the lamellar phase of the lipoplexes, while their acyl chain length-
dependent transfer from the complex enables adaptation of the hexagonal phase. Within the
endosomal compartment, this transition appears instrumental in causing the dissociation and
cytosolic release of the ODNs for their nuclear homing.
PEG interference on the delivery of lipoplexes
91
INTRODUCTION
Short single-stranded DNA fragments like oligonucleotides(ODNs) provide a means for
modulating gene expression, following their appropriate targeting to and hybridization with a
given gene sequence. Both gene therapeutic and cell biological approaches will benefit from
such applications. With the development of suitable delivery systems, the delivery efficiency of
ODNs, which spontaneously very poorly if at all translocate across the plasma membrane barrier,
has been greatly improved (1, 2). Although readily applicable in vitro, a programma-ble or
controllable mode of delivery is particularly crucial for in vivo application, which demands
stringent requirements for particle stability during circulation, and specific delivery to selected
target tissue and/or cells.
Cationic lipid-DNA complexes (‘lipoplexes’) have been successfully used for gene delivery,
and also cellular delivery of ODNs can be greatly improved in this manner, although the overall
mechanism of delivery is still poorly defined. To overcome such drawbacks as relatively short
circulation time in vivo, unspecific binding of serum components or an undesired interaction
with non-target cells (3, 4), lipoplexes are often coated with polymers, such as polyethylene
glycol (PEG) (5-7). However, little insight is available as to how ‘PEGylation’ affects the overall
cationic lipid-mediated delivery of a gene or ODNs into cells. In this regard, highly relevant
issues concern the effect of the polymer on the physical properties of the lipid-DNA or lipid-
ODN complex, and how such properties as well as the presence and membrane-anchorage of
PEG as such, affects the (intra-)cellular interactions and processing, relevant to the eventual
delivery of plasmid or ODN. Recent work has emphasized the close relationship between
structure and function (i.e., transfectability) of lipid-DNA complexes. Thus the evidence
indicates that the lipid-DNA complexes are highly ordered structures, and that an inverted
hexagonal phase (HII) of the complexes strongly promotes transfection efficiency (8, 9). By
contrast, a lamellar phase (L�) of the complexes correlates with stable particles, displaying
substantially lower transfection potency. However, it is not unlikely that the additional inclusion
of distinct lipids in the lipoplex, including the incorporation of PEGylated lipid may perturb the
delicate balance of the transfection-supporting phase (10). In addition, the bulky presence of
PEG at the interface of interacting membranes will also pose as a steric barrier, which could,
among others, frustrate the lamellar to hexagonal phase transition by precluding tight interaction
of opposed membranes, necessary for such a transition to occur (11). Accordingly, the PEG-lipid
derivatives should eventually depart from the lipoplexes, when the complex has reached the
desired site of delivery. Such a dissociation could be accomplished, for example, when the
Chapter 5
92
polymer is either ‘attached’ to membranes via cleavable bonds, such as an S-S bound, or via
lipid anchors, which in a time-dependent manner can exchange out of the lipoplexes. The latter
class of lipids include Ceramide-PEG, which in studies largely based on the use of liposomes,
displayed exchange properties, the kinetics of which were governed by the length of the acyl
chain (12).
In the present work we have examined the structural and functional consequences of
incorporating PEG-lipids in cationic lipid/ODN complexes, and the ensuing effect on cellular
interaction and processing of such complexes, including ODN delivery. As demonstrated by
small angle X-ray scattering and cryo electron microscopy, the presence of PEG-lipids interferes
with the phase properties of the lipoplexes by stabilizing a lamellar-like morphology. This phase
displays an inherent particle stability that strongly impedes intracellular ODN release, leading to
particle entrapment in the endo-lysosomal pathway without cellular transfection.
MATERIALS AND METHODS
Materials.
Poly(ethylene) glycol (PEG), Mw 2000 D, covalently attached to distearoyl-
phosphatidylethanolamine (DSPE) [DSPE-PEG 2000], was obtained from Avanti Polar
Lipids(Alabaster, USA), and dioleoyl-phosphatidylethanolamine (DOPE) and N-Rh-PE were
purchased from the same source. CeramideC8-PEG, CeramideC14-PEG and CeramideC20-PEG
(Mw PEG is 2000 D) were obtained from Northern lipids(Vancouver, Canada). The cationic
lipid SAINT-2 (N-methyl-4(dioleyl)methylpyridiniumchloride) was synthesized as described in
detail elsewhere(13). Antisense ODN complementary to the mRNA (sequence see Gene Bank
accession L25438) of the receptor of the corticotropin releasing factor (CRF-R), targeted to bp
474-490, and a 17-mer randomized-sequence of ODN were designed and manufactured by
Biognostik(Göttingen, Germany), with the sequences: 5’GGA TGA AAG CCG AGA TG 3’ and
5’-ACT ACG ACC TAC GTG AC-3’, respectively. The randomized ODN sequence, labeled at
the 5’-end with FITC, was used for cellular binding and uptake studies. All ODNs were thioated
and purified by high-performance liquid chromatography, cross-flow dialysis and ultrafiltration.
All chemicals were from Sigma(Missoursi, USA), unless stated otherwise.
Preparation of lipid vesicles and ODN-containing lipoplexes.
The lipids were dissolved in chloroform/methanol(1:1, volume ratio). SAINT-2 and
DOPE(1:1, molar ratio) with or without various lipid-PEG2000 analogues (at concentrations of 5
or 10 mol %, relative to the total lipids) were mixed, and the solvent was removed by
evaporation under a stream of nitrogen, followed by placing the vial under vacuum for at least
PEG interference on the delivery of lipoplexes
93
1hr. The lipids were then resuspended in Millipore water at stock concentrations of 1 or 0.5 mM,
and sonicated to clarity in a bath sonicator in a closed vial. When required for monitoring the
(intra)cellular fate of the lipoplexes by fluorescence microscopy, 0.5 mol% N-Rh-PE was
included in the lipid mixture. Insertion of the PEG-lipids into the lipoplexes was accomplished
by two procedures. In procedure I, lipoplexes were prepared with PEG-lipid-containing SAINT-
2/DOPE liposomes and ODNs as follows. 20nmol of PEG-lipid-containing liposomes,
suspended in 200 �l CHO-SFM medium (Life Technology, Breda, the Netherlands), were mixed
with 0.1nmol ODNs, diluted in 200 �l of the same medium. After 20 min at room temperature,
the mixture was diluted with 600 �l pre-warmed medium, and added to the cells. Alternatively in
procedure II, preformed lipoplexes of SAINT-2/DOPE and ODNs were coated with PEG (lipid-
PEG coating). In this case 20nmol SAINT-2/DOPE liposomes and 0.1nmol ODNs were mixed in
100 �l CHO-SFM medium and incubated for 20 min at room temperature. Then 2nmol lipid-
PEG was added and the mixture was incubated at 60�C for 1hr, cooled to 37 oC, and diluted
with 900 �l medium prior to addition to the cells. Following lipoplex assembly, the packing
efficiency of the ODNs was examined by determining ODN accessibility towards oligreen,
using the Oligreen� ssDNA Quantitation kit (Molecular probe, OR, USA). The assay was
performed according to instructions, provided by the manufacturer.
ODN release assay.
Lipoplexes were prepared as follow: 20 nmoles of liposomes (with or witout PEG-lipid, see
above) were mixed with 0.1 nmole of ODNs in 120 �l 150 mM NaCl /10 mM Hepes, pH 7.4,
and incubated for 20 minutes at room temperature. The lipoplexes were then diluted in 880 �l of
an Oligreen solution 1X (Molecular Probe, OR, USA). Fluorescence was subsequently
monitored at excitation and emission wavelengths of 485 and 520 nm, respectively. After 100
seconds, 100 nmoles of vesicles consisting of DOPE/dioleylphosphatidylcholine (DOPC) and
dioleoylphosphatidylserine(DOPS), molar ratio 2:1:1, were added to trigger ODN release. After
400 seconds Triton X-100 was added at a final concentration of 0.2%, reflecting the level of
fluorescence obtained after complete dissociation. Data of release were calculated from the
fluorescence level obtained at 400 secs, corrected for the background value obtained prior to
addition of anionic vesicles, relative to the fluorescence obtained upon total release (Triton
value).
Chapter 5
94
Fluorescence microscopy studies with FITC-dextran and fluorescently labeled –lipid/ODN
complexes.
The (intra)cellular fate of ODNs and lipoplexes was determined by monitoring the fate of
FITC-labeled ODNs and N-Rh-PE labeled lipoplexes by epifluorescence or confocal laser
scanning microcopy. CHO cells were grown on coverslips in 6-well plates and treated with
various lipoplexes as described, and incubated during a time interval as indicated. The cells
were rinsed twice with HBSS, and analyzed directly or fixed for 10 min in 2.5%
paraformaldehyde in PBS, washed and mounted on microscope slices for examination.
The endosomal-lysosomal pathway in living CHO cells was labeled with FITC-dextran (MW
71,600, Sigma) by a 12hr incubation with the probe at 2mg/ml. Subsequently, the cells were
washed and incubated with N-Rh-PE-labeled lipid-ODN complexes for 5hr at 37�C.
Fluorescence microcopic examination of the samples was carried out using a TCS Leica SP2
confocal laser scanning microscope (Wetzlar,Germany).
Cellular binding and uptake studies.
4x105 CHO-K1 cells/well were seeded in 6-well plates. After 24hr, when the cells had reached
70-80% confluency, the various complexes of FITC-ODNs and lipids were prepared as described
above, added to the cells and incubated in 5%CO2 / 95% air at 37�C during time intervals as
indicated. Following the desired incubation times, the cells were rinsed twice with HBSS (Life
Technologies, Paisley, Scotland), trypsinized and resuspended in medium prior to quantifying
fluorescence by fluorescence-activated cell sorting(FACS) measurements.
Transfection of pGFP with SAINT-2/DOPE with or without PEG-lipids.
1.2�105 CHO cells were seeded per well in 12-well plates the day before transfection. Cells
were transfected with pGFP-lipid complexes (0.5µg/10 nmol lipid) in 500µl CHO-SFM
medium. The complexes were removed after a 4 hr incubation and fresh medium with 10%FCS
was added. 48 hour post-transfection, the transfection efficiency was evaluated by FACS, as
described (14).
Antisense assay.
The antisense effect of ODNs delivered by SAINT-2/DOPE with or without lipid-PEGs was
examined by Western immunoblot. 106 CRF-R-expressing and control CHO cells were seeded in
10cm dishes and grown for 24 h. The cells were then treated with the various complexes (5nmol
ODNs/100nmol lipid) for 5 h, after which period the complexes were removed and fresh
medium was given. The cells were harvested 72 h after ODN treatment, lysed, and the
membranes were isolated as described (15). Samples (25 �g of protein) were then analyzed on
PEG interference on the delivery of lipoplexes
95
12.5% SDS-PAGE( Bio-Rad, Hercules, CA), blotted on pure nitrocellulose membrane(Trans-
Blot� Transfer medium, Bio-Rad, Hercules, CA) and probed with goat anti-rat CRF-R(1:500,
Santa Cruz), followed by horseradish peroxidase conjugate rabbit anti-goat antibody( Sigma,
Steinheim, Germany). The blot was processed with ECLTM ( Amersham pharmacia biotech.,
Buckingham, England) according to the manufacturer’s instructions.
Cryo electron microscopy of lipoplexes.
The morphology of lipid-ODN complexes was determined by transmission cryo-EM. Two �l
of the various samples, indicated in the legends, were applied on glow discharged holey carbon-
coated grids, and the excess of liquid was blotted away by Waterman paper. The specimen was
frozen in liquid ethane and then mounted in a GatAn(mol 626) CRYO-STAGE and examined in
a Philips CM 120 cryo electron microscope, operating at 120 kV.
Small angle X-ray scattering (SAXS) analyses.
To determine the lipoplex structure, SAXS measurements of PEG-devoid lipoplexes and PEG-
containing lipoplexes were performed at 25 �C using a NanoStar device (Brucker AXS and
Anton Paar) with a ceramic fine-focus X-ray tube, operating in a point focus mode. The tube was
powered with a Kristalloflex K760 generator at 35kV and 40mA. The primary beam was
collimated using cross-coupled Göbel mirrors and a 0.1-mm pinhole providing a CuK� radiation
beam( the wavelength �=0.154 nm) with a full-width at half-maximum of about 0.2 mm in
diameter at the sample position. The sample-detector distance was 0.65 m. The use of a Hi-Star
position-sensitive area detector (Siemens AXS) allowed recording the scattering intensity in the
q-range of 0.5 to 3.5 nm-1. The scattering vector q is defined as q = 4�/� sin(�/2), where � is the
scattering angle. The measurements of the samples were performed using a sample cell of 2mm
thickness covered by two thin kapton films. For sample preparation, 1000 nmol SAINT-2/DOPE
were gently mixed with 5nmol ODNs in 60�l 150mM NaCl/10mM HEPES. After 20min at
room temperature, the samples were analyzed by SAXS measurements.
RESULTS
The presence of PEGylated lipids does not affect ODN-lipoplex association
It was of obvious relevance to first determine whether the inclusion of PEGylated lipids
affected the efficiency of ODN-cationic lipid complex association. To this end ODNs were
complexed with various lipid mixtures and the pool of free and lipid- associated ODNs was
determined by the oligreen assay. Consistent with previous observations, at a charge ratio of
5:1(+/-) SAINT2/DOPE almost 95 % of the ODN fraction became complexed. Inclusion of 5-
Chapter 5
96
10 mole % each of the PEGylated lipids DSPE-PEG or Ceramide-PEG (procedure I) did not
significantly change the fraction of oligogreen accessible ODNs. This also holds when
PEGylated lipids are incorporated by exogenous addition at elevated temperature, following
complex assembly (procedure II). Thus irrespective of the presence of PEGylated lipids or the
assembly procedure, in the PEGylated ODN-cationic lipid complexes more than 90 % of the
added ODN fraction became associated with the lipoplexes and was effectively shielded from the
external environment (not shown).
We next examined the effect of PEG on the cellular uptake and intracellular traffick-ing of
ODNs.
Fig.1 The effect of DSPE-PEG on lipoplex-mediated uptake of ODNs . CHO cells were incubated with FITC-ODN containing complexes, prepared asdescribed in Methods. The cell associated fluorescence was measured by FACS after5h (gray bars) and overnight (O/N;black bars). The data are expressed relative to thecell associated fluorescence obtained for SAINT-2/DOPE/ODNs after an O/Nincubation, which was set at 100 %. Cells were also incubated with SAINT-2/DOPEcomplexes in the absence (no PEG) and in the presence of 10 mol% free DSPE-PEG2000. Alternatively, cells were incubated with PEG-lipid-containing complexesprepared with the indicated lipid analogue and concentration, prepared according toeither procedure I (PEG-lipid present in liposomes) or II (external coating; seeMethods). Data are the mean values (�SD) of three determinations.
The insertion of DSPE-PEG into lipoplexes inhibits the cellular association of ODNs in vitro.
The presence of PEG conveys a relative inertness to particles such as liposomes, composed of
phospholipids, in terms of their susceptibility towards interacting with factors like serum or
when engaging in intermembrane interactions (16-19). We therefore investigated the effect of
PEG, covalently coupled to DSPE, on the cationic lipid-mediated uptake of ODNs by CHO cells,
both as a function of preparation procedure and incubation time. As shown in Fig. 1, membrane-
020406080
100120140
no PE-PEG 10% freePE-PEG
5% PE-PEG/S/D
10% PE-PEG/S/D
10% PE-PEG
coating
Cel
lula
r- a
ssoc
iate
d O
DN
s(%
)
5h O/N
PEG interference on the delivery of lipoplexes
97
inserted but not free PEG-lipid effectively inhibited the cellular association of PEGylated
complexes in a PEG concentration dependent manner, while qualitatively similar effects were
observed when comparing the 5h and overnight incubations. Note that although the potentially
effective PEG-lipid concentration available for membrane insertion is higher upon exogenous
addition (PE-PEG II) than when preparing complexes with PEGylated liposomes (PE-PEG I),
the inhibition of complex-cell association in either case is very similar (approx. 70 % after an
overnight incubation). The latter suggests that both procedures apparently give rise to a minimal
surface density packing of PEG-lipids that suffices for obtaining an effective optimal inhibition
of lipoplex-cell surface interaction, since it is likely that the densities of PEG-lipids in either
complex differ.
DSPE-PEG inhibits cellular delivery and prevents intracellular dissociation of ODNs fromlipoplexes
As demonstrated previously (1), after internalization of SAINT-2/DOPE lipoplexes, the
associated ODNs readily dissociate and accumulate in the nuclei. Thus in the absence of PEG-
lipid, effective delivery of the FITC-labeled ODNs is observed, following an incubation for
either 5 (Fig. 2A) or 24 h (not shown). Note that at these conditions ODNs and the lipids, marked
by N-Rh-PE-labeling (red), become separated (insert Fig. 2A). A virtually identical image was
obtained (not shown; cf 2A) when 10% DSPE-PEG was included in the medium. By contrast,
when 10% DSPE-PEG had been incorporated into the lipoplexes by exogenous addition
(procedure II, coating), the ODNs were mostly seen associated with the plasma membrane, often
showing a clustered appearance, whereas nuclear staining was virtually negligible (Fig. 2B).
When lipoplexes were used prepared from liposomes that contained 10% DSPE-PEG (procedure
I), the uptake of ODNs was similarly strongly inhibited (Fig. 2C). However, rather than a
localization at the cell surface, in this case a fine particulate distribution is seen, largely limited
to intracellular compartments that are clearly distinguishable from the nucleus. Interestingly, the
uptake increased when the DSPE-PEG concentration was reduced to 5%, but no shift in
intracellular ODN distribution was observed (Fig. 2D). In fact, even at a density as low as 1 mol
% DSPE-PEG nuclear accumulation of ODNs was still effectively prevented (not shown).
Accordingly, the presence of the PEGylated lipids displayed two effects. First, in a PEG-lipid
concentration dependent manner, the net uptake of lipoplexes is inhibited. Second, once
intracellularly, the presence of the PEG-lipids effectively inhibits complex dissociation. To
support the latter notion, PEGylated lipoplexes were prepared that contained both fluorescently-
tagged ODN (FITC) and N-Rh-PE (0.5 mole %), as a marker of the lipoplex lipid phase. The
Chapter 5
98
merged pictures reveal (Fig. 2E) that dissociation of the complex does not take place (c.f.2A),
even at DSPE-PEG concentrations as low as 1 mole %. Thus escape of ODNs from lipoplexes
was abolished when DSPE-PEG was present. Finally, the distribution pattern of the internalized
complexes, seen in 2C and 2D, suggest that the ODNs are trapped in the endo-lysosomal
pathway. Indeed, when loading this pathway with FTIC-Dextran (Fig. 2F), followed by Rh-PE-
labeled complexes, a colocalization of both probes is observed, implying that the particles are
processed towards lysosomes, which typically localize in perinuclear regions of the cells.
Apparently, the hydrolytic activity of the lysosomes did not suffice to release ODNs or
fragments thereof after long incubation times, since as noted above, even after 24 h the
distribution of the FITC-labeled ODNs was identical to that observed after 5 h (not shown; cf.
Figs .2D and F).
These data indicate that although DSPE-PEG may represent a useful tool for inhibiting ODN-
lipoplex interaction with cells (thereby prolonging the circulation time of particles in vivo), the
lipid analogue is not suitable for targeting (by coupling of tissue-specific antibodies) or the
controlled delivery of the therapeutic cargo. Thus, once reaching the cellular destination, the
PEGylated lipid remains firmly associated with the lipoplex, which precludes ODN release and
causes the complex to be processed into the degradation compartment. This notion emphasizes
the need for exchangeable derivatives. In this regard, claims have been made (12) that,
depending on the length of their fatty acyl chain, PEGylated ceramides may display the desired
exchange properties, necessary for the process that eventually leads to a destabilization of the
lipoplex structure that favors ODN release.
Fig.2 Intracellular localization of ODNs delivered by DSPE-PEG containinglipoplexes. CHO cells were incubated with FITC-labeled ODN containing lipoplexes,with or without DSPE-PEGs for 5 h. The ODN localization was visualized by theconfocal laser scanning microscopy. A. Cells were incubated with FITC-ODNcontaining SAINT-2/DOPE lipoplexes. Note that most ODNs were seen in the nuclei(bright green), while the carrier, marked by N-Rh-PE, remained localized in theperinuclear region (insert). B. Cells were incubated with DSPE-PEG coated SAINT-2/DOPE/ODN complexes (procedure II, using 10 mole % PEGylated lipid). Note thevirtual absence of ODN nuclear localization in this case. C. Cells were incubated withFITC-ODNs complexed with SAINT-2/DOPE/10mol%DSPE-PEG liposomes (procedureI). D. Cells were incubated with FITC-ODNs complexed with SAINT-2/DOPE/5mol%DSPE-PEG. Note the absence of nuclear localization of ODNs. E. Cellswere incubated as D. The lipid phase of the complex was visualized by including 0.5mol % N-Rh-PE (red). Note that the ODNs and lipids were colocalizing in the cytosol(yellow dots), but the dissociation of ODNs (green) and lipids (red) was not apparent. F.Cells were pre-incubated with FITC-Dextran(green) to mark the endosomal-lysosomalpathway. After washing, the cells were incubated with N-Rh-PE-labeled ODN-containing lipoplexes (red) that has been prepared from DSPE-PEG (5 mole %)-containing SAINT-2/DOPE liposomes (procedure I). Note that the complexes aremostly colocalizing with dextran at the perinuclear region.
PEG interference on the delivery of lipoplexes
99
40�m20�m
200�m
40�m 40�m
20�m
A
FE
DC
B
Chapter 5
100
Fig.3 The effect of Ceramide-PEG coating on the uptake of ODNs . CHO cells wereincubated with preformed SAINT-2/FITC-ODN complexes, that were subsequentlycoated (procedure II) with 10mol% ceramideC8, ceramideC14 or ceramideC20-PEG2000. The cell associated fluorescence was measured by FACS after an incubation
for 2h(gray bars),5h(black bars), or
overnight(O/N)(whitebars). The cell associatedfluorescence obtainedafter an O/N incubationwith control lipoplexes,i.e., without ceramide-PEG, was set at 100%.Data are the mean values(�SD) of five differentexperiments, carried outin duplicate.
Dissociation of PEGylated lipids is required for nuclear delivery of ODNs
Lipids may display a facilitated exchange when the length of one of their acyl chains is
shortened (20). This will diminish hydrophobic interactions within the lipidic core of a
membrane and cause an increase in their relative water solubility. Thus ODN lipoplexes were
prepared using liposomes that contained PEGC8, C14 and C20-ceramide species. The presence
of Cer-PEG affected the net cell-association of Cer-PEG containing lipoplexes by the cells,
similarly as observed for DSPE-PEG. The highest uptake was seen for short-chain Cer-
containing lipoplexes, suggesting that uptake efficiency might be related to the relative
differences in the kinetics of PEG-Cer exchange, the more rapid exchange (occurring during the
incubation of the lipoplexes with the cells) leading to a higher uptake (Fig. 3). The PEG-Cer-
mediated delivery patterns of ODNs, obtained after a 5h (Fig.4) or overnight incubation, as
examined by microscopy, were very similar to those obtained for DSPE-PEG containing
lipoplexes. Although the PEGylated complexes are efficiently internalized by the cells, no
significant intracellular release of the FITC-labeled ODNs, as observed for PEG-lipid devoid
complexes (Fig.4 A), was apparent. It should be noted however that the Cer-PEG was present
upon complex assembly, implying a distribution of the lipid analogue throughout the complex.
Hence replenishment may have occurred when peripheral Cer-PEG would be released from the
complex as a result of monomeric transfer (21, 22). Therefore, we next examined the effect on
delivery when the Cer-PEG had been inserted exogenously. Thus, 10 mole % CeramideC8-PEG,
CeramideC14-PEG or CeramideC20-PEG was incorporated exogenously into preformed
0
20
40
60
80
100
120
no Cer-PEG 10% Cer8-PEG coating
10% Cer14-PEG coating
10% Cer20-PEG coating
cellu
lar-
asso
ciat
ed O
DN
s(%
)
2h4hO/N
PEG interference on the delivery of lipoplexes
101
Fig.4 PEG-ceramide inhibits ODNdelivery when incorporated inliposomes prior to complex assemblyCHO cells were incubated with FITC-ODNs complexed with SAINT-2/DOPEand containing either 5mol%ceramideC8-PEG, ceramideC14-PEG orceramideC20-PEG. After 5hrs, the ODNlocalization was determined byfluorescence microscopy. Note that inthe absence of ceramide-PEG ODNswere mostly localized in nucleus (greennuclei), whereas the ODNs were largelylocalized in the cytosol when deliveredby ceramide-PEG containingcomplexes.
Fig.5 Effect of exogenous coating of lipoplexes with ceramide- PEG on ODNdelivery CHO cells were incubated with SAINT-2/DOPE-FITC-ODN complexes(A,E)which had been coated (procedure II) with10mol%ceramideC8-PEG (B,F),ceramideC14-PEG (C,G) or CeramideC20-PEG (D,H). After 5h (A-D) or an overnight(O/N) incubation (E-G), the localization of ODNs was visualized by fluorescencemicroscopy (right panel). The left panel shows the corresponding phase contrastimages.
5hr incubation O/N incubation
A.
B.
G.
F.
E.
C.
D. H.
A B
C D
Chapter 5
102
lipoplexes, as described in the Method session (procedure II), and the PEGylated lipoplexes were
incubated with cells for 2h, 5h and overnight. In this case nuclear accumulation of ODNs was
seen after 2 h for control lipoplexes and for lipoplexes coated with CeramideC8-PEG. With the
CerC14 and C20 derivatives, complexes could only be detected in close association with the
plasma membrane (data not shown). As demonstrated in Fig. 5, when extended to a 5h
incubation, the nuclear labeling of cells treated with control (A) and Cer-C8 lipoplexes (B) was
prominently apparent, it gradually appeared in cells treated with CeramideC14-PEG complexes
(C), whereas for Cer-C20 complexes, the ODNs were still largely associated with internalized
and cell surface-bound complexes (D). After an overnight incubation (Fig. 5E-H), effective
nuclear delivery of ODNs was seen for both the CeramideC8 and C14 derivatives, whereas some
delivery was apparent in case of complexes that contained the CeramideC20 derivative. The time
and ceramide-PEG species dependence of nuclear delivery implied a chain-length dependent
dissociation of the Cer-PEG from the lipoplexes, thereby activating the capacity of lipoplex-
mediated delivery of ODNs.
This possibility was next examined by simulating such a release by incubating PEGylated
ODN lipoplexes with anionic lipid vesicles (23).
PEG-lipids modulate the release of ODN from lipoplexes.
As shown in table 1 the presence of PEG-lipids prevents the release of the ODNs from the
lipoplexes, induced by the interaction of anionic vesicles with the lipoplexes. In a density-
dependent manner, the presence of DSPE-PEG effectively prevents the release of ODNs upon
addition of anionic vesicles, irrespective of whether the PEGylated lipid was incorporated in
vesicles prior to lipoplex assembly or when pre-assembled lipoplexes had been coated
subsequently. Note that addition of free DSPE-PEG did not affect the release. Also coating
(procedure II) or prior incorporation of PEG-ceramide derivatives (procedure I) effectively
prevented ODN release. Interestingly, in the presence of non-incorporated ceramide analogues,
effective inhibition of release was also observed when either ceramide-C 8 or C14 was included
in the mixture, but not in case of the Cer20PEG analogue. These differences very likely reflect
differences in the transfer properties as free monomers of the short chain derivatives (C8 and
C14) versus the essentially non-exchangeable properties of the C20 and DSPE-PEG derivative
(12), which leads to rapid integration into the lipoplex of the former, but not of the latter.
PEG interference on the delivery of lipoplexes
103
Lipid-PEGcoating
SAINT2/DOPE/lipid-PEG
Free PEG
5 % DSPE-PEG 42,0 52,5 92,910 % DSPE-PEG 12,5 14,3 106,15 % CER8-PEG 30,1 66,2 50,05 % CER14-PEG 29,0 58,9 48,65 % CER20-PEG 17,1 18,9 89,2
Effect of PEGylated lipids on the release of ODNs from lipoplexes in the presenceof anionic vesicles. PEG-coated lipoplexes were incubated with anionic vesicles(DOPE/DOPC/DOPS, 2/1/1). The release of ODNs ( expressed as percentage) wasmonitored by measuring fluorescence, arising when Oligreen binds to the liberatedODNs, as described in Methods. For calibration, the 100% value was set to thefluorescence obtained upon total disruption of the complexes with Triton X-100. Thepercentage of release is expressed as the ratio of fluorescence, measured after 5minutes of incubation with anionic vesicles, relative to total release. With control,PEG-lipid devoid complexes, a release of 95% was obtained. Error is within 5 %.
PEGylation interferes with functional delivery of plasmids and ODNs into eukaryotic cells.
The data thus far demonstrate that PEGylated lipids inhibit or delay lipoplex interaction with
cell surfaces and/or intracellular delivery of ODNs. To better define the functional consequences,
we determined the effect of the various PEG derivatives on the delivery and expression of the
reporter gene pGFP and the antisense effect of ODNs on CRF-receptor expression. As shown in
Fig. 6, the transfection efficiency of pGFP was virtually abolished when pGFP had been
delivered by SAINT-2/DOPE/5mol%DSPE-PEG complexes, prepared with PEG-lipid
containing liposomes (procedure I). When the complexes of pGFP and SAINT-2/DOPE were
externally coated with 10mol%DSPE-PEG (procedure II), an inhibition of more than 85 % in
transfection efficiency was still observed. As anticipated, the lowest inhibition of transfection
was observed following external coating of the complexes with C8 and C14-Cer PEG (Fig. 6),
i.e., PEGylated lipids that show a time-dependent increase in exchange.
To correlate the intracellular localization of ODNs with their potential antisense effect, the
efficiency of CRF-R down regulation was examined by Western immunoblot. As shown in Fig
7, down regulation of CRF-R expression was seen when antisense ODNs were delivered with
SAINT-2/DOPE, compared to the levels of expression seen in untreated cells, cells treated with
complexes containing mismatch ODNs or in cells treated with antisense ODNs alone. When the
same amount of antisense ODNs were delivered with 5mol%DSPE-PEG containing complexes,
the antisense effect on CRF-R was completely abolished.
Chapter 5
104
As demonstrated above, the delivery of fluorescently tagged ODNs was time-dependent, when
entrapped in complexes, coated with PEG-ceramides of different fatty acyl chain length. Thus,
after a 5 hr incubation, ODN delivery was largely restored when complexes had been coated
(procedure II) with CerC8-PEG, partly restored when coated with CerC14-PEG, whereas nuclear
delivery still did not occur after this time interval in case of CerC20-PEG. Consistently, as
shown in Fig 7, the antisense effect of ODN in down-regulating CRF-R expresion was inversely
proportional to the fatty acyl chain length of the employed PEG Ceramide. Thus relative to
untreated cells (lane 5) and ODN treatment by delivery in PEG-devoid complexes (lane 1), an
effective reduction of CFR-R expression was seen with CerC8-PEG coated complexes. The
efficiency was diminshed when employing CerC14-PEG coated complexes, whereas no
significant effect was seen with CerC20-PEG coated complexes. These functional data are
therefore fully consistent with the observations, obtained when employing fluorescently tagged
markers to monitor the effect of PEGylated lipids on the (intra-)cellular processing of ODN-
containing lipoplexes (Figs 4 and 5).
In spite of the presence of PEGylated lipids like DSPE-PEG or Cer-PEG, lipoplexes are (at
least partly) internalized by cells, implying that the presence of PEG does not necessarily prevent
lipoplex-cell membrane (receptor) interaction, necessary for endocytic internalization, the
pathway along which SAINT-2 containing lipoplexes deliver plasmids (unpublished
observations ). However, once internalized, this interaction apparently does not suffice for ODNs
to be released and requires the dissociation of the PEGylated lipid from the complex , implying
that additional factors must be involved, like the intimacy of complex-endosomal membrane
interaction (as noted above), and in addition lipid structural requirements (e.g. the lamellar to
hexagonal phase changes) that may co-determine overall nucleic acid release efficiency (8, 9).
The effect of PEGylated lipids on lipoplex structure was therefore examined next.
The presence of PEGylated lipids stabilized the lamellar phase of ODN complexes
In recent years it has become apparent that the lipid phase of lipoplexes may play an important
role in bringing about lipoplex-mediated transfection. Specifically, lamellar L� complexes bind
stably to membranes while HII complexes are unstable and such features are thought to be
instrumental in productive transfection (8, 9). Accordingly, this could imply that lipoplexes
prepared from PEG-liposomes and ODNs display a lipid phase that lacks the ability to disturb
the endosomal membrane and/or to release ODNs. As shown previously (9), in water SAINT-
2/DOPE liposomes appear as unilamellar vesicles with a diameter of 100-200nm. However,
PEG interference on the delivery of lipoplexes
105
when suspended in salt, a transition from the lamellar to the hexagonal phase takes place. As
shown in Fig 8, the turbidity of a suspension of SAINT-2/DOPE vesicles in water increased
more than 20-fold upon addition of salt, reflecting vesicle clustering upon charge neutralization
and a concomitant transition to the hexagonal phase upon membrane-membrane interaction (9).
The presence of ODNs did not interfere with this transition. However, when 5mol%DSPE-PEG
was present in the SAINT-2/DOPE vesicles under otherwise the same conditions, the
aggregation of the lipid vesicle was evidently prevented. Accordingly, since the hexagonal phase
transition requires close interactions between opposed membranes, the data suggest that the
PEGylated lipid stabilizes the lamellar phase by preventing such interactions.
Fig. 6 Effect of PEGylation on pGFP transfection. CHO cells were transfected withpGFP, complexed with SAINT-2/DOPE, with or without lipid-PEG as described inMethods. The transfection efficiency was determined by FACS, and expressed as thepercentage of GFP positive cells. In A, the plasmid was complexed with liposomesconsisting of SAINT-2/DOPE/PEG-lipid (5 mole %). In B, transfection was carried withcomplexes that had been coated with the various PEG-lipids, after complex formation.Note that transfection is progressively abolished when the complexes contain therelatively non-exchangeable PEG-lipids DSPE-PEG and ceramideC20-PEG. The resultsare the mean values ±SD of two experiments, carried out in duplicate.
0.0
10.0
20.0
30.0
40.0
50.0
60.0
70.0
S/D S/D/C8-PEG
S/D/C14-PEG
S/D/C20-PEG
S/D/PE-PEG
GFP
pos
itive
cel
ls(%
)
A
0.0
10.0
20.0
30.0
40.0
50.0
60.0
70.0
S/D freePE-PEG
C8-PEGcoating
C14-PEGcoating
C20-PEGcoating
PE-PEGcoating
GFP
pos
itive
cel
ls(%
)B
Chapter 5
106
Fig. 7 Down-regulation of CRF-receptor by antisense ODN, delivered bylipoplexes, with or without lipid-PEGs. CHO cells, which stably expressed the CRF-receptor (lane 5, ‘untreated cells’), were treated with antisense-containing lipoplexes,containing the various PEGylated lipid derivatives. The antisense effect was evaluatedby immuno Western blotting. The cells were incubated with lipoplexes or freeantisense ODNs for 5 hrs. Then fresh serum-containing medium was added and after72hr CRF-R expression was examined. The cell membranes were isolated, andproteins were separated on PAGE, probed with goat anti-CRF-R and then with HRP-conjugated rabbbit anti-goat antibodies. The blot was processed with ECL. In A, areduced CRF-R expression is seen following lipoplex-mediated antisense delivery (1.AS+S/D), compared to untreated cells (5). Also note that the antisense effect wasabolished when non-exchangeable DSPE-PEG was included in the lipoplex formulation(2. AS+S/D/DSPE-PEG). MAS+S/D (3) reflects treatment of the cells with mismatchantisense, while AS (4) indicates treatment with free antisense.In B, the effect of PEG-ceramides on lipoplex-mediated antisense delivery wasexamined. Antisense ODNs were associated with cationic lipids (1. AS+S/D) or Cer-PEG-coated complexes (2. AS+S/D+CerC8-PEG, 3. AS+S/D+Cer14-PEG, 4.AS+S/D+CerC20-PEG), compared to untreated cells (5). Note that the effect of down-regulation correlates with the length of the ceramide acyl chain length.
Fig8. Effect of DSPE-PEG on vesicleaggregation in water andphysiological salt solutions. SAINT-2/DOPE (S/D) and SAINT-2/DOPE/5mol%DSPE-PEG(S/D/DSPE-PEG) vesicles, prepared in water, weresuspended in salt solutions (HBS), andthen mixed with ODNs. The turbiditywas monitored as a function of time ata wavelength of 350nm. The maximalturbidity levels obtained were plottedand the time required to reach thisplateau value is indicated on top of thebars.
1 2 3 4 5
A
1 2 3 4 5
B
turbidity
0
0.2
0.4
0.6
0.8
1
S/D/DSPE-PEG HBSODN S/D HBS
ODN S/D ODNHBS
8.8
2.2 9.86
2.3 2.2 1.58
PEG interference on the delivery of lipoplexes
107
Fig. 9 Morphology of ODNlipoplexes as revealed by cryo-EM. The various ODN lipoplexeswere investigated by cryo-EM, asdescribed in Methods. In A and B,control complexes consisting ofSAINT-2/DOPE/ODN are shown.Note the striated structureswhich are typical for thehexagonal morphology. Thesecomplexes tended to aggregateand the average size was about100-200nm. In C and D, DSPE-PEG-containing complexes(procedure I) are shown, whichappeared as small and roundparticles with an average size of30-50nm. Striated structureswere far less frequent and oftenparticles were seen with a fewlamellae (arrow). Such complexesdid not tend to aggregate. In Eand F, the preformed SAINT-2/DOPE/ODN complexes hadbeen coated with 10mol% DSPE-PEG (Procedure II). Mostcomplexes showed the striatedstructures typically seen prior tocoating, as in A and B. In G,
SAINT-2/DOPE/DSPE-PEGvesicles in water are shown,revealing a mixture of particleswith different sizes with little ifany electron dense materialinside. When 150mMNaCl/10mM HEPE, pH 7.4 wasadded, the giant liposomes
disappeared and the unilamellar liposome and particles with invaginations coexisted.Clearly there was no appearance of a hexagonal texture. Bars are 100nm.
To further clarify and support these observations, the morphology of cationic lipid-ODN
complexes was investigated by cryo-EM and SAXS. Control SAINT-2/DOPE-ODN complexes
appeared as fingerprint-like structures initially showing particles with a diameter of 100-200 nm,
which cluster into larger complexes as a function of time (> 15 min; Figs.9 A and B). The finger-
print structure is typical of the hexagonal phase. To support this conclusion, we analyzed the
samples by SAXS. As shown in Fig.10A, three diffraction maxima were apparent at q=0.105,
A B
C D
E F
G H
Chapter 5
108
0.181 and 0.201(A)-1 , implying that the location of the peaks is in the ratio of 1: �3: �4, i.e.,
fully consistent with a hexagonal morphology. The periodicity of the phase is about 7nm (Fig. 10
A). By contrast, the morphology of DSPE-PEG-containing complexes is different. Cryo EM
revealed the presence of small homogenous complexes with a diameter of 30-50nm, which did
not display a tendency to aggregate (Figs 9C and D). The structures appeared less densely
packed than PEG-devoid complexes, as reflected by a less electron-dense appearance of the
images, while often only a few membrane layers were seen, surrounding an aqueous volume, i.e.,
an internal space lacking significant internal structure. These structures more closely resemble a
lamellar than a hexagonal organization. Indeed, as shown in Fig. 10B, the SAXS diffraction
pattern differed considerably from that obtained for the non-PEGylated complexes (Fig.10A).
Only one peak at q=0.111(A)-1 was obtained, corresponding to a distance of 5.6nm. This distance
is compatible with a lamellar phase of similar periodicity, observed for lamellar complexes
obtained for plasmid-containing lipoplexes (8, 9). A similar periodicity was obtained when
ODNs were substituted for plasmid or when the concentration of DSPE-PEG was reduced from
5 to 1 mole%.
Fig. 10 The structure of cationic lipid-ODN complexes, determined by smallangle X-ray scattering (SAXS)The cationic lipid-ODN complexes were prepared as described in Methods and thestructures of the complexes were determined by small angle X-ray scattering. A showsthe diffraction pattern of control SAINT-2/DOPE/ODN complexes, revealing threepeaks at q=0.105, 0.180 and 0.210(Å)-1 , which indicate a hexagonal morphology witha periodicity of 7.0 nm. B shows the diffraction pattern obtained with SAINT-2/DOPE/DSPE-PEG/ODN complexes. In this case only 1 peak at q=0.111 (Å)-1 wasobtained . This peak corresponds to a distance of 5.7nm.
0 0.1 0.2 0.3
10
100
q-vector, (Å) -1
ODNs-SAINT-2/DOPE
0 0.1 0.2 0.3
10
q-vector, (Å) -1
ODNs-SAINT-2/DOPE/DSPE-PEG
0.4
A B
PEG interference on the delivery of lipoplexes
109
When 10% DSPE-PEGs was incorporated into preformed SAINT-2/DOPE-ODN complexes,
their presence evidently controlled the size of the complexes by preventing clustering, observed
to occur for complexes devoid of PEG-lipid. The PEG-lipid being restricted to the outer
periphery of the complex only, the typical fingerprint structure was largely maintained in this
case (Figs.9 E and F). However, as shown above (Figs. 1 & 2), in line with the impeding effect
of externally exposed non-exchangable PEG-lipid, such complexes give rise to poor release of
ODNs.
To further define the effect of the PEGylated lipid on hexagonal phase formation, we
examined the lipid phase of SAINT-2/DOPE/5mol%DSPE-PEG vesicles in water, which consist
of a mixture of particles of various sizes and shapes, as shown in Fig. 9G. Following their
incubation in a physiological salt environment, the transition into a hexagonal texture was not
seen (Fig. 9 H). Rather, vesicles in salt appeared as well-defined unilamellar liposomes and
particles with invaginations, indicating that DSPE-PEG apparently prevents the SAINT-2/DOPE
from adopting a hexagonal morphology by interfering with the close approach of opposed
membranes.
DISCUSSION
The present work demonstrates that as a function of their structure, PEGylated lipid analogues
can strongly interfere with the functional properties of lipoplexes, in terms of antisense or gene
delivery. This interference is only partly due to an inhibition of complex internalization by cells,
which has prompted their development and application in the first place. Rather, our data reveal
that PEG-lipid analogs strongly interfere with structural features of the complex, involving a
stabilization of the lamellar phase and precluding an intimate interaction with the endosomal
membrane, thereby impeding cytosolic release of ODN or genes. The latter was apparent from
the strong diminishment in the down-regulation of a target membrane receptor and a decrease in
expression of a reporter gene. Clearly, the bilayer organization per se does not affect particle
internalization, but non-bilayer features appear crucial in events that subsequently govern
intracellular release of ODNs or reporter genes, events that appear to be triggered when the
complexes are processed along the endosomal track (24, 25, unpublished observations).
Importantly, the capacity to release ODN or plasmid does not solely rely on the fact that the
lipoplexes have overall adopted the hexagonal phase, and that they have acquired access into the
endosomal compartment. Thus our data suggest (Figs. 9E and F) that direct membrane-
membrane interactions between such complexes and the endosomal membrane are necessary, a
Chapter 5
110
step which is prevented when PEGylated lipids are present. In that case, the complexes arrive in
the lysosomal compartment, without the occurrence of cytosolic release of ODN or plasmid.
Rather unexpectedly, we observed a relatively poor effect of PEGylated lipids to interfere with
internalization a such of lipoplexes in vitro. Thus a significantly diminished internalization was
apparent only when the concentration of PEGylated lipid amounted 5-10 mole %, irrespective of
whether the lipid was included in the liposomes used for complex preparation, or when inserted
following complex assembly. For example, for the cationic SAINT complexes we observed that
inclusion of 5 mole % DSPE-PEG, although causing a virtually complete inhibition of nuclear
ODN delivery, resulted in an uptake only slightly less than that of control complexes (Fig.1). In
this context, it has also been reported that the protective effect of PEG on liposomes can been
influenced by lipid composition, charge ratio, PEG molecular weight (10) and serum (26).
Possibly, the cationic charge promotes cell surface association an effect which is not abolished
by the presence of the PEGylated lipid.
Currently, two procedures have been described to incorporate PEGylated lipids into gene
delivery complexes. One involves prior incorporation of lipid-PEGs into liposomes, i.e., before
lipoplex assembly triggered upon addition of DNA (21, 27). Alternatively, lipid-PEGs are
inserted into the preformed lipoplexes or liposomes (17, 28, 29). Indeed, exogenous addition led
to an effective integration of the lipid into the SAINT-2 complexes, which at such conditions,
was limited to the outer periphery of the complex only (Figs. 9 and 10). Importantly, only such
PEGylated complexes appear to display a fertile use in delivery, provided that the PEG-lipids are
exchangeable. Thus even when exchangeable C8-Cer is contained in the entire complex, no
delivery is observed and the lipoplexes reach the lysosomes (Fig.4). In fact when complexes
were prepared according to the latter procedure, we noted a careful control of the PEG derivative
on the size of the complexes formed. Small particles, with diameters between 50-80 nm were
obtained, showing a fine punctate appearance within the cell and a processing along the
endocytic pathway to its end point, the lysosomes (Fig.2). Thus although ODNs gained access
into the cells, the non-exchangeable behavior of DSPE-PEG and incomplete exchange of short
chain PEG ceramide derivatives, presumably in conjunction with their rapid size-dependent
processing into the endosomal track (30) precluded intracellular release. Since DSPE-PEG is
frequently used to stabilize liposomes or lipoplexes in order to achieve effective delivery of
drugs, the present observations discourage the use of DSPE-PEG for this purpose.
When located in the complex’ periphery by exogenous insertion, Cer-PEG derivatives may
still inhibit cellular association of the complexes. Yet the efficiency of inhibition clearly showed
a dependence on the kinetics of PEG-Cer exchange (table), and in parallel to that the Cer-PEG
PEG interference on the delivery of lipoplexes
111
lipid dependent down regulation of receptor expression (Fig. 7). Thus C8-ceramide-PEG
displayed little if any inhibition of both parameters while with increasing fatty acyl chain length
the effective inhibition increased, and became compatible with effects of the non-exchangeable
DSPE-PEG in case of CerC20-PEG. The data suggest that early after the onset of the incubation
C8-CerPEG readily transfers from the complex to, presumably, the cell surface, since after only
two hours a substantial nuclear accumulation of ODNs was apparent.
As noted, the localization of the PEG-ceramide (lipoplex surface localization vs distribution in
the entire complex) and the analogue’s density are crucial in this regard. The latter in particular
determines the efficiency of uptake (cf. Fig.1), but also at lower mole percentages, intracellular
release can still be substantially delayed (Figs. 4 and 5), implying a distinction between the
effect of the PEGylated lipids on internalization on the one hand, and ODN release on the other.
Thus the data would suggest that electrostatic interactions between the complexes and the cell
surface need not necessarily be prevented by the presence of the PEGylated lipids, which is in
line which similar observations reported by others (7, 31). Indeed, the primary and secondary
energy minimum for such interactions, taking place when membranes come in close proximity,
are localized in the range of distances between 3-10 nm (32). For non-lamellar transitions to take
place the intermembrane distance should likely not exceed a distance of 1 nm, an event that
appears to be precluded by the steric interference of the PEGylated lipid. It is in this context
particularly interesting to note that the release of ODNs from PEGylated lipoplexes is also
inhibited by adding anionic liposomes (table), a feature that might mimic the mechanism
involved in nucleic acid release from lipoplexes in the endosomal compartment (33). This
observation could suggest that a simple flip-flop and electrostatic displacement of ODNs
following their substitution by PS in the cationic lipid complex does not occur or, alternatively,
that such a translocation requires the complex to adopt the non-lamellar phase at the complex
surface as well, a lamellar phase being stabilized when coated with PEGylated lipid (Figs. 8 and
9).
Rather, the present work emphasizes an absolute requirement for dissociation of the PEG-lipid
analogue to abrogate its effect on the ability of cationic lipids in conjunction with DOPE, to
adopt ODN- or plasmid dissociation promoting non-bilayer phases of the lipoplex. Depending on
the length of the acyl chain, ceramide-PEG may dissociate from liposomes or lipoplexes over
time, thus triggering the transfection-competence of lipoplexes they arrive at their site of
destination (12,29, 34, 35). In this context it is finally relevant to note that the presence of 5 mole
% DSPE-PEG prevents interactions between SAINT-2/DOPE vesicles (Fig. 8), irrespective of
the presence of salt, charge neutralization representing a strongly promoting step in hexagonal
Chapter 5
112
phase formation of SAINT-containing lipoplexes (9). Indeed, within such structures a lamellar
phase is the most prominent structure present (Fig. 9). For undergoing the hexagonal phase, i.e.,
conditions for productive release, tight intermembrane interactions are needed. Such interactions
only occur when the PEG-lipids dissociate effectively from the lipoplexes, even very low
concentrations (1 mole %) being sufficient to effectively interfere with the dissociation of
plasmids and ODNs from such complexes.
ACKNOWLEDGEMENTThis work was supported by a grant from The Netherlands Organization for ScientificResearch (NWO)/NDRF Innovative Drug Research (940-70-001).
REFERENCES:1. Shi, F., Nomden, A., Oberle, V., Engberts, J. B. and Hoekstra, D. (2001) Efficient
cationic lipid-mediated delivery of antisense oligonucleotides into eukaryotic cells:down-regulation of the corticotropin-releasing factor receptor, Nucleic Acids Res.29, 2079-2087.
2. DeLong, R. K., Yoo, H., Alahari, S. K., Fisher, M., Short, S. M., Kang, S. H., Kole,R., Janout, V., Regan, S. L. and Juliano, R. L. (1999) Novel cationic amphiphiles asdelivery agents for antisense oligonucleotides, Nucleic Acids Res. 27, 3334-3341.
3. Bally, M. B., Harvie, P., Wong, F. M. P., Spencer, K., Wasan, E. K. and Reimer, D.L. (1999) Biological barriers to cellular delivery of lipid-based DNA carries, Adv.Drug Deliv. Rev. 38, 291-315.
4. Chesnoy, S. and Huang, L. (2000) Structure and function lipid-DNA complexes forgene delivery, Annu. Rev. Biophs. Biomol. Struct. 29, 27-47.
5. Blume, G., Cevc, G., Crommelin, M. D., Bakker-Woudenberg, I. A., Kluft, C. andStorm, G. (1993) Specific targeting with poly(ethylene glycol)-modified liposomes:coupling of homing devices to the ends of the polymeric chains combines effectivetarget binding with long circulation times, Biochim Biophys Acta. 1149, 180-184.
6. Maruyama, K., Takahashi, N., Tagawa, T., Nagaike, K. and Iwatsuru, M. (1997)Immunoliposomes bearing polyethyleneglycol-coupled Fab' fragment showprolonged circulation time and high extravasation into targeted solid tumors invivo, FEBS Lett. 413, 177-180.
7. Song, L. Y., Ahkong, Q. F., Rong, Q., Wang, Z., Ansell, S., Hope, M. J. and Mui, B.(2002) Characterization of the inhibitory effect of PEG-lipid conjugates on theintracellular delivery of plasmid and antisense DNA mediated by cationic lipidliposomes, Biochim Biophys Acta. 1558, 1-13.
8. Koltover, I., Salditt, T., Radler, J. O. and Safinya, C. R. (1998) An invertedhexagonal phase of cationic liposome-DNA complexes related to DNA release anddelivery, Science 281, 78-81.
9. Smisterova, J., Wagenaar, A., Stuart, M. C., Polushkin, E., ten-Brinke, G., Hulst,R., Engberts, J. B. and Hoekstra, D (2001) Molecular shape of the cationic lipidcontrols the structure of cationic lipid/dioleylphosphatidylethanolamine-DNAcomplexes and the efficiency of gene delivery, J. Biol. Chem. 276, 47615-47622.
10. Holland, J. W., Cullis, P. R. and Madden, T. D. (1996) Poly(ethylene glycol)-lipidconjugates promote bilayer formation in mixtures of non-bilayer-forming lipids,Biochemistry 35, 2610-2617.
PEG interference on the delivery of lipoplexes
113
11. Cullis, P. R., Tilcock, C. P. and Hope, M. J. (1991) Lipid Polymorphism, inMembrane fusion (Wilschut, J. and Hoekstra, D., Ed), pp35-64, Marcel Dekker,New York.
12. Wheeler, J. J., Palmer, L., Ossanlou, M., MacLachlan, I., Graham, R. W., Zhang, Y.P., Hope, M. J., Scherrer, P. and Cullis, P. R. (1999) Stabilized plasmid-lipidparticles: construction and characterization, Gene Ther. 6, 271-281.
13. Meekel,A.A.P., Wagenaar,A., Smisterova,J., Kroeze,J., Haadsma,P., Bosgraaf,B.,Stuart,M.C.A., Brisson,A., Ruiters,M.H.J., Hoekstra,D. et al. (2000) Synthesis ofpyrimidinium amphiphiles used for transfection and some characteristics ofamphiphile/DNA complex formation, Eur. J. Org. Chem. 665-673.
14. Audouy, S., Molema, G., de-Leij, L. and Hoekstra, D. (2000) Serum as a modulatorof lipoplex-mediated gene transfection: dependence of amphiphile, cell type andcomplex stability, J. Gene Med. 2, 465-476.
15. Kramer,E.M., Koch,T., Niehaus,A. and Trotter,J. (1997) Oligodendrocytes directglycosyl phosphatidylinositol-anchored proteins to the myelin sheath inglycosphingolipid-rich complexes, J. Biol. Chem. 272, 8937-8945.
16. Silvius, J. R.and Leventis, R. (1993) Spontaneous interbilayer transfer ofphospholipids: dependence on acyl chain composition, Biochemistry 32, 13318-13326.
17. Uster, P. S., Allen, T. M., Daniel, B. E., Mendez, C. J., Newman, M. S.and Zhu, G.Z. (1996) Insertion of poly(ethylene glycol) derivatized phospholipid into pre-formedliposomes results in prolonged in vivo circulation time, FEBS Lett. 386, 243-246.
18. Ishida, T., Iden, D. L. and Allen, T. M. (1999) A combinatorial approach toproducing sterically stabilized (Stealth) immunoliposomal drugs, FEBS-Lett. 460,129-133.
19. Chiu, G. N., Bally, M. B. and Mayer, L. D. (2001) Selective protein interactions withphosphatidylserine containing liposomes alter the steric stabilization properties ofpoly(ethylene glycol). Biochim. Biophys. Acta 1510, 56-69.
20. Kok, J. W. and Hoekstra, D. (1999) Fluorescent lipid analogues: applications in celland membrane biology, in Fluorescent and Luminescent Probes for BiologicalActivity (Mason, W. T., Ed) 2nd ed., pp136-155, Academic Press, London.
21. Maurer, N., Wong, K. F., Stark, H., Louie, L., McIntosh, D., Wong, T., Scherrer, P.Semple, S. C. and Cullis, P. R. (2001) Spontaneous entrapment of polynucleotidesupon electrostatic interaction with ethanol-destabilized cationic liposomes,Biophys. J. 80, 2310-2326.
22. Hoeksta, D. and Martin, O. C. (1982) Transbilayer redistribution ofphosphatidylethanolamine during fusion of phospholipid vesicles. Dependence onfusion rate, lipid phase separation and formation of non-bilayer structures,Biochemistry 21, 6097-6103.
23. Xu, Y. and Szoka, F. C. (1996) Mechanism of DNA release from cationicliposome/DNA complexes used in cell transfection, Biochemistry 35, 5616-5623
24. Zabner, J., Fasbender, A. J., Moninger, T., Poellinger, K. A. and Welsh, M. J. (1995)Cellular and molecular barriers to gene transfer by a cationic lipid, J. Biol. Chem.270, 18997-19007.
25. Friend, D. S., Papahadjopoulos, D. and Debs, R. J. (1996) Endocytosis andintracellular processing accompanying transfection mediated by cationicliposomes, Biochim. Biophys. Acta., 1278, 41-50.
26. Johnstone, S. A., Masin, D., Mayer, L. and Bally, M. B. Surface-associated serumproteins inhibit the uptake of phosphatidylserine and poly(ethylene glycol)liposomes by mouse macrophages, Biochim. Biophys. Acta. 1513, 25-37.
27. Zhang, Y. P., Sekirov, L., Saravolac, E. G., Wheeler, J. J., Tardi, P., Clow, K., Leng,E., Sun, R., Cullis, P. R. and Scherrer, P. (1999) Stabilized plasmid-lipid particles
Chapter 5
114
for regional gene therapy: formulation and transfection properties, Gene Ther. 6,1438-1447.
28. Mok, K. W., Lam, A. M. and Cullis, P.R. Stabilized plasmid-lipid particles: factorsinfluencing plasmid entrapment and transfection properties, Biochim. Biophys.Acta. 1419, 137-150.
29. Anwer, K., Kao, G., Proctor, B., Rolland, A. and Sullivan, S. (2000) Optimization ofcationic lipid/DNA complexes for systemic gene transfer to tumor lesions, J. DrugTarget. 8, 125-135.
30. Zuhorn, I. S., Visser, W. H., Bakowsky, U., Engberts, J. B. and Hoekstra, D. (2002)Interference of serum with lipoplex-cell interaction: modulation of intracellularprocessing, Biochim. Biophys. Acta. 78207, 1-12.
31. Harvie, P., Wong, F. M. and Bally, M. B. (2000) Use of poly(ethylene glycol)-lipidconjugates to regulate the surface attributes and transfection activity of lipid-DNAparticles, J. Pharm. Sci. 89, 652-663.
32. Hoekstra, D and Wilschut, J. (1996) Water transport, in Biological membranes,(Benga, G., Ed.) Vol. 1, pp 144-176, CRC Press, Boca Raton.
33. Zelphati, O. and Szoka, F. C. (1996) Mechanism of oligonucleotide release fromcationic liposomes, Proc. Natl. Acad. Sci. U. S. A. 93, 11493-11498.
34. Webb, M. S., Saxon, D., Wong, F. M., Lim, H. J., Wang, Z., Bally, M. B., Choi, L. S.,Cullis, P. R. and Mayer, L. D. (1998) Comparison of different hydrophobic anchorsconjugated to poly(ethylene glycol): effects on the pharmacokinetics of liposomalvincristine, Biochim. Biophys. Acta. 1372, 272-282.
35. Monck, M. A., Mori, A., Lee, D., Tam, P., Wheeler, J. J., Cullis, P. R. and Scherrer,P. (2000) Stabilized plasmid-lipid particles: pharmacokinetics and plasmid deliveryto distal tumors following intravenous injection, J. Drug Target. 7, 439-452.
115
Chapter 6
In situ entry of oligonucleotides into brain cells can
occur through a nucleic acid channel;
options for effectuating antisense therapy in the brain
Fuxin Shi1, Jerome Swinny2, Eric Ronken3 and Dick Hoekstra1
1Department of Membrane Cell Biology, Faculty of Medical Sciences, University of GroningenAntonius Deusinglaan 1, 9713 AV Groningen, The Netherlands; 2Department of ElectronMicroscopy, Faculty of Medical Sciences, University of Groningen, The Netherlands; 3SolvayPharmaceuticals, Research Laboratories, Weesp, The Netherlands.
Submitted
Chapter 6
116
Abstract
Because of failure of traditional treatments of brain tumors, and given a gradually increasing
number of neurodegenerative diseases, brain tissue has become a challenging therapeutic target.
Since antisense oligonucleotides (ODNs) are readily internalized by neuronal cells in culture,
these compounds could possibly serve as novel, therapeutic agents to meet such a challenge. In
previous in vitro work, using cell culture systems, we have demonstrated that antisense
efficiency depends on effective cytosolic delivery of ODNs, and their homing into the nucleus.
Intracellular delivery requires a vector such as cationic liposomes, since free ODNs remain
largely trapped in the endocytic pathway, following cellular uptake. Here we studied the cellular
uptake properties of ODNs by explants of rat brain (brain slices), and by in vivo brain tissue after
administration of ODNs by bolus injection. In contrast to in vitro uptake, it is shown that in brain
slices, ODNs were taken up by neuronal and non-neuronal cells, irrespective of their packaging
with cationic liposomes. In either case, a diffuse distribution of ODNs was seen in the cytosol
and/or nucleus. Uptake of ODNs by brain slices as a result of membrane damage, potentially
arising from the isolation procedure, could be excluded. Interestingly, internalization was
inhibited following treatment of the tissue with antibody GN-2640, directed against a nuclei acid
channel, present in rat kidney cells. Our data support the view that an analogous channel is
present in brain tissue, allowing entry of free ODNs, but not plasmids. Indeed, for delivery of the
latter and accomplishment of effective transfection, cationic lipids were needed for gene
translocation into both brain slices and brain tissue in vivo. These data imply that for antisense
therapy to become effective in brain, cationic lipid-mediated delivery will only be needed for
specific cell targeting but not for delivery per se in order to accomplish nuclear delivery of
ODNs into brain cells, and subsequent downregulation of disease-related targets.
Oligos enter brain cells through a nucleic acid channel
117
Introduction
Antisense oligonucleotides (ODNs) offer an interesting potential as effective therapeutic
agents. By virtue of their ability to hybridize to complementary mRNA, antisense ODNs can
induce its effective degradation, thereby impeding the de novo biosynthesis of target proteins.
Antisense sequence, stability, and cellular uptake, among others, determine the antisense
efficiency. In vitro, the uptake properties of antisense ODNs have been well characterized. When
added to cultured cells, free antisense ODNs are endocytosed but remain often largely trapped in
the endocytic compartments. Their translocation from endosomal compartments is therefore
essential, which can be promoted by ODN complexation with cationic lipids. The latter facilitate
transfer of ODNs across the endosomal membrane, and via transport through the cytosol, ODNs
passively accumulate into the nuclei. Here they acquire access to target mRNAs via interaction
with the nuclear matrix, leading to an inhibition of the synthesis of target proteins (1, 2). Not
only do cationic liposomes greatly enhance the antisense effect by promoting endocytic escape
of ODNs in vitro (3, 4), they are also used to prolong the plasma half-lives of oligonucleotides in
vivo (5, 6).
The rapid development of in vivo models and pharmacokinetic distribution studies suggest that
oligonucleotide delivery displays a wide spectrum of tissue distribution, ODNs being taken up in
particular by organs of the reticuloendothelial system (RES) such as liver, spleen, kidney and
lungs. However, brain tissue has proven to be essentially inaccessible for ODNs, following
systemic administration. Accordingly, brain is a particularly challenging organ for drug
targeting, also in light of the increasing number of neurodegenerative diseases that have become
apparent in recent years. Moreover, unlike the success of treatments of a variety of tumors,
relatively little progress has been made in the treatment of brain tumors over the last decade.
Although ODNs hardly cross the blood brain barrier, they readily accumulate in neuronal tissue
when injected directly into brain (7). Furthermore, it has also been claimed that oligonucleotides,
complexes with liposomes that were conjugated with transferrin receptor antibodies, OX-26,
could cross the blood brain barrier and distribute over a wide area in the brain (8, 9), although
data on intracellular ODN uptake have not been reported in these studies. It has been shown that
following a bolus injection of FITC-oligonucleotides into intracerebro-ventricles, the FITC
signal is primarily localized to the ventricle, while continuous infusion of FITC-oligonucleotides
into the ventricle causes a wide distribution in the brain, except in the white matter (10).
However, little insight is available as to how ODNs enter cells in the brain, and whether delivery
vectors will facilitate oligonucleotides in penetrating the extracellular matrix or plasma
membranes, as observed in vitro.
Chapter 6
118
Hence, based upon previous work on in vitro delivery of ODNs into neuronal cells, we
examined here whether and how ODNs could acquire access to brain cells, using an ex vivo
model involving brain slices. The data were verified by investigating directly the fate of a bolus
injection in the brain in vivo. In contrast to the in vitro findings, we observed that ODNs were
taken up through the whole slice without the need of carriers. We demonstrate that facilitated
diffusion of ODNs through a brain cell localized nucleic acid channel can account for the high
uptake of ODNs in brain slices. In comparison, a GFP-reporter plasmid did not acquire access
via the channel, but rather, required cationic liposomes for cellular entry and subsequent
expression. Similar properties of ODN uptake characteristics, as observed in brain slices, were
apparent in vivo, following a bolus injection of ODNs in rat brain.
Material and Methods
Materials
The cationic lipid N-methyl-4(dioleyl)methylpyridiniumchloride (SAINT-2) was synthesized
as described in detail elsewhere(11). Dioleoylphosphatidylethanolamine (DOPE) was purchased
from Avanti Polar Lipids (Alabaster, USA). Antisense ODNs and FITC-labeled ODNs were
designed and manufactured by Biognostik (Göttingen, Germany), with the sequence ATC CAT
GCC TGC CT targeted to rat 5-HT1A mRNA (sequence see Gene Bank, accession J05276,
targeted to bp115-128; ref. 1). All ODNs were thioated and purified by high-performance liquid
chromatography, cross-flow dialysis and ultrafiltration. Polyclonal antibodies, GN-2640, raised
against a peptide of the 45kDa pore forming subunit of the rat kidney nucleic acid channel were
a kind gift from Dr. Basil Hanss from Mt. Sinai School of Medicine, New York, U.S.A. The
plasmid pEGFP-N1, containing the gene encoding for green fluorescent protein (GFP) under the
control of the CMV promoter, was obtained from Clonetech. Rhodamine-labeled plasmids were
obtained from GTS INC (San Diego, CA, U.S.A).
All chemicals were from Sigma (Missoursi, USA), unless stated otherwise.
Preparation of brain slices.
8-day-old Sprague-Dawley rat pups were decapitated and the brains were aseptically removed.
The cerebellum was dissected in ice-cold preparation medium (MEM containing 2 mM glutamax
I, pH 7.3) and the meninges were carefully removed. Sagittal slices of 400µm were cut using a
McIllwain tissue chopper, separated with fine forceps and transferred onto humidified
transparent membranes (Millicell-CM, Millipore). They were cultured on a liquid layer of MEM
containing HBSS (25%), horse serum (25%), glutamax I (2mM) and NaHCO3 (5mM), HEPES
Oligos enter brain cells through a nucleic acid channel
119
(10mM), pH7.3, in a humidified atmosphere with 5% CO2 at 37�C. Cultures were maintained for
up to 12 days in vitro (DIV). Medium was changed every 2-3 days.
Uptake of ODNs and plasmids, and the transfection of pGFP in brain slices.
One nano mole ODN or 5 µg pGFP in 50µl DMEM were added on top of the mounted brain
slices for the indicated time. Then the brain slices were washed and supplemented with fresh
medium. Alternatively, the same amount of ODNs or pGFP was mixed with 75 nmol (+/- charge
ratio 2.5:1) SAINT-2/DOPE in 10µl 5% glucose, and after 20 minutes the lipoplexes were added
on top of the mounted brain slice at 37�C. After 4 hrs, lipoplexes were removed and the medium
was refreshed. The delivery efficiency was determined at 4 and 24 hr, and the transfection
efficiency was analyzed after 2 and 6 days by (epi-)fluorescence microscopy (Olympus, Japan)
or by TCS Leica SP2 confocal laser scanning microscopy (Wetzlar, Germany). The integrity of
the plasma membranes of the cells in the brain slices was verified by Hoechst blue staining
(20µg/ml, Molecular probe). The nuclei were counterstained with propidium iodide (2.5µg/ml) in
the fixed brain slices. ODNs labeled with FITC, plasmids labeled with rhodamine, and SAINT-
2/DOPE, traced with rhodamine-PE, were used to monitor their uptake by fluorescence
microscopy.
Western blotting of the nucleic acid channel
The expression of the nucleic acid channel was examined by Western Immunoblot. Cerebella
samples at P8 (8 days after birth) were prepared from fresh rat cerebellum, lysed in sample
buffer (3ml/mg, 5%SDS, 5%mercaptoethanol, 8M urea, 6.25mM Tris-HCl, pH 6.8, and 0.01%
bromophenol blue). 4, 9 and 15 µg of cerebellum samples were separated on 12.5% SDS-PAGE
(Bio-Rad, Hercules, CA), blotted on a pure nitrocellulose membrane (Bio-Rad, Hercules, CA)
and probed with rabbit GN-2640 (1:75). The blot was then probed with horseradish peroxidase-
conjugated donkey anti-rabbit IgG (1:3000, Amersham Biosciences, Buckinghamshire,
England), and processed using ECL (enhanced chemiluminescence) plus (Amersham Pharmacia
Biotech) according to the manufacturer’s instruction.
Inhibition of nucleic acid channel activity in brain slices.
Brain slices were prepared as above. The nucleic acid conducting channels in the brain slices
were blocked either with specific polyclonal antibody, GN-2640, or with pre-immune serum at
4ºC for 2 hr at the indicated concentration. Alternatively, the channels were blocked with
heparan sulfate or L-malic acid at a concentration range of 20 to 200µg/ml for 1 hr. Then one
nmol ODNs, suspended in a volume of 50ul, were added to the brain slices, and after 30min the
uptake of ODNs was analyzed by confocal microscopy.
Chapter 6
120
Bolus injection of ODNs into rat brain in vivo.
A guide cannula was implanted into the right nucleus accumbens of Wistar rats. FITC-ODNs
(0.1mM) diluted with 4.5% NaCl (4:1) was infused for 4 minutes into the nucleus accumbens
(0.75µl/min) through a cannula after a brief anesthesia with isoflurane. After 4 or 24 hours the
animals were decapitated, and the brain was dissected and frozen on dry ice. Subsequently,
cyosections (20µm) were cut, collected and evaluated by fluorescence microscopy. Alternatively,
the cannula was inserted into the lateral ventricle of adult Wistar rats under anesthesia. 5ug
pGFP alone or complexed with 100nmol SAINT-2/DOPE in a volume of 15µl were bolus
injected into ICV via a cannula (2 animals). GFP expression was examined after 48 h.
Results and Discussion
Uptake of ODNs by brain slices does not depend on cationic lipids
Free ODNs can be endocytosed by cultured cells and most of them end up in endocytic
compartments. By contrast, when complexed with cationic lipids, the uptake of ODNs is not only
enhanced but in this case translocation of ODNs from the endosomes into the cytosol occurs,
which leads to nuclear uptake, a prerequisite for downregulation of protein targets. To verify our
previous data on the cellular uptake of ODN lipoplexes and the functional down regulation of
membrane receptors in neuronal cells in vitro, the next challenge was to investigate biological
consequences of a similar approach in vivo. Therefore, we studied the uptake properties of ODNs
by brain slices in the presence and absence of cationic lipids. In preliminary experiments, using
freshly prepared brain slices, we observed a substantial internalization of FITC-labeled ODNs,
even in the absence of SAINT-2/DOPE liposomes, and its diffuse distribution in cytosol and
nucleus. Accordingly, taking our previous data obtained in cell cultures into account, these
observations raised concern as to the integrity of the plasma membrane of the relevant cells. To
exclude therefore that ODN uptake could be due to damage of the plasma membrane during slice
preparation, the integrity of the membrane was verified by Hoechst staining. As shown in
Fig.1A, when added to brain slices after their fixation with 3% paraformaldehyde, which is used
as a positive control, Hoechst permeated into virtually every cell in the slice. When stained
immediately after isolation, damage of the brain slices is particularly apparent at the periphery of
the slices (Fig.1 B). After 4 days of recovery in culture, the damage to the brain slices was
essentially minimal and only occasional penetration of the Hoechst dye could be detected (Fig.1
C). Therefore, all further experiments on the brain slices were carried out with preparations that
had been cultured for four days. When such slices were treated with FITC-labeled ODNs and
ODNs complexed with cationic lipids (SAINT-2/DOPE, 1:1) for 4 hr, qualitative analysis of the
Oligos enter brain cells through a nucleic acid channel
121
specimen by microscopy revealed that the uptake of ODNs per se (Fig.1 D) or ODNs complexed
with cationic lipids (Fig.1 E) was almost identical. Furthermore, ODNs were taken up through
the whole slice as readily determined by serial scanning by confocal microscopy (date not
shown), while the uptake of ODNs continued to increase when monitored over a time interval of
24 hr (Fig.1 F). Hence, these data not only imply that cationic lipids did not significantly
improve the uptake of ODNs by the brain slices, but rather that ODNs were internalized by cells
present in the slices without a need for a vector for internalization and, presumably, for their
release from endocytic compartments. A priori, having demonstrated previously that
endocytosed ODNs remain trapped within the endocytic compartments, these data could suggest
that ODNs taken up by the cells in the brain slices did not substantially enter the cells by
endocytosis. Rather direct access into the cytosol might have been acquired by direct
translocation across the plasma membrane. Having thus acquired access to the cytosol, the ODNs
will then readily accumulate in the nucleus via diffusion through the nuclear pore (12, 13). It was
therefore important to more carefully define the intracellular localization of internalized ODNs,
incubated with the slices as such or when complexed with cationic lipids.
Free ODNs are internalized by brain cells and acquire access to the nucleus
As demonstrated in Fig. 2, when FITC-ODNs alone (Fig.2A, green) or complexed with
cationic lipids (Fig.2 C) were applied to the brain slices, more than 50% of the cells showed high
uptake of ODNs in the cytosol and/or nuclei. Accumulation of free ODNs in the nuclei (yellow)
could be clearly discerned when the latter had been counterstained with propidium iodide (red)
(Fig.2A). Occasionally, an almost exclusive localization in the cytosol was seen, while the ODNs
had not yet reached the nucleus (insert Fig. 2A). Overall, the uptake of ODNs did not show a cell
type-specific preference, as inferred from ODN internalization versus morphology of the cells in
the brain slices.
Chapter 6
122
Fig. 1 Uptake of ODNs by brain slices. 400nm brain slices were prepared as describedin Methods. To evaluate their plasma membrane integrity, the slices were incubatedwith the membrane permeating dye Hoechst (20µg/ml) at various conditions. A. Thebrain slice was fixed and permeabilized, and stained with Hoechst, the sample servingas a positive control. B. A brain slice was stained with Hoechst immediately afterpreparation C. A brain slice was kept in culture for four days and then stained withHoechst. Note that after 4 days in culture only few plasma membrane damaged cellswere apparent. D. Free FITC-ODNs (1 nmol) were incubated with the brain slice for 4hr. E. FITC-ODNs (1 nmol), complexed with cationic lipids (75 nmol), were incubatedwith the brain slice for 4 hr. Note that visual inspection reveals that the uptake of freeODNs or ODNs complexed with cationic lipids is qualitatively very similar. F. FreeFITC-ODNs (1 nmol) were incubated with brain slices for 24 hr and a gradual increasein uptake of ODNs was apparent.
When complexed with cationic lipids, efficient delivery of labeled ODNs into cells in the
brain slices and their localization into cytosol and nucleus, were also readily observed (Fig.2C).
N-Rh-PE-labeled cationic liposomes were rapidly taken up by the brain slices and they appeared
as fine dots in intracellular compartments without showing any diffusion in the cytoplasm or
nucleus (Fig. 2 B). This would suggest that the cells most likely internalized the liposomes via an
endocytic internalization mechanism, the particles becoming entrapped in endosmal/lysosomal
compartments, consistent with previous observations (3, 14). Note that FITC-labeled ODNs do
escape from the complexes and reach the cytosol prior to reaching the nucleus, without the lipids
reaching this organelle, since red fluorescence could not be detected in the nucleus (Fig. 2 B and
C). Taken together, these data, in conjunction with those presented in Figure 1 clearly indicate
that ODNs, irrespective of their complexation with cationic lipid complexes, readily acquire
B CA
FD
alone
E
complexes
4h 4h 24h
Oligos enter brain cells through a nucleic acid channel
123
Fig. 2 Cellular uptake of FITC-ODNs and cationic lipids, visualized at highmagnification. Free FITC-labeled ODNs (A, 1 nmol) or complexe with cationicliposomes (C, 75 nmol), labeled by the fluorescent lipid analogue N-Rh-PE wereincubated with brain slices as described and examined by fluorescence microscopy.A. The nuclei were counterstained with propidium iodide (2.5 µg/ul, red). Note that aprominent localization of ODNs was seen in both the cytosol (green) and nuclei(yellow), whereas in some cells ODNs appeared only in the cytosol as diffusefluorescence (insert in A; nuclei were not counterstained). B. Localization of cationiclipids, traced with N-Rh-PE (red), following an incubation of the liposomes with brainslices for 4 hr. The cationic liposomes, reflected by N-Rh-PE fluorescence, appeared asfine dots in the cytosol, presumably reflecting their localization in the endocyticpathway, without any significant diffuse fluorescence in the cytoplasm. C. FITC-ODNswere complexed with N-Rh-PE labeled cationic liposomes and were subsequentlyincubated with the brain slice for 4 hr at 37°C. Note that the ODNs (green)accumulated in the nuclei, while the cationic lipids (red) remained in the cytosol, someintact complexes (yellow) remaining clustered at the extracelluar matrix.
Fig. 3 Uptake of free ODNs can occur at 4 oC. The internalization of FITC-ODNs bybrain slices was carried out at 4˚C (A) or 37 ˚C (B) as described in Methods for anincubation period of 2hr. Note that although it was less than at 37 ˚C (B), at 4˚C aprominent internalization of ODNs could be seen. The data imply that a mechanismother than endocytosis is involved in ODN internalization in brain slices.
A B C
A
B
Chapter 6
124
access to cells present in brain slices to very similar extents. Moreover, free ODNs do not
become entrapped in the endocytotic pathway, but rather show a very similar distribution, i.e., in
the cytosol and nucleus (Fig. 2) as that observed when delivered via cationic lipids. In relation to
the notion that ODNs, present in endosomes do not substantially leak into the cytosol, the data
strongly support the view that free ODNs, when incubated with brain slices must acquire
cytosolic access by direct translocation across the plasma membrane, which is in marked contrast
to an almost exclusive endocytic uptake of ODNs, when added to cells in culture. Accordingly,
in order to translocate across the plasma membrane, a membrane localized transporter, present in
cells that constitute the brain slices, would be a likely mechanism, given that permeation across
perturbed plasma membranes can be excluded (see above). In fact, recently the presence of such
nucleic acid channels has been demonstrated to be localized in the plasma membrane of rat
kidney cells (15, 16). The potential presence of such a channel in brain cells was therefore
examined.
Entry of ODNs into cells in brain slices can be accomplished via nucleic acid channels
If direct uptake across the plasma membrane of brain cells would occur, and if endocytosis
would not significantly contribute to ODN uptake, as characterized in the previous section, the
effect of temperature might discriminate between either mechanism, endocytosis showing a
prominent temperature dependence (17-19). Interestingly, when brain slices were incubated with
ODNs at 4�C for 2 hours, a substantial uptake could still be seen (Fig. 3 A), although it was less
than that at 37�C (Fig. 3 B). Nevertheless, the experiment shows ODN uptake to occur at
conditions, i.e., at 4 oC, where endocytosis is effectively inhibited, implying that entry into cells
must take place via direct passage across the plasma membrane.
Fig. 4 Expression of the nucleic acid channel inrat brain. The expression of the nucleic acidchannel was examined with immuno-Westernblot, as described in Methods. 4, 9 and 15 µg ofthe lysate of rat brain was separated and blottedon PAGE. One band at 83 kDa was detected,using the specific antibody GN-2640 in a dose-dependent manner.
Further support for a non-endocytic pathway of entry of free ODN by brain slices was derived
from inhibition experiments using heparan sulfate. This large polyanion is similar in molecular
175836247
4 9 15 µg
Oligos enter brain cells through a nucleic acid channel
125
weight and valency to oligonucleotides, and can compete with ODNs for binding sites on the cell
surface, causing an inhibition of endocytosis of ODNs (20). A dose-dependent inhibition of the
uptake of ODNs by heparan sulfate could be seen (Fig. 5 E and F), although the inhibition was
modest. Thus at the highest dose used in this study (200 µg/ml; Fig. 5 F), the inhibition of uptake
was approximately 30 % (n=3). Together, the uptake at low temperature and the (modest)
inhibition of ODN internalization by brain slices at 37 oC in the presence of heparan sulfate are
in line with a non-endocytic entry pathway of free ODNs in brain cells. Rather, taking also into
account the intracellular distribution, a direct crossing of ODNs across the plasma membrane
appeared the most likely path of entry. It was therefore of interest to investigate whether such
ODN transfer could occur via a plasma membrane-localized channel. Recently, such channels
have been identified for the first time in rat kidney cell membranes, but whether such channels
are ubiquitously distributed is unknown.
Fig. 5 Modulation of ODN uptake in brain slices, following pretreatment with ODNchannel antibodies. One nmol FITC-ODNs was incubated with a brain slice for 30min(A). Before addition of the same amount of FITC-ODNs, the brain slice was first treatedwith pre-immune serum (B) or with the channel antibody GN-2640, at a dilution ofantibody of 1:10 (C) or 1:100 (D). Note that a strong inhibition of the uptake of ODNscould be seen in C and D, in which the nucleic acid channel had been pre-blockedwith the channel specific antibody whereas the pre-immune serum (B) was withouteffect. E and F. The brain slices were pre-incubated with heparan sulfate before theaddition of the same amount of FITC-ODNs. Note that the inhibitory effect of heparansulfate was dose dependent, showing no effect at 20 ug/ml ( E ), while at 200 ug/ml,an inhibition of approx. 30 % was apparent (F ). G and H. The brain slices weretreated with L-malate prior to the addition of ODNs (1 nmole). L-Malate acid (G. 20ug/ml; H. 200ug/ml) did not interfere with the uptake of ODNs by the brain slices(compared to A).
A B C D
E F G H
Chapter 6
126
The nucleic acid channel is a heteromultimeric complex, which consists of at least two protein
subunits, i.e., a 45-kDa pore forming subunit (p45) and a 36-kDa regulatory subunit (p36). The
latter has been identified as a cytosolic malate dehydrogenase (cMDH), which seems to confer
the selectivity of the pore protein to ODNs. Both phosphodiester and phosphothioate ODNs can
be translocated via this channel (15). To investigate the potential presence of this channel in
brain cells, a cellular lysate of rat brain was prepared and the sample was analyzed by Western
blotting, using an antibody, GN-2640, raised against a peptide of the 45-kDa pore-forming
subunit, as detected in rat kidney. As shown in Fig. 4, the antibody staining clearly revealed the
presence of two protein bands in a dose-dependent manner, one showing a molecular weight of
at 83kD. It is tempting to suggest that these bands presumably represent hetero-dimers of the
putative ODN channel protein, which may thus bear a remarkable similarity with the
heteromultimeric complex identified in rat kidney. To obtain further support for this notion, we
next examined whether the channel's activity could be blocked upon binding of the antibody.
Interestingly, an antibody concentration dependent inhibition of the uptake of ODNs was
observed (Fig. 5 C, D). As shown in Figure 5B, relative to control uptake of ODNs, no effect
was seen when the brain slices had been preincubated with pre-immune serum. By contrast, an
inhibition of ODN uptake by more than 50% occurred, as determined by fluorescence
measurements in three independent experiments, when the slices had been preincubated with
GN-2640 antibodies at 1:10 dilution (Fig. 5 C). Hence, these data imply that a functionally active
ODN translocator is present on brain cells, as localized in brain slices.
L-malate, which is a substrate for cytosolic malate dehydrogenase, has been identified as a
regulatory subunit on the nucleic acid channel, and upon reconstitution of the channel in
liposomes, L-malate was shown to block the reconstituted activity (16). However, when L-
malate was preincubated with brain slices at concentrations ranging from 20 to 200µg/ml prior to
addition of the ODNs, no significant effect on ODN uptake was apparent in this case (not
shown). Hence, unlike an effective inhibition by the antibody raised against the pore-forming
subunit no effect was seen when modulating the regulatory subunit of this channel with its
substrate. However, the apparent discrepancy may be explained by taking into account that the
L-malate-induced inhibition of the channel activity was reported for the liposomally
reconstituted protein (16). Not unlikely, at such conditions, at least part of the channels may be
randomly reconstituted implying a partly exposure of the cMDH such that it may react with L-
malate, when the latter is added to the incubation medium. Using brain slices, like in our
Oligos enter brain cells through a nucleic acid channel
127
experiments, exogenously added L-malate will not readily gain access to cMDH, which may
well be hidden in the channel complex, facing in vivo more likely the cytosolic site of the plasma
membrane.
Plasmid uptake does not occur via ODN channels and requires cationic-lipid mediated
delivery.
Since the data obtained thus far strongly suggest the presence of an ODN transfer channel in
the plasma membrane of brain cells, as detected in brain slices, we next examined whether
plasmids could also be translocated via the channel protein. Thus, rhodamine-labelled plasmids
encoding GFP were incubated with brain slices, either at 4�C or 37�C. However, except for some
fuzzy fluorescence at the cell's surface, no significant internalization of plasmid was apparent at
either condition, as illustrated in Fig.6A for the interaction at 37�C. Consistent with this
observation, neither was the expression of GFP fluorescence, acting as a reporter gene, apparent
at either condition (Fig.6C). This would imply that the nucleic acid channel is not adapted in
carrying large size plasmids across the plasma membranes of brain cells. However, substantial
internalization of rhodamine-labelled plasmids could be detected when complexed with cationic
liposomes, and a subsequent incubation with the brain slices at 37�C (Fig. 6 B). As a result, a
relative high GFP expression was seen at the periphery of the slice, culminating in the
transfection of approximately 100 cells per slice (Fig. 6 D). Taken together, while relatively
small ODNs are readily translocated via ODN channel proteins in brain cells present in brain
slices, the internalization of larger sized plasmids requires cationic lipids for delivery at elevated
temperature. This suggests, as demonstrated before for in vitro cultures of a variety of cells,
including neuronal cell lines, that also in cells in brain slices endocytic entry, followed by
permeation across endosomal membranes represents the entry pathway for lipoplexes.
The remarkable observation that ODNs, by contrast, might enter brain cells in slices via an
ODN channel protein, raised the intriguing question whether entry in vivo could be accomplished
by a similar mechanism. Hence to further evaluate the extent to which in vitro versus in vivo data
could be extrapolated and compared, we carried out similar experiment in vivo.
Similar mechanisms operate for ODN and plasmid entry into brain slices and brain tissue
in vivo.
To determine whether ODN uptake in brain slices reflected an analogous mechanism by which
ODNs could be taken up in vivo as well, FITC-ODNs were locally infused into the right nucleus
Chapter 6
128
accumbens in rats. After 4 hr, massive uptake of FITC-ODNs in cells localized near the point of
injection was observed (Fig. 7 A arrow). Most interestingly, FITC-ODNs were not entrapped in
vesicular compartments within the cell, but rather, showed a fairly diffuse appearance. This
indicates that endocytosis is not the only mechanism for the uptake of ODNs in rat brain. This
observation is consistent with the uptake of ODNs by the brain slices but contrasts with in vitro
uptake by cell culture systems. Over a period of another 24 h, the number of cells that showed
incorporation of ODNs further increased (Fig. 7 B), although the region that became stained was
largely restricted to that near the injection site, implying that further penetration of ODNs into
the brain was rather poor. In contrast to the relatively rapid translocation of free ODNs into brain
cells in vivo, plasmids did not penetrate when infused in free form (data not shown). By contrast
but in line with the observations on brain slices, pGFP complexed with cationic lipids transfected
around 100 cells near the injection site (Fig. 7 C). Thus in terms of a requirement for a vector
and the efficiency of transfection, the brain explant system and in vivo brain tissue displayed a
remarkable similar behavior.
In summary, the data presented in this work demonstrate that brain cells, both in vivo and in
brain explants, contain nucleic acid channel proteins in their plasma membranes, which
effectively carry free ODNs directly into the cytosol. This direct translocation facilitates their
homing into the nucleus, a prerequisite for their ability to express antisense activity. In previous
work, we observed that cultured cells do not express such an activity, delivery requiring a vector
such as cationic lipids, which allows cytosol-nuclear transfer of ODNs after endocytic
internalization and subsequent perturbation of endosomal membrane integrity. At present it is
unclear why brain cells in the slices apparently express the channel protein, like in vivo, whereas
neuronal cells in culture, such as raphe cells, RN46 and septal neurons, SN48, do not (1). Yet for
therapeutic purposes, the consequence of the presence of a channel for ODNs but not plasmids,
as shown here, could be of potential interest. Although antisense ODNs hardly cross the blood
brain barrier, they readily accumulate in brain tissue when directly injected into rat brain. Thus
when locally applied, specific antisense molecules may readily acquire access to brain cells for
downregulation of specific target proteins, the specificity being dictated by the target, rather than
by the nature of the brain cells. On the other hand, if the latter, i.e, specific cellular targeting is
required, the use of an appropriately targeted vector will be unavoidable. As shown in the present
work, also vector-mediated delivery can be readily accomplished in brain slices and in in vivo
brain tissue, although an apparent limitation is clearly the restriction in diffusion of the injected
sample from the injection site.
Oligos enter brain cells through a nucleic acid channel
129
Fig. 6 The uptake and transfection of pGFP in brain slices. Brain slices were treatedwith Rhodamine-labeled pGFP alone or when complexed with cationic lipids, preparedas described in the Methods. The transfection efficiency was examined 6 days aftertreatment. When pGFP alone was applied onto the brain slice, no significant uptake ofrhodamine-labelled plasmid (red) could be observed (A) and no GFP transfection couldbe detected (C). When pGFP complexed with cationic lipids was applied, the uptake ofrhodamine-labeled pGFP could be observed (B, red) and around 100 cells, localizednear the periphery of the slice, expressed GFP (D, green).
Fig. 7 Infusion of ODNs or pGFP into rat brains in vivo. FITC-ODNs were infused intorat brains as described in Methods. The rat was sacrificed 4 or 24 hr after infusion,and brains were sectioned. The pictures were taken from areas that were near theinfusion point. Note that after 4 hours (A) FITC-ODNs had prominently accumulatedwithin the cells (arrow) and were also found in association with the extracellular space(arrow head). After 24 hr the pool of extracellular ODNs (arrow head, B) haddecreased, while concomitantly an increase in cellular uptake of ODNs (arrow) wasapparent. C. pGFP complexed with cationic lipids was infused into the brain of the ratas described in Methods. The GFP expression was examined after two days.Substantial expression, i.e., around 100 transfected cells, were seen directly adjacentto the injection site. No pGFP expression could be detected when pGFP alone wasapplied (data not shown).
A B
DC
A B C
Chapter 6
130
In this context it is interesting to note that in terms of efficiency cationic lipid mediated
delivery of ODNs is nearly identical to the uptake of free ODNs in the brain slices. There are
several possibilities that could account for the similarity in uptake at either condition. First, the
entire complex of cationic lipid and ODNs is endocytosed, followed by the release of ODNs
from endosomes and the nucleotide’s accumulation in the nucleus (Fig.2C), which has been
well-documented for in vitro uptake by cell cultures (3, 21). However, we also observed that
cationic complexes were poorly penetrating, whereas ODNs were seen throughout the slice (Fig
1. E). Accordingly, it is also possible that ODN complex/plasma membrane interaction may
cause a destabilization of the complexes by charge neutralization and induce the release of
ODNs from the complexes. Dissociated ODNs may diffuse into the slice and subsequently enter
the cells through the nucleic acid channel. The possibility of the latter scenario is suggested by
observations that the cationic lipids were mostly seen at the surface or at the periphery of the
slice (Fig. 2 B, C).
Evidently, these and other issues require further work. Yet, the unanticipated presence of
nucleic acid channels on brain cells raises novel challenges and opportunities in the development
of therapeutic strategies in the treatment of brain diseases. In this context, it will be of particular
interest to develop devices that cross the blood brain barrier so that ODN-based approaches
could rely on non-invasive protocols.
AcknowledgementsWe gratefully acknowledge Dr. Basil Hanss (Division of Nephrology, Mt. Sinai School ofMedicine, New York, U. S. A.) for providing us GN2640 antibody and helpfuldiscussion on the nucleic acid channel. This work was supported by a grant from TheNetherlands Organization for Scientific Research (NWO)/NDRF Innovative DrugResearch (940-70-001). Anno Wagenaar and Professor Jan Engberts are thanked forhelpful discussions and for synthesis of SAINT-2.
Oligos enter brain cells through a nucleic acid channel
131
References:1. Shi, F., Visser, W. H., de Jong, N. M. J., Liem, R. S. B., Ronken, E. and Hoekstra,
D (2003) Antisense oligonucleotides reach mRNA targets via the RNA matrix;downregulation of the 5-HT1A receptor. Exp. Cell Res. in press.
2. Lorenz, P., Baker, B. F., Bennett, C. F. and Spector, D. L. (1998). Phosphorothioateantisense oligonucleotides induce the formation of nuclear bodies. Mol. Biol. Cell 9:1007-1023.
3. Shi, F., Nomden, A., Oberle, V., Engberts, J. B. and Hoekstra, D. (2001). Efficientcationic lipid-mediated delivery of antisense oligonucleotides into eukaryotic cells:down-regulation of the corticotropin-releasing factor receptor. Nucleic. Acids Res.,29: 2079-2087.
4. DeLong, R. K., Yoo, H., Alahari, S. K., Fisher, M., Short, S. M., Kang, S. H., Kole,R., Janout, V., Regan, S. L. and Juliano, R. L. (1999) Novel cationic amphiphiles asdelivery agents for antisense oligonucleotides. Nucleic Acids Res., 27: 3334-3341.
5. Blume, G., Cevc, G., Crommelin, M. D., Bakker-Woudenberg, I. A., Kluft, C. andStorm, G. (1993) Specific targeting with poly(ethylene glycol)-modified liposomes:coupling of homing devices to the ends of the polymeric chains combines effectivetarget binding with long circulation times. Biochim. Biophys. Acta, 1149: 180-184.
6. Maruyama, K., Takahashi, N., Tagawa, T., Nagaike, K. and Iwatsuru, M. (1997)Immunoliposomes bearing polyethyleneglycol-coupled Fab' fragment showprolonged circulation time and high extravasation into targeted solid tumors invivo. FEBS Lett., 413: 177-180.
7. Chauhan, N. B. (2002) Trafficking of intracerebroventricularly injected antisenseoligonucleotides in the mouse brain. Antisense Nucleic Acid Drug Dev., 12: 353-357.
8. Shi, N., Boado, R. J. and Pardridge, W. M. (2000) Brain-specific expression of anexogenous gene after i.v. administration. Proc. Natl. Acad. Sci. U. S. A., 97: 14709-14714.
9. Penichet, M. L., Kang, Y. S., Pardridge, W. M., Morrison, S. L. and Shin, S. U.(1999) An antibody-avidin fusion protein specific for the transferrin receptor servesas a delivery vehicle for effective brain targeting: initial applications in anti-HIVantisense drug delivery to the brain. J. Immunol., 163: 4421-4426.
10. Zhang, S. P., Zhou, L. W., Morabito, M., Lin, R. C. and Weiss, B. (1996) Uptake anddistribution of fluorescein-labeled D2 dopamine receptor antisenseoligodeoxynucleotide in mouse brain. J. Mol. Neurosci., 7: 13-28.
11. Meekel, A.A.P., Wagenaar, A., Smisterova, J., Kroeze, J., Haadsma, P., Bosgraaf,B., Stuart, M.C.A., Brisson, A., Ruiters, M. H. J., Hoekstra, D. and Engberts, J. B.(2000). Synthesis of pyrimidinium amphiphiles used for transfection and somecharacteristics of amphiphile/DNA complex formation. Eur. J. Org. Chem., 2000:665–673.
12. Shoeman, R. L., Hartig, R., Huang, Y., Grub, S., Traub, P. (1997). Fluorescencemicroscopic comparison of the binding of phosphodiester and phosphorothioate(antisense) oligodeoxyribonucleotides to subcellular structures, includingintermediate filaments, the endoplasmic reticulum, and the nuclear interior.Antisense Nucleic. Acid. Drug Dev. 7: 291-308.
13. Hartig, R., Shoeman, R. L., Janetzko, A., Grub, S. and Traub, P. (1998) Activenuclear import of single-stranded oligonucleotides and their complexes with non-karyophilic macromolecules. Biol-Cell, 90: 407-426.
14. Zuhorn, I. S., Kalicharan, R. and Hoekstra, D. (2002). Lipoplex-mediatedtransfection of mammalian cells occurs through the cholesterol-dependentclathrin-mediated pathway of endocytosis. J. Biol. Chem. 277: 18021-18028.
Chapter 6
132
15. Hanss, B., Leal-Pinto, E., Bruggeman, L. A., Copeland, T. D. and Klotman, P. E.(1998) Identification and characterization of a cell membrane nucleic acid channel.Proc. Natl. Acad. Sci. U. S. A., 95: 1921-1926
16. Hanss, B., Leal-Pinto, E., Teixeira, A., Christian, R. E., Shabanowitz, J., Hunt, D.F. and Klotman, P. E. (2002) Cytosolic malate dehydrogenase confers selectivity ofthe nucleic acid-conducting channel. Proc. Natl. Acad. Sci. U. S. A., 99: 1707-1712.
17. Drin, G., Cottin, S., Blanc, E., Rees, A. R. and Temsamani, J. (2003) Studies on theinternalization mechanism of cationic cell-penetrating peptides. J. Biol. Chem. 278:31192-31201.
18. Fittipaldi, A., Ferrari, A., Zoppe, M., Arcangeli, C., Pellegrini, V., Beltram, F. andGiacca, M. (2003) Cell membrane lipid rafts mediate caveolar endocytosis of HIV-1Tat fusion proteins. J. Biol. Chem., 278: 34141-34149.
19. Le, P. U. and Nabi, I. R. (2003) Distinct caveolae-mediated endocytic pathwaystarget the Golgi apparatus and the endoplasmic reticulum. J. Cell Sci., 116: 1059-1071.
20. Benimetskaya, L., Loike, J. D., Khaled, Z., Loike, G., Silverstein, S. C., Cao, L., el-Khoury, J., Cai, T. Q. and Stein, C. A. (1997) Mac-1 (CD11b/CD18) is anoligodeoxynucleotide-binding protein. Nat. Med. 3: 414-420.
21. Zelphati, O. and Szoka, F. C. Jr (1996) Mechanism of oligonucleotide release fromcationic liposomes. Proc. Natl. Acad. Sci. U. S. A., 93: 11493-11498.
133
Chapter 7
Cationic liposome-mediated delivery of proteins intoeukaryotic cells: entry along the pathway of caveolae-
mediated endocytosis
Fuxin Shi1, Xuedong Yan2 and Dick Hoekstra1
Department of Cell Biology, sections Membrane Cell Biology1 and Liposome Research2,Faculty of Medical Sciences, University of Groningen; Antonius Deusinglaan 1, 9713 AV
Groningen, The Netherlands
Submitted
Chapter 7
134
Abstract
For therapeutic or cell biological purposes, direct delivery of peptides and proteins within cells
may represent a useful alternative for delivery of plasmids and/or antisense oligonucleotides for
similar purposes, the latter requiring gene expression and effective downregulation of mRNA
expression, respectively. Here we have investigated whether liposomes prepared from the
cationic lipid SAINT-2 and dioleoylphosphatidylethanolamine (DOPE), representing potent
carriers for nucleic acids, could also be applied as protein carriers. Using a variety of proteins,
including an antibody and the enzyme β-galactosidase, we demonstrate that proteolipid particles
can be readily formed as a result of charge neutralization when liposomes and proteins are
mixed. Our data show that the proteolipid particles are effectively internalized by the cells along
a pathway that does not involve endocytosis, as revealed by the ineffectiveness of potassium-
depletion and endocytosis mutants to interfere with particle internalization. Rather, entry via
caveolae appears a major route of cellular penetration for the proteolipid complexes, as
suggested by colocalization studies with a lipid marker that associates with caveolae. Following
internalization, the proteins appear functionally intact, as indicated by the expression of
intracellularly introduced β-galactosidase activity. The caveolar pathway, which appears to be
governed by the size of the particles, being in the order of 700 nm, may be advantageous since it
would avoid the potential rapid degradation of internalized proteins in the endosomal/lysosomal
pathway.
Caveolae pathway of lipoprotein particles
135
Introduction
Cationic lipids are widely employed to deliver plasmids and antisense oligonucleotides into
eukaryotic cells to either express a desired protein or modulate intrinsic protein expression,
respectively, thereby contributing towards the understanding of protein functioning and/or the
development of appropriate (gene) therapeutics. However, for many purposes including
therapeutics and for reasons of fundamental cell biological interest, it would be equally
appealing to directly deliver proteins into cells. Thus far, this technology has been relatively
poorly developed and evaluated.
Peptide carriers such as HIV-1 Tat, which can be genetically or chemically hybridized to
proteins, have been reported to mediate intracellular delivery of various oligopeptides and
proteins as large as 120 kDa (1, 2). Also, peptides such as the antennapedia homeodomain and
arginine-rich peptides display similar properties (3-5). A common feature of these peptide
carriers, which are largely derived from protein transduction domains of viruses, is that they are
highly basic and hydrophilic. The mechanism of internalization of such peptides does not
involve endocytosis, as cellular entry is not impeded at 4 °C (5, 6). However, an obvious
limitation of such carriers is that they require crosslinking to the target peptide or protein, which
is not suitable for many proteins without severely compromising their structural or functional
properties. This is particularly the case when, as for the TAT peptide, denaturation of the protein
may be needed to enhance the accessibility of the TAT domain in order to facilitate delivery (7).
To eliminate the need for cross-linking the artificial peptide Pep-1 has been synthesized, which
consists of three domains. These include (i) a hydrophobic domain capable of interacting with
the to-be-delivered protein, (ii) a hydrophilic domain derived from simian virus 40 to improve
the intracellular delivery and the solubility of carrier peptide, and (iii) a spacer domain which
connects the two domains (8). Not unexpectedly, the mechanism of Pep-1-mediated delivery is
similar to that of delivery via the TAT-peptide, the entry occurring independent of endocytosis.
In general, the success of peptide-mediated delivery critically relies on the exposure of the
peptide in the associated or hybridized protein complex, which is obviously necessary for
appropriate interaction with the cell’s surface.
Also liposomes of various composition have been exploited as delivery vehicle for proteins (9,
10), for example for the purpose of vaccine development (11-14). Rather than via cross-linking,
the interaction of the protein with liposomes usually relies on hydrophobic interactions or charge
attraction. The latter is in particular the case when using cationic lipid delivery systems (15),
although delivery of net-positively charged proteins is poorly accomplished with such systems.
Also, little is known about the mechanism of cationic lipid-mediated delivery of proteins. When
Chapter 7
136
complexed with DNA, such carriers have been reported to be processed along the pathway of
receptor-mediated endocytosis (16-19). In previous work, we have shown that a cationic lipid
complex, consisting of a pyridinium amphiphile SAINT-2 effectively delivers chromosomes,
plasmids, and oligonucleotides into eukaryotic cells (20-22). This prompted the current study in
which we examined whether this formulation is also capable of delivering intact proteins into
cells. Our data reveal the effective delivery of a variety of proteins along a pathway, which
appears to rely on caveolar-mediated entry. This mode of entry is of particular interest, since in
this manner a means might be provided for the delivered protein to escape potential breakdown
in the lysosomal degradation pathway.
Material and Methods
Materials.
The cationic lipid SAINT-2(N-methyl-4(dioleyl) Methylpyridiniumchloride) was synthesized
as described in detail elsewhere(23) . Dioleoyl-phosphatidylethanolamine (DOPE) was
purchased from Avanti Polar Lipids(Alabaster, USA). β-Galactosidase and FITC-dextran (M.W
9,400 and 71,600) were purchased from Sigma (Missoursi, USA), and Cy3-antibody (goat anti-
mouse) was purchased from Jackson Immuno Research (West Grove PA, USA). All other
chemicals were from Sigma (Missoursi, USA), unless stated otherwise.
Cell culture.
Chinese hamster ovary (CHO) cells were grown in CHO-S-SFM medium (Gibco)
supplemented with 10% heat inactivated fetal calf serum, 2mM L-glutamine and penicillin
(50U/ml) /streptomycin (50µg/ml) in 5% CO2/95% air at 37�C.
Preparation of cationic liposomes.
The lipids were dissolved in chloroform/methanol(1:1, volume ratio). SAINT-2 and DOPE
were mixed (1:1, molar ratio), and the solvent was removed by evaporation under a stream of
nitrogen, followed by placing the vial under vacuum for at least 1hr. The lipids were then
resuspended in Millipore water at stock concentrations of 1 mM, and sonicated to clarity in a
bath sonicator in a closed vial.
Preparation of cationic proteolipid particles.
Various amounts of proteins and cationic liposomes, suspended in a total volume of 50µl,
were diluted with an equal volume of culture medium, and then gently mixed. Assembly of the
proteolipid particles was monitored by fluorescence microscopy and by native protein gel
electrophoresis. 0.4 µg Cy3-antibodies were mixed with 2 to 20 nmol SAINT-2/DOPE in 50 µl
Caveolae pathway of lipoprotein particles
137
culture medium for 10 min at room temperature. They were applied to a microscope slide and
examined by fluorescence microscope (Olympus, Japan). One microgram of β-galactosidase,
goat antibody and BSA alone or complexed with 20nmol cationic lipids was loaded onto the
native 12.5% Tris glycine gel and run at 80 V, using Tris glycine as running buffer (250mM
Tris, 250 mM glycine). The gel was stained with Coomassie Blue (Bio-Rad).
Particle size and zeta potential of the proteolipid particles (1 µg protein and 100 nmol SAINT-
2/DOPE) was measured with a Nicomp Model 380 Submicron Particle Analyzer System (Santa
Barbara, CA, USA).
Intracellular delivery of protein.
CHO cells were grown on coverslips in a 12-well plate to 70% confluence. The cells were
washed with HBSS (Gibco), and incubated with proteolipid particles for the indicated time-
intervals. The cells were washed twice, followed by β-galactosidase staining or analysis by
confocal microscopy (TCS Leica SP2 confocal laser scanning microscope, Wetzlar, Germany).
β-galactosidase staining.
The β-galactosidase assay was performed using chlorophenol red-D-galactopyranoside as the
substrate. The enzyme activity was determined spectrophotometrically. The cells from a 12-well
plate were lyzed with 500 µl Lysis buffer (Promega). In a 96-well plate, 15 µl cell lysate was
diluted to 100 µl PBS/0.5%BSA and incubated with 150 µl substrate (1mg/ml chlorophenol red
galactopyranoside in a buffer consisting of 60mM sodium dibasic phosphate buffer, pH 8, 1mM
magnesium sulfate, 10mM KCl and 50 mM β-mercapto-ethanol). The activity of β-galactosidas
was measured at 550nm. The intracellular enzyme activity was examined with X-gal staining.
Briefly, cells were washed, fixed with 0.5% glutardialdehyde in PSB for 5 min at room
temperature and washed again. Then cells were incubated for 2 hr at 37˚C with X-gal staining
solution (10 mg X-gal in 0.5 ml dimethylformamide, 9.5 ml staining buffer, consisting of 0.01%
sodium desoxycholate, 0.02%Nonidet P40, 1mM MgCl2, 5mM K3[Fe(CN6) ],
5mMK4[Fe(CN6)]). Then X-gal solution was removed, and the cells were washed twice with
HBSS and examined by microscopy (Olympus, Japan).
Caveolae staining.
CHO cells were incubated with 0.4µg Cy3-antibody/20nmol SAINT-2/DOPE particles for 4
hours at 37˚C, washed 5 times with HBSS, and then chased for 10 min at 37˚C to allow plasma
membrane bound particles to be internalized. Then the cells were incubated with the caveolae
marker BODIPY-LactosylCeramide(1 µM; Molecular Probes) for 30min at 37˚C, and potential
colocolization of antibody and lipid was determined by confocal microscopy.
Chapter 7
138
Potassium depletion.
CHO cells, grown on coverslips, were rinsed three times in K+-free buffer (1mg/ml D-glucose,
140mM NaCl, 1mM CaCl2, 1mM MgCl2, pH 7.4), followed by an incubation for 4hr at 37˚C
with proteolipid particles prepared as above, except that all solution were K+-free. Control
experiments were carried out in an identical manner, except that all solutions contained 10mM
KCl.
Expression of Eps 15 mutants in CHO cells.
GFP-constructs of the dominant negative Eps15, EH21 and DIII, and control D3∆2, were kind
gifts of Alexandre Benmerah and Alice Dautry-Varsat (Institut Pasteur, Paris, France). CHO
cells were transfected with 1µg of the constructs, complexed with 15 nmol SAINT-2/DOPE, for
4 hours after which the medium was refreshed. Two days after transfection the cells were treated
for 4 hr with Cy3-antibody proteolipid particles, prepared as described above. The results were
analyzed by confocal microscopy.
Microscopy and colocalization studies.
Conventional epifluorescence microscopy was performed with an AX70 fluorescence
microscope (Olympus). Photographs were taken with a digital camera (3.3MegaPixel-camera
Color View II) and analyzed with AnalySIS docu software by Soft Imaging System (Münster,
Germany). In any given experiment, all photomicrographs were exposed and printed identically.
Confocal microscopy was carried out for multicolor and colocalization studies with a TCS Leica
SP2 confocal laser scanning microscope (Wetzlar,Germany).
Results
SAINT2/DOPE-mediated translocation of macromolecules into cells
To obtain insight into the potential of SAINT2/DOPE liposomes as a versatile macromolecule
carrier and translocator, we first examined the delivery efficiency of different sizes of FITC-
dextran. In free form, small size FITC-dextran (9kD) was readily internalized by CHO cells (Fig.
1 A and B), and no significant increase in internalization was seen when complexed with
cationic liposomes (40 µg/20nmol) (Fig. 1 C and D). However, whereas free FITC-dextran could
be discerned as fine punctate fluorescence (Fig. 1 A and B), complexed FITC-dextran was found
in larger clusters within the cell, presumably reflecting accumulating liposomal particles loaded
with FITC-dextran (Fig. 1 C and D). Interestingly, as shown in Figs. 1E and F, large size FITC-
dextran (71kD, 40 µg) is poorly taken up by the cells when applied in free form, showing only a
faint back ground fluorescence, even after an incubation of 16 hrs (Fig.1F). By contrast, when
complexed with the cationic liposomes (20nmol), substantial internalization was apparent in
Caveolae pathway of lipoprotein particles
139
virtually all cells, with a significant accumulation of FITC-dextran in perinuclear regions of the
cell (Fig. 1 H).
The finding that SAINT2-mediated delivery is apparently not restricted to that of DNA and
oligonucleotides prompted us to next examine whether proteins could be similarly introduced
into cells.
Fig. 1 Effect of cationic lipids on FITC-dextran uptake. A and B. 40 µg FITC-dextran(9kDa) was incubated with CHO cells for 4 (A) and 16 h (B). The uptake of dextran canbe visualized by the appearance of fine dots. In C and D 40 µg FITC-dextran (9 KDa)was complexed with 20 nmol cationic liposomes prior to the incubation with the cells.Note that in this FITC-dextran can also be seen as larger aggregates within the cells. Eand F. 40 µg FITC-dextran (71 KDa) was incubated with CHO cells for 4 (E) or 16 h(F). Note that little if any uptake of the larger size dextran(71 kDa) is apparent. G andH. 40 µg FITC-dextran (71 Kda) complexed with 20 nmol cationic liposomes wasincubated with the cells for 4 (G) or 16 h (H). Note that cationic lipids stronglyenhanced the cellular uptake of 71 KDa dextran. In each panel the left picturerepresents the fluorescence micrograph, the right image represents the correspondingphase contrast.
A B
C D
E F
G H
4hr 16hr
Chapter 7
140
Interaction of proteins with cationic lipids: proteolipid particle assembly
To investigate whether cationic SAINT-2 lipids displayed the ability to deliver proteins into
cells, we first determined the efficiency of interaction of several proteins with these cationic
lipids by fluorescence microscopy. To this end, Cy3-labeled-IgG antibody (0.4µg; molecular
weight 150 kDa), was mixed with increasing amounts of SAINT-2/DOPE liposomes, prepared as
described. As shown in Fig.2, free Cy3-IgG is visualized by fluorescence microscopy as a
uniform haze without detectable particles. Following mixing with the cationic liposomes, distinct
intensely fluorescent particles appeared and the hazy fluorescence became reduced with
increasing liposome concentration (2-20 nmol; Fig. 2 b-d). When mixed at the highest liposome
concentration, the highest level of particle homogeneity was attained while the hazy fluorescence
had decreased to background level (Fig. 2 d).
Fig. 2 Formation of proteolipid particlesoccurs upon mixing of proteins andcationic liposomes. Cy3-labeled IgGantibodies (0.4 µg) were mixed with 2,10 and 20 nmol SAINT-2/DOPE and thesamples were subsequently examinedunder the fluorescence microscope (A-D)or analysed by a protein gel shift assay(E) . A. Free Cy3-antibodies arevisualized by fluorescence microscopy asa uniform haze, and particles are notdetectable. B-D. Upon addition of
various amounts (upper right corner) of cationic liposomes, distinct intensefluorescent particles were formed and the hazy fluorescence was reduced. D. At 20nmol cationic liposomes, distinct particles appeared and the hazy fluorescencedecreased to the background level.
Fig. 3 One microgram of β-galactosidase,goat IgG and BSA alone or complexed with20 nmol SAINT-2/DOPE(+) were applied ona gel as described in Methods. Note that all3 proteins avidly associated with thecationic lipids
Next, the ability of cationic lipids to complex different proteins was analyzed by a gel shift
assay. 1µg of the protein of interest, alone or complexed with 20 nmol cationic lipids, was
applied to a polyacrylamide gel. The results in Fig. 3 show that the β-galactosidase, goat IgG and
BSA bands disappeared following their incubating with cationic lipids, indicating their efficient
B
0 nmol
10 nmol
2 nmol
20 nmol
A
C D
B-gal B-gal(+) Ab Ab(+) BSA BSA(+)
Caveolae pathway of lipoprotein particles
141
interaction with the liposomes. At this ratio of protein over cationic liposomes, the majority of
particles were within a size range of 724±400nm, as determined by light scattering analysis. For
the free antibody solution in water, a zeta potential of –6.0 mV was determined which switched
to 27.9 mV, when the protein was complexed in proteolipid particles. Since the zeta potential of
cationic liposomes per se in water was 40.4 mV, these data suggest that charge neutralization
presumably represents the driving force for the formation of the proteolipid particles.
Having determined that the SAINT-2 cationic lipid is capable of forming proteolipid particles,
we subsequently examined whether the particles were internalized by eukaryotic cells, and
whether the introduced proteins maintained functionality, such as enzyme activity in case of the
β-galactosidase proteolipid particles.
Intracellular delivery of proteins
To investigate functional protein delivery per se, 40 µg β-galactosidase (the molecular weight
of β-galactosidase is 116 kDa) was complexed with increasing amounts of cationic liposomes.
The proteolipid particles thus formed were subesequently incubated with CHO cells for 4 hours,
after which the cells were lyzed and the amount of cell-associated β-galactosidase was analyzed.
As shown in Fig. 4A, following an incubation of the enzyme with the cells in the absence of
cationic liposomes, no significant enzyme activity was detectable in the cell lysates. By contrast,
the cellular uptake of β-galactosidase increased in a dose-dependent manner when complexed
with cationic liposomes. Note that at optimal conditions (20 nmol of lipid) approximately 1.5 %
of the total fraction of β-galactosidase activity added, became cell associated. Since enzyme
determination in cell lysates does not necessarily discriminate between activity present within
the cells or at the cell surface, cellular uptake of β-galactosidase complexed with cationic lipids
was also examined by microscopy (Fig. 4 B). 2.5 µg β-galactosidase alone or complexed with 20
nmol cationic lipids were incubated with CHO cells. Following an incubation with the substrate
X-gal, which is converted into a dark blue product, the cells were microscopically examined.
Product formation in terms of intracellular blue-staining was not detectable when the cells had
been incubuted with β-galactosidase alone (upper pannel), while extensive staining was apparent
in all cells when they had been incubated with the proteolipid particles (lower pannel). The data
also indicate that blue staining will only be accomplished by in situ conversion of the substrate
by the enzyme, i.e., significant intracellular diffusion of extracellularly produced product can be
excluded.
Chapter 7
142
Fig. 4 SAINT2/DOPE-mediated delivery of β-galactosidase into CHO cells. A. 40 µg β-galactosidase was complexed with increasing amounts of cationic liposomes, andincubated with CHO cells for 4 hours, after which the cells were lyzed and the amountof β-galactosidase was analyzed. Data (+ SD) are the mean values of threedeterminations. B. 2.5 µg β-galactosidase alone or complexed with 20nmol cationiclipids was incubated with CHO cells, as described in A. Enzyme activity was visualizedby X-gal staining as described in Methods. No detectable cellular staining was seenwhen the cells had been incubated with free β-galactosidase (left pannel), whileextensive staining was apparent when the enzyme had been delivered as proteolipidparticle (right pannel).
Fig. 5 SAINT2/DOPE-mediated delivery of antibodies. A. Cy3-labelled antibodies (red),nonrelated to CHO cells, were incubated with the cells for 24 h. Nuclei were stainedwith Hoechst (blue). Note that the free Ab neither entered nor became associated withthe cells. B and C. Cy3-labeled antibodies (0.4 ug) were complexed with 20nmolcationic lipids and incubated with CHO cells for 4 hr. Significant amounts ofantibodies were taken up by all cells in the population. The contours of individual cellsare indicated.
Fig. 6 Expression of dominant negative Eps15 mutants DIII or EH21 does not inhibitthe internalization of proteolipid particles. Cy3-labelled proteolipid particles wereincubated with CHO cells, which had been transfected with plasmids expressing GFP-tagged dominant negative mutants EH21 (A), and DIII (B) or the control mutant D∆/2(C) of Eps15, as described in Methods. The expression of GFP-tagged Eps15(green)was visualized in the left panel, the internalization of protein particles (red) in thesame field was visualized in the middle panel, and merged pictures are shown in theright panel. The arrows in the middle panel indicate cells, expressing Eps15 mutants.Note that cells expressing Eps 15 mutants (green) show essentially as much uptake ofproteolipid particles (red dots) as the untransfected cells do.
Fig. 7 Colocalization of intracellular protein particles with a caveolae marker. A. CHOcells were treated with the caveolae marker BODIPY-LactosylCeramide (1 uM; green)for 30min at 37˚C. Lactosylceramide is taken up by cells via caveolae (fine dotsunderneath the plasma membrane) and eventually accumulates in the Golgi complex(perinuclear clustered fluorescence). B. CHO cells were incubated with 0.4 µg Cy3-antibody proteolipid particles (red), prepared as described in Methods. After 4 hours at37˚C, the cells were washed, and chased for 10 min at 37˚C to allow plasmamembrane bound particles to become internalized. Then 1 µM BODIPY-lactosylCeramide (green) was added to visualize the caveolae. Note that proteinparticles were seen colocalizing with lactosylceramide in compartments near theplasma membrane surface (yellow dots).
Caveolae pathway of lipoprotein particles
143
Fig. 4
Fig. 5
Fig. 6
Fig. 7
A B C
A
C
B
A B
b-gal. uptake
0
0.5
1
1.5
2
0 1 2 10 20 40
cationic liposomes(nmol)
A
B
Chapter 7
144
Antibodies do not cross the plasma membrane, unless the membranes become permeabilized
following fixation. It was therefore of interest to examine whether large size antibodies, like
Cy3-labeled IgG (Mw 150 kDa), could also acquire intracellular access by cationic-liposome
mediated translocation. As demonstrated and confirmed in Fig.5A, red (Cy3-)labeled antibodies
(Fig. 5, red), unrelated to CHO cells, were not able to associate or enter CHO cells in a
detectable amount, even after 24 hr of incubation. In contrast, a significant amount of antibodies
could be seen in essentially all cells after a 4 h incubation period, when they had been complexed
with cationic liposomes (Fig. 5 B and C).
Clearly, the data presented thus far indicate that the proteolipid particles, assembled upon
incubating a variety of proteins with cationic liposomes, effectively deliver complexed proteins
into eukaryotic cells, which in free form did not gain cellular entry. The expression of
intracellularly delivered �-galactosidase suggests that following complexation and delivery,
functional activity can be maintained.
Finally, the toxicity of the proteolipid particle delivery system was verified by comparing
cellular morphology of control and treated cells and by carrying out an MTT assay. Under the
conditions used in this study, significant toxicity could neither be seen nor measured (not
shown).
In previous work, we have shown that lipoplexes, consisting of plasmids or oligonucleotides
complexed with cationic lipids, are largely internalized along the pathway of cholesterol-
dependent clathrin-mediated endocytosis. Since (i) the size of SAINT2/DOPE lipoplexes is in
the order of 200 nm, (ii) the current proteolipid particles are in the order of 750 nm while (iii)
particle size may be a governing factor in the routing of cellular internalization (24), we next
investigated the mechanism of uptake of the proteolipid particles.
Intracellular uptake of proteolipid particles involves a non-clathrin-mediated pathway
A straightforward experiment was carried out to classify the uptake of the proteolipid
particles. When Cy3-antibody particles were incubated with CHO cells at 4˚C, or with ATP-
depleted cells (following treatment with N3F) at 37 oC, no detectable fluorescence became
apparent within the cells (data not shown), implying that the uptake is an active and energy-
dependent process. Since phagocytosis primarily occurs in professional phagocytic cells like
macrophages, and little if any of such activity is present in cells like CHO, we focussed on the
possibility of entry via coated and/or non-coated endocytosis, including caveolae-mediated
internalization.
Caveolae pathway of lipoprotein particles
145
Clathrin-mediated endocytosis is effectively inhibited upon potassium depletion (25, 26), and
consistently, clathrin-mediated uptake of lipoplexes is impeded under such conditions (16). Upon
depleting potassium from CHO cells, as described in Methods, no significant inhibition of
internalization of the proteolipid particles was seen, compared to untreated CHO cells (data not
shown). In fact if anything, the uptake of the complexes in the potassium-depleted cells seemed
slightly higher (data not shown) than in control cells. For further support of this observation, we
next examined the internalization of proteolipid particles in CHO cells, which overexpressed Eps
15 mutants. Eps15 plays a role in the plasma membrane docking of the adaptor protein AP2,
which interacts with clathrin and results in the formation of coated pits at the plasma membrane.
Clathrin-mediated endocytosis can be inhibited by expression of dominant negative mutants of
Eps 15, EH21-GFP and DIII-GFP, which encode the AP2 binding sites of Eps15 but not the
domains for correct coated pit targeting of Eps15. As a control the D3∆2 construct was
expressed, which lacks AP2 binding sites (27, 28). As reported previously, whereas receptor-
mediated internalization of the transferrin (Tf)/Tf-receptor complex is effectively inhibited in the
EH21 and DIII mutants (not shown, cf. 16), internalization of the proteolipid complexes (red
fluorescence) in either transfected mutant (Fig. 6 A/B, green) and control cells (Fig. 6C,green)
or in nontransfected cells, was essentially indistinguishable. Together, these data suggest that
internalization of the proteolipid complexes, in marked contrast to the internalization of the
lipoplexes, did not occur along the pathway of clathrin-mediated endocytosis.
Recently, we observed that in non-professional phagocytic cells particles exceeding 500 nm in
size are preferentially internalized by along the caveolae-mediated pathway (24). We therefore
investigated whether the uptake of Cy3-antibody-proteolipid particles, which display a particle
size around 700 nm, co-internalized with Bodipy-lactosylceramide, a caveolae marker. To
exclude the non-specific association of protein particles with caveolae markers prior to
internalization, the following procedure was applied. The Cy3-protein particles (red) were
incubated with CHO cells for 4 hr to allow for uptake. Subsequently, the cells were extensively
washed, which was followed by a brief chase to ensure complete disappearance of proteolipid
particles from the cell surface. After the chase, the cells were labeled with the caveolae marker
Bodipy-Lactosylceramide (green), and its localization relative to that of the protein complexes
was examined by confocal miscroscpy. It has been reported that lactosylceramide is internalized
via caveolae and sorted to the Golgi complex, and our data are consistent with such a pathway in
CHO cells (Fig. 7 A), the clustered intracellular fluorescence representing the Golgi area.
Interestingly, a major fraction of the proteolipid particles colocalized with lactosylceramide in
submembrane compartments (yellow dots) (Fig. 7 B).
Chapter 7
146
Discussion
In the present study we have demonstrated that liposomes, prepared from the cationic lipid
SAINT2 and DOPE also effectively complexes a variety of macromolecules, other than DNA
and oligonucleotides, such as dextrans and proteins. When incubated with cells, such complexes,
similarly as the lipoplexes, are gaining intracellular access, while the protein’s function such as
enzyme activity, may still be maintained. Interestingly, our data suggest that the pathway of
internalization differ from that of lipoplexes, which enter via clathrin-mediated endocytosis,
whereas the much larger proteolipid particles appear to enter along the caveolae-mediated
pathway. Particle size parameters may explain this difference in entry pathway, as proposed
elsewhere (24).
Upon small-size FITC-dextran delivery (9kD), diffuse fluorescence in cytosol and
occasionally within the nuclear space could be discerned, although a major fraction was
apparently confined to defined intracellular compartments, as reflected by a localized and
prominent punctate fluorescence. Irrespective of the presence of the cationic lipid, the overall
intracellular distribution of this FITC dextran species was quite similar, except for the presence
of some larger intracellular clusters when applying the liposomal carrier. Evidently, entry of the
larger species required liposomes, emphasizing the translocation properties of the system.
Qualitatively, these data are consistent with those previously reported by Zelphati et al.(15), who
employed a novel cationic lipid analogue, trifluoroacetylated lipopolyamine, for the purpose of
protein delivery, since commercially available reagents like DOTAP, DMRIE, Trans-IT, FuGene
6, Transfast, Lipofectamine and lipofectin proved to be ineffective. Here we have employed
SAINT2, a compound of which we previously demonstrated that it also effectively introduces
plasmids, artificial chormosomes and oligonucleotides into cells, demonstrating its usefulness as
a highly versatile delivery vehicle.
The experiments indicate that charge attraction represents a major driving force for the
formation of proteolipid particles and their subsequent internalization by the cells. Indeed, the
proteins used in the present work are slightly negatively charged in water. Due to the adopted
protein/lipid ratio and given the positive charge of the cationic lipid, the proteolipid particles
bear a net positive charge, which was lower than that of the cationic lipids alone in water. This
indicates that charge neutralization occurs during the formation of the particles, while the net
positive charge, as occurs for lipoplexes, presumably favors their facilitated interaction with the
cell surface. Evidently, there are many factors that could interfere with such an interaction,
including charge differences between proteins employed, which will affect the net positive
charge of the proteolipid particle as well as the organization of the protein in the lipid
Caveolae pathway of lipoprotein particles
147
environment. In addition, hydrophobic interactions should be taken into account (15, 29, 30), a
parameter that may also affect the stability of the proteolipid particle and thereby the dissociation
of protein from the delivery vehicle.
Insight into the organization of the protein with the cationic lipids is of importance for two
reasons. If the protein associates with the cationic liposomes, it may interfere with cell surface
association of the complex by steric interference. In this way it could also interfere with the
structural organization of the lipids, required for accomplishing actual delivery (see below). The
latter may also become important when the protein forms a stabilized particle with the cationic
liposome in stead, simultaneously leading to a potential interference with the disassembly of the
liposomal structure.
Clearly, large complexes arise, given their average diameter of 700 nm as measured by
dynamic light scattering, but further work is needed to obtain detailed insight into the structure
of the proteolipid particles. This knowledge will be imperative to further appreciate the
subsequent events, following particle association at the cell surface, which concerns the fate of
the delivered protein. Our data suggest that following internalization, which occurs by virtually
every cell in the population, the protein does not gain massive access as a free diffusable
compound into the cytosol. Rather, although some cytosolic background staining is apparent,
much of the protein is localized in defined intracellular structures, where it colocalizes with the
lipid, as revealed when inserting a fluorescent lipid analogue into the complex (not shown; cf.
22). In principle, this feature is very reminiscent of cationic-lipid mediated delivery of plasmids,
which analogously shows a major accumulation of the lipoplexes in intracellular compartments,
without a clearly detectable escape into the nucleus, as commonly observed for the much smaller
oligonucleotides (22). In spite of the localized appearance of the protein, it can be functionally
active, as demonstrated by the intracellular delivery of functionally active β-galactosidase.
Whether this activity originates from complex-dissociated enzyme, present in the cytosol, or
from enzyme still associated with the proteolipid particles but captured in the cytosol-localized
compartments, remains to be determined. Nevertheless, the diffused pattern of overall blue
staining indicates that reaction products may become localized to and/or produced within the
cytosol.
Whether and if so, how the enzyme would have gotten access to the cytosol is an equally
interesting issue. Previously, we have determined that SAINT2 lipoplexes are internalized along
the pathway of clathrin-mediated endocytosis and that release of the plasmid requires the
adaptation of hexagonal lipidic phases of the complex, which presumably are instrumental in
perturbing the endosomal membrane in order to accomplish release of either plasmids or
Chapter 7
148
oligonucleotides (31). Release as such likely involves charge neutralization, an event that can be
mimicked in vitro. Whether this scenario is also operating in proteolipid delivery and whether
such a release occurs from within the endosomal compartment is less clear. The evidence present
in this work support the view that the protein particles are not internalized via the clathrin coated
pathway as revealed by the insensitivity of protein delivery toward potassium and the
observation that delivery is neither impeded in mutants defective in clathrin-mediated
endocytosis. Rather, our data are consistent with a caveolae-mediated pathway of entry which is
supported by the colocalization of the proteolipid particles with a lipid analogue which
specifically marks caveolar entry (16, 24, 32). Moreover, in recent work we demonstrated that
particles with a diameter exceeding 400 nm favor a pathway of caveolae mediated uptake,
smaller particles, like lipoplexes of approx 200-300 nm, entering via clathrin-mediated
endocytosis. These data are therefore fully compatible with the notion that the defined
cytoplasmic structures in which the proteolipid particles localize may in fact represent
caveosomes. If so, we submit that such structures may then function as slow release
compartments for protein delivery, since it is presumed that caveolae-mediated internalization
my avoid delivery to the lysosomes, and hence degradation, representing a pathway also reported
to be exploited by certain viruses and bacteria in eukaryotic cells (33, 34). Clearly, although
further work is required, this study may rationalize future directions and developments in
controlled protein delivery based upon the application of cationic lipid delivery vehicles.
AcknowledgementsThis work was supported by a grant from The Netherlands Organization forScientific Research (NWO)/NDRF Innovative Drug Research (940-70-001).Anno Wagenaar and Professor Jan Engberts are thanked for helpfuldiscussions and for providing us with SAINT-2.
Caveolae pathway of lipoprotein particles
149
References:1. Fawell, S., Seery, J., Daikh, Y., Moore, C., Chen, L. L., Pepinsky, B. and Barsoum,
J. (1994) Tat-mediated delivery of heterologous proteins into cells. Proc. Natl.Acad. Sci. U. S. A., 91: 664-668.
2. Schwarze, S. R., Ho, A., Vocero-Akbani, A. and Dowdy, S. F. (1999) In vivo proteintransduction: delivery of a biologically active protein into the mouse. Science, 285:1569-1572.
3. Derossi, D., Joliot, A. H., Chassaing, G. and Prochiantz, A. (1994) The third helix ofthe Antennapedia homeodomain translocates through biological membranes. J.Biol. Chem., 269: 10444-10450.
4. Rojas, M., Donahue, J. P., Tan, Z. and Lin, Y. Z. (1998) Genetic engineering ofproteins with cell membrane permeability. Nat. Biotechnol., 16: 370-375.
5. Futaki, S., Suzuki, T., Ohashi, W., Yagami, T., Tanaka, S., Ueda, K. and Sugiura,Y. (2001) Arginine-rich peptides. An abundant source of membrane-permeablepeptides having potential as carriers for intracellular protein delivery. J. Biol.Chem., 276: 5836-5840.
6. Vives, E., Brodin, P. and Lebleu, B. (1997) A truncated HIV-1 Tat protein basicdomain rapidly translocates through the plasma membrane and accumulates inthe cell nucleus. J. Biol. Chem., 272: 16010-16017.
7. Bonifaci, N., Sitia, R. and Rubartelli, A. (1995) Nuclear translocation of anexogenous fusion protein containing HIV Tat requires unfolding. AIDS., 9: 995-1000.
8. Morris, M. C., Depollier, J., Mery, J., Heitz, F. and Divita, G. (2001) A peptidecarrier for the delivery of biologically active proteins into mammalian cells. Nat.Biotechnol., 19: 1173-1176.
9. Witschi, C. and Mrsny, R. J. (1999) In vitro evaluation of microparticles andpolymer gels for use as nasal platforms for protein delivery. Pharm. Res., 16: 382-390.
10. Yeh, M. K., Davis, S. S. and Coombes, A. G. (1996) Improving protein delivery frommicroparticles using blends of poly(DL lactide co-glycolide) and poly(ethyleneoxide)-poly(propylene oxide) copolymers. Pharm. Res., 13: 1693-1698.
11. Li, W. M., Dragowska, W. H., Bally, M. B. and Schutze-Redelmeier, M. P. (2003)Effective induction of CD8+ T-cell response using CpG oligodeoxynucleotides andHER-2/neu-derived peptide co-encapsulated in liposomes. Vaccine, 21: 3319-3329.
12. Irie, T., Watarai, S. and Kodama, H. (2003) Humoral immune response of carp(Cyprinus carpio) induced by oral immunization with liposome-entrapped antigen.Dev. Comp. Immunol., 27: 413-421.
13. Zho, F. and Neutra, M. R. (2002) Antigen delivery to mucosa-associated lymphoidtissues using liposomes as a carrier. Biosci. Rep., 22: 355-369.
14. Ninomiya, A., Ogasawara, K., Kajino, K., Takada, A. and Kida, H. (2002) Intranasaladministration of a synthetic peptide vaccine encapsulated in liposome togetherwith an anti-CD40 antibody induces protective immunity against influenza A virusin mice. Vaccine, 20: 3123-3129.
15. Zelphati, O., Wang, Y., Kitada, S., Reed, J. C., Felgner, P. L. and Corbeil, J. (2001)Intracellular delivery of proteins with a new lipid-mediated delivery system. J. Biol.Chem., 276: 35103-35110.
16. Zuhorn, I. S., Kalicharan, R. and Hoekstra, D. (2002) Lipoplex-mediatedtransfection of mammalian cells occurs through the cholesterol-dependentclathrin-mediated pathway of endocytosis. J. Biol. Chem., 277: 18021-18028.
17. Wrobel, I. and Collins, D. (1995) Fusion of cationic liposomes with mammaliancells occurs after endocytosis. Biochim. Biophys. Acta. 1235: 296-304.
Chapter 7
150
18. Zabner, J., Fasbender, A. J., Moninger, T., Poellinger, K. A. and Welsh, M. J. (1995)Cellular and molecular barriers to gene transfer by a cationic lipid. J. Biol. Chem.,270: 18997-19007.
19. Friend, D. S., Papahadjopoulos, D. and Debs, R. J. (1996) Endocytosis andintracellular processing accompanying transfection mediated by cationicliposomes. Biochim. Biophys. Acta, 1278: 41-50.
20. Oberle, V., de-Jong, G., Drayer. J. I. and Hoekstra, D. Efficient transfer ofchromosome-based DNA constructs into mammalian cells. Biochim-Biophys-Actain press
21. van-der-Woude, I., Wagenaar, A., Meekel, A. A., ter-Beest, M. B., Ruiters, M. H.,Engberts, J. B. and Hoekstra, D. (1997) Novel pyridinium surfactants for efficient,nontoxic in vitro gene delivery. Proc. Natl. Acad. Sci. U. S. A., 94: 1160-1165.
22. Shi, F., Nomden, A., Oberle, V., Engberts, J. B. and Hoekstra, D. (2001) Efficientcationic lipid-mediated delivery of antisense oligonucleotides into eukaryotic cells:down-regulation of the corticotropin-releasing factor receptor. Nucleic Acids Res.,29: 2079-2087.
23. Meekel,A.A.P., Wagenaar,A., Smisterova,J., Kroeze,J., Haadsma,P., Bosgraaf,B.,Stuart,M.C.A., Brisson,A., Ruiters,M.H.J., Hoekstra,D. and Engberts, J. B. (2000)Synthesis of pyrimidinium amphiphiles used for transfection and somecharacteristics of amphiphile/DNA complex formation, Eur. J. Org. Chem. 665-673
24. Rejman, J., Oberle, V., Zuhorn, I. S. and Hoekstra, D. Size-dependentinternalization of particles via the pathway of clathrin- and caveolae-mediatedendocytosis. Biochem. J., in press.
25. Larkin, J. M., Brown, M. S., Goldstein, J. L. and Anderson, R. G. (1983) Depletionof intracellular potassium arrests coated pit formation and receptor-mediatedendocytosis in fibroblasts. Cell, 33: 273-285.
26. Hansen, S. H., Sandvig, K. and van-Deurs, B. (1993) Clathrin and HA2 adaptors:effects of potassium depletion, hypertonic medium, and cytosol acidification. J. CellBiol., 121: 61-72.
27. Benmerah, A., Lamaze, C., Begue, B., Schmid, S. L., Dautry-Varsat, A. and Cerf-Bensussan, N. (1998) AP-2/Eps15 interaction is required for receptor-mediatedendocytosis. J. Cell Biol., 140: 1055-1062.
28. Benmerah, A., Bayrou, M., Cerf-Bensussan, N. and Dautry-Varsat, A. (1999)Inhibition of clathrin-coated pit assembly by an Eps15 mutant. J. Cell Sci., 112:1303-1311.
29. Zuhorn, I. S., Visser, W. H., Bakowsky, U., Engberts, J. B. and Hoekstra, D. (2002)Interference of serum with lipoplex-cell interaction: modulation of intracellularprocessing. Biochim. Biophys. Acta., 1560: 25-36.
30. Audouy, S., Molema, G., de-Leij, L. and Hoekstra, D. (2000) Serum as a modulatorof lipoplex-mediated gene transfection: dependence of amphiphile, cell type andcomplex stability. J. Gene Med., 2: 465-476.
31. Smisterova, J., Wagenaar, A., Stuart, M. C., Polushkin, E., ten-Brinke, G., Hulst,R., Engberts, J.B. and Hoekstra, D. (2001) Molecular shape of the cationic lipidcontrols the structure of cationic lipid/dioleylphosphatidylethanolamine-DNAcomplexes and the efficiency of gene delivery. J. Biol. Chem., 276: 47615-47622.
32. Puri, V., Watanabe, R., Singh, R. D., Dominguez, M., Brown, J. C., Wheatley, C. L.,Marks, D. L. and Pagano, R. E. (2001) Clathrin-dependent and -independentinternalization of plasma membrane sphingolipids initiates two Golgi targetingpathways. J. Cell Biol., 154: 535-547.
33. Shin, J. S., Gao, Z. and Abraham, S. N.(2000) Involvement of cellular caveolae inbacterial entry into mast cells. Science, 289: 785-788.
34. Pelkmans, L., Puntener, D. and Helenius, A.(2002) Local actin polymerization anddynamin recruitment in SV40-induced internalization of caveolae. Science, 296:535-539.
151
Chapter 8
Summary and perspectives
Chapter 8
152
Over the past decades, antisense technology has been proclaimed to potentially represent an
extremely powerful tool to modulate and hence regulate gene expression. The onset was
triggered by observations reported in 1981 on the occurrence of natural antisense RNAs,
operating within eukaryotic cells (1,2). Since then a scientific avalanche has been generated in
this research area, culminating in the marketing of the first commercial antisense drug Vitravene.
The naturally occurring antisense RNAs are 35-150 nucleotides in length and are able to post-
transcriptionally inhibit the expression of the targeted mRNA. Although numerous examples of
antisense activity have been reported in bacteria, in this thesis the focus is on the development
and potential application of antisense technology in eukaryotic cells. Although relatively simple
and straightforward in theory, we will first discuss some of the crucial hubs that are usually
encountered in the experimental application of antisense methodology.
The first challenge is how to select a proper antisense sequence. mRNA is not a random coil
but is folded into a three-dimensional structure, and accordingly, antisense sequences require to
be target to the accessible regions on the mRNA. Currently antisense sequences are largely
selected empirically, i.e., at the level of trial and error. However, computer-facilitated
predication of the mRNA structure appears a useful tool for selecting antisense sequences to the
“open” regions on the RNA. Compared to other methods, such as mRNA walking,
oligonucleotide array and RNase H scanning, this approach is relatively time saving and
effective. Even though the maximal antisense effect is not necessarily guaranteed with this
method, a good correlation has been found so far between antisense sequences and antisense
effects, based upon the predicted structure of the mRNA. In this thesis, research is described
which relies on the use of several antisense sequences targeted to the CRF-receptor or 5-HT1a,
guided by computer-aided antisense sequence design and selection technology, called RADAR
(Rational algorithmic Design of ANTISENSE reagents). Indeed, the results obtained here let us
conclude that this methodology represents a fertile and effective approach in the design of
appropriate antisense sequences.
The second challenge one is confronted with involves the stabilization of antisense molecules,
once they are exposed to the extra- and/or intracellular environment. To resolve such issues, two
approaches have been particularly successful. Thus, chemical modification - without
significantly affecting efficiency- may confer resistance of the nucleotides towards nuclease
degradations, while the use of carriers can protect the antisense probes from a direct contact to
nuclease or other perturbing compounds.
Several classes of nucleotide analogues have been generated to improve the resistance to
nuclease and enhance its affinity to mRNA. It should be noted, however, that any modification
Summary and perspectives
153
on the nucleotide backbone or sugar might attenuate or abolish the antisense effect by preventing
the activation of RNase H, which is a key factor in causing degradation of the targeted mRNA,
following binding of the antisense. In our work, oxygen on the nucleotide backbone of the
antisense molecules has been substituted for sulfur. This modification confers resistance towards
nucleases, while the property to recruit RNase H is maintained, as demonstrated (chapter 3 and
4). Potential binding of proteins to such 'sulfur' modified antisense nucleotides should be
critically evaluated, as such a phenomenon has given rise to mis-interpretation of the actual
antisense action. On the other hand, this modification drastically improves the pharmacological
profile by preventing the clearance of antisense molecules from the circulation by filtration via
the kidney.
The third challenge concerns the delivery of antisense molecules to the targeted mRNA. The
majority of the antisense fraction, when applied systemically, is captured by reticuloendothelial
organs, like liver, spleen and lung, while in tissue culture, antisense molecules are largely
sequestered in endocytic compartments. By using appropriate delivery vehicles, direct deposition
of antisense molecules into non-reticuloendothelial tissue such as brain or tumors might be
accomplished, or means could thus be provided to promote intracellular uptake and translocation
from endocytic compartments, which will be required for encountering the targeted mRNA. In
general, cationic liposomes display a high capacity to complex DNA by charge neutralization,
show a low immunogenicity in comparison with viral vectors, but their delivery efficiency is
relatively low. To improve the latter, much effort is undertaken in devising and testing newly
synthesized cationic lipid analogues, and formulations derived thereof. In tissue culture cationic
liposomes can improve the cellular uptake of antisense molecules over a thousand fold, promote
their translocation over endosomes, and therefore enhance the antisense efficiency, as
demonstrated in Chapter 3. How does an antisense molecule reach the targeted mRNA and
interfere with its function? We have found that antisense molecules exclusively accumulate in
the nucleus and become bound to the RNA matrix after release from the endosomal
compartments, and transfer to and translocation into the nuclear compartment. This latter step
appears imperative in their ability to induce the degradation of the targeted RNA molecules, as
investigated in Chapter 4. However, cationic lipoplexes (consisting of cationic liposomes and
nucleotides) tend to aggregate in vivo, which induces their disposition in the lung, whereas the
net positive charge will facilitate uptake by the liver. Accordingly, modification of the surface of
lipoplexes is required to prevent the aggregation and shield the positive charge to improve the
pharmacological profiles. An attempt has been made in this thesis to fulfil this goal by utilizing
polyethylene glycol-coupled lipid analogues. A programmable and functional delivery of
Chapter 8
154
oligonucleotides might be realized by modifying the cationic lipoplexes with exchangeable
polythylene glycol lipids. This work is described in chapter 5.
To develop nucleic acids for silencing gene expression at the posttranscriptional level, RNA
interference (RNAi), a powerful new tool for analyzing gene knockdown phenotypes in living
mammalian cells, represents an alternative to antisense strategy. The term RNAi was first coined
after the discovery in 1998, that the injection of double stranded RNAs (dsRNA) into C. elegans
interfered with the expression of specific genes (3). Since then, work in this area and the
appearance of ensuing papers, has been booming. Several advantages of RNAi make researchers
in this field believe that RNAi will greatly exceed the potential of other posttranscriptional
regulators, like antisense and ribozyme technology. RNAi's are generated from the long dsRNA
molecule by endogeous ribonuclease (4, 5). Interestingly, this enzyme also includes a helicase
domain that unwinds the two strands of RNA molecules, allowing the antisense strand to bind to
the targeted RNA molecule. Then the targeted RNA molecule is hydrolyzed by a crudely defined
multi-subunit RNA-induced silencing complex (RISC). Therefore, in essence, the action of
RNAi relies upon an antisense mechanism, since ultimately a single-stranded RNA molecule
binds to the targeted RNA molecule by Watson-Crick base pairing, and recruits a ribonuclease to
degrade the targeted RNA. However, in spite of the compelling similarity between antisense and
RNAi, there are several important differences. (i) RNAi is generated from 200- and 500- bp
precursors, which are processed to segments of 22 bp, which result in a highly specific
suppression of a targeted gene. By contrast, antisense RNA is a single strand, 35-150 nucleotides
in length and is transcribed from the same DNA molecule as its targeted RNA. (ii) The naturally
occurring RNAi is exclusively found in eukaryotes as the oldest and most ubiquitous antiviral
system, while the majority of antisense RNAs are found in prokaryotes. This might imply that by
nature RNAi is more suitable than antisense RNAs for silencing genes in eukaryotic cells. (iii)
Self-amplification and a “cell to cell” spreading of RNAi results in a long-lasting suppression of
the targeted gene in plants and worms, while antisense RNA represents a rather transient
suppression of the targeted gene in prokaryotes. As yet, this amplification and “systemic”
spreading of RNAi have not been observed in mammals. (iv) The target sequence of RNAi is
exclusively localized to an encoding region on the targeted mRNA, while the target sequence of
antisense RNA can be either on an encoding region or a non-coding region.
For the potential therapeutic use of “artificial” RNAi and antisense RNA, the same challenges
are encountered. First, they have molecular weights that usually exceed 1000 Da, which thus
brings about a significant delivery problem. Second, the suppression of gene expression with
both approaches is transient in mammals. Third, the selection of proper sequences and length of
Summary and perspectives
155
RNAi or antisense are crucial in order to achieve specificity. Fourth, they are both RNAs, which
implies that both are susceptible to ribonuclease degradation. Synthetic small antisense or RNAi
molecules are both ideal candidates to fulfil the short-term suppression of gene expression in
mammalian cells, because dsRNA> 30nt will provoke the global inhibition of gene expression
via the activation of dsRNA-dependent protein kinase and interferon response. The single
stranded antisense molecules confer better gene suppression when applying chemically modified
nucleotides, while double stranded RNAi molecules are not tolerant to such modification.
However, it should be noted that mammalian cells lack the mechanism to support the
amplification of RNAi-mediated silencing, as observed in C. elegans, thus DNA vector-based
strategies are preferred to express small RNAi for long-term inhibition of gene expression in
mammals. Mammalian cells also lack the mechanism to support “systemic” spread of RNAi, as
is observed in plants. Thus the delivery of RNAi with carriers is needed to silence gene
expression in mammals.
Optimization and simplification of the design of an efficient antisense and interference RNA
molecule, and a better description of the systemic nature of their response in whole animals,
combined with the ongoing improvement of in vivo nucleic acid delivery technologies will
enable to translate both the antisense and interference RNA technology into medical use in the
near future.
Chapter 8
156
References:
1. Tomizawa, J., Itoh, T., Selzer, G. and Som, T. (1981) Inhibition of ColE1 RNAprimer formation by a plasmid-specified small RNA. Proc. Natl. Acad. Sci. U. S. A.78, 1421-1425.
2. Stougaard, P., Molin, S. and Nordstrom, K. (1981) RNAs involved in copy-numbercontrol and incompatibility of plasmid R1. Proc. Natl. Acad. Sci. U. S. A. 78, 6008-6012.
3. Fire, A., Xu, S., Montgomery, M. K., Kostas, S. A., Driver, S. E. and Mello, C. C.(1998) Potent and specific genetic interference by double-stranded RNA inCaenorhabditis elegans. Nature 391, 806-811.
4. Grishok, A., Tabara, H. and Mello, C. C. (2000) Genetic requirements forinheritance of RNAi in C. elegans. Science 287, 2494-2497.
5. Zamore, P. D., Tuschl, T., Sharp, P. A. and Bartel, D. P. (2000) RNAi: double-stranded RNA directs the ATP-dependent cleavage of mRNA at 21 to 23 nucleotideintervals. Cell 101, 25-33.
157
Samenvatting
Samenvatting
158
De ontdekking van de natuurlijke aanwezigheid van antisense
oligonucleotiden in eukaryote cellen, die de expressie van specifieke eiwitten
kunnen reguleren, heeft in de afgelopen tientallen jaren gezorgd voor een
intensieve speurtocht naar de mogelijkheid om deze moleculen toe te passen
als geneesmiddel of als moleculair gereedschap in celbiologisch onderzoek.
Antisense oligonucleotiden of antisense RNAs bestaan uit enkelvoudige
nucleinezuur ketens met een lengte van 35 tot 150 nucleotiden. In het
laboratorium kunnen dergelijke structuren worden gemaakt door chemische
synthese. Wanneer de juiste nucleinezuur volgorde wordt gekozen dan kan een
gegeven antisense molecuul binden aan een selectief mRNA wat
verantwoordelijk is voor de productie van een specifiek eiwit. Als gevolg
daarvan kan de synthese van een eiwit, met name in het geval van een daaraan
gekoppeld ziekteproces, onderdrukt worden, waardoor een niet gewenste
ontsporing in de cel geen kans krijgt zich verder te ontwikkelen. Dit laatste
wordt aangeduid als antisense therapie. Anderzijds kan onderdrukking van de
expressie van een eiwit interessante inzichten verschaffen over de functie van
dat eiwit in de cel, iets wat bijzonder relevant is in het licht van de onlangs
opgehelderde nucleotidenvolgorde van het menselijk genoom. Immers, alhoewel
de volledige genetisch code inmiddels ontrafeld is, is de aard en functie van de
gecodeerde eiwitten daarmee nog geenszins opgelost. Antisense technologie zou
daarbij zeer behulpzaam kunnen zijn. Hoewel het principe van antisense
technologie, dat willen zeggen de blokkering van het aanmaken van een nieuw
specifiek eiwit door te interfereren met de machinerie (mRNA) die
verantwoordelijk is voor deze aanmaak dus betrekkelijk eenvoudig is, zijn in de
praktische uitvoering daarvan de nodige barrières te overwinnen.
De eerste uitdaging is om de meest geschikte nucleotiden volgorde van een
antisense molecuul te selecteren, waardoor er ook daadwerkelijk een remming
in de eiwit productie van het doel-eiwit op zal treden. Daarbij moet rekening
worden gehouden met het feit dat het mRNA, waartegen het antisense gericht
moet zijn, een goed gedefinieerde driedimensionale structuur bezit. Met andere
woorden, het doelgebied in de structuur moet wel toegankelijk zijn voor het
antisense molecuul. Deze antisense volgordes worden tegenwoordig grotendeels
proefondervindelijk vastgesteld. Echter, met behulp van computertechnolgie is
Samenvatting
159
het mogelijk om betere structurele voorspellingen te doen, waardoor het
eenvoudiger wordt om geschikte doelstructuren in de mRNA structuur te
voorspellen. In feite is deze methode te prefereren boven andere technieken, die
vaak tijdrovender zijn.
In dit proefschrift wordt onderzoek beschreven naar het effect van
verschillende antisense volgordes op de expressie en het functioneren van een
aantal eiwitten, die een functie vervullen als receptor op het oppervlak van
neuronale cellen in kweek. Een van die eiwitten, de zogenaamde CRF receptor,
speelt een rol bij psychiatrische aandoeningen zoals stress. De gebruikte
antisense moleculen werden geselecteerd op basis van de genoemde computer
technolgie en de verkregen resultaten bevestigen de effectiviteit van deze
technologie voor het identificeren van geschikte oligonucleotide als antisense
probes.
De activiteit van dergelijke probes kan uiteraard pas worden vastgesteld
nadat die stoffen toegang hebben gekregen in de cel en nadat ze daar zijn
gearriveerd niet onmiddelijk worden herkend als vreemde stoffen en als gevolg
daarvan worden afgebroken. De biologisch stabiliteit van antisense moleculen
is dus een belangrijke vereiste. Door chemische modificatie van de structuur
kan die stabiliteit van antisense moleculen worden verkregen, zonder dat
daarmee de effectiviteit van hun werking wordt beïnvloed. In hoofdstuk 3 wordt
beschreven hoe het vervangen van zuurstof door zwavel die gewenste stabiliteit
geeft terwijl bovendien het antisense effect van het molecuul intakt blijft.
Hoe slaagt een antisense molecuul erin om in de cel zijn doel, nl. een gegeven
mRNA dat verantwoordelijk is voor de productie van een specifiek eiwit, te
bereiken? Met name voor het toepassen van antisense technologie in
proefdieren (in vivo) is dit geen triviaal probleem. Wanneer antisense moleculen
in vivo worden geinjiceerd, dan worden ze doorgaans vrij snel uit het bloed
verwijderd door organen die behoren tot het reticuloendotheliale systeem, zoals
de lever en milt. In gekweekte cellen worden antisense moleculen opgenomen
via het proces van endocytose, waardoor ze worden afgevoerd naar het afbraak
systeem van de cel, de lysosomen. Door geschikte dragers of ‘carriers’ te
gebruiken die antisense moleculen aan zich binden en die bovendien een
weefsel-specifiek herkenningssysteem bevatten, is het wellicht mogelijk om ze
Samenvatting
160
specifiek af te geven aan een van tevoren vastgesteld orgaan, zoals hersenen of
een tumor weefsel. Een dergelijke carrier zou ook behulpzaam kunnen zijn bij
het verhogen van de opname van antisense moleculen door cellen en wellicht in
staat zijn om afgifte in het cytoplasma van de cel te bevorderen, hetgeen een
belangrijke stap is om een interactie te kunnen aangaan met het mRNA, dat
zich deels hier en in de kern van de cel bevindt. Liposomen, gemaakt van
cationische lipiden, blijken uitstekend geschikt voor dat doel en dat onderzoek
wordt beschreven in hoofdstuk 4. Het blijkt dat de positief geladen lipide
moleculen waaruit deze carriers bestaan zeer goed negatief geladen
oligonucleotide moleculen kunnen binden. Tevens hebben dergelijke moleculen
eigenschappen die berusten op een structurele verandering die belangrijk lijkt
te zijn voor het destabilizeren van intracellulaire (endosomale) membranen,
hetgeen nodig is voor het vrijkomen van de antisense moleculen in het
cytoplasma. Deze studies, die worden beschreven in hoofdstuk 3 en 4 zijn van
belang om de antisense technologie verder te optimaliseren.
Wanneer het antisense molecuul eenmaal toegang heeft gekregen tot het
cytoplasm, dan blijkt het vrij snel te accumuleren in de kern van de cel. In
hoofdstuk 4 wordt beschreven dat dit een belangrijke stap is in het
mechanisme waarmee antisense moleculen in staat zijn om uiteindelijk de
expressie van een eiwit te remmen. Vastgesteld kon worden dat via binding aan
de RNA matrix uiteindelijk het mRNA wordt afgebroken en dat er niet slechts
sprake is van de blokkering van expressie via sterische interferentie, waardoor
een eiwit nog steeds deels gemaakt zou kunnen worden.
Complexen die bestaan uit cationische lipiden en antisense moleculen
hebben de neiging om in vivo te aggregeren. Dat fenomeen werd geconstateerd
in experimenten met proefdieren. Om dit proces van complex aggregatie,
waardoor het transport naar gewenste weefsels sterk beperkt zou kunnen
worden en mogelijk zou kunnen leiden tot toxisch effecten, te beperken,
werden mogelijkheden onderzocht om dit tegen te gaan. Daarbij werd gebruik
gemaakt van polyethyleenglycol-gekoppelde lipiden die werden ingebouwd in de
antisense-bevattende lipide complexen. De eigenschappen van dergelijke
complexen werden nader bestudeerd in hoofdstuk 5 en er kon worden
aangetoond dat er op deze wijze inderdaad een programmeerbare aflevering van
Samenvatting
161
antisense oligonucleotiden gerealiseerd kan worden. Dat dit uiteindelijk leidt
tot een sterke vermindering van de productie en belangrijker, van de functie
van specifieke eiwitten zoals van de 5HT1a serotonine receptor, wordt in detail
beschreven in hoofdstuk 4.
Het uiteindelijk doel van deze en soortgelijke studies is om antisense
technologie toe te passen in vivo, met mogelijk een therapeutisch doel. Om
inzicht te krijgen in de potentie daarvan werden experimenten uitgevoerd met
hersenweefsel, zogenaamde brain slices, die als model systeem voor in vivo
hersenmateriaal fungeren. Tamelijk onverwacht werd vastgesteld dat
hersencellen in staat zijn om zeer efficient antisense oligonucleotiden op te
nemen zonder dat daar een carrier voor nodig is. Dat bleek ook het geval te zijn
wanneer antisense oligonucleotiden direct in de hersenen van ratten werden
ingebracht. Met behulp van nader onderzoek werd vastgesteld dat
hersencellen, het juiste type is nog niet geidentificeerd, over een nucleotide
transport eiwit beschikken dat in staat is antisense oligonucleotiden
rechtstreeks in het cytoplasma te transporteren, zoals wordt aangetoond in
hoofdstuk 6. Dit transport eiwit is tot nu toe alleen vastgesteld in de nier en
blijkt niet in staat te zijn om genen (plasmiden), die ook uit nucleotiden zijn
opgebouwd, te transporteren. Voor dat doel en voor de aflevering van antisense
oligonucleotiden aan specifieke cellen in de hersenen zijn carriers zoals
cationische liposomen nodig.
Tenslotte is bestudeerd of de beschreven carrier, zoals toegepast in de
beschreven antisense studies, ook in staat is om eiwitten in de cel te
transporteren. Dat zou nieuwe mogelijkheden kunnen bieden voor bijvoorbeeld
het bestuderen van de functionele eigenschappen van intracellulaire eiwitten
met behulp van antilichamen of voor het inbrengen van therapeutisch
relevante eiwitten. In hoofdstuk 7 wordt aangetoond dat met behulp van
dezelfde carrier als voor het antisense, ook talloze eiwitten, inclusief
antilichamen, in de cel kunnen worden gebracht.
De betekenis en het nut van verdere ontwikkeling van de antisense
technologie in het licht van zeer recente bevindingen op een parallel gebied,
namelijk dat wat betreft de ontdekking en toepassing van RNAi hetgeen
Samenvatting
162
uiteindelijk resulteert in overeenkomstige effecten als gevonden voor antisense
constructen, wordt bediscussieerd in hoofdstuk 8.
163
Acknowledgements
Wish to thank those who accompany me for so many years.
To my supervisor, Prof. Dick Hoekstra, I am so grateful to be one of your
Ph.D students and work closely with you in these past 5 years. As trained to be
an otolaryngologist in China, I made a big switch to be a scientist in Holland. It
turns out to be such a memorial experience to work in your lab. Your
inspiration always lighted up my way to go further; your trust ensured me a
great freedom to do research; your enthusiasm led me through the winter time
of the project; certainly your Dutch Samenvatting enabled me to finish my
dissertation. In deed it’s you who guide me today to a glory doctorate title.
To my colleagues in Membrane Cell Biology, Luc, Inge, Kasper, Karin, Jenny,
Ina, Joke, Bert, John, and Timen, Zuzana, Arjen, Jan Willem, Hans, Sven, Wia,
Jan Wijbenga, and Tini, I thank all of you for the time that you spent with me.
You were never too busy to teach me lab techniques and show your concerns
for my life. My sincere appreciation goes to Anita and Willy, two technicians in
this project. Your fundamental contribution speeded up my research and
ensured this project to be finished in a due time. To my long-lasting roommates
in 661 and 1025, Volker, Olaf, Asia, Sandrine, and Delphine, you have made
me so happy to be in the office. Volker, my dear friend and colleague, your
164
jokes and concerns will never be forgotten. Gerry, I am so glad that I could be
acquainted with you from the beginning. Tounsia, Nicole, Dona, Jarmila, Eve,
Testuo, Erna, Cobbi, Cecile …, the former members in the lab, the time that we
were together was so cheerful and it will be memorized forever in my mind.
Jan Engberts and Anno, thank you for your helpful discussion and synthesis
of SAINT compounds. Marc and Evgeny, thank you for your help in cryo-EM
and X-ray. Robert Liem, thank you for your EM work and your concern to my
welfare.
To dear NDRF members, I had a good time to meet you every half a year and
to partake in the hospitality of Solvay. A special thank word goes to Eric.
Without your contribution and help, the in vivo work presented in my
dissertation will not be possible. Natasja and Andrew, thank you for your
cooperative work in Chapter 4 and Chapter 6.
To my dear Chinese friends in Groningen, your names will remind me in the
future the joys and the sorrows we have had in Groningen…. A key note thank
goes to my best friends, Yan Xuedong and Sun Rui, Qian Cheng and Yu Lili,
Tang Lixia and Gao Hui…..
Ba and Ma (parents in Chinese), without your constant support and concern,
I will not be at the place where I am. Ma, you devote all your life to our family,
and now it has been extended to grandchilden. Boyang, my beloved daughter,
your smiles and your giggles are the sunshine to me. Xiaoqin, image the path
we have had together for many years, especially those in a foreign land with the
achievements of bringing both our daughter and our Ph.D dissertations to the
world. Isn’t it a tough job? Life is not easy, but our persistence is always the
way for us to overcome difficulties. Certainly, I will not forget my brother, Shi
Jingjiang, who influences my life and career so much.
March 2004, Groningen
165
Publications1. Fuxin Shi, Anita Nomden, Volker Oberle, Jan B. F. N. Engberts and Dick Hoekstra.
Efficient cationic lipid-mediated delivery of antisense oligonucleotides into eukaryotic
cells: down-regulation of the corticotropin-releasing factor receptor. Nucleic Acids
Research, 2001, Vol. 29, 2079-2087
2. Fuxin Shi, Luc Wasungu, Anita Nomden, Marc C. A. Stuart, Evgeny Polushkin, Jan B.
F. N. Engberts and Dick Hoekstra. Interference of poly(ethylene glycol)-lipid analogues
with cationic-lipid-mediated delivery of oligonucleotides; role of lipid exchangeability
and non-lamellar transitions. Biochemistry Journal, 2002, Vol. 366, 333-341.
3. Fuxin Shi, Robert S.B. Liem, Willy H. Visser, Natasja M. J. de Jong, Eric Ronken and
Dick Hoekstra. Antisense phosphothioate oligonucleotides interact via theRNA matrix to
reach the target mRNA. Downregulation of the 5-HT1A receptor. Experimental Cell
Research 291 (2003) 313-325.
4. Fuxin Shi, Jerome Swinny, Eric Ronken and Dick Hoekstra. Oligonucleotides enter cells
mainly through a nucleic acid channel on the explant of rat brain. Submitted.
5. Fuxin Shi, Xuedong Yan and Dick Hoekstra. Cationic liposome-mediated delivery of
proteins into eukaryotic cells: entry along the pathway of caveolae-mediated endocytosis.
Submitted.
6. Fuxin Shi, Dick Hoekstra. Does the delivery of oligonucleotides matter? Make sense of
antisense oligonucleotides. Submitted.
7. J. D. Swinny, D. Kalicharan, E. H. Blaauw, J. IJkema-Paassen, F. Shi, A. Gramsbergen,
J. J. L. van der Want. Corticotropin releasing factor types 1 and 2 are differentially
expressed in pre- and postsynaptic elements in the postnatal developing rat cerebellum.
European Journal of Neuroscience 2003, 18(3): 549-62.
8. J. D. Swinny, D. Kalicharan, F. Shi, Albert Gramsbergen, J. J. L. van der Want. The
Postnatal development expression pattern of urocortin in the rat oligocerebellar system.
(Conditionally accepted by Journal of comparative Neurology).
9. X. Wang, Fuxin Shi, B. Poolman1and G. T. Robillard. Unidirectional permeability of
hydrophobin SC3 membrane formed at an oil/water interface. submitted.