development of high throughput screening systems of...
TRANSCRIPT
Development of High Throughput Screening Systems
of Mutant Libraries for Sugar Modifying Enzymes
(Glucose Oxidase and Cellulases)
Von der Fakultät für Mathematik, Informatik und Naturwissenschaften der RWTH
Aachen University zur Erlangung des akademischen Grades einer Doktorin der
Naturwissenschaften genehmigte Dissertation
vorgelegt von
Diploma de Licenta
Raluca Ostafe
aus Timisoara, Romania
Berichter: Universitätsprofessor Dr. rer. nat. Rainer Fischer
Universitätsprofessor Dr. Dirk Prüfer
Tag der mündlichen Prüfung:
21.05.2013
Diese Dissertation ist auf den Internetseiten der Hochschulbibliothek online verfügbar.
TableofContentsI Introduction .......................................................................................................................................... 6
I.1 The Importance of Enzymes in Industry .......................................................................................... 6
I.2 Glucose Oxidase ............................................................................................................................... 7
I.3 Cellulases ......................................................................................................................................... 9
I.4 Directed Evolution ......................................................................................................................... 10
I.5 Diversity Generation ...................................................................................................................... 11
I.6 High Throughput Screening (HTS) .................................................................................................. 13
I.7 Fluorescence Assays ...................................................................................................................... 15
I.7.1 Fluorescence assays for GOx ................................................................................................... 16
I.7.2 Fluorescence assays for cellulases .......................................................................................... 17
I.8 Enzyme‐induced Polymerization ................................................................................................... 19
I.9 Project Objectives .......................................................................................................................... 21
I.10 Project Overview ......................................................................................................................... 21
II Materials and Methods ....................................................................................................................... 24
II.1 General Recipes for Media ............................................................................................................ 24
II.1.1 LB media (Bertani 1951) ...................................................................................................... 24
II.1.2 YNB CAA Glu/Gal, Raf (Karst et al. 1999) ............................................................................ 24
II.1.3 YPD (Kaiser et al. 1994) ....................................................................................................... 24
II.1.4 BMGY/BMMY (Invitrogen 2010) ......................................................................................... 24
II.2 Directed Evolution of Glucose Oxidase .......................................................................................... 25
II.2.1 ABTS MTP assay .................................................................................................................. 25
II.2.2 Synthesis of fluorescein tyramide ....................................................................................... 25
II.2.3 Creation of mutant libraries ................................................................................................ 25
II.2.4 Transformation of S. cerevisiae EBY 100 cells..................................................................... 26
II.2.5 Fluorescence staining of yeast cells without compartmentalization ................................. 26
II.2.6 Enzymatic assay for FACS .................................................................................................... 26
II.2.6.1 Agar plate ABTS assay .................................................................................. 27
II.2.6.2 MTP analysis after FACS sorting ................................................................... 28
II.2.7 DNA isolation from yeast .................................................................................................... 28
II.2.8 Cloning in the pICZalpha A vector ....................................................................................... 29
I . Introduction P a g e | 4
II.2.9 Transformation of Pichia pastoris ....................................................................................... 29
II.2.10 Selection of the highest producer ....................................................................................... 30
II.2.11 Fermentation in Pichia pastoris .......................................................................................... 30
II.2.12 Protein purification ............................................................................................................. 30
II.2.13 Determination of kinetic properties of the mutants .......................................................... 31
II.3 Screening System for GOx Activity based on Alginate Beads ........................................................ 31
II.3.1 Synthesis of tyramine alginate ............................................................................................ 31
II.3.2 Modification of lyticase ....................................................................................................... 31
II.3.3 Polymerization of beads ..................................................................................................... 32
II.4 In Vitro Translation (IVT) of GOx .................................................................................................... 32
II.5 Cellulase Screening System ............................................................................................................ 33
II.5.1 APCC synthesis .................................................................................................................... 33
II.5.2 APCC assay in MTPs ............................................................................................................ 34
II.5.3 FACS assay and sorting ........................................................................................................ 34
II.5.4 Agar plate assay for cellulases ............................................................................................ 35
III Results and discussion ........................................................................................................................ 36
III.1 Directed Evolution of Glucose Oxidase ....................................................................................... 36
III.1.1 Cloning in vector pCTCON2 .................................................................................................. 36
III.1.2 Creation of mutant libraries ................................................................................................ 40
III.1.3 Enzymatic assay for FACS .................................................................................................... 43
III.1.4 FACS Sorting ........................................................................................................................ 47
III.1.4.1 Survival of yeast cells .................................................................................. 47
III.1.4.2 Reference libraries ...................................................................................... 48
III.1.4.3 Sorting the consensus libraries ................................................................... 53
III.1.5 MTP assays .......................................................................................................................... 54
III.1.6 Cloning in P. pastoris ........................................................................................................... 57
III.1.7 Purification .......................................................................................................................... 58
III.1.8 Kinetic characterization ...................................................................................................... 60
III.1.9 Thermal stability measurements ........................................................................................ 62
III.2 In Vitro Translation (IVT) of Glucose Oxidase ............................................................................. 63
III.3 Alginate Bead Selection System for GOx .................................................................................... 66
I . Introduction P a g e | 5
III.4 Directed Evolution of Cellulases ................................................................................................. 69
III.4.1 Enzyme assay ...................................................................................................................... 69
III.4.2 FACS assay ........................................................................................................................... 71
IV Final Discussion ................................................................................................................................... 74
IV.1 Glucose Oxidase Screening Systems ........................................................................................... 74
IV.1.1 In vitro translation ............................................................................................................... 76
IV.1.2 Alginate bead selection system for GOx ............................................................................. 77
IV.1.3 Cellulase assay .................................................................................................................... 78
V Conclusions ......................................................................................................................................... 79
VI List of Abbreviations ........................................................................................................................... 81
VII Literature ....................................................................................................................................... 84
I . Introduction P a g e | 6
I Introduction
I.1 The Importance of Enzymes in Industry
For centuries, humankind has used enzymes for a variety of purposes such as brewing,
baking and the production of alcohol. As an example, the first production of cheese can be
traced back to 3000 BC in Egypt (Simoons 1971). However, modern enzyme technology is
considered to have started in 1874 when a Danish scientist made the first enzyme preparation
for an industrial process by extracting dried calf stomach with a saline. The main focus of
enzyme scientists in the nineteenth century was fermentation. Liebig and Pasteur raised an
interesting debate, Liebig arguing that fermentation products are obtained by a chemical
process and that the yeast in the fermentation media are unnecessary, and Pasteur arguing
that the process could not occur unless viable organisms were present. The dispute was settled
by the Buchner brothers who demonstrated that yeast extract can convert glucose to ethanol
just like living yeast cells proving that the conversion is due to the enzymes present in the cells.
The term “enzyme” was actually coined in 1876 by Kuhne derived from the Greek “en” meaning
“in” and “zyme” meaning “yeast”. Numerous discoveries were made over the following decades
but the real turning point for the use of enzymes in industry was made in the early 1960s when
glucoamylase was shown to break down starch completely into glucose, and within a few years
the entire glucose production industry switched from acid hydrolysis to enzyme hydrolysis
(Shanmugam 2009). In the last 20–30 years, enzymes have found their way into a large number
of industrial processes including the production of pharmaceuticals, food, detergents, animal
feed, ethanol and even diagnostics. Today it is estimated that the global industrial enzyme
market is worth approximately 3.3 billion USD per year (BIO030F 2011).
World enzyme demand grew at close to a double‐digit pace from 2003 to 2008, helped
significantly by the rapid increase in world energy prices which made enzyme‐related processes
and products more cost effective (Anon. 2009). According to the “European Markets for
Enzymes in Industrial Applications” report (M145‐39 2007) the major objectives in biocatalyst
development are:
1. The production of biocatalysts which are better, faster, less expensive and more versatile
I . Introduction P a g e | 7
2. The development of biocatalysts that can catalyze a greater range of reactions, have
higher temperature stability and improved solvent compatibility.
I.2 Glucose Oxidase
Glucose oxidase (GOx) (‐D‐glucose: oxygen‐1‐oxidoreductase; EC 1.1.3.4) is found mainly
on the surface of fungi (where it helps protect against bacterial infections) and also in honey
where it acts as a natural preservative (White, Subers, and Schepartz 1963). Interestingly, most
of the known natural GOx enzymes come from fungi (Kriechbaum et al. 1989; Murray et al.
1997; Witt et al. 2000). However, recent discoveries have shown that GOx is also present in the
saliva of some insect larva where it inhibits, by an unknown mechanism, the production of
natural insecticidal substances by plants (Musser et al. 2005).
More than 90% of GOx publications focus on enzymes produced by Aspergillus niger and
Penicillium amagasakiense. The A. niger GOx is a homodimeric enzyme with a molecular weight
of 110–325 kDa, depending on the degree of glycosylation (Kriechbaum et al. 1989). Each
monomer folds into two distinct domains, one binding non‐covalently but tightly the flavin
adenine dinucleotide (FAD) cofactor and the other binding the β‐D‐glucose substrate (Swoboda
1969). The dimer contains two disulfide bonds and two free sulfhydryl groups (Tsuge, Natsuaki,
and Ohashi 1975). The glycosylated enzyme can have a carbohydrate content of up to 16%
(w/w) consisting mainly of N‐linked and O‐linked mannose (80% (w/w) attached to asparagine,
threonine and serine residues via glycosidic bonds (Takegawa et al. 1989). GOx catalyzes the
oxidation of β‐D‐glucose to D‐glucono‐1,5‐lactone. Molecular oxygen acts as the electron
acceptor and is reduced to hydrogen peroxide as shown in Figure 1 (Pazur and Kleppe 1964).
The Michaelis–Menten constant (kM) of the A. niger GOx for β‐D‐glucose varies from 11 to
40 mM (30°C, pH 5.0). This enzyme has a high specificity for β‐D‐glucose (kM = 20 mM, kcat = 400
s‐1), but 2‐deoxy‐D‐glucose can also be accepted as substrate (kM = 29 mM, kcat = 200 s‐1). GOx
has a pH optimum in the 5.5–6.0 interval (Hatzinikolaou et al. 1996) and shows 20% reduced
activity at physiological pH (Bright and Appleby 1969). The optimal temperature of this enzyme
is 20°C, but the enzyme is stable at temperatures below 50°C (Kalisz et al. 1991). GOx is
I . Introduction P a g e | 8
inhibited by hydrogen peroxide (one of the reaction products), halide ions and heavy metals
(Rogers and Brandt 1971).
The enzyme species most widely used for industrial applications comes from A. niger
probably because of the known gene sequence and 3D protein structure but also because of its
commercial availability (Leskovac et al. 2005). The natural host is not ideal for the industrial
manufacture of GOx because of the production of numerous other contaminants (Witteveen,
Veenhuis, and Visser 1992). To overcome this problem Saccharomyces cerevisiae has been
selected as a heterologous host for production of recombinant GOx, because it has a well‐
defined secretory system (Frederick et al. 1990). The secretion of foreign proteins can be
achieved by fusing the leader peptide of the yeast α‐factor pre‐pro‐sequence to the
heterologous gene (Brake et al. 1984). The enzyme produced by S. cerevisiae shows extensive
N‐linked hyperglycosylation in comparison to the A. niger GOx (De Baetselier et al. 1991). The
expression of GOx in Escherichia coli generates a non‐glycosylated enzyme, but this usually
accumulates within inclusion bodies (Witt, Singh, and Kalisz 1998).
Figure 1 ‐ Glucose oxidase reaction. The product of the reductive half‐reaction,
D‐gluconolactone, hydrolyses spontaneously to gluconic acid. In the oxidative half‐
reaction, reduced flavin is reoxidized to FAD by molecular oxygen (Witt et al. 2000).
I . Introduction P a g e | 9
I.3 Cellulases
Lignocelluose is one of the most abundant renewable energy sources on Earth.
Approximately 100 billion tons of carbon per year is thought to be fixed by photosynthesis and
half is in the form of cellulose (Ryu and Mandels 1980). Interest in the decomposition of
cellulose to yield soluble sugars can be traced back to the 1950s, and many advances in
cellulose research have been reported since then with applications in the food, brewing, wine,
animal feed, pulp/paper, agriculture and renewable energy industries (Bhat 2000). The
importance of such enzymes is highlighted by the fact that cellulases, hemicellulases and
pectinases account for 20% of the global enzyme market (Mantyla, Paloheimo, and Suominen
1998).
Cellulose is a linear biopolymer composed of glucose units linked by β(1‐4) glycosidic bonds.
Cellulase activity is a complex process because the natural degradation of cellulose biomass
requires a mixture of hydrolytic enzymes comprising endo‐acting (endoglucanases) and exo‐
acting cellulases (cellobiohydrolases) (Dashtban, Schraft, and Qin 2009). Both enzymes are
inhibited by their products, so the efficient hydrolysis of cellulose requires the presence of a
β‐glucosidase that cleaves the final glycosidic bonds of cellobiose and produces glucose (Maki,
Leung, and Qin 2009). In addition, non‐hydrolytic helper proteins known as swollenins form
part of the hydrolytic complex, and act to make the cellulose substrate more accessible to the
endo and exo‐acting enzymes (Saloheimo et al. 2002).
The decomposition of cellulose into soluble sugars can be achieved chemically or
biologically. In the chemical process, the raw material must be treated with strong acids,
resulting in product losses and equipment damage. Enzyme‐based approaches are therefore
advantageous in principle, but the cost/efficiency ratio of enzymes is not yet competitive (Van
de Vyver et al. 2011).
Many fungi and bacteria possess cellulosic activity but Trichoderma species seem to
produce the most effective cellulase complexes which also happen to be resistant to chemical
inhibitors. These cellulases also have disadvantages such as the inability to digest lignin,
inactivation under typical bioreactor conditions (50°C for 48 h), low specific activity and
I . Introduction P a g e | 10
feedback inhibition by sugars (Ryu and Mandels 1980). All these properties are therefore
important targets for enzyme improvement.
I.4 Directed Evolution
Directed evolution is a protein engineering method used to generate proteins with novel or
improved functions compared to their natural counterparts. Directed protein evolution differs
from natural Darwinian evolution because it requires external intelligence rather than pure
chance, it does not create new species (macroevolution) but improvements within a species
(microevolution), and it does not take millions of years but happens rapidly (Stemmer and
Holland 2003). The first laboratory experiments in this field were performed by Spiegelman
who introduced the concept of “evolution in a test tube” (Mills, Peterson, and Spiegelman
1967). Since then the protein engineering field has blossomed, with success stories spanning a
wide range of applications, including biotransformation for chemical synthesis (Jaeger and
Eggert 2004), bioremediation (Furukawa 2003), biosensors (Doi and Yanagawa 1999), vaccines
(Locher et al. 2004), therapeutic proteins (Kurtzman et al. 2001) and protein structure function
relationships (Lee et al. 2000).
A typical directed evolution experiment includes four major steps (Matsuura and Yomo
2006): 1) diversity generation, 2) cloning and expression of variants, 3) screening for improved
variants, and 4) isolating the gene encoding the improved protein variant (Figure 2).
Figure 2 ‐ General overview of a directed evolution experiment
(modified after Matsuura and Yomo 2006).
I . Introduction P a g e | 11
The fundamental challenges in directed evolution experiments are how to create unbiased
diversity and how to scan through a statistically meaningful fraction of the resulting sequence
space (Tuck Seng Wong, Zhurina, and Schwaneberg 2006).
I.5 Diversity Generation
Many mutagenesis methods have been developed over the last 30 years although
comprehensive comparative studies have not been carried out. PCR‐based methods, especially
epPCR (error prone PCR) are used the most frequently. Taq DNA polymerase (from Thermus
aquaticus) lacks a 3’→5’ proofreading exonuclease activity, and is used in epPCR because it has
an intrinsic high error rate of 8.0 x 10‐6 mutations/bp/duplication (Cirino, Mayer, and Umeno
2003). This achieves approximately one substitution per 125,000 bp per PCR cycle, which is
insufficient to randomize genes that seldom exceed 2000 bp in length. The error rate of the
polymerase can be increased by using MnCl2, unbalanced nucleotide concentrations, longer
extension times, and increased concentrations of MgCl2 (Cadwell and Joyce 1992). Different
polymerases with higher error rates can also be used, like the mutazyme polymerase (Hogrefe
et al. 2002) or an engineered Pfu polymerase (Pfu‐Pol (exo‐) D473G) (Biles and Connolly 2004).
In theory, the concentration of nucleotides or MnCl2 could be controlled to adjust the error rate
to achieve an average of one, two or more amino acid substitutions per protein sequence.
The advantages of epPCR methods are their robustness and simplicity. But there are also
disadvantages like the high number of transitions which give hot and cold spots for
mutagenesis in a protein sequence, the lack of consecutive nucleotide exchange which makes it
impossible to obtain some amino acids substitutions, and the large number of stop codons and
disruptive amino acids like Gly and Pro (Tuck Seng Wong, Roccatano, and Schwaneberg 2007).
Due to the redundancies in the genetic code and to the biased mutational spectra of DNA
polymerase, epPCR methods are limited in their ability to create diversity on the gene level
(Patrick, Firth, and Blackburn 2003).
The inherent bias of epPCR methods can be reduced but the high resulting diversity begs
the question as to whether it is desirable to create extremely large random libraries rather than
more focused high‐quality ones. This means that the libraries should contain a low proportion
I . Introduction P a g e | 12
of the template gene and few stop codons, they can be focused on certain types of amino acid
exchanges such as charged or hydrophobic residues, or they can be focused onto localized parts
of the protein sequence.
Another popular method for diversity generation is DNA shuffling (Muller and Arndt 2007;
Jackel, Kast, and Hilvert 2008). The principle is based on recombining different gene sequences
using multiple PCR cycles (Muller and Arndt 2007; Jackel, Kast, and Hilvert 2008). The most
common form of this method starts from one or multiple genes that are digested by DNase to
form double stranded DNA fragments of 10–50 bp. Multiple rounds of strand separation, re‐
annealing and amplification then reassemble the full‐length gene. More than 60% identity is
usually needed between the sequences that are recombined (Powell et al. 2001). Multiple
variations of this method exist with improved parameters (Reiter et al. 2007; Coco et al. 2001;
Fujii, Kitaoka, and Hayashi 2006; Herman and Tawfik 2007). However, all these methods involve
a certain degree of self‐hybridization among the starting genes, which lowers the quality of the
mutant library due to the high proportion of starting sequences (Reetz 2008).
Rational and semi‐rational library design can also be used for directed evolution but this
requires some structural knowledge in order to make appropriate decisions about the target
residues. Once a target has been decided, the normal approach is to saturate that position with
all 19 possible amino acid substitutions (Hogrefe et al. 2002).
Another semi‐rational approach is based on consensus engineering, which is generally used
to increase protein stability. This is based on the hypothesis that the consensus residue at any
position in a group of aligned homologs is beneficial in terms of protein stability because it has
been conserved during the natural evolution of the protein (Lehmann & Wyss, 2001). This
remains an imprecise technique because all the residues present in a consensus sequence may
not necessarily increase protein stability or activity. For example, in a study of the GroEL
chaperones, 34 amino acid substitutions which occur with less than 35% frequency in their
respective positions in the alignment were chosen for analysis. From these only 13 proved
beneficial, 5 were neutral and 16 were destabilizing. However when amino acids that occurred
with less than 20% frequency were selected, 13 of 18 positions proved beneficial (Wang et al.
1999). The consensus approach is emerging and some fine tuning may be needed, but it has the
I . Introduction P a g e | 13
potential to achieve improvements in a protein that would otherwise take months or even
years using classical directed evolution involving multiple cycles of evolution and the screening
of millions of mutants.
I.6 High Throughput Screening (HTS)
In directed evolution, the production of an extensive and unbiased library is important, but
the availability of a genuinely high‐throughput screening system for the targeted activity
remains a significant bottleneck (T S Wong et al. 2005). Although HTS assays have been
developed to screen for binding interactions (Fields and Song 1989; Feldhaus et al. 2003;
Georgiou 2000; Roberts 1999) selection based on enzyme activity is much more challenging.
The most common screening methods are: (1) colorimetric or fluorometric assays in
microtiter plates (MTPs) (Cohen et al. 2001); and (2) solid‐phase screening on agar plates, filters
or membranes (Lin and Cornish 2002). MTP methods are generally expensive, laborious and
achieve only a moderate throughput (103‐105), and although solid‐phase methods are cheaper,
they are less sensitive and more difficult to implement although with a slightly higher
throughput (104‐106) (M. Olsen, Iverson, and Georgiou 2000). Flow cytometry (Figure 3) is
emerging as the most promising approach because ultra‐high‐throughput screening can be
Figure 3 – Flow chart for FACS screening of gene libraries.
I . Introduction P a g e | 14
achieved (>107 per hour) (Aharoni et al. 2005) even though signal detection can be limited to
fluorescence and accuracy can be impaired by dye diffusion (S. Becker et al. 2004).
The flow cytometer was developed in the 1970s and rapidly became an essential instrument
for biological research, with many applications in medicine (Craig and Foon 2008; Campana and
Coustan‐Smith 2004; Lenkei et al. 1998). State‐of‐the art flow cytometers can analyze up to 13
parameters (forward scatter, side scatter, and 11 fluorescence colors) per cell at rates of up to
100,000 cells per second (Givan 2001). More recently, the technique has been used in the areas
of bacterial genetics and protein engineering (Daugherty, Iverson, and Georgiou 2000).
Regardless of the screening or selection method, it is necessary to maintain a link between
the genotype and phenotype of the target enzymes in order to isolate improved mutants
(Mastrobattista et al. 2005). In the case of MTP assays, this connection is maintained because
individual wells can be matched to the corresponding clones. Linking is more difficult to achieve
in FACS analysis and usually involves some form of in vitro compartmentalization. This is why
flow cytometry screening for improved mutants has been restricted to the intracellular
expression of the mutated protein and to cases where the diffusion of the fluorescence product
can be controlled (Kawarasaki et al. 2003) or where it can be trapped on the cell surface (M. J.
Olsen et al. 2000).
Griffiths and Tawfik have developed in vitro compartmentalization methods involving the
restriction of cells and genes in aqueous microdroplets dispersed in water‐in‐oil emulsions
(Tawfik and Griffiths 1998). This means the gene and the product of the enzyme reaction are
entrapped within artificial compartments that mimic the role of the cell membrane during
natural evolution. Furthermore, fluorescent aqueous droplets in water‐oil‐water double
emulsions can be sorted by FACS and can be recovered intact without content loss (Bernath et
al. 2004). Cells expressing the gene/enzyme of interest are encapsulated in the water‐oil‐water
compartments together with the components necessary for the fluorescence assay. After the
fluorescent product develops the droplets can be sorted using the flow cytometer (Figure 4).
The authors proved the feasibility of the technology using two examples: directed evolution
of thiolactonases where the enzyme was expressed in E. coli cells subsequently entrapped in
double emulsions (Aharoni et al. 2005), and directed evolution of β‐galactosidase where the
I . Introduction P a g e | 15
enzyme was expressed directly in double emulsions by in vitro translation (Mastrobattista et al.
2005).
I.7 Fluorescence Assays
The first rule of directed evolution is “you get what you screen for” (Arnold and Volkov
1999), thus screening platforms must be designed carefully based on the precise catalytic
activity that is sought, preferably using natural substrates. The use of artificial substrates can
select for proteins that bind artificial substrates in preference to natural ones, e.g. the directed
evolution of glycosyltransferase CstII identified a mutant that bound to the fluorescent
indicator dye strongly but was no more active than the wild‐type enzyme in the presence of the
natural substrate (Aharoni et al. 2006).
Since this project involved the development of a HTS assay based on FACS detection, it is
also very important that the assay is compatible with flow cytometry detection systems. One of
Figure 4 The principle of FACS using water‐in‐oil‐in‐water double emulsions. The cells or genes
expressing the gene of interest are encapsulated in the double emulsions together with the
components necessary for the detection of enzyme activity. After the fluorescence product is
formed, the droplets can be sorted by the flow cytometry. Multiple rounds of sorting can be
employed in order to increase the purity of the final population of droplets. Modified after
(Mastrobattista et al. 2005; Aharoni et al. 2005).
I . Introduction P a g e | 16
the products (or substrates) must be distinguished by the FACS lasers and fluorescence filters
and must not cross‐react with any other components used in the screening process.
I.7.1 Fluorescence assays for GOx
There are numerous fluorescence assays in the literature for the detection of GOx activity.
However most are incompatible either with the FACS detection system or with the cell system
itself. The drawbacks of most of these assays are discussed in one of our previous papers (R.
Prodanovic et al. 2012). As discussed above, the connection between the gene and the product
of the enzymatic reaction detected in the assay must be maintained, either by constraining in
the wells of the MTP or using emulsions in the FACS screening system. In the latter case the
product of the enzymatic reaction is a fluorescent soluble compound that, if not enclosed, it
would diffuse away from the cell responsible for producing it. An example of this type of sorting
is discussed in our ViPer assay paper (R. Prodanovic et al. 2012) and other publications (Aharoni
et al. 2005; Mastrobattista et al. 2005). Another way of maintaining the connection between
genotype and phenotype is to use a fluorescence assay that covalently stains the cells
responsible for the enzymatic activity. The tyramide fluorescein assay stains yeast cells by
binding the fluorescence‐labeled tyramine to tyrosine residues on the cell surface (Figure 5).
Since the encapsulation efficiency of yeast cells in double emulsions is less than 20%, covalent
staining of the yeast surface increases the sorting speed by at least 10‐fold.
I . Introduction P a g e | 17
I.7.2 Fluorescence assays for cellulases
Cellulase assays can be classified in three main groups: 1) those detecting product
formation; 2) those monitoring substrate consumption; and 3) those monitoring changes in the
physical properties of the substrate (P. Y.‐H. Zhang et al. 2006).
Most assays belong to the first group, generally involving the detection of reducing sugars.
The most common is the filter paper assay, which is the IUPAC standard procedure for the
measurement of cellulase activity. Whatman filter paper is used as a substrate and the assay
detects the formation of reducing sugars using 3,5‐diniotrosalycilic acid (DNS) (Miller 1959). The
samples need to be boiled extensively to promote color development (Y. H. P. Zhang, Hong, and
Ye 2009). This method has been optimized for the MTP format (Xiao, Storms, and Tsang 2004)
but is generally non‐reproducible, time‐consuming, labor‐intensive and has a low sensitivity
(Dashtban et al. 2010).
Methods based on the detection of reducing sugars have also been developed to screen for
endoglucanase activity, using carboxymethyl‐cellulose substrates. The reaction is performed at
pH 4.8 and 50°C (Mandels, Andreotti, and Roche 1976) after which the reducing sugars are
detected either by HPLC (Fujita et al. 2002), the GOx/peroxidase reaction (Trinder 1969) or
Figure 5 Tyramine‐fluorescein labeling of proteins.
I . Introduction P a g e | 18
using a suitable colorimetric method such as the Somogyi‐Nelson reaction in which copper
oxide acts as an inorganic oxidant (Green, Clausen, and Highley 1989).
Invitrogen (Carlsbad, USA) markets a cellulase activity assay kit based on the detection of
the reducing sugars using a coupled GOx/peroxidase reaction involving Amplex red. However
the products of the cellulose hydrolysis reaction are not glucose but cellobiose and other
polysaccharides. Since GOx is highly specific for glucose and has very low activity with other
sugars (Keilin and Hartree 1952) this assay does not provide an accurate determination of
cellulase activity unless the specific target is β‐glucosidase activity.
Other types of cellulase assays, mostly used for exoglucanases and β‐glucosidases, employ
an artificial substrate that releases a chromophore or a fluorophore after hydrolysis, such as
4‐methylumbelliferyl‐β‐cellobiose (Boschker, Cappenberg, and Using 1994), Lucifer‐yellow‐
cellulose (Tang, Rowell, and Cumming 1995), resorufin‐β‐D‐cellobioside (Coleman, Studler, and
Naleway 2007) and fluorescent microfibrils (Helbert et al. 2003). However, all these assays use
unnatural substrates and therefore pose the risk of selecting for enzymes that happen to favor
these substrates over natural cellulose.
Other assays use agar plates containing cellulosic substrates and colonies with cellulase
activity consume the cellulose around them. The agar plates are then stained with dyes such as
Gram’s iodine (Kasana et al. 2008), hexadecyltrimethyl ammonium bromide (Hankin and
Anagnostakis 1977) or Congo Red (Teather and Wood 1982) allowing the identification of halos
formed around microbes containing cellulase activity. These methods can be used to screen
large numbers of colonies but they are only plus/minus (qualitative) procedures since no linear
relationship can be established between halo zones and enzyme activities (Dashtban et al.
2010).
The highest throughput cellulase assay currently available is based on the MTP format, but
these employ unnatural substrates or detect reducing sugars using a GOx‐coupled reaction,
with the disadvantages noted above. GOx/peroxidase‐coupled assays are better because any
natural or artificial substrate can be used. GOx is highly specific for glucose, but this sugar is
formed in low amounts during cellulose hydrolysis, so this type of coupled assay does not give a
true representation of the cellulase activity. By adapting a novel GOx fluorescence assay (ViPer),
I . Introduction P a g e | 19
previously developed for the detection of GOx activity, the fluorescent products of the reaction
were found to be compatible with the HTS of double emulsions using FACS (R. Prodanovic et al.
2012). For the cellulase assay, the same principle was applied but substituting GOx for hexose
oxidase (HOx), because this has a much broader substrate specificity and can use D‐glucose, D‐
galactose, maltose, cellobiose and lactose as substrates (Hansen and Stougaard 1997). This
makes it possible to detect a much wider range of sugars formed during cellulase activity,
starting from any natural or artificial cellulosic substrate.
I.8 Enzyme‐induced Polymerization
Alginates are linear unbranched polymers containing β(1,4)‐linked D‐mannuronic acid (M)
and α(1,4)‐linked L‐guluronic acid (G) residues (Figure 6a) (Heng, Chan, and Wong 2003). They
represent a class of copolymers usually found in the brown algae such as Macrocystis pyrifera
and Laminaria hyperborean, but also produced by bacteria such as Azotobacter vinelandii, A.
chroococcum and several Pseudomonas species. The ratio of alternating residues differs
according to the source organism, and the different monomer ratios lead to preferred
copolymer conformations (Sabra, Zeng, and Deckwer 2001). The free carboxylic acids make
hydrogen bonds with water molecules (pKa M 3.38, pKa G 3.65). Ca2+ ions can replace this
hydrogen bonding, zipping the guluronate chains together, but not the mannuronate chains
(Figure 6b) (Draget et al. 2003).
The main application of alginate is as a matrix polymer for the microencapsulation of drugs
(Tonnesen and Karlsen 2002), proteins (Coradin, Nassif, and Livage 2003), cells (Smidsrod and
Skjak‐Braek 1990) and DNA (Quong et al. 1998).
I . Introduction P a g e | 20
Microencapsulation in calcium alginate beads has been explored in various techniques, each
with advantages and disadvantages. The most common preparation method for calcium
alginate microspheres is extruding sodium alginate solution as droplets into calcium chloride
solution, which generates beads larger than 100 µm (Heng, Chan, and Wong 2003). Smaller
particles can be generated by emulsification methods (Chan, Heng, and Wan 1997). Alginate
microspheres have pore sizes of 6.8–16.6 nm (Klein, Stock, and Vorlop 1983), depending on the
diameter of the beads. Producing microbeads in which the capsule membrane has a
controllable weight cut off can be achieved by coating the bead surface with poly‐L‐lysine (PLL),
by adjusting the alginate‐PLL reaction time, the molecular weight of the PLL, the concentration
of PLL and by mixing PLL preparations with different molecular weights (King et al. 1987).
Peroxidase is often used for the polymerization of phenol derivatives under mild conditions
(Fukuoka, Uyama, and Kobayashi 2004; Kurisawa et al. 2005). Sakai et al. developed a phenol‐
substituted alginate which could form cell‐enclosing capsules for biomedical applications.
Phenol‐substituted alginates can be synthesized from alginate and tyramine using aqueous
phase carbodiimide activation chemistry. Horseradish peroxidase (HRP) causes the gelation of
the phenol alginate in the presence of peroxide (Sakai and Kawakami 2007) (see Figure 7). We
can couple the GOx reaction to the gelation of tyramine alginates thus creating a new screening
system based on GOx activity.
Figure 6 ‐ a. Structural unit of alginates. b. Ca2+ ion complexes with the
guluronate chains (Draget et al. 2003).
I . Introduction P a g e | 21
I.9 Project Objectives
From an industrial point of view both GOx and cellulase are important enzymes, with a
combined market value of more than 7 million USD per annum (Godfrey and West 1996;
Newman and Turner 2005). High specific activity is the most desirable property because less of
the enzyme is needed to complete the process and this can significantly reduce production
costs. As well as meeting these industrial criteria by developing better catalysts, this project
also set out to establish novel HTS systems for two different enzyme classes (oxidases and sugar
hydrolases). We selected GOx and cellulase as our model enzymes, reflecting their importance
as industrial biocatalysts. We set out to create a reliable screening system because this remains
the bottleneck in any directed evolution approach (Gumulya and Reetz 2011).
I.10 Project Overview
We aimed to develop a novel HTS system for the selection of enzymes with tailored
properties, based on FACS in double emulsions (Figure 8).
Figure 7 a. Formation of alginate tyramine conjugate. b. The
gelation of alginate–tyramine conjugates by HRP‐catalyzed oxidative
coupling of phenols (Sakai and Kawakami 2007).
I . Introduction P a g e | 22
GOx is used in the food industry to remove oxygen from beverages (Labuza and Breene
1989), as a source of hydrogen peroxide for food preservation (C. M. Wong, Wong, and Chen
2008), for improving dough quality (Rasiah et al. 2005) and as a stabilizer for some food
additives (James and K Simpson 1996). In pharmaceutical and analytical biochemistry, GOx is
used for the quantitative determination of glucose levels (M Rossi, D Quach, and Rosenzweig
2004) and in the medical field it is an important component of biosensors (G Wagner et al.
1998) and generators using blood glucose as the energy source (Pizzariello, Stred’ansky, and
Miertus 2002). The improvement of GOx activity would therefore benefit a broad range of
applications, but this project focused on the properties of GOx required for its use in the
production of miniaturized biofuel cells. We aimed (1) to increase electron transfer from the
electrode to the enzyme (increase enzyme activity); and (2) to improve GOx activity under
physiological conditions, which differ from the optimal conditions of the wild type enzyme
(increased activity at pH 7.4, reduced value of kM for glucose).
Cellulases are widely used in food processing (Kasai et al. 2006), textile manufacture and in
laundry detergents (Kudriashova et al. 2009), they are used in the pulp and paper industry
(Prasetyo et al. 2011), in the conversion of biomass to biofuels (McMillan et al. 2011) and in a
pharmaceutical context to remove undigested plant material (phytobezoars) from the gut
(Fernández Morató et al. 2009). The improvement of cellulases would therefore also provide
benefits in many industries, thus we sought to develop a FACS‐based screening system using
natural cellulase substrates.
In fulfilling these objectives we have developed HTS systems for both enzymes, establishing
fluorescence assays tailor‐made for screening the desired activities, but the principles can be
applied to all glycosidases and all oxidases. Improved GOx mutants were identified by screening
gene libraries created using the consensus approach.
I . Introduction P a g e | 23
Figure 8 Flowchart of the PhD thesis work. The first step focused on technology development, e.g. finding the best way
to create mutant gene libraries, developing a reliable HTS system and finding suitable expression hosts for the selected
enzymes. The second step focused on applying the developed technology to screen mutant gene libraries.
I I . Materials and Methods P a g e | 24
II MaterialsandMethods
II.1 General Recipes for Media
II.1.1 LB media (Bertani 1951)
We prepared 0.5% (w/v) yeast extract, 1% (w/v) peptone, 1% (w/v) NaCl, pH 7.4. The media
was autoclaved at 120°C for 20 min. Agar plates were prepared by adding 1.5% agar before
autoclaving. After the media cooled to 60°C, we added 100 µg/mL ampicillin, mixed thoroughly
and poured the media into Petri dishes.
II.1.2 YNB CAA Glu/Gal, Raf (Karst, Sadeghi, and Menees 1999)
We prepared 0.51% (w/v) ammonium sulfate, 0.16% (w/v) yeast nuclear base without
ammonium sulfate or amino acids, 0.5% (w/v) casamino acids. The media was autoclaved at
120°C for 20 min. Glucose solution was autoclaved separately and added to the media at a final
concentration of 2% (w/v). Gal/Raf media was prepared by filter sterilizing the solutions and
adding to final concentrations of 2% (w/v) galactose and 1% (w/v) raffinose. Chloramphenicol
(50 µg/mL) was added to the media if necessary. Agar plates were prepared as above.
II.1.3 YPD (Kaiser, Michaelis, and A. Mitchell 1994)
We prepared 1% (w/v) yeast extract, 2% (w/v) peptone, 0.008% (w/v) adenine hemisulfate.
The media was autoclaved at 120°C for 20 min. Glucose solution was autoclaved separately and
added to the media at a final concentration of 2% (w/v). Zeocin was added to the media if
necessary at different concentrations. Agar plates were prepared as above.
II.1.4 BMGY/BMMY (Invitrogen 2010)
We prepared 1% (w/v) ammonium sulfate, 0.34% (w/v) yeast nuclear base without
ammonium sulfate or amino acids, 1% (w/v) yeast extract and 2% (w/v) peptone. The media
was autoclaved at 120°C for 20 min before supplementing with 100 mM potassium phosphate
buffer pH 6.0 (autoclaved) and 0.00004% (w/v) biotin (filter sterilized). Glycerol solution was
autoclaved separately and added to the media at a final concentration of 2% (v/v). To prepare
I I . Materials and Methods P a g e | 25
BMMY medium, we added methanol to a final concentration of 0.5% (v/v). Chloramphenicol
(50 µg/mL) was added to the media if necessary. Agar plates were prepared as above.
II.2 Directed Evolution of Glucose Oxidase
II.2.1 ABTS MTP assay
For rapid confirmation of GOx activity, we used the MTP plate format of the ABTS assay
modified after (Zhu et al. 2006). Samples (1–10 µL) were added to an ABTS mix comprising 300
mM glucose, 1 U/mL HRP and 4 mM ABTS. Absorbance was monitored at 405 nm for 10 min.
The absorption coefficient of oxidized ABTS is εABTS ox = 36.8 cm2/µmol.
II.2.2 Synthesis of fluorescein tyramide
Fluorescein tyramide was synthesized as described (Van Gijlswijk et al. 1997) using 20 mM
5/6‐carboxyfluorescein succinimidyl ester (Pierce, Illinois, USA) in dimethylformamide (DMF)
mixed with 25 mM triethylamine and 20 mM tyramine HCl (Sigma‐Aldrich Chemie GmbH,
Munich, Germany) also in DMF. The mix was kept at room temperature for 2 h in the dark and
ethanol was added in order to obtain a 2 mM fluorescein tyramide solution which was kept at
4°C until required. Carboxy coumarin tyramide was obtained using the same synthesis protocol
starting from N‐succinylimidil 7‐methoxy coumarin‐3‐carboxylate (Pierce, Illinois, USA).
II.2.3 Creation of mutant libraries
The consensus library was created using the Quick Change Multi site‐directed mutagenesis
kit (Agilent Technologies, Santa Clara, USA) comprising PCR enzymes and E. coli XL10 Gold ultra‐
competent cells. Site‐directed primers were designed at 16 relevant positions and were
synthesized by Eurofins MWG Operon (Ebersberg, Germany). The PCR was performed according
to the manufacturer’s instructions using the GOx sequence cloned in pCTCON2 (containing a
c‐myc tag) as the template (400 pg/µL) and the primers adjusted to 200 nM. The cycling
conditions were 95°C for 1 min, 1 cycle; 95°C for 1 min/55°C for 1 min/65°C for 16.5 min, 30
cycles; 65°C for 20 min, 1 cycle, after which the products were digested with DpnI for 3 h at
37°C and stored at 4°C until required.
I I . Materials and Methods P a g e | 26
Single stranded PCR products were introduced into E. coli XL10 Gold cells according to the
manufacturer’s protocol and the transformants were plated on LB Amp agar and grown at 37°C
for 16 h. The cells from all the plates were collected in liquid LB Amp media and re‐grown for
another 16 h at 37°C, 160 rpm. The DNA was then isolated using the Macherey‐Nagel Plasmid
DNA kit (Düren, Germany).
II.2.4 Transformation of S. cerevisiae EBY 100 cells
The S. cerevisiae EBY 100 strain and the pCTCON2 vector were kindly provided by Dane
Wittrup (MIT, MA, USA). The DNA was introduced into yeast cells (Gietz and Schiestl 2007)
using a 2.5‐h 42°C heat shock. After transformation, the cells were grown directly in YNB‐CAA
glucose medium for 48 h at 27°C, 160 rpm. To induce GOx expression, the cells were
transferred to YNB‐CAA Gal/Raf media and grown at 27°C, 160 rpm, for 16‐18 h and then used
for FACS experiments.
II.2.5 Fluorescence staining of yeast cells without compartmentalization
Yeast cells expressing GOx were washed three times in PBS and diluted to 109 cells/mL
(OD600 = 1 is equivalent to 2 x 107 cells/mL). A 25 µL suspension containing 107 cells was mixed
with 10 U/mL HRP (AppliChem GmbH, Gatersleben, Germany) and with the specified
concentrations of tyramide‐fluorescein or tyramide‐AMCA.
II.2.6 Enzymatic assay for FACS
Yeast cells expressing GOx were washed three times in PBS and diluted to 109 cells/mL
(OD600 = 1 is equivalent to 2 x 107 cells/mL). A 25‐µL suspension containing 107 cells, 20 µM
tyramide‐fluorescein, 10 U/mL HRP (AppliChem GmbH, Germany) and 5 U/mL almond
β‐glucosidase (Sigma‐Aldrich Chemie GmbH, Germany) was added to 500 µL 1.5% Abil Em 90
(Tego, Germany) in LMO (Sigma‐Aldrich Chemie GmbH, Germany). The mix was emulsified
using a MICCRA D‐1 dispenser at 8000 rpm for 3 min on ice and supplemented with
octylglucoside (Sigma‐Aldrich Chemie GmbH, Germany) from a 37 mM solution in ethanol. The
emulsion was vortexed for 30 s and incubated for 30 min at room temperature.
I I . Materials and Methods P a g e | 27
Cells were recovered from the emulsion by stirring (5 min, 1200 rpm) with 1 mL LMO
containing 3% (v/v) Span 80 (Sigma‐Aldrich Chemie GmbH, Munich, Germany). The emulsion
was centrifuged (2 min, 14,000 x g), the supernatant discarded and the cells were rinsed three
times with LMO containing 3% (v/v) Span 80, twice with PBS containing 0.5% (v/v) Triton X‐100
(Sigma‐Aldrich Chemie Gmbh, Munich, Germany) and finally three times with 0.1% (w/v) BSA in
PBS. The cells were resuspended in 100 µL of the BSA solution, 30 µg/mL of mouse anti c‐myc
antibody (Abcam, Cambridge, UK) was added and the mixture was shaken gently for 1 h at
30°C. The cells were washed again three times with 0.1% (w/v) BSA in PBS and resuspended in
100 µL of the same solution. The secondary anti‐mouse IgG antibody was added and the
mixture was incubated at 4°C for 1 h. The secondary antibody was either goat anti‐mouse Fc
IgG phycoerythrin (Pierce, Illinois, USA) (1:100 dilution) or goat anti‐mouse Fc IgG Dylight 650
(Abcam, Cambridge, UK) (1:50 dilution). Before FACS analysis, the cells were washed three
times in 0.1% (w/v) BSA in PBS and finally resuspended in 1 mL PBS.
The cells were analyzed using a BD FACS Diva (NJ, USA) flow cytometry system with the
trigger parameter set on forward scattering. The analysis rate was 1000–5000 events/s and the
sorting speed was 10–100 events/s. The 488 and 647 nm laser excitation wavelengths were
used for detection and emissions were detected using 530 and 670 nm filters. The positive cells
were gated on a fluorescence double plot as specified in each experiment. The cells were
sorted in the single cell mode onto YNB‐CAA glucose chloramphenicol agar plates or into MTP
wells containing 100 µL YNB‐CAA glucose liquid medium plus 50 µg/mL chloramphenicol.
II.2.6.1 Agar plate ABTS assay
After FACS, the cells were grown at 27°C for 3 d, replica plated onto YNB‐CAA Gal/Raf
medium and cultivated for another 24 h. Screening medium was prepared by mixing 2% agar
with an equal volume of ABTS solution containing 333 mM glucose, 1.75 U/mL HRP and 7 mM
ABTS. This was poured over the agar cell plates. Green halos were observed around colonies
with GOx activity.
I I . Materials and Methods P a g e | 28
II.2.6.2 MTP analysis after FACS sorting
We used a previously‐described protocol (Zhu et al. 2006; Baron et al. 1994; Sun et al. 2001)
modified as follows. The cells were sorted into MTP wells and grown for 2 d at 30°C, 450 rpm,
80% humidity. The plates were sealed with Parafilm to prevent evaporation. We then
transferred 5 µL of the cell culture to new MTP wells containing 25 µL YNB CAA Glu medium
and the cells were allowed to grow for a further 24 h before adding 80 µL YNB CAA Gal/Raf
medium and growing for another 24 h.
We resuspended 5 µL of the above cells in 70 µL PBS and the OD600 was determined, then
70 µL ABTS solution was added and the kinetics were measured at 405 nm every 20 s for 10
min. Two measurements were taken from each culture, one using 4 mM ABTS solution
containing 250 mM glucose and 1 U/mL HRP, and one with the same components but only 5
mM glucose.
For each well measurement, we calculated the slope of the linear region. The slopes were
normalized to the OD600 in each well and the values were compared with the values of the cells
expressing the wild type enzyme. The best mutants were selected and re‐screened using
columns 1 and 12 of an appropriate number of 96‐well plates filled with 20 µL of YNB CAA Glu
chloramphenicol medium. These were inoculated with 2 µL of a saturated culture containing
the mutants while two wells were used for the standard cells expressing the wild type enzyme.
The plates were wrapped in Parafilm and incubated at 30°C and 450 rpm for 24 h at 80%
humidity.
Columns 2 to 11 of the rescreening plates were filled with 25 µL of YNB CAA Glu
chloramphenicol medium. Using an 8‐channel pipette, the cultures in columns 1 and 12 were
resuspended and 3 µL of each suspension was transferred from column 1 to columns 2–6 and
from column 12 to columns 7–11. The screening protocol described above was repeated.
II.2.7 DNA isolation from yeast
The DNA from mutants with higher activity than the wild type variant were isolated as
described (Singh and Weil 2002) with the following modifications. The cells were grown for 1 d
in 5 mL YNB CAA Glu medium at 27°C, 160 rpm. They were centrifuged at 11 000 x g for 1 min
and the supernatant was discarded. We added 500 µL lysis buffer (50 mM Tris‐HCl pH 7.5, 10
I I . Materials and Methods P a g e | 29
mM EDTA, 0.5% (v/v) β‐mercaptoethanol) and 100 U/mL lyticase from Arthrobactor luteus
(Sigma‐Aldrich Chemie Gmbh, Munich, Germany) and the reaction was incubated at 37⁰ C for 2
h. The cells were centrifuged as above, plasmid DNA was isolated using the Macherey‐Nagel
Plasmid DNA kit (Düren, Germany) and the DNA was sent for sequencing.
II.2.8 Cloning in the pICZalpha A vector
The best mutants were re‐cloned in vector pICZalpha A (Invitrogen, Darmstadt, Germany).
The cloning was performed using standard restriction/digestion and PCR methods (94°C for 4
min, 1 cycle; 94°C for 1 min/55°C for 1 min/72°C for 2.3 min, 30 cycles; 72°C for 10 min, 1 cycle)
using 0.04 U/µL Pfu DNA polymerase (Fermentas GmbH, St. Leon‐Rot, Germany), 0.2 mM dNTP
mix (Fermentas GmbH, St. Leon‐Rot, Germany), 0.5 µM forward primer 5’‐ATC TCT CGA GAA
AAG AAG CAA TGG CAT TGA AG‐3’, 0.5 µM reverse primer 5’‐GAA GCT CTA GAG CTC ACT GCA
TGG‐3’ and 50 ng pCTCON2 plasmid containing mutant sequences as the template. The primers
were ordered from Eurofins MWG Operon (Ebersberg, Germany). The purified fragments were
digested using XhoI/XbaI (New England Biolabs GmbH, Frankfurt am Main, Germany) and
ligated in to the pICZalpha A vector. The ligation mixture was directly introduced into E. coli
XL10 Gold cells (Agilent Technologies, Santa Clara, USA). Plasmid DNA was isolated using the
Macherey‐Nagel Plasmid DNA kit (Düren, Germany) and the DNA was sent for sequencing.
II.2.9 Transformation of Pichia pastoris
Competent Pichia pastoris KM71H cells (Invitorgen, Darmstadt, Germany) were prepared
according to the published protocol (D. M. Becker and Guarente 1991). The DNA isolated from
E. coli cells was digested with PmeI (New England Biolabs GmbH, Frankfurt am Main, Germany)
and purified using the Macherey‐Nagel DNA clean‐up kit (Düren, Germany). Up to 5 µL of
solution containing 500–1000 ng of DNA was added to 40 µL of frozen competent cells on ice
and left to thaw. The cells were mixed with the DNA and transferred to 2‐mm gap
electroporation cuvettes for electroporation using an Eppendorf multiporator (Eppendorf AG,
Hamburg, Germany) at 2500 V for 5 ms. Ice‐cold 1 M sorbitol was added and the cells were
transferred to 3 mL YPD medium and left to recover for 1 h at 27°C, 160 rpm. The cells were
I I . Materials and Methods P a g e | 30
then centrifuged at 3000 x g for 5 min, resuspended in 1 mL medium and plated on YPD plates
containing 250 µg/mL zeocin and incubated for 3 d at 27°C.
II.2.10 Selection of the highest producer
We transferred 96 colonies from the transformation plates into MTP wells containing 100
µL BMGY medium. The cells were grown to saturation (24 h) at 30°C, 450 rpm, 80% humidity
and then diluted five times in 30 µL BMGY medium on new MTPs and grown for an additional
24 h. We added 80 µL of BMMY and the cells were incubated under the same conditions for an
additional 24 h to express GOx. The activity of each colony was determined using the ABTS MTP
protocol described above with the ABTS mix containing 250 mM glucose. The best producing
clone of each mutant was selected.
II.2.11 Fermentation in Pichia pastoris
Cells from agar plates were inoculated into 25 mL BMGY medium and incubated for 24 h at
27⁰C, 160 rpm. For the optimization of fermentation, the cells were centrifuged at 3000 x g, 10
min and resuspended in 5 mL BMMY medium (OD600 ~30) and incubated at 27⁰C, 160 rpm.
Every 24 h, we added 0.5 % (v/v) methanol to the fermentation medium and the GOx activity
was measured.
Once the optimal conditions had been determined, the fermentation was scaled up to 2 L
by transferring the 25 mL initial culture into 2 L BMGY medium followed by incubation for an
additional 24 h. After centrifugation as above the cells were transferred to 400 mL BMMY
medium. This protocol was modified from the manufacturer’s instructions (Invitrogen 2010).
II.2.12 Protein purification
After fermentation for 4 d according to Invitrogen recommendations, cells were pelleted by
centrifugation at 11,000 x g for 10 min using a Beckman Coulter Avanti J26 XP centrifuge
(Krefeld Germany). The supernatant was collected and filtered through a 0.22‐µm PTFE filter
(Carl Roth GmbH, Germany) and the filtrate was concentrated to 5–10 mL using a Viva Flow 50
system with a 50‐kDa membrane (Sartorius AG, Germany). The concentrate was dialyzed
against 10 mM phosphate buffer (pH 6.0) overnight at 4°C, and loaded onto a 20 ‐ mL Fast Flow
I I . Materials and Methods P a g e | 31
DEAE Sepharose column (GE Healthcare Europe GmbH, Germany) using the ÄKTApurifier (GE
Healthcare Europe GmbH, Germany). The protein was purified using a linear gradient from 10
to 300 mM phosphate buffer pH 6 over 15 column volumes. We tested 50‐mL fractions using
the ABTS assay and those with separate peaks of GOx activity were collected and concentrated
to 5 mL using 10‐kDa ultrafiltration columns (Sartorius AG, Germany).
II.2.13 Determination of kinetic properties of the mutants
The kinetic characteristics of each GOx variant were determined using triplicate MTP ABTS
assays with glucose concentrations ranging from 1.2 to 266 mM, at pH 5.5 and 7.4. The slope of
each measurement was calculated over the linear region and fitted onto Michaelis‐Menten
curves to allow the kM and kcat values to be determined. Lineweaver‐Burk, Eadie‐Hofstee and
Hanes‐Woolf plots were also constructed, and the outliers were identified and removed. We
determined kcat values by measuring absorbance at 280 nm (the absorption of 1.5 mg/mL GOx
is considered equivalent to 1 AU based on the sequence, as calculated using ProtParam).
II.3 Screening System for GOx Activity based on Alginate Beads
II.3.1 Synthesis of tyramine alginate
Tyramide alginate was synthesized as previously described (Sakai and Kawakami 2007). We
mixed 50 mL 1% (w/v) sodium alginate in water with 50 mL 9% (w/v) tyramine hydrochloride in
water containing 50.4 mM EDAC and 25.2 mM NHS. The mix was stirred for 20 h at room
temperature and dialyzed against water for 48 h with 5–6 changes. The resulting polymer was
lyophilized and kept at room temperature until needed. All chemicals were ordered from
Sigma‐Aldrich Chemie GmbH (Munich, Germany).
II.3.2 Modification of lyticase
The protocol we used was previously described (Carraway and Triplett 1970; Anjaneyulu
and Staros 1987) with the following modifications. We dissolved 2 mg poly‐L‐lysine in 1 mL PBS
and slowly added 0.8 mg EDAC (4 mM final concentration) and 2.2 mg sulfo‐NHS (10 mM) with
constant mixing. The solution was shaken gently at room temperature for 15 min and 2 mg/mL
I I . Materials and Methods P a g e | 32
of Arthrobactor luteus lyticase was added and incubated for 4 h at room temperature. The
solution was dialyzed against PBS and concentrated on a 10‐kDa ultrafiltration centrifugation
column (Sartorius AG, Goettingen, Germany). The concentration of enzyme was measured at
280 nm and the samples were aliquotted and kept at –20⁰C until needed. All chemicals were
purchased from Sigma‐Aldrich Chemie GmbH (Munich, Germany).
II.3.3 Polymerization of beads
We prepared a solution comprising 2% (w/v) tyramine alginate, 20 U/mL HRP and 250 mM
glucose, and mixed 200 µL with either GOx xpressing cells (105 S. cerevisiae EBY 100 cells
expressing GOx cloned in the pCTCON2 vector), a reference library or soluble GOx (5 U/mL). The
mixture was transferred to 600 µL 3% (v/v) Span 80 in LMO and stirred with a magnetic stirrer
at 10,000 rpm for 5 min and at 2000 rpm for 1 h. The beads and cells were removed from the
emulsion by washing three times with 1% (v/v) Triton X‐100 and twice with 50 mM Tris‐HCl (pH
7.4) interspersed with 2‐min centrifugation steps at 200 x g. The beads and cells were shaken
gently for 1 h in 0.15% (w/v) poly‐L‐lysine and were then transferred into the lyticase lysis
buffer (50 mM Tris HCl pH 7.4, 10 mM EDTA, 0.5 % (w/v) β‐mercaptoethanol) containing 15
µg/mL poly‐L‐lysine cross‐linked lyticase. The samples were incubated for 1 h at 27⁰C and
plated on YNB CAA Glu medium. Colonies were counted and a replica plate was prepared to
assess the positive and negative cells using the ABTS agar plate assay. All chemicals were
obtained from Applichem (Darmstadt, Germany) except the LMO which was acquired from
Sigma‐Aldrich Chemie GmbH (Munich, Germany).
II.4 In Vitro Translation (IVT) of GOx
GOx sequences were expressed using the EasyXpress insect cell protein synthesis kit from
Qiagen (QIAGEN GmbH, Hilden, Germany). The wild‐type GOx was cloned in vector pIX4 with
and without the pro‐‐factor signal sequence. Cloning was carried out using the standard
restriction/digestion and PCR methods (94°C for 4 min, 1 cycle; 94°C for 1 min/55°C for 1
min/72°C for 2.3 min, 30 cycles; 72°C for 10 min, 1 cycle) in a mix comprising 0.04 U/µL Pfu DNA
polymerase (Fermentas GmbH, St. Leon‐Rot, Germany) 0.2 mM dNTP mix (Fermentas GmbH,
St. Leon‐Rot, Germany), 0.5 µM forward primer 5’‐TCA TCG AGG AGG TCT CCC ATG AGA TTT
I I . Materials and Methods P a g e | 33
CCT TCA ATT TTT AC‐3’ for the pro‐‐factor sequence or 5’‐ATC ATC GAG GAG GTC TCC CAT
GAG CAA TGG CAT TGA AGC CAG‐3’ for the mature protein, 0.5 µM reverse primer 5’‐ATC ATC
AGA TCT TTA CTG CAT GGA AGC ATA ATC TT‐3’, and 50 ng AF GOX pUC57 as the template
plasmid. The primers were ordered form Eurofins MWG Operon (Ebersberg, Germany). The
purified fragments were digested using BseRI/BglII (New England Biolabs GmbH, Frankfurt am
Main, Germany) and ligated into the pIX4 vector. The ligation mixture was directly introduced
into E. coli XL10 Gold cells (Agilent Technologies, Santa Clara, USA). Plasmid DNA was isolated
using the Macherey‐Nagel Plasmid DNA kit (Düren, Germany) and the DNA was sent for
sequencing.
We also tested the TnT® T7 Insect Cell Extract Protein Expression System (Promega GmbH
Mannheim, Germany). In this case the wild‐type GOx sequence was cloned with and without
the pro‐‐factor signal sequence into vector pF25A, using standard restriction/digestion and
PCR methods (94°C for 4 min, 1 cycle; 94°C for 1 min/55°C for 1 min/72°C for 2.3 min, 30 cycles;
72°C for 10 min, 1 cycle) in a mix comprising 0.04 U/µL Pfu DNA polymerase (Fermentas GmbH,
St. Leon‐Rot, Germany) 0.2 mM dNTP mix (Fermentas GmbH, St. Leon‐Rot, Germany), 0.5 µM
forward primer 5’‐ATC ATC GCG ATC GCA TGA GAT TTC CTT CAA TTT TTA C‐3’ for the pro‐‐
factor sequence or 5’‐ATC ATC GCG ATC GCA TGA GCA ATG GCA TTG AAG CCA G‐3’ for the
mature protein, 0.5 µM reverse primer 5’‐ATC ATC GTT TAA ACT CAC TGC ATG GAA GCA TAA
TCT T‐3’ and 50 ng AF GOX pUC57 as the template plasmid. The primers were ordered form
Eurofins MWG Operon (Ebersberg, Germany). Fragment purification, ligation, transformation,
DNA isolation and sequencing were carried out as described above.
IVT was carried out according to the manufacturer’s instructions and GOx activity was
detected using the ABTS assay (section II.2.1).
II.5 Cellulase Screening System
II.5.1 APCC synthesis
We synthesized 3‐carboxy‐7‐(4’‐aminophenoxy)‐coumarine (APCC) as previously described
(Khanna, Chang, and Ullman 1989) although the reduction from 3‐carboethoxy‐7‐(4’‐
nitrophenoxy)coumarine to 3‐carboethoxy‐7‐(4’‐aminophenoxy) coumarine was achieved using
I I . Materials and Methods P a g e | 34
iron powder in ethanol (Agrawal and Tratnyek 1996) after characterization by NMR
spectroscopy (Setsukinai et al. 2000). The substrate for APCC synthesis (3‐carboethoxy‐7‐
(hydroxy) coumarine) was synthesized as previously described (Chilvers et al. 2001).
II.5.2 APCC assay in MTPs
Fluorescence‐based assays were carried out in 96 MicroWell™ Nunclon™Δ Optical Black Flat
Bottom Plates (Greiner Bio One GmbH, Frickenhausen, Germany) using a BioTec Synergy 2
Multi‐Mode Microplate Reader (Bad Friedrichshall, Germany) in a total reaction volume of 50
µL. The reaction mix comprised APCC (500 µM), VbrPOx (0.1 U/mL), Hox (3.4 U/mL), CMC
(0.3%), NaBr (50 mM) in Tris H2SO4 (50 mM, pH 7.4), supplemented with either different
concentrations of purified A. niger cellulase (Sigma‐Aldrich Chemie Gmbh, Munich, Germany) or
107 cells expressing cellulase Cel5A. HOx was kindly provided by Danisco Deutschland GmbH
(Frankfurt, Germany). S. cerevisiae YPH500 cells (Agilent Technologies, Santa Clara, USA)
expressing enzyme Cel5A using the pESC‐Trp vector were previously cloned in our laboratory.
The assay was monitored at 360/460 nm excitation/emission wavelengths. Some assays used
APF (aminophenoxifluorescein) (Invitrogen Karlsruhe, Germany) in place of APCC and were
monitored at 485/528 nm excitation/emission wavelengths.
II.5.3 FACS assay and sorting
The in vitro compartmentalization of yeast cells was carried out as previously described,
with modifications (Aharoni et al. 2005). The washed yeast cells, together with the reaction
components used in the MTP assay described above (25 µL) were added to 250 µL of the ice
cold oil phase, comprising 1.5% (v/v) ABIL EM 90 (Tego, Germany) in LMO. The two phases were
homogenized on ice in a 2 mL round bottom cryotube for 3 min at 8000 rpm using a T18 basic
ULTRA‐TURRAX® from IKA (Germany). The second water phase (250 µL) containing 1.5% (w/v)
carboxy‐methylcellulose sodium salt (CMC) and 1% (v/v) Triton X‐100 in PBS was added to the
primary emulsion and homogenized on ice for 3 min at 7000 rpm. Emulsions were diluted with
PBS and analyzed using a BD FACS Diva (NJ, USA) flow cytometer. The trigger parameter was set
to forward scattering. The analysis rate was 8000–20,000 events/s and the sorting speed was
10–100/s. We used 488 and 375 nm lasers for excitation and 460 and 520 nm filters for
I I . Materials and Methods P a g e | 35
detection. The positive emulsion droplets were gated on the FL3‐DAPI, FL1‐FITC plot and sorted
in single‐cell mode directly onto YNB‐CAA Glu agar plates supplemented with 25 µg/mL
chloramphenicol. After 2 days at 27°C, cellulase expression was induced by replica plating and
the cells were cultivated on YNB CAA Gal/Raf containing 0.1% CMC (1 day, 27°C).
II.5.4 Agar plate assay for cellulases
The agar plate method to detect cellulase activity was carried out as described, with
modifications (Farkas, Lisikova, and Biely 1985). Cells growing on YNB CAA Gal/Raf 0.1% CMC
plates were overlaid with 0.1% Congo Red (Sigma‐Aldrich Chemie Gmbh, Munich, Germany)
and incubated for 30 min at room temperature. The plates were washed with 1 M NaCl and 5%
acetic acid. Cells expressing cellulase activity were revealed by halos.
I I I . Results and discussion P a g e | 36
III Results and discussion
III.1 Directed Evolution of Glucose Oxidase
III.1.1 Cloning in vector pCTCON2
Transformation efficiency is one of the bottlenecks in any directed evolution experiment.
Although the highest transformation efficiencies are achieved in E. coli, GOx accumulates within
inclusion bodies because it is an eukaryotic enzyme with two disulfide bonds and multiple
glycans (Witt, Singh, and Kalisz 1998). An eukaryotic host would be more suitable, and S.
cerevisiae is a good candidate because of its transformation efficiency, the stability of episomal
plasmid DNA and its fast growth.
The linkage between the protein of interest and its corresponding gene must be maintained
throughout the experiment. It would be ideal to have the protein displayed on the surface of
the yeast cell especially in the case of FACS screening systems. For this purpose we used the
pCTCON2 yeast display vector and S. cerevisiae strain EBY100.
Figure 9 Expression of GOx constructs using the
EBY100 strain and pCTCON2 vector. GOx is fused to an
N‐terminal Aga2p tag and a C‐terminal c‐myc tag.
Modified after Boder (Boder and Wittrup 1997).
I I I . Results and discussion P a g e | 37
The yeast display system uses the S. cerevisiae α‐agglutinin receptor to display heterologous
proteins on the cell surface (Boder and Wittrup 1997). The system comprises two subunits:
AGA1 and AGA2. The EBY100 strain expresses the AGA1 peptide (which binds to β‐glucan) on
the surface of the yeast cell using the Gal1 promoter (Lu et al. 1995). The AGA2 subunit is
encoded by the pCTCON2 vector and is expressed as an N‐terminal fusion to the protein of
interest. The N‐terminal region of AGA2 binds to AGA1 on the surface of the yeast cell, by
forming two disulfide bonds (Cappellaro et al. 1991). GOx was expressed also containing a C‐
terminal c‐myc tag (Figure 9).
The GOx surface display construct was obtained starting from the wild‐type GOx synthetic
gene ordered from Gene Script (accession number P13006). After cloning, the constructs were
confirmed by sequencing and used to transform EBY 100 cells. The presence of GOx activity on
the cell surface was confirmed using the ABTS assay (Zhu et al. 2006).
A fermentation curve was created to determine the optimal time for expression. One yeast
colony was picked from the agar plate and inoculated into YNB‐CCA Glu medium. The culture
was later centrifuged and resuspended in YNB‐CAA galactose/raffinose medium at an OD600 of
0.4, and the GOx activity and OD600 were measured every 2 h in the supernatant and cell
suspension. The fermentation curve in Figure 10 shows that maximum expression occurs
between 15 and 25 h. The enzyme activity then reaches a plateau and declines to undetectable
levels by 30 h. Thus for all further experiments, the conditions were adjusted so that the
expression was performed for 16–18 h.
I I I . Results and discussion P a g e | 38
Cells expressing GOx were tested for enzyme molecules leaking from the surface. The cells
were washed several times in PBS and the activity was determined after each washing step.
Figure 11 shows that the supernatant from the original expression culture medium contains
significant GOx activity, but after the first wash there is no remaining activity and after the third
wash the activity of the cells remains constant, proving that GOx molecules remaining on the
cells are tightly bound to the surface. In order to avoid any cross talk between droplets, the
cells were always washed three times to make sure that all the non‐bound GOx molecules were
removed. The FACS screening system discussed in the following chapters involves the
compartmentalization of cells in water‐in‐oil emulsions, and the presence of soluble enzyme
would lead to cross‐contamination and a loss of connection between genotype and phenotype.
Cells must therefore be washed thoroughly before compartmentalization so that no soluble
GOx is present in the solution even when the yeast surface display system is used.
Figure 10 Fermentation curve of wild‐type GOx expressed in S. cerevisiae EBY 100 cells. At
each specified time point 1 mL of culture was removed from the fermentation flask and used to
determine OD600 nm and for the ABTS assay (section II.2.1).
I I I . Results and discussion P a g e | 39
The GOx cDNA was also modified by adding a 3’sequence encoding a C‐terminal c‐myc tag.
This allows the quantitation of enzyme molecules expressed on the surface of the yeast cells by
using anti‐c‐myc fluorescent antibodies. In most directed evolution experiments it is quite
difficult to determine the specific activity of the enzyme until the protein is purified. By
detecting the amount of enzyme molecules on one fluorescence channel and the activity of the
enzyme on another, we could theoretically measure the specific activity of the enzyme and sort
not only plus/minus variants but also mutants with higher activity.
The fermentation conditions were optimized to maximize the stable expression of the
enzyme, which occurs between 16 and 24 h and then declines. Expression was therefore
stopped at this time point.
Because the enzymes were expressed with two tags (an N‐terminal AGA1 tag for surface
display and a C‐terminal c‐myc tag for antibody staining) we were concerned that the tags
would inhibit enzyme activity. Indeed the tagged variants did show a lower activity, but the
ratios between the activities of the wild‐type and mutant enzymes remained constant making
this construct suitable for directed evolution experiments.
Figure 11 GOx activity in the supernatant after multiple rounds of
washing. 1 mL of fermentation medium was removed 12 h after the
induction of enzyme activity. The cells were centrifuged, the supernatant
collected and the cells were resuspended in same amount of fresh PBS. After
each washing step the activity was measured, using the ABTS MTP assay
(section II.2.1) both in the supernatant and in the cell resuspension.
I I I . Results and discussion P a g e | 40
III.1.2 Creation of mutant libraries
Mutant gene libraries were created using the consensus approach (Martin Lehmann et al.
2002). The libraries were designed starting from A. niger GOx, accession number P13006, and
using this as a BLAST query against the NCBI database (http://blast.ncbi.nlm.nih.gov/Blast.cgi).
The results were analyzed and all sequences with an identity >80% and <40% were discarded.
The remaining sequences were compared pairwise to ensure that none showed >90% similarity,
and where this occurred only one sequence was kept for further analysis. The final result was a
list of seven protein sequences (Table 1), which were used for multiple sequence alignment
using BioEdit (Ibis Biosciences, CA, USA). The alignments were inspected and adjusted manually.
Table 1 Sequence similarity table
A2I7K9 P13006 Q9HFQ1 A2QZ31 Q70FC9 B8MDS4
_
Q92452
A2I7K9 100 66 66 48 90 90 99
P13006 100 85 49 67 61 66
Q9HFQ 100 50 66 60 66
A2QZ31 100 47 46 48
Q70FC9 100 83 90
B8MDS4 100 91
Q92452 100
A positional frequency report was generated using the seven selected sequences, showing
for every position in the alignment the number of times that a certain amino acid appears in the
series (Figure 12). The following positional frequencies were obtained: 0% (0 out of 7), 14.29%
(1 out of 7), 28.57% (2 out of 7), 42.86% (3 out of 7), 57.14% (4 out of 7), 71.43% (5 out of 7),
86.71% (6 out of 7) and 100% (7 out of 7). A score of 0% means that for the specified position
the amino acid does not appear at all, 14.29% means that the amino acid appears in 1 of the 7
proteins in the set at that position and so on. For the gene libraries, we initially selected the
positions that had an 86.71% frequency of certain amino acids, and only those with a different
amino acid in the starting gene (P13006). These amino acids were changed at each position
with the most frequent amino acid from the series (with the amino acid present at that position
in all the other six proteins used in the alignment, but not in the starting gene). There were nine
positions with these properties and they were changed as follows: S53F, A192T, E310D, I403L,
I I I . Results and discussion P a g e | 41
L429I, I474L, H510N, M528L, M556V1. If only these positions were used for the library, the
library size would be 29 = 512 mutants. Since the screening capabilities of FACS are far beyond
this number, we also looked at amino acid positions with other frequencies: 71.42% and
57.14%. Using the same overall approach, the following additional positions were chosen:
V106I, V293I, I597V, R37K, E374D, R537K and N278D. The final number of substituted positions
was 16, leading to a library of 65,536 mutants. In order to ensure the entire library is screened,
100 copies of each clone should be analyzed, making the library size in the range of 107.
After the positions were chosen, site‐directed mutagenesis primers were designed and used
to generate multiple mutants. The PCR products were introduced into XL10 Gold chemical
competent cells as provided with the kit. Five random clones were isolated and sent for
1The numbers associated with the mutation position of each amino acid are the numbers corresponding to the mature protein sequence not the alignment sequence.
Figure 12 Part of positional frequency report. The first column represents the number of the positions as they
appear from the sequence alignment. Row 1 represents the types of amino acids present in the sequence (all possible
20 amino acids plus gaps). The report gives for every position the occurrence of each of the 20 possible amino acids.
I I I . Results and discussion P a g e | 42
sequencing to verify the distribution of mutations in the clones. As shown in Table 2, nearly all
the mutations are represented, the mutations are uniformly distributed throughout the gene
and no wild‐type clones were detected. The DNA was extracted from the E. coli libraries and
used to transform S. cerevisiae EBY 100 cells. These yeast libraries were used for the FACS
experiments.
Table 2 Random sequence of clones from the consensus library
Mutations Clone1 Clone2 Clone3 Clone4 Clone5
R37K X X
S53F X X
V106I X
A192T X X
N278D X
V293I X
E310D X
E374D X X X
I403L X X
L429I X X
I474L X
H510N X
M528L
R537K X x X
M556V X
I597V
Using the consensus approach we obtained 105 different GOx clones multiplied to a library
containing approximately 107 clones to ensure the entire diversity was available for screening.
This is an optimal size for the FACS screening system because the entire population can be
screened in one experiment to increase reproducibility.
I I I . Results and discussion P a g e | 43
III.1.3 Enzymatic assay for FACS
The tyramide fluorescein assay, which covalently stains the surface of yeast cells, has
previously been tested using yeast cells expressing GOx (positive cells) compared to negative
control cells containing an empty plasmid. Different mixtures of positive and negative cells
were tested. It can clearly be seen that by increasing the percentage of positive cells in the mix,
the percentage of stained cells are also increases (Figure 13). It also can be seen that the entire
population of cells shows an increase in florescence. This can be explained by the diffusion of
the product form the active cells towards the negative cells.
In order to prevent this cross‐talk, the assay was adapted for single emulsions, with an
encapsulation efficiency of almost 100%. The emulsion components can be removed after the
reaction is complete and the cells can be recovered and sorted using FACS without emulsions,
thus retaining a high sorting speed. The initial protocol required the cells expressing the
enzyme to be mixed with the components used in the enzymatic reaction (glucose, HRP and
tyramide‐fluorescein) followed by a compartmentalization step that enclosed all these
components in single emulsions. Since the compartmentalization step takes at least 3 min, it
means that the reaction starts before the compartments are formed, allowing a certain amount
of cross‐talk and thus generating a high background signal. One way to solve this problem is the
addition of ascorbic acid to the reaction mix (Radivoje Prodanovic et al. 2011). Ascorbic acid
delays the start of the reaction by acting as a sacrificial substrate for HRP, which oxidases the
ascorbic acid first and moves on to the tyramide substrate only when the ascorbic acid is
depleted. The initiation of the fluorescence reaction start can thus be controlled by adding
different amounts of ascorbic acid. The tyramide‐fluorescein assay cannot be tested using the
Figure 13 Tyramide fluorescein assay using different mixtures of yeast cells expressing GOx (positive cells) and controls lacking the
GOx gene (negative cells).
I I I . Results and discussion P a g e | 44
MTP format because it requires a fluorescent molecule already present in the reaction medium
to bind the cell surface. In other words, the fluorescence is not dependent on the enzymatic
reaction, but the fluorescent molecules are instead covalently linked to enzyme molecules on
the surface of the cells.
We therefore used the APCC assay to optimize the reaction conditions. As shown in Figure
14, increasing the amount of ascorbic acid delays the formation of the fluorescent product.
When ascorbic acid and glucose are present at the same concentration, no fluorescent product
can be detected. To start the fluorescence reaction after the emulsification process is complete,
the ratio between the concentrations of glucose, ascorbic acid and GOx need to be precisely
controlled, which is challenging when dealing with libraries in which the GOx activity is
unknown. Figure 14 indicates that the optimal reaction contains 0.75 mM ascorbic acid, i.e. the
product forms after 10 min and the signal is high, but this is not the case when cells are used in
the system. In the presence of yeast cells, the amount of glucose that needs to be used in order
to detect a positive reaction is much higher (at least 20 mM as compared to 1.5 mM in the case
of soluble enzyme). This is due to the fact that the cells themselves consume glucose as a
carbon source, depleting the glucose from the reaction mix. The problem is further amplified by
Figure 14 APCC assay in MTP with commercial GOx using 0.5 mM APCC, 1.5 mM glucose, 5U/mL GOx,
10U/mL HRP and the specified concentrations of ascorbic acid.
I I I . Results and discussion P a g e | 45
the fact that in the emulsion (the reaction compartment) the quantity of cell / volume is more
than 50%. All the drawbacks presented above make the system difficult to use and to adjust.
The presence of ascorbic acid, and the encapsulation of the initial reaction within single
water‐in‐oil emulsions, reduces the cross‐talk between positive and negative cells. When used
to sort reference and epPCR libraries, we enriched the positive population from 4.7 to 33‐fold,
but it proved impossible to achieve a >40% proportion of positive cells (R. Prodanovic et al.
2012). In order to overcome the above challenges we have modified the assay to allow the
external delivery of glucose. This completely avoids any cross‐talk between the droplets
because the reaction only starts after the compartments are formed. In addition, antibodies are
used to detect GOx molecules present on the surface of the cells in order to determine the
specific activity of the enzyme. This makes it possible to use this assay not only as a plus/minus
screening system but also to select higher activity mutants.
External delivery is achieved by pre‐mixing all the components needed for the enzyme
reaction, except glucose, which is delivered to the system when the compartments have
formed thus allowing the reaction to start. A schematic representation of the process is
provided in Figure 15. Glucose is delivered through the oil phase as β‐octylglucoside. This
compound has a detergent structure and accumulates at the oil‐water interface. In the aqueous
phase, β‐glucosidase added at the beginning of the emulsification process catalyzes the release
of glucose from the β‐octylglucoside, thus making glucose available for the GOx reaction in the
aqueous phase. Once the glucose becomes available, the GOx reaction can start leading to the
formation of peroxide. Peroxide, in the presence of HRP, oxidizes the phenolic rings of both
tyramine (from the tyramide‐fluorescein substrate) and tyrosine (from the proteins on the
surface of the yeast cell), creating phenolic radicals. These phenolic radicals polymerize
amongst themselves and stain the surface of the cells. When the reaction is complete, the oil
phase is removed and the cells are analyzed and sorted by FACS.
There are many advantages to this system. First, the cross‐talk between different cells is
completely eliminated. Second, the concentration of glucose available to the system can be
precisely controlled. This is important if we wish to select for kM mutants, for example, when
the screening needs to be performed under glucose limiting conditions. Also, the time of the
I I I . Results and discussion P a g e | 46
reaction can be controlled, which could theoretically allow us to catch the enzymatic reaction
on the linear slope, not the plateau. This feature, combined with the staining of enzyme
molecules available on the surface of the cells with fluorescence‐labeled antibodies recognizing
the c‐myc tag, would allow sorting on the basis of specific enzyme activity because it would be
possible to determine the enzyme activity per number of enzyme molecules.
Figure 15 The tyramide‐fluorescein screening system (section II.2.5).
I I I . Results and discussion P a g e | 47
III.1.4 FACS Sorting
III.1.4.1 Survival of yeast cells
We tested the influence of different assay components and of FACS sorting itself on the
survival of the yeast cells. First, in order to determine how FACS sorting affects cell survival,
yeast cells were stained with 7‐aminoactinomycin D (7‐AAD), a fluorescent dye that intercalates
in double stranded DNA (Steensma, Timm, and Witzig 2003). To access the DNA, the membrane
needs to be compromised therefore all cells that are stained by this compound are considered
to be dead. Figure 16 shows a comparison between cells stained with 7‐AAD and unstained
controls, with the former appearing in gate P4 and partially in P5. This means that the cells in
gate P6 are not stained at all and consequently they are alive. Therefore 96 cells from gate P6
were sorted directly into MTP wells and were grown for 2 days at 27⁰C, 160 rpm in YNB CAA Glu
medium (Invitrogen 2002) and the number of wells that had growing cells was counted. The
results showed that 58 clones survived sorting, representing a 60% survival rate.
The effect of the emulsion and reaction components on the survival of the yeast cells are
presented in Table 3. None of the compounds had a significant influence on cell survival, with
the exception of the oil used for the emulsification process. Since each mutant in the library is
represented by at least 100 copies, even after FACS sorting and emulsification there should still
Figure 16. Flow cytometric analysis of EBY100 cells transformed with the empty
pCTCON2 vector stained with 7‐aminoactinomycin D (7‐AAD). Bivariate histograms showing
the correlation between side scattering (measurement of size) and fluorescence of 7‐AAD.
I I I . Results and discussion P a g e | 48
be at least 20 surviving copies of each clone. Therefore the entire diversity is represented and
none of it is lost due to the experimental conditions of screening.
Table 3 Influence of different component used in the screening process on cell survival
Number
of cells
Survival
rate
Cells without any other compound 1568 100%
Cells plus oil 1152 73%
Cells plus β‐octylglucoside 1528 97%
Cells plus β‐glucosidase 1544 98%
Tyramide fluorescein 1502 95%
III.1.4.2 Reference libraries
In order to test the efficiency of the enzyme assay, we initially sorted a series of artificial
reference libraries made by mixing different proportions of yeast cells expressing the wild type
enzyme and control cells transformed with the empty vector. First, the cells were stained with a
mouse anti‐c‐myc antibody and then with an anti‐mouse phycoerythrin (PE)‐labeled secondary
antibody. Positive and negative cells were mixed in different concentrations according to their
optical density, washed three times with PBS to remove unbound GOx molecules, mixed with
the reaction components and emulsified for 3 min at 8000 rpm in mineral oil plus Abil EM.
When the emulsions were complete, 185 nmol of β‐octylglucoside was added through the oil
phase (5 µL of 37 mM β‐octylglucoside dissolved in ethanol). The emulsions were shaken gently
at room temperature for 30 min after which the oil phase was removed and the cells were
analyzed by FACS.
As shown in Figure 17, there is a clear correlation between the percentage of positive cells
added to the mix and the number of positive cells observed after the reaction. The cells present
in gate P4 were sorted from the 0.01%, 0.1%, 1% and 5% libraries, using yield mode for the
0.01% and 0.1% libraries due to the very low number of cells present in the sorting gate. Single
cell mode was used for the 1% and 5% libraries.
I I I . Results and discussion P a g e | 49
Yield mode increases the speed of sorting and reduces the number of aborted events but
the purity of the final sorted population is lower. Single cell mode offers the highest purity but
considerably reduces the sorting speed due to the large number of aborted events. The sorted
cells were grown for 3 days at 27⁰C, replica‐plated on YNB CAA Gal/Raf plates and incubated for
another 24 h before carrying out the ABTS agar plate assay. Figure 18 shows an example of cells
sorted from the 5% reference library using the ABTS agar plate assay to detect positive clones
before and after sorting. The results from all libraries are summarized in Table 4.
Figure 18 ABTS agar plate assay for GOx activity of cells before and after
sorting of a 5% reference library.
Figure 17 FACS recordings of different mixes of referent libraries. Bivariate histograms showing correlation between FITC
(measuring enzyme activity) and PE fluorescence (measuring the amount of GOx molecules on the surface of the yeast cell
through the c‐myc epitope). (See section II.2.6 of Materials and Methods)
I I I . Results and discussion P a g e | 50
As shown in Table 4, we achieved high enrichment factors for all libraries proving that the
sorting system is effective. The purity of populations varied depending on the proportions in
the starting library and the sorting mode. The purity obtained for the 0.01% and 0.1% libraries
was quite low, partly reflecting the use of yield sorting mode, which means that some events
lacking all the optimal parameters are sorted nevertheless. Alternatively, the minimal signal
from the small number of positive cells may be close to background, making the sorting process
vulnerable to measurements errors. This phenomenon also affects the 1% library, with a final
purity of ~50%. Optimal sorting occurs when starting with a 3–5% positive population, resulting
in 92% purity. This means that the best strategy for sorting libraries with a <5% starting
concentration of positive cells would be to carry out a pre‐sorting step in yield mode to enrich
for positive cells, followed by sorting in single‐cell mode to achieve >90% enrichment.
Table 4 Results from the sorting of different reference libraries
As stated above, the number of surface enzyme molecules was determined by antibody
staining while the enzyme activity was determined using the fluorescein‐tyramide assay. The
emission spectra of PE and fluorescein overlap, i.e. fluorescein excited at 488 nm emits in the
PE region (Figure 19 a). In order to avoid any compensation problems we decided to use other
combinations of dyes. One option was amino‐methyl‐coumarin‐acetate‐tyramide (AMCA‐Tyr)
as the substrate instead of fluorescein tyramide. Even though the emission spectra of the two
fluorophores do not overlap (Figure 19 b), the AMCA staining is much less intense than the
corresponding fluorescein staining (Figure 20).
Reference library Before sorting After sorting Enrichment
0.01% 0.01% 6.89% 689x
0.1% 0.1% 7.1% 71x
1% 1.4% 46.34% 33.1x
5% 3.3% 92.6% 28x
I I I . Results and discussion P a g e | 51
Figure 20 FACS recordings of yeast cells stained with tyramide fluorescein (top) and tyramide AMCA (bottom).
Double plots representing the forward scattering versus the FITC and AMCA fluorescence, respectively
Figure 19 (a). Emission spectra of FITC and PE when using a 488 nm excitation laser. (b). AMCA and PE emission
spectra at 375 nm (left) and 488 nm (right) excitation. (c). FITC and Alexa fluor 647 emission spectra at 488 nm and
647 nm excitation. Data obtained using http://www.bdbiosciences.com/research/multicolor/spectrum_viewer/
I I I . Results and discussion P a g e | 52
It appears that a 10‐fold higher concentration of AMCA‐tyramide is required to achieve
staining comparable with that of the fluorescein derivate. Therefore by using AMCA‐tyramide
instead of fluorescein‐tyramide the sensitivity of the assay would be reduced significantly,
which is not at all desirable. We therefore tested a different secondary antibody, labeled with
the alternative fluorophore Dylight 647. As shown in Figure 19 c, there is no cross‐talk with
fluorescein but the label is more sensitive than AMCA‐tyramide.
To improve the screening system even more, we decided to change the order of the
staining procedures. In the original method, the cells were stained with the antibodies first and
then compartmentalized within emulsions for the enzyme reaction. This led to the loss of
antibody fluorescence, probably because the cells were washed using oils and detergents that
could displace the antibody‐antigen bonds when the emulsion components were removed. In
order to avoid this problem, we decided to perform the enzyme reaction first and apply the
antibodies when the emulsion was removed. The fluorescent staining from the enzyme assay is
unaffected by the washing steps because the fluorophore is covalently bonded to cell‐surface
proteins.
We have already demonstrated by sorting the reference libraries that the screening system
is suitable for plus/minus type of sorting with a screening capacity of 107cells/hour, but a
system that could select cells based on differential activity would be even more valuable. We
therefore introduced certain modifications such as the use of different combinations of
fluorescent dyes to simultaneously measure GOx activity and the abundance of the GOx
protein. The best pairing was fluorescein‐tyramide for the detection of GOx activity and
antibodies coupled to DyLight 647 to detect GOx molecules, because these do not show any
cross‐talk between channels and have a high sensitivity. This combination of dyes was used to
compare wild‐type GOx with a mutant that was five times more active. The greatest difference
in activity was observed at low β‐octylglucoside concentrations, which shows that the reaction
stops in the linear region at this concentration and that it should be possible to distinguish
between mutants with different activities. However, due to high standard deviations in
expression levels between cells and the small differences in fluorescence levels, the current
sensitivity for discrimination between mutants is approximately a five‐fold difference.
I I I . Results and discussion P a g e | 53
III.1.4.3 Sortingtheconsensus libraries
We pursued the goal of quantitative screening for GOx activity by testing real libraries at
different glucose concentrations. As shown in Figure 21, more cells can be detected at higher
glucose concentrations because under these conditions the system can detect even low activity
clones. When lower concentrations of glucose are available, only the most active clones
become positive. We therefore used 1.85 nmol β‐octylglucoside for further library screening.
The libraries were sorted twice under these conditions and ~100 clones from the sorted
population were tested in MTP assays. The rest of the clones were seeded into medium and
GOx expression was induced before the next round of sorting. Figure 22 shows that the
population of higher activity cells (gated at P1) increases after each round of sorting.
Figure 22 FACS recordings of a consensus library during multiple rounds of sorting (section II.2.6 ). These are double plots
of FITC fluorescence (corresponding to enzyme activity) versus APC fluorescence (corresponding to enzyme abundance).
Figure 21 FACS recordings of consensus library screens at three different β‐octylglucoside
concentrations using the protocol from section II.2.6 but varying the concentration of
β‐octylglucoside added to the oil phase. These are double plots of FITC fluorescence (corresponding
to enzyme activity) versus APC fluorescence (corresponding to enzyme abundance).
I I I . Results and discussion P a g e | 54
The FACS recordings showed that high concentrations of glucose result in a larger number
of cells from the library showing enzyme activity, indicating that under glucose‐limiting
conditions only higher‐activity clones are detected. Performing multiple rounds of sorting on
this library showed that the percentage of higher‐activity clones increases after each round,
indicating that this system could be used to screen for clones with higher activity.
III.1.5 MTP assays
The cells collected in MTP wells were analyzed using the ABTS soluble assay with glucose
concentrations of 5 mM (below kM of the wild‐type enzyme) and 250 mM (saturating
conditions). This helped to determine whether the mutants have an improved kM or kcat. The
measurements at 5 mM glucose report any improvement in kM/kcat and those at 250 mM report
any improvement in kcat.
The cells were grown for 3–4 days after FACS sorting in 100 μL YNB CAA Glu medium, 450
rpm, 30°C, 80% humidity, then inoculated into 100 μL of the same medium and incubated
under the same conditions for 24 h in order to avoid any growth deviation. In the final step, 5
μL of the saturated culture was inoculated into 25 μL fresh medium and incubated for another
24 h before 80 μL of Gal medium was added. Enzyme activity was measured after 8 h. Each MTP
included three wild type clones for expression and activity standardization. Clones were
considered to have no activity if the values were 10 times below those of the wild type enzyme.
All the values were normalized to the OD600 of the cells in each well. The data obtained from
these experiments are presented in Figure 23. The proportion of cells with no activity decreases
after each round of sorting and the number of higher‐activity clones increases. This is in
accordance with the results obtained by FACS analysis (Figure 22).
We selected approximately 50 clones that showed improved activity for re‐analysis. Five
copies of each (one per well) were grown in fresh MTPs in order to control for variation in cell
growth and expression. The standard deviation of all measurements was 2.4–17 %. The best
five mutants were picked, sequenced and prepared for further analysis. The mutations
obtained from the sequencing results are compared with the parent strain in Table 5.
I I I . Results and discussion P a g e | 55
Table 5 Sequences of the improved mutants compared with the wild type gene. Light blue represents mutations
common to the parent sequence and in purple represents new mutations
Mutations
Starting
parent A2 A2‐1 AS1F5 F9 F9‐1
T30V X X X X X X
R37K X X X
I94V X X X X X X
V106I X X
A162T X X X X X
V293I X
E310D X
R537K X X X
M556V X X X X
Three hundred clones with supposedly higher activity were sorted from the consensus
library and analyzed using the ABTS MTP screening system. This assay was optimized for MTPs
so that the standard deviation never exceeded 17%, making the screening system reproducible
and reliable. The sorted mutants were compared with the wild type clone and we found that
the number of active clones increased in each round (as shown with the reference libraries) and
that the number of higher‐activity clones increased, showing that the assay is quantitative.
Figure 23 Activity of GOx mutants after each round of sorting in MTP assays. We collected
100 mutants after each round of sorting and the GOx activity was assessed for each individual
clone at two glucose concentrations, 250 mM (saturation conditions, showing the kcat) and 5 mM
representing the kM/kcat (see section II.2.6.2)
I I I . Results and discussion P a g e | 56
These results match the FACS recordings, which also show a larger proportion of higher‐activity
clones after each round of sorting. The sequences of the five most active mutants showed that
certain mutations were present in several of the mutants and could be important for increasing
GOx activity.
In order to confirm the quantitative nature of the assay, we compared two clones with
known activity (wild type vs. A2, which is five times more active). The two cell populations were
emulsified with the substrates as before and fed with β‐octylglucoside at different
concentrations (185, 18.5 and 1.85 nmol). The emulsions were incubated for 30 min then the
emulsion components were removed and the cells were stained with antibodies as described
above prior to FACS analysis. There appeared to be no difference between the wild type and A2
clones at high concentrations of glucose (Figure 24).
Figure 24 FACS analysis of wild‐type and A2 clones using the tyramide‐fluorescein assay at different
concentrations of β‐octylglucoside. The bivariate histograms show FITC fluorescence corresponding to the
enzyme activity and APC fluorescence corresponding to the enzyme concentration.
I I I . Results and discussion P a g e | 57
However, at lower substrate concentrations we observed a clear difference between the
two clones, probably because in the latter case the reaction stops before it reaches the plateau,
therefore we can detect the slope of the reaction. This suggests that it is possible to identify
differences in activity between clones in a library.
III.1.6 Cloning in P. pastoris
The level of GOx expression in the EBY 100 cells is 1–2 mg/L, with 75% of the molecules
linked to the cell surface by disulfide bonds. The amount of enzyme present in the extracellular
medium (~0.5 mg/L) is too low to determine the kinetic properties accurately. The enzyme can
be removed from the cell surface by treating cells with a reducing agent such as DTT or
β‐mercaptoethanol, but this would also destroy the intramolecular disulfide bonds and
denature the enzyme. An additional drawback of the surface display system is that even if the
enzyme could be purified in sufficient amounts, the N‐terminal and C‐terminal tags might alter
its kinetic properties. Therefore we needed to express these mutants in larger amounts without
tags, which is why the P. pastoris expression system was chosen for the purification of GOx
mutants. We selected P. pastoris KM71H strain together with the pICZalphaA vector because
this has a methanol inducible promoter (Invitrogen 2010).
Five mutants plus the wild type clone were inserted into the pICZalpha A vector and the
sequences were verified. Large quantities of the mutant plasmids were isolated from E. coli and
linearized to promote integration into the yeast genome as single or multiple copies. Higher
copy numbers tend to promote higher levels of recombinant protein expression. The best
performing clones were selected from among 96 colonies using a MTP assay to detect higher
production, and were used for scaled‐up fermentation.
Fermentation conditions were optimized for the wild‐type clone to determine the most
suitable time for expression. The cells were inoculated into BMGY glycerol medium (Invitrogen
2010) to an OD600 of 30 and then the glycerol medium was removed by centrifugation and
replaced with BMGY methanol medium at an OD600 of ~80 (0.1‐0.2 of the glycerol medium
volume). The cells were adapted to the glycerol medium gradually, starting with one colony
seeded into 25 mL medium for 24 h and progressing to higher volumes each with a 24‐h growth
phase. The cells were not allowed to enter the stationary phase because this inhibits
I I I . Results and discussion P a g e | 58
recombinant protein expression following induction. After transfer to expression medium, the
GOx activity was monitored every 24 h and the medium was supplemented with 0.5% methanol
for induction every day. As shown in Figure 25, the largest amount of protein is obtained after
fermentation for 4 d. These conditions were therefore chosen for the purification of mutant
enzymes. We were able to detect 60–80 mg/L of protein under these conditions.
III.1.7 Purification
After the fourth day of expression, fermentation was stopped by centrifuging the cells at
11,000 x g for 10 min. We obtained approximately 30–40 mg of GOx from each mutant. The
supernatant was collected and filtered to remove any remaining cells, but 30% of enzyme
activity remained connected to the cell surface. We attempted to recover some of these
proteins by incubating the cells in concentrated salt solution, but the yield was too low and we
therefore proceeded with the supernatant alone. The proteins in the supernatant were
concentrated to a volume of ~20 mL by cross‐flow ultrafiltration using a 50‐kDa membrane and
were dialyzed against 10 mM phosphate buffer (pH 6.0) before separation on a DEAE 20 mL
column using a 10–250 mM phosphate buffer gradient (pH 6.0) over 15 column volumes (CVs).
The GOx peak eluted at ~26% buffer B (130 mM phosphate) as shown in Figure 26.
Figure 25 Optimization of fermentation conditions for the wild type enzyme using P. pastoris
KM71H cells (section II.2.11). After induction, 1 mL of fermentation medium was removed every day
and the activity of GOx was measured using the ABTS MTP assay (section II.2.1).
I I I . Results and discussion P a g e | 59
All the mutants presented similar peaks and eluted at similar time points. After separation,
the fractions corresponding to the GOx peak were pooled (~50 mL) and concentrated to 5 mL
using 10 kDa ultrafiltration centrifugation columns. The protein concentration was measured by
absorbance at 280 nm and the same amount of each mutant was loaded on a SDS
polyacrylamide gel which confirmed similar purity levels for each clone (Figure 27).
Approximately 30 mg of each enzyme was isolated using this optimized purification method
and the purity as determined by SDS‐PAGE was in each case sufficient for kinetic analysis.
Figure 27 Analysis of purified mutants by SDS‐PAGE (8% acrylamide),
with 0.3 mg protein loaded on each lane to compare purity. The amount of
protein was estimated by absorbance at 280 nm.
Figure 26 Separation of GOx mutants by DEAE chromatography using the concentrated
extracellular medium from Pichia fermentation loaded onto a HiPrep 16/20 DEAE FF column (GE
Healthcare) using 10 mM phosphate pH 6.0 as buffer A and 500 mM phosphate pH 6.0 as buffer B
with a gradient form 0% B to 60 % B over 15 CV and a flow rate of 2 mL/min (section II.2.12).
I I I . Results and discussion P a g e | 60
III.1.8 Kinetic characterization
The kinetic characterization of the mutants was performed in triplicate MTP ABTS assays,
with measurements in the range of 1.2–266 mM glucose at pH 5.5 and 7.4. The pH optimum of
GOx is 5.5, but miniature biofuel cells operating in human blood would need to be highly active
at physiological pH. Therefore, the screening strategy was devised to select clones with higher
activity at pH 7.4.
The data were fitted on Michaelis‐Menten curves but the Lineweaver‐Burk, Eadie‐Hofstee
and Hanes‐Woolf curves were also plotted to check for any irregularities. We decided to
exclude data showing high measurement errors and data representing outliers from these
curves but only one point needed to be removed since all the other measurements had
standard deviations around 5% and the data points were fitted on all equations with a
correlation coefficient of 0.99. Figure 28 shows the curves for the wild‐type enzyme in PBS but
nevertheless this also represents the appearance of the curves for all mutants.
Figure 28 Kinetic data for wild type GOx in PBS
I I I . Results and discussion P a g e | 61
The kcat was determined by measuring the enzyme concentrations (absorbance at 280 nm)
and the absorption of 1.5 mg/mL of GOx was considered to yield one active unit (AU) as
calculated according to the sequence using ProtParam. The kinetic data of all mutants were
measured in triplicate in two different days and the final values, which are averages of all six
measurements, are presented in Table 6. The maximum standard deviation was 6%.
Table 6 Kinetic constants for all mutants Acetate buffer pH 5.5
kM average kcat average Specificity
constant
(mM) s‐1 mM‐1∙s‐1
Wild type 28.26 189.38 6.70
Parent 15.13 266.37 17.60
Commercial 23.96 376.03 15.68
A2 18.93 578.32 30.55
A21 15.75 466.18 29.59
F5 16.05 368.16 22.93
F9 13.05 350.97 26.88
F19 15.06 321.53 21.34
PBS buffer pH7.4
kM average kcat average Specificity
constant
(mM) s‐1 mM‐1∙s‐1
wt 23.19 130.16 5.61
parent 14.71 227.99 15.49
commercial 18.86 372.01 19.72
A2 13.08 432.17 33.11
A21 10.93 350.77 32.09
F5 14.26 308.44 21.62
F9 10.37 228.36 22.02
F19 9.53 266.80 27.99
All the mutants showed a 3–4‐fold increase in activity compared to the wild type enzyme at
pH 5.5, and a 3.6–6‐fold increase at pH 7.4. This greater improvement at pH 7.4 reflects the
screening conditions we imposed. All the mutants were also 1.3–1.9‐fold more active than the
I I I . Results and discussion P a g e | 62
commercial variant of the enzyme, but this is not necessary a relevant comparison because the
commercial enzyme is produced in A. niger and has different glycosylation pattern. The industry
fermentation and purification processes for the production of GOx are also different from our
methods, and the greatly increased scale could influence the determination of kinetic constants
such as kcat. The commercial amino acid sequence is the same as the wild type variant that we
used in this project. This variant is expressed in the same vector and strain system as the rest of
the mutants and purified by the same protocol making the comparison between wild type and
mutant strains relevant and important.
III.1.9 Thermal stability measurements
The thermal stability of the GOx mutants was determined by incubating the enzyme in 50
mM acetate buffer (pH 5.5) at 60°C in the absence of substrate. Aliquots from this solution
were withdrawn periodically and the residual activity was measured using the ABTS assay. The
percentage of residual activity at time t was defined as A(t). This was compared to the activity
when t = 0, which was arbitrarily considered 100%. By plotting t using an exponential equation
it was possible to determine the inactivation rate constant (kd) as follows: ∙
Figure 29 shows the data for mutant F9 but the other mutants were similar. Each point
represents an average of three measurements. The half times for thermal stability were
calculated by setting A(t) = 0.5. The values for all mutants are summarized in Table 7.
I I I . Results and discussion P a g e | 63
Table 7 Thermal stability half times of GOx mutants
t1/2 (min)
Parent 9.00
Wild type 10.50
A2 11.74
A2‐1 13.86
AS1F5 11.74
F9 15.75
F9‐1 19.80
Commercial 22.35
All the mutants showed higher thermal stability than the parent and wild‐type enzymes, but
mutant F9‐1 was the best‐performing mutant with double the wild‐type half time of
inactivation.
III.2 In Vitro Translation (IVT) of Glucose Oxidase
Thus far we have presented screening systems based on in vivo expression platforms. Our
data show that the encapsulation of yeast cells in double emulsions is <10% efficient. This
problem was solved in the case of GOx by using an enzyme assay that directly stains the surface
Figure 29 Thermal inactivation curve for mutant F9.
I I I . Results and discussion P a g e | 64
of the yeast cell, but the same approach could be more difficult to achieve for other enzymes
and a fluorescence assay with a soluble product seems more feasible. In addition to the low
encapsulation efficiency of double emulsions, further difficulties with yeast in vivo expression
systems include the low transformation efficiencies, low survival rates in different conditions
thus narrowing the screening possibilities, and the long generation times (Gietz & Schiestl,
2007;(Lindegren and Haddad 1954). The 2–3 hour generation interval means 2–3 days are
required to recover yeast cells after each round of sorting restricting progress to two rounds of
evolution per week.
One solution that could help address all these problems is the use of in vitro
transcription/translation systems. This would allow multiple rounds of evolution to be
performed in one day. Most in vitro systems are based on E. coli extracts, but this is not
appropriate for GOx because it is a eukaryotic glycosylated enzyme with disulfide bonds
(Frederick et al. 1990). Previous attempts to express GOx in vivo in E coli resulted in the
formation of inclusion bodies that had to be solubilized to regain enzyme activity (Witt, Singh,
and Kalisz 1998). This would not be appropriate for our system because the gene libraries need
to be taken directly from the PCR and compartmentalized in emulsions together with the
transcription/translation mix and the enzyme reaction components. Translation in vitro usually
takes 4 h (Qiagen 2006) which could be followed by FACS. The DNA from the positive droplets
would be recovered and amplified by PCR and a new round of evolution could be initiated
immediately.
We have chosen insect cell extracts as a more suitable IVT system because a variant of GOx
was found to be natively expressed in the salivary glands of some insect larva suggesting
compatibility (Musser et al. 2005). The first system that we tried was the insect cell
transcription/translation kit from Qiagen (Qiagen 2006). Although the final goal would be
transcription directly form a PCR product template, the proper vectors can increase the yield of
expressed proteins. Therefore, the wild‐type GOx gene was cloned in vector pIX4 in two
formats, with and without an N‐terminal ‐factor (AF) pro‐peptide sequence to direct the
protein into the secretory pathway. The manufacturers suggest that some signal peptides may
not be compatible with the insect cell‐free extract, therefore we followed the manufacturer’s
I I I . Results and discussion P a g e | 65
recommendations and included the following reactions (1) both GOx constructs with and
without additional FAD in the reaction, (2) the positive luciferase control supplied with the kit,
(3) the negative control without any DNA (also with and without FAD), and (4) a reaction with a
defined amount of GOx added to determine whether any IVT kit components interfere with
GOx activity. The FAD was added to improve GOx expression because it is a flavin enzyme and
the presence of FAD is essential in the folding process.
GOx expression could not be detected in any of the samples, whereas luciferase activity
could be detected in the control sample. The reactions containing added GOx showed normal
activity meaning that the components of the kit were not responsible for the loss of enzyme
activity. The amount of enzyme that could theoretically be produced by this kit was too low for
detection by Coomassie Brilliant Blue but sufficient for detection by western blot, but no GOx
antibodies were available and the enzyme was expressed without a tag. Therefore we could not
determine whether or not the enzyme was expressed, or whether it was expressed but
incorrectly folded hence inactive. We therefore attempted to denature and refold the enzyme
by incubating with Triton X‐100 and sonicating for 10 min (Witt, Singh, and Kalisz 1998). A slight
increase in absorbance was observed in the samples containing the AF variant but there was no
increase over time, suggesting the activity remains close to zero (Figure 30).
Because no clear positive results were obtained using the Qiagen kit we used an alternative
commercial kit incorporating a coupled transcription translation which is more advantageous
Figure 30 ABTS measurements of IVT samples (section II.2.1).
I I I . Results and discussion P a g e | 66
for screening (Promega 2008). As stated above, the ideal screening process would create the
libraries by PCR and compartmentalize the PCR products in the emulsion for the measurement
of expression and activity. Where transcription and translation are uncoupled, transcription
must be carried out separately and the mRNA must be collected for the translation reaction,
which means that the emulsions would have to be broken after the first step and recreated
around individual RNA molecules thus destroying the link between the gene and product.
We cloned the wild type GOx gene in vector pF25A with and without the AF and carried out
the same reactions as described above for the Qiagen kit. As above, we were unable to detect
GOx activity in any of the samples. We carried out an additional experiment which included
biotinylated lysine tRNA molecules in the translation reaction, allowing the detection of any
newly‐synthesized protein by western blot. However no band corresponding to GOx could be
distinguished in any sample.
Therefore we can conclude that the insect cell IVT system is not suitable for the expression
of GOx, at least in its current formulation. It may be possible to achieve GOx production by
modifying certain parameters, such as the DNA concentration, the time of expression and the
temperature, all of which affect translation efficiency. Furthermore, the structure and spacing
of elements such as the ribosome binding site and 5’ untranslated region can affect translation,
so different vector systems could be tested. However, IVT systems are expensive and in order
to try out all the different possibilities it would be advisable to develop cheaper and more
controllable homemade systems.
III.3 Alginate Bead Selection System for GOx
Natural evolution shows that the true power of screening involves selection methods that
trigger the survival of the fittest. We have therefore attempted to develop such a selection
system based on GOx activity, involving the encapsulation of yeast cells inside single emulsions
together with a tyramide‐alginate, HRP and glucose in such a way that there is only one cell per
compartment. When compartmentalization is complete, the tyramide phenolic groups are
polymerized by GOx leading to formation of solid beads around the cells with GOx activity.
When the emulsion components are removed, cells with GOx activity will be enclosed within a
I I I . Results and discussion P a g e | 67
solid bead whereas those lacking GOx activity will be free in the medium. The addition of a cell
wall degrading enzyme such as lyticase will then kill all the cells that are not protected by beads
therefore leaving in solution only those cells without GOx activity. This type of screening
platform would be more useful still in combination with IVT because a solid compartment
would form around DNA molecules expressing GOx activity allowing them to be separated from
non‐enclosed DNA.
In an initial set of experiments, we aimed to prove that we can generate beads using HRP
and externally provided peroxide. Alginate beads were also prepared using soluble GOx and
cells expressing the enzyme. When the experiment was carried out using 10% FITC‐stained cells
expressing active GOx, and 90% non‐stained cells lacking GOx, the beads formed mostly around
the cells expressing GOx (Figure 31).
Figure 31 Alginate‐tyramide reaction based on GOx activity using a mixture
of 10% active cells stained with FITC and 90% non‐active cells without staining
(section II.3.3).
I I I . Results and discussion P a g e | 68
The second step was to determine optimum conditions for lyticase activity. Most available
protocols use both chemical and enzymatic degradation to kill the cells, but our aim was to find
appropriate conditions in which the cytotoxicity was entirely attributed to the lyticase activity.
Therefore we tested different protocols to determine the influence of each component present
in the digestion buffer. The best protocol used a lysis buffer comprising 50 mM Tris‐HCl (pH
7.5), 10 mM EDTA and 0.5% (v/v) β‐mercaptoethanol. The lysis buffer did not affect the survival
of the cells, but the addition of lyticase killed 97% of the cells.
The assay was tested on reference libraries containing 5% positive cells and was compared
to pure cultures containing all positive or all negative cells. The first experiment killed more
than 95% of the cells regardless of their origin, suggesting the beads were not sufficient to
protect the cells from the lyticase activity probably because the pore size was large enough to
allow the relatively small lyticase enzyme (~40 kDa) to pass through. To reduce the pore size,
the beads were incubated in poly‐L‐lysine solution. This reduces the pore size because the
solute is positively charged, and therefore coats the surface of the negatively‐charged beads
(Klein, Stock, and Vorlop 1983). In addition the enzyme molecules were cross‐linked with poly‐
L‐lysine molecules using the EDAC coupling method (Anjaneyulu and Staros 1987). This
increased the size of the lyticase without affecting its catalytic activity.
The improved protocols were tested on the pure cultures of positive and negative cells,
resulting in survival rates of 11% and 7.7%, respectively. Although the survival rate of the
positive cells was reproducibly 1.4‐fold higher, this was not sufficient for a HTS system.
Interestingly, when soluble GOx was added to the cells, the survival rates increased to 26–33%.
This suggests that the assay is not sensitive enough for the levels of GOx present on the cell
surface and that higher GOx expression levels or more tyramide residues are needed to achieve
stronger alginate polymerization.
I I I . Results and discussion P a g e | 69
III.4 Directed Evolution of Cellulases
III.4.1 Enzyme assay
The enzyme assay for the detection of cellulase activity is based on the ViPer assay
previously developed for GOx (R. Prodanovic et al. 2012). The substrate 3‐carboxy‐7‐(4’‐
aminophenoxy)coumarin (APCC) can be prepared from an fluorescent compound (3‐
carboethoxy‐umbelliferone) and an activated carboxylic aryl group (p‐fluoronitrobenzene). The
aryl group acts as a cap that is sensitive to oxidation and quenches the fluorescent compound.
In the presence of peroxidase, the aryl group is removed from the capped fluorogenic substrate
and the fluorescent compound (3‐carboxy‐7‐hydroxycoumarin) is released. The compound was
synthesized in our laboratory as previously described (Khanna, Chang, and Ullman 1989).
As for the GOx assay, an emulsion is required to maintain the connection between the gene
and the product of the enzyme reaction (3‐carboxy‐7‐hydroxycoumarin). This product is
negatively charged and therefore does not diffuse from the emulsion, and is compatible with
FACS. Unlike other cellulase assays based on fluorescence, this one can use any natural
cellulase substrate. In theory any form of cellulose would be suitable, but we chose
carboxymethylcellulose (CMC) because it is soluble and similar to natural cellulose. The reaction
scheme is shown in Figure 32.
I I I . Results and discussion P a g e | 70
The reaction was initially tested in a MTP format using commercial A. niger cellulase,
resulting in a 25‐fold increase in fluorescence over the control reaction without cellulase. The
MTP assay was also tested using S. cerevisiae YDH 500 cells expressing cel5A (kindly provided by
Dr. U. Commandeur) compared to an empty vector control. The four reactions are compared in
Figure 33, revealing a lag phase at the beginning of the reaction that can be attributed to the
time required for cellulase to hydrolyze CMC into units that can be oxidized by HOx. The empty
cells also show a slight increase in fluorescence that probably reflects the presence of sugars on
the surface of the cells, which are oxidized by HOx and give a marginal increase over
background.
Figure 32 Schematic representation of ViPer assay for cellulases.
I I I . Results and discussion P a g e | 71
III.4.2 FACS assay
After successfully establishing a MTP format assay, we next adapted it to work under
emulsion conditions. The empty vector cells do not create a problematic background signal
because their fluorescence remains constant over time. As shown in Figure 34, the reaction is
much slower within the emulsions and takes more than 6 h to reach a steady state. This is
challenging because the cells have a low survival rate after 6 h under emulsion conditions which
interferes with the subsequent FACS analysis.
Figure 33 Detection of cellulase activity in MTPs using the ViPer assay.
I I I . Results and discussion P a g e | 72
Various methods were tested to increase the assay sensitivity, culminating in the use of a
different substrate, namely aminophenoxyfluorescein (APF) (Invitrogen, Carlsbad, USA), whose
fluorescence intensity is tenfold higher than APCC. Using this substrate, a significant difference
between the positive and negative cells was observed after 2 hours incubation (Figure 35).
Reference libraries were sorted under the P2 gate directly onto YNB‐CAA Glu agar plates.
Cells were grown at 27°C for 3 d after sorting and then transferred by replica platting onto YNB
CAA Gal/CMC plates for the induction of enzyme activity. They were incubated for a further 16–
24 h at 27°C and then tested using a Congo Red assay (Nakazawa et al. 2008) which revealed a
12‐fold enrichment in one round of sorting, as shown by halos around the positive colonies in
Figure 36. The assay is therefore highly efficient for the screening of cellulase diversity libraries.
Figure 34 FACS analysis of the ViPer cellulase assay over time, in emulsions containing negative cells (S. cerevisiae
YPH500 transformed with empty vector pESC‐Trp), positive cells (S. cerevisiae YPH500 transformed with vector pESC‐Trp
containing cel5A cellulase) and the soluble enzyme (Sigma cellulase from A. niger). These are double plots of UV and FITC
fluorescence with APCC used as the substrate. The enzyme activity was observed on the UV2 channel.
I I I . Results and discussion P a g e | 73
A novel florescent assay for the detection of cellulase activity was developed by modifying
the previously published ViPer assay (R. Prodanovic et al. 2012) used for the detection of GOx.
The assay was modified by using HOx to detect both monosaccharides and polysaccharides
resulting from cellulase activity and thus has the advantage of compatibility with any natural or
artificial cellulose substrate. The MTP format assay revealed a 25‐fold difference in
fluorescence between the positive and control reaction when using either soluble cellulase of
cells expressing cellulase. The emulsion screening format showed a clear difference between
positive and negative cells despite the long lag phase (4–6 h) before sufficient fluorescent
product accumulates for detection. In order to increase the sensitivity of the assay so that more
cells survive for FACS analysis, the APCC fluorogenic substrate was replaced with APF. Using this
Figure 36 Reference libraries containing positive and negative cells expressing cellulase activity before and
after the FACS sorting. Cellulase activity is detected by the formation of halos in the Congo Red agar plate assay.
Figure 35 FACS analysis of the ViPer cellulases assay over time in emulsions containing negative cells (S.
cerevisiae YPH500 transformed with empty vector pESC‐Trp), positive cells (S. cerevisiae YPH500 transformed
with vector pESC‐Trp containing cel5A cellulase) and a reference library with 5% positive cells. These are bivariate
histograms of forward scattering (FSC) and FITC.
I V . Final Discussion P a g e | 74
improved system, reference libraries were sorted and the active cellulase population was
increased 12‐fold in one round of sorting.
IV Final Discussion
IV.1 Glucose Oxidase Screening Systems
For the directed evolution of GOx, the wild‐type gene was cloned in a surface display vector
so that the expressed enzyme molecules were covalently linked onto the surface of the cell.
This maintains the connection between phenotype and genotype at all times during the
experiment. The connection is further maintained by the emulsion compartments, but if the
enzyme is not initially linked covalently to the cell surface then some enzyme molecules
produced by one cell can end up in another compartment making the detection system
inaccurate. Also, since we were aiming to detect not only the presence or absence of GOx
activity but also a specific activity, the amount of enzyme molecules present on the cell and
responsible for the reaction need to be quantified as precisely as possible. This is not possible
with non‐surface display systems because the cells must be washed extensively during
emulsification, enzyme reaction and antibody staining, resulting in the loss of many enzyme
molecules and of the correlation between enzyme abundance and activity in the detection step.
Surface display is therefore essential if the tyramide‐fluorescein assay is used to quantify the
enzyme as well as its activity. The use of both N‐ and C‐terminal tags may inhibit enzyme
activity but the impact appeared to be similar across all enzyme variants because the relative
activity of the wild‐type and different mutants remained the same even if the absolute activity
was reduced.
We increased the accuracy of the assay further by delivering the substrate once the
compartments had formed. In the initial assay procedure, glucose was included in the reaction
mix and the reaction started before emulsification was completed. The fluorescein tyramide
radicals were therefore already formed before compartmentalization was complete and this
lead to cross‐staining of a different cell than the one producing the radical, reducing the
accuracy of the assay and preventing the recovery of >40% positive cells in a sorted population.
I V . Final Discussion P a g e | 75
In addition, the concentration of glucose available in the assay could never be precisely
controlled since the yeast cells fed on the glucose. These problems were solved by the external
delivery of the substrate in the form of β‐octylglucoside.
We also found the optimal combination of fluorescent dyes for the simultaneous
measurement of enzyme activity and abundance. The best pairing was fluorescein to measure
activity and DyLight647 to measure abundance. Both dyes are intense enough to ensure
sensitivity and there is no overlap between their excitation and emission spectra.
By testing the assay with reference libraries, we found that it was possible to enrich the
active population of cells. Where the initial percentage of positive cells is <5%, the best strategy
is to sort initially using yield mode to recover the entire positive population, followed by single
cell mode to enrich for positive cells. We used the assay to distinguish between the wild‐type
enzyme and a mutant with five times the wild‐type activity, and found that the populations of
the two clones could be distinguished at low concentrations of glucose suggesting that it should
be possible to sort the highly active clones from a mutant library.
The GOx gene library was created using the consensus approach by site directed
mutagenesis at 16 amino acid positions. The library contained 105 different GOx mutants that
were multiplied 100‐fold to generate a library of 107 clones to ensure all diversity was screened.
This is necessary because of statistical sampling errors and because some cells are killed during
the assay (70% survival rate under reaction conditions and 60% survival rate during FACS).
The library was sorted using the tyramide fluorescein FACS based assay for two consecutive
rounds using low glucose concentrations and stringent gating in order to increase the likelihood
of selecting high‐activity clones. One hundred clones were analyzed after each sorting round
and were compared to the wild type enzyme using a MTP format assay. The results showed
that the percentage of active clones increased after each round of sorting, reaching 100% after
the second round, agreeing with the reference library experiments. The percentage of higher
activity clones also increased after each round, showing that the assay was suitable for the
quantitative screening of enzyme activity.
The mutants with the highest activity were collected and re‐analyzed. The five most
promising candidates were re‐cloned in P. pastoris expression vectors to remove the tags and
I V . Final Discussion P a g e | 76
produce large amounts of native protein for accurate kinetic characterization. Optimized
fermentation conditions yielded 30–40 mg of protein per liter of fermentation medium,
although 30% remained tightly attached to the cell surface and this could not be recovered. The
mutants were purified from the fermentation media by ultrafiltration and DEAE
chromatography resulting in approximately 10–20 mg of highly pure protein.
The kinetic parameters of all the mutants and the wild‐type enzyme were determined using
the ABTS assay and the data were fitted on the Michaelis‐Menten equation. The screening was
performed at pH 7.4 to favor the recovery of mutants that would function at the physiological
pH of human blood. All five mutants showed higher activity than wild‐type at their native pH
but the difference was more pronounced at pH 7.4. The best‐performing mutant was A2 with a
1.7‐fold lower kM, a 3.3‐fold higher kcat and a 5.8‐fold higher specificity constant. Mutant F9‐1
showed twice the thermostability as the parent and the wild type enzymes. The stability of the
A2 mutant was similar to wild type suggesting that a combination of the mutations present in
these mutants could produce a synergic mutant with higher activity and thermostability.
IV.1.1 In vitro translation
Despite the success of in vivo expression systems for GOx, their drawbacks include the low
transformation efficiency of yeast cells (maximum 106/µg of DNA) which reduces diversity
unless a high number of transformation reactions are performed. Additionally the 2–3 day
recovery time after sorting (reflecting the long generation interval of yeast cells) suggests that
only two rounds of evolution can be achieved per week. The potential of directed evolution can
be increased by eliminating these bottlenecks and reducing the screening time.
Another problem with cell‐based systems is the variable survival rates under different
conditions, narrowing thus the properties that can be tested by screening. The high‐throughput
in vivo formats assay the properties of each individual cell, whereas the MTP format screens
populations of (supposedly identical) cells. MTP screening always involves a master plate from
which clones can be recovered if cells are killed during screening. This makes it possible to
screen e.g. for thermostability and oxidative stability because the integrity of the cell is not
important and killed cells can be replaced. We encountered these while working with GOx but
they also apply to other enzyme classes.
I V . Final Discussion P a g e | 77
One solution to all these problems would be the development of an assay based on in vitro
translation (IVT), which would circumvent the limitations of transformation, reduce the
generation time to a few hours thus allowing up to three rounds of evolution per day and all
direct HTS for thermostability because DNA is not affected at high temperatures in the same
way as cells. For GOx screening, we were able to develop an assay based on the covalent
staining of the yeast cells surface but such adaptations may not be possible for many other
enzymes. Developing a fluorescence assay with a soluble product would be simpler, but double
emulsions would be required to maintain the connection between the gene and the enzyme
activity during screening. However the encapsulation efficiency of yeast cells in double
emulsions is ~10% and even E. coli cells cannot exceed ~25%, meaning that a lot of empty
droplets would be present in the mix. IVT systems would increase the encapsulation efficiency
and allow the use of smaller and more homogeneous emulsions.
Unfortunately, our attempts to express GOx by IVT using insect cell extracts were not
successful. However, there are numerous potential solutions such as varying the DNA
concentration, using different expression vectors or changing the temperature. Another option
is to use different cell free extracts, e.g. yeast cells, wheat germ or E. coli.
IV.1.2 Alginate bead selection system for GOx
We developed a novel selection system based on GOx activity by using alginate modified
with tyramide residues. The reaction involves the detection of hydrogen peroxide formed by
GOx, and this is used by HRP to polymerize the phenolic groups of tyramide resulting in the
formation of solid compartments around cells expressing GOx. When lyticase is added, cells
that are not protected by the alginate compartments are killed. We were able to demonstrate
the formation of stable compartments preferentially around cells expressing GOx despite a
certain amount of cross‐talk. However no enrichment for GOx‐positive cells was observed in
different reference libraries after lyticase treatment and even population of 100% GOx‐positive
cells showed less than 3% survival suggesting that the alginate beads did not protect the cells
against lyticase activity.
We addressed this challenge by using poly‐L‐lysine to reduce the pore size of the beads and
to increase the size of lyticase by cross‐linking. Unfortunately, even after these improvements
I V . Final Discussion P a g e | 78
there was no enrichment of the reference libraries and the survival rate in the 100% GOx‐
positive cells remained at 3%. However, by adding soluble GOx the survival rate increased to
33%, suggesting that the GOx activity on the cell surface is not sufficient to induce significant
polymerization of the tyramide‐alginate.
IV.1.3 Cellulase assay
We developed a new florescence assay for the detection of cellulase activity by modifying
the ViPer GOx assay described before. Instead of GOx for the detection of glucose monomers
resulting from cellulase activity, we used HOx. This has broader substrate specificity than GOx
and can oxidize both monosaccharides and oligosaccharides. Glucose is an uncommon product
of cellulase degradation so this improvement in the assay is considerable, theoretically allowing
the use of any cellulosic substrate. However, we used CMC to test the assay because it is
soluble and similar to natural cellulose.
Initial MTP tests using soluble cellulase and the fluorogenic substrate APCC revealed a 25‐
fold difference in fluorescence between reactions containing soluble cellulase and controls
lacking the enzyme. A lag phase was observed before the detection of fluorescence, which can
be attributed to the time needed for the cellulase to produce saccharides that can be oxidized
by HOx. Similar results were obtained when the soluble cellulase was replaced with yeast cells
expressing cellulase. The control cells (empty vector) also generated some background
fluorescence probably reflecting the oxidation of sugar residues on the cell surface, but unlike
the cellulase‐expressing cells this fluorescence did not increase over time, resulting in a clear
difference between positive and negative cells by the assay end‐point.
This optimized assay was used to detect cellulase activity inside double emulsions. When
soluble cellulase was used we could detect fluorescence immediately after emulsification even
though the signal was relatively low. When positive cells were used, a clear difference between
positive and negative cells was observed only after 4 to 6 h of incubation, at which point the
survival of the cells was compromised. We therefore replaced the APCC substrate with APF,
which has a ten‐fold higher quantum yield making the assay ten times more sensitive. With this
substrate, fluorescence could be detected after 2 h of incubation. A 5% reference library was
V . Conclusions P a g e | 79
sorted using this assay and 12‐fold enrichment of the positive population was achieved after
one sorting round, validating the use of this screening system.
V Conclusions
We have developed a new and improved version of the previously‐ tyramide‐fluorescein
assay for GOx (Radivoje Prodanovic et al. 2011) that avoids cross‐talk between droplets.
We developed a new procedure for the external delivery of glucose through the oil
phase in the form of ‐octylglucoside, allowing substrate delivery after
compartmentalization is completed.
We modified the assay to include the surface display of GOx variants bearing an epitope
that allows quantitation using antibodies.
The sorting of reference libraries using the new assay resulted in up to 90% enrichment
for positive cell populations.
Libraries containing 105 different GOx mutants were created using the consensus
approach and 107 (100 copies of each) were screened using the new fluorescence assay.
A purity of 99% positive clones was obtained after only two rounds of sorting.
The proposed assay can be used to screen for higher‐activity clones, not only as a
plus/minus system. This was demonstrated after two rounds of sorting, by increasing
the percentage of higher‐activity clones six‐fold.
By sorting the consensus libraries, we isolated five mutants with 3–5.8‐fold more activity
than the wild type GOx.
One of the mutants demonstrated twice the thermal stability of the wild‐type enzyme.
An alginate/tyramide‐based selection system was developed for the detection of GOx
activity.
The assay showed promising results because beads formed preferentially around cells
expressing GOx activity.
However when lyticase was added to the system, all the cells were killed suggesting that
the beads do not protect the cells against lyticase activity.
V . Conclusions P a g e | 80
Cells survive only when high concentrations of soluble GOx are added to the
compartments, suggesting that more GOx needs to be expressed on the cell surface or
more tyramine molecules are needed on the modified alginate.
A new high‐throughput fluorescence assay based on FACS screening in double emulsions
was developed for the detection of cellulase activity, which can be used with any natural
cellulosic substrate.
The assay is based on the detection of cellulase hydrolysis products using an enzymatic
cascade which leads to the formation of a fluorescent product.
Using this assay, reference cellulase libraries were screened and 60% enrichment for
positive cells could be achieved after only one round of sorting.
V I . List of Abbreviations P a g e | 81
VI ListofAbbreviations
ABTS 2,2′‐azino‐bis(3‐ethybenz‐thiazoline‐6‐sulfonic acid
3D tree dimensional (associate with protein structure)
7‐AAD 7‐Aminoactinomycin D
AF pre‐pro‐alpha factor signal sequence
AMCA amino‐methyl‐coumarin‐acetate
AMCA‐Tyr amino‐methyl‐coumarin‐acetate Tyramine
APCC 3‐Carboxy‐7‐(4’‐aminophenoxy)‐coumarine
APF Aminophenoxy fluorescein
BLAST basic local alignment search tool – an algorithm for comparing primary biological
sequence information
BMGY Buffered Glycerol Yeast complex medium (see II.1.4)
BMMY Buffered Methanol Yeast complex medium (see II.1.4)
bp base pair
CMC carboxy‐methyl cellulose
CV column volume
d day
DMF dimethylformamide
DMSO dimethylsulfoxide
DNA 2’‐Desoxynucleotide‐5’‐triphosphate
DNase deoxyribonuclease I
DNS 3,5‐diniotrosalycilic acid (referred either to this chemical or to the method that use it)
dNTP desoxyribonucleicacid
DTT dithiothreitol
EDAC 1‐Ethyl‐3‐[3‐dimethylaminopropyl]carbodiimide hydrochloride
EDTA ethylenediaminetetraacetic acid
epPCR Error‐prone polymerase chain reaction
ER endoplasmic reticulum
V I . List of Abbreviations P a g e | 82
FACS fluorescence activated cell sorting
FAD flavin adenine dinucleotide
Fc IgG fragment c of IgG
FCS forward scattering
FITC Fluorescein isothiocyanate
Gal galactose
Glu glucose
GOx glucose oxidase
h hour
Hox hexose oxidase
HPLC high performance liquid chromatography
HRP horseradish peroxidase
HTS high throughput screening
HTS high throughput screening
IgG immunoglobulin G
IUPAC international union of pure and applied chemistry
IVT in vitro translation
kcat catalytic constant
kM Michaelis‐Menten constant
LB medium Luria‐Bertani medium
LMO light mineral oil
min minute
mM millimolar
mol mol
MTP microtiter plate
NHS N‐hydroxysuccinimide
nm nanometer
OD optical density
PAGE polyacrylamide gel electrophoresis
V I . List of Abbreviations P a g e | 83
PBS phosphate buffer saline
PCR polymerase chain reaction
Pfu‐Pol Pyrococcus furiosus DNA polymerase
pH Pondus hydrogenii ("power of hydrogen") – pH is defined as the decimal logarithm of
the reciprocal of the hydrogen ion activity
PLL poly‐L‐lysine
PTFE polytetrafluoroethylene – in this thesis, the reference is to the material used to made
microbiological filters
Raf raffinose
RNA ribonucleic acid
rpm rotation per minute
SDS sodium dodecyl sulfate
Span 80 a commercial name of a nonionic surfactant (Sorbitane monooleate)
Taq DNA Pol Thermus aquaticus DNA polymerase
Tris‐HCl a buffer made with Tris(hydroxymethyl) aminomethane hydrochloride and
hydrochloric acid
Triton X‐100 comercial name of a nonionic surfactant (octyl phenol ethoxylate)
tRNA transfer RNA
U units
UV ultraviolet
v/v volulme / volume – referred to solutions (mixtures) preparation
Vmax maximum reaction speed
VnBrPO vanadium bromo perodidase
w/v weight / volume – referred to solutions (mixtures) preparation
w/w weight / weight – referred to solutions (mixtures) preparation
Wt wild type
YNB CAA
Glu/Gal, Raf
Yeast Nitrogen Base Casamino Acids media(see II.1.2)
YPD yeast extract/peptone/dextrose medium (see II.1.3)
V I I . Literature P a g e | 84
VII Literature
Agrawal, Abinash, and Paul G. Tratnyek. 1996. “Reduction of Nitro Aromatic Compounds by Zero‐Valent Iron Metal.” Environmental Science & Technology 30 (1) (January): 153–160. doi:10.1021/es950211h. http://dx.doi.org/10.1021/es950211h.
Aharoni, Amir, Gil Amitai, Kalia Bernath, Shlomo Magdassi, and Dan S Tawfik. 2005. “High‐throughput Screening of Enzyme Libraries: Thiolactonases Evolved by Fluorescence‐activated Sorting of Single Cells in Emulsion Compartments.” Chemistry & Biology 12 (12) (December): 1281–9. doi:10.1016/j.chembiol.2005.09.012. http://www.ncbi.nlm.nih.gov/pubmed/16356845.
Aharoni, Amir, Karena Thieme, Cecilia P C Chiu, Sabrina Buchini, Luke L Lairson, Hongming Chen, Natalie C J Strynadka, Warren W Wakarchuk, and Stephen G Withers. 2006. “High‐throughput Screening Methodology for the Directed Evolution of Glycosyltransferases.” Nature Methods 3 (8): 609–614. doi:10.1038/NMETH899.
Anjaneyulu, P S, and J V Staros. 1987. “Reactions of N‐hydroxysulfosuccinimide Active Esters.” International Journal of Peptide and Protein Research 30 (1) (July): 117–24. http://www.ncbi.nlm.nih.gov/pubmed/3667072.
Anon. 2009. World Enzymes Market. http://www.reportlinker.com/p0148002‐summary/World‐Enzymes‐Market.html.
Arnold, F H, and A A Volkov. 1999. “Directed Evolution of Biocatalysts.” Current Opinion in Chemical Biology 3 (1) (February): 54–9. http://www.ncbi.nlm.nih.gov/pubmed/10021399.
De Baetselier, A, A Vasavada, P Dohet, V Ha‐Thi, M De Beukelaer, T Erpicum, L De Clerck, J Hanotier, and S Rosenberg. 1991. “Fermentation of a Yeast Producing A. Niger Glucose Oxidase: Scale‐up, Purification and Characterization of the Recombinant Enzyme.” Bio/technology (Nature Publishing Company) 9 (6) (June): 559–61. http://www.ncbi.nlm.nih.gov/pubmed/1367226.
Baron, A J, C Stevens, C Wilmot, K D Seneviratne, V Blakeley, D M Dooley, S E Phillips, P F Knowles, and M J McPherson. 1994. “Structure and Mechanism of Galactose Oxidase. The Free Radical Site.” Journal of Biologycal Chemistry 269 (40) (October): 25095–25105. http://www.jbc.org/cgi/content/abstract/269/40/25095.
Becker, D M, and L Guarente. 1991. “High‐efficiency Transformation of Yeast by Electroporation.” Methods in Enzymology 194 (January): 182–7. http://www.ncbi.nlm.nih.gov/pubmed/2005786.
Becker, Stefan, Hans‐Ulrich Schmoldt, Thorsten Michael Adams, Susanne Wilhelm, and Harald Kolmar. 2004. “Ultra‐high‐throughput Screening Based on Cell‐surface Display and Fluorescence‐activated Cell Sorting for the Identification of Novel Biocatalysts.” Current Opinion in Biotechnology 15 (4) (August): 323–9. doi:10.1016/j.copbio.2004.06.001. http://www.ncbi.nlm.nih.gov/pubmed/15296929.
Bernath, Kalia, Mingtan Hai, Enrico Mastrobattista, Andrew D Griffiths, Shlomo Magdassi, and Dan S Tawfik. 2004. “In Vitro Compartmentalization by Double Emulsions: Sorting and Gene Enrichment by Fluorescence Activated Cell Sorting.” Analytical Biochemistry 325 (1) (February 1): 151–7. http://www.ncbi.nlm.nih.gov/pubmed/14715296.
Bertani, G. 1951. “Studies on Lysogenesis. I. The Mode of Phage Liberation by Lysogenic Escherichia Coli.” Journal of Bacteriology 62 (3) (September): 293–300.
V I I . Literature P a g e | 85
http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=386127&tool=pmcentrez&rendertype=abstract.
Bhat, M K. 2000. “Cellulases and Related Enzymes in Biotechnology.” Biotechnology Advances 18 (5) (August): 355–83. http://www.ncbi.nlm.nih.gov/pubmed/14538100.
Biles, Benjamin D, and Bernard A Connolly. 2004. “Low‐fidelity Pyrococcus Furiosus DNA Polymerase Mutants Useful in Error‐prone PCR.” Nucleic Acids Research 32 (22) (January): e176. doi:10.1093/nar/gnh174. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=545472&tool=pmcentrez&rendertype=abstract.
BIO030F. 2011. Enzymes in Industrial Applications: Global Markets. Boder, Eric T., and K Dane Wittrup. 1997. “Yeast Surface Display for Screening Combinatorial
Polypeptide Libraries.” Nature Biotechnology 15 (6) (June): 553–7. doi:10.1038/nbt0697‐553. http://dx.doi.org/10.1038/nbt0697‐553.
Boschker, H T, T E Cappenberg, and A Sensitive Method Using. 1994. “A Sensitive Method Using 4‐methylumbelliferyl‐beta‐cellobiose as a Substrate to Measure (1,4)‐beta‐glucanase Activity in Sediments.” Applied and Environmental Microbiology 60 (10) (October): 3592–6. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=201860&tool=pmcentrez&rendertype=abstract.
Brake, A J, J P Merryweather, D G Coit, U A Heberlein, F R Masiarz, G T Mullenbach, M S Urdea, P Valenzuela, and P J Barr. 1984. “Alpha‐factor‐directed Synthesis and Secretion of Mature Foreign Proteins in Saccharomyces Cerevisiae.” Proceedings of the National Academy of Sciences of the United States of America 81 (15) (August): 4642–6. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=391546&tool=pmcentrez&rendertype=abstract.
Bright, H J, and M Appleby. 1969. “The pH Dependence of the Individual Steps in the Glucose Oxidase Reaction.” The Journal of Biological Chemistry 244 (13) (July 10): 3625–34. http://www.ncbi.nlm.nih.gov/pubmed/5794229.
Cadwell, R C, and G F Joyce. 1992. “Randomization of Genes by PCR Mutagenesis.” PCR Methods and Applications 2 (1) (August): 28–33. http://www.ncbi.nlm.nih.gov/pubmed/1490172.
Campana, Dario, and Elaine Coustan‐Smith. 2004. “Minimal Residual Disease Studies by Flow Cytometry in Acute Leukemia.” Acta Haematologica 112 (1‐2) (January): 8–15. doi:10.1159/000077554. http://www.ncbi.nlm.nih.gov/pubmed/15178999.
Cappellaro, Corinna, Karin Hauser, Vladimir Mrsa, Manfred Watzele, Gabriele Watzele, Claudia Gruber, Widmar Tanner, and V Mrśa. 1991. “Saccharomyces Cerevisiae a‐ and Alpha‐agglutinin: Characterization of Their Molecular Interaction.” The European Molecular Biology Organization Journal 10 (13) (December): 4081–8. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=453156&tool=pmcentrez&rendertype=abstract.
Carraway, K L, and R B Triplett. 1970. “Reaction of Carbodiimides with Protein Sulfhydryl Groups.” Biochimica Et Biophysica Acta 200 (3) (March 31): 564–6. http://www.ncbi.nlm.nih.gov/pubmed/5461755.
V I I . Literature P a g e | 86
Chan, L W, P W Heng, and L S Wan. 1997. “Effect of Cellulose Derivatives on Alginate Microspheres Prepared by Emulsification.” Journal of Microencapsulation 14 (5): 545–55. doi:10.3109/02652049709006808. http://www.ncbi.nlm.nih.gov/pubmed/9292431.
Chilvers, K.F., J.D. Perry, A.L. James, and R.H. Reed. 2001. “Synthesis and Evaluation of Novel Fluorogenic Substrates for the Detection of Bacterial Beta‐galactosidase.” Journal of Applied Microbiology 91 (6) (December): 1118–1130. doi:10.1046/j.1365‐2672.2001.01484.x. http://doi.wiley.com/10.1046/j.1365‐2672.2001.01484.x.
Cirino, Patrick C, Kimberly M Mayer, and Daisuke Umeno. 2003. “Generating Mutant Libraries Using Error‐prone PCR.” Methods in Molecular Biology (Clifton, N.J.) 231 (January): 3–9. doi:10.1385/1‐59259‐395‐X:3. http://www.ncbi.nlm.nih.gov/pubmed/12824595.
Coco, W M, W E Levinson, M J Crist, H J Hektor, A Darzins, P T Pienkos, C H Squires, and D J Monticello. 2001. “DNA Shuffling Method for Generating Highly Recombined Genes and Evolved Enzymes.” Nature Biotechnology 19 (4) (April): 354–9. doi:10.1038/86744. http://www.ncbi.nlm.nih.gov/pubmed/11283594.
Cohen, N, S Abramov, Y Dror, and A Freeman. 2001. “In Vitro Enzyme Evolution: The Screening Challenge of Isolating the One in a Million.” Trends in Biotechnology 19 (12) (December): 507–10. http://www.ncbi.nlm.nih.gov/pubmed/11711194.
Coleman, D.J., MJ Studler, and JJ Naleway. 2007. “A Long‐wavelength Fluorescent Substrate for Continuous Fluorometric Determination of Cellulase Activity: resorufin‐beta‐D‐cellobioside.” Analytical Biochemistry 371 (2) (December 15): 146–53. doi:10.1016/j.ab.2007.08.027. http://dx.doi.org/10.1016/j.ab.2007.08.027.
Coradin, T, N Nassif, and J Livage. 2003. “Silica‐alginate Composites for Microencapsulation.” Applied Microbiology and Biotechnology 61 (5‐6) (June): 429–34. doi:10.1007/s00253‐003‐1308‐5. http://www.ncbi.nlm.nih.gov/pubmed/12700874.
Craig, Fiona E, and Kenneth A Foon. 2008. “Flow Cytometric Immunophenotyping for Hematologic Neoplasms.” Blood 111 (8) (April 15): 3941–67. doi:10.1182/blood‐2007‐11‐120535. http://www.ncbi.nlm.nih.gov/pubmed/18198345.
Dashtban, Mehdi, Miranda Maki, Kam Tin Leung, Canquan Mao, and Wensheng Qin. 2010. “Cellulase Activities in Biomass Conversion: Measurement Methods and Comparison.” Critical Reviews in Biotechnology 30 (4) (December): 302–9. doi:10.3109/07388551.2010.490938. http://www.ncbi.nlm.nih.gov/pubmed/20868219.
Dashtban, Mehdi, Heidi Schraft, and Wensheng Qin. 2009. “Fungal Bioconversion of Lignocellulosic Residues; Opportunities & Perspectives.” International Journal of Biological Sciences 5 (6) (January): 578–95. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2748470&tool=pmcentrez&rendertype=abstract.
Daugherty, P S, B L Iverson, and G Georgiou. 2000. “Flow Cytometric Screening of Cell‐based Libraries.” Journal of Imunological Methods 243 (1‐2) (September 21): 211–27. http://www.ncbi.nlm.nih.gov/pubmed/10986416.
Doi, N, and H Yanagawa. 1999. “Design of Generic Biosensors Based on Green Fluorescent Proteins with Allosteric Sites by Directed Evolution.” FEBS Letters 453 (3) (June 25): 305–7. http://www.ncbi.nlm.nih.gov/pubmed/10405165.
Draget, Kurt Ingar, Bjørn T Stokke, Yoshiaki Yuguchi, Hiroshi Urakawa, and Kanji Kajiwara. 2003. “Small‐angle X‐ray Scattering and Rheological Characterization of Alginate Gels. 3. Alginic
V I I . Literature P a g e | 87
Acid Gels.” Biomacromolecules 4 (6): 1661–8. doi:10.1021/bm034105g. http://www.ncbi.nlm.nih.gov/pubmed/14606893.
Farkas, V., Maria Lisikova, and P. Biely. 1985. “Novel Media for Detection of Microbial Producers of Cellulase and Xylanase.” FEMS Microbiology Letters 28 (2) (June): 137–140. doi:10.1111/j.1574‐6968.1985.tb00779.x. http://doi.wiley.com/10.1111/j.1574‐6968.1985.tb00779.x.
Feldhaus, Michael J, Robert W Siegel, Lee K Opresko, James R Coleman, Jane M Weaver Feldhaus, Yik A Yeung, Jennifer R Cochran, et al. 2003. “Flow‐cytometric Isolation of Human Antibodies from a Nonimmune Saccharomyces Cerevisiae Surface Display Library.” Nature Biotechnology 21 (2) (February): 163–70. doi:10.1038/nbt785. http://www.ncbi.nlm.nih.gov/pubmed/12536217.
Fernández Morató, J, L Ilzarbe Sánchez, J Bessa Caserras, and J Mateu de Antonio. 2009. “[Cellulase Treatment in 3 Cases of Large Phytobezoars].” Farmacia Hospitalaria : Órgano Oficial De Expresión Científica De La Sociedad Española De Farmacia Hospitalaria 33 (2): 100–3. http://www.ncbi.nlm.nih.gov/pubmed/19480798.
Fields, S, and O Song. 1989. “A Novel Genetic System to Detect Protein‐protein Interactions.” Nature 340 (6230) (July 20): 245–6. doi:10.1038/340245a0. http://www.ncbi.nlm.nih.gov/pubmed/2547163.
Frederick, K R, J Tung, R S Emerick, F R Masiarz, S H Chamberlain, A Vasavada, S Rosenberg, S Chakraborty, L M Schopfer, and L M Schopter. 1990. “Glucose Oxidase from Aspergillus Niger. Cloning, Gene Sequence, Secretion from Saccharomyces Cerevisiae and Kinetic Analysis of a Yeast‐derived Enzyme.” The Journal of Biological Chemistry 265 (7) (March 5): 3793–802. http://www.ncbi.nlm.nih.gov/pubmed/2406261.
Fujii, Ryota, Motomitsu Kitaoka, and Kiyoshi Hayashi. 2006. “RAISE: a Simple and Novel Method of Generating Random Insertion and Deletion Mutations.” Nucleic Acids Research 34 (4) (January): e30. doi:10.1093/nar/gnj032. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=1380258&tool=pmcentrez&rendertype=abstract.
Fujita, Yasuya, Shouji Takahashi, Mitsuyoshi Ueda, Atsuo Tanaka, Hirofumi Okada, Yasushi Morikawa, Takashi Kawaguchi, Motoo Arai, Hideki Fukuda, and Akihiko Kondo. 2002. “Direct and Efficient Production of Ethanol from Cellulosic Material with a Yeast Strain Displaying Cellulolytic Enzymes.” Applied and Environmental Microbiology 68 (10) (October): 5136–41. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=126432&tool=pmcentrez&rendertype=abstract.
Fukuoka, Tokuma, Hiroshi Uyama, and Shiro Kobayashi. 2004. “Polymerization of Polyfunctional Macromolecules: Synthesis of a New Class of High Molecular Weight Poly(amino Acid)s by Oxidative Coupling of Phenol‐containing Precursor Polymers.” Biomacromolecules 5 (3): 977–83. doi:10.1021/bm034490+. http://www.ncbi.nlm.nih.gov/pubmed/15132690.
Furukawa, Kensuke. 2003. “‘Super Bugs’ for Bioremediation.” Trends in Biotechnology 21 (5) (May): 187–90. http://www.ncbi.nlm.nih.gov/pubmed/12727376.
G Wagner, J, D W Schmidtke, C P Quinn, T F Fleming, B Bernacky, and A Heller. 1998. “Continuous Amperometric Monitoring of Glucose in a Brittle Diabetic Chimpanzee with a Miniature Subcutaneous Electrode.” Proceedings of the National Academy of Sciences of
V I I . Literature P a g e | 88
the United States of America 95 (11): 6379–6382. http://www.mendeley.com/research/continuous‐amperometric‐monitoring‐glucose‐brittle‐diabetic‐chimpanzee‐miniature‐subcutaneous‐electrode/.
Georgiou, G. 2000. “Analysis of Large Libraries of Protein Mutants Using Flow Cytometry.” Advances in Protein Chemistry 55 (January): 293–315. http://www.ncbi.nlm.nih.gov/pubmed/11050937.
Gietz, R Daniel, and Robert H Schiestl. 2007. “High‐efficiency Yeast Transformation Using the LiAc/SS Carrier DNA/PEG Method.” Nature Protocols 2 (1) (January): 31–4. doi:10.1038/nprot.2007.13. http://www.ncbi.nlm.nih.gov/pubmed/17401334.
Van Gijlswijk, R P, H J Zijlmans, J Wiegant, M N Bobrow, T J Erickson, K E Adler, H J Tanke, and A K Raap. 1997. “Fluorochrome‐labeled Tyramides: Use in Immunocytochemistry and Fluorescence in Situ Hybridization.” The Journal of Histochemistry and Cytochemistry: Official Journal of the Histochemistry Society 45 (3) (March): 375–82. http://www.ncbi.nlm.nih.gov/pubmed/9071319.
Givan, A L. 2001. “Principles of Flow Cytometry: An Overview.” Methods in Cell Biology 63 (January): 19–50. http://www.ncbi.nlm.nih.gov/pubmed/11060835.
Godfrey, T., and S. West. 1996. Introduction to Industrial Enzymology. Ed. T. Godfrey and S. West. 2nd ed. Macmillan Press.
Green, F, C A Clausen, and T L Highley. 1989. “Adaptation of the Nelson‐Somogyi Reducing‐sugar Assay to a Microassay Using Microtiter Plates.” Analytical Biochemistry 182 (2) (November 1): 197–9. http://www.ncbi.nlm.nih.gov/pubmed/2610335.
Gumulya, Yosephine, and Manfred T Reetz. 2011. “Enhancing the Thermal Robustness of an Enzyme by Directed Evolution: Least Favorable Starting Points and Inferior Mutants Can Map Superior Evolutionary Pathways.” Chembiochem : a European Journal of Chemical Biology 12 (16) (September 13): 2502–2510. doi:10.1002/cbic.201100412. http://www.ncbi.nlm.nih.gov/pubmed/21913300.
Hankin, L, and S L Anagnostakis. 1977. “Solid Media Containing Carboxymethylcellulose to Detect CX Cellulose Activity of Micro‐organisms.” Journal of General Microbiology 98 (1) (January): 109–15. http://www.ncbi.nlm.nih.gov/pubmed/401863.
Hansen, O. C., and P. Stougaard. 1997. “Hexose Oxidase from the Red Alga Chondrus Crispus.” The Journal of Biological Chemistry 272 (17): 11581–11587.
Hatzinikolaou, D G, O C Hansen, B J Macris, A Tingey, D Kekos, P Goodenough, and P Stougaard. 1996. “A New Glucose Oxidase from Aspergillus Niger: Characterization and Regulation Studies of Enzyme and Gene.” Applied Microbiology and Biotechnology 46 (4) (November): 371–81. http://www.ncbi.nlm.nih.gov/pubmed/8987726.
Helbert, William, Henri Chanzy, Tommy Lykke Husum, Martin Schülein, and Steffen Ernst. 2003. “Fluorescent Cellulose Microfibrils as Substrate for the Detection of Cellulase Activity.” Biomacromolecules 4 (3): 481–7. doi:10.1021/bm020076i. http://www.ncbi.nlm.nih.gov/pubmed/12741760.
Heng, P W S, L W Chan, and T W Wong. 2003. “Formation of Alginate Microspheres Produced Using Emulsification Technique.” Journal of Microencapsulation 20 (3): 401–13. doi:10.1080/0265204031000093069. http://www.ncbi.nlm.nih.gov/pubmed/12881119.
Herman, Asael, and Dan S Tawfik. 2007. “Incorporating Synthetic Oligonucleotides via Gene Reassembly (ISOR): a Versatile Tool for Generating Targeted Libraries.” Protein
V I I . Literature P a g e | 89
Engineering, Design & Selection : PEDS 20 (5) (May): 219–26. doi:10.1093/protein/gzm014. http://www.ncbi.nlm.nih.gov/pubmed/17483523.
Hogrefe, Holly H, Janice Cline, Geri L Youngblood, and Ronda M Allen. 2002. “Creating Randomized Amino Acid Libraries with the QuikChange Multi Site‐Directed Mutagenesis Kit.” BioTechniques 33 (5) (November): 1158–60, 1162, 1164–5. http://www.ncbi.nlm.nih.gov/pubmed/12449398.
Invitrogen. 2002. “pYD1 Yeast Display Vector Kit” Catalog no. ———. 2010. EasySelect TM Pichia Expression Kit For Expression of Recombinant Proteins Using
pPICZ and pPICZα in Pichia Pastoris. Computer Programs in Biomedicine. Vol. 18. Carlsbad. doi:Manual part no. 25‐0172.
Jackel, C, P Kast, and D Hilvert. 2008. “Annual Review of Biophysics.” In Annual Review of Biophysics, 153–173. Palo Alto: Annual reviews.
Jaeger, Karl‐Erich, and Thorsten Eggert. 2004. “Enantioselective Biocatalysis Optimized by Directed Evolution.” Current Opinion in Biotechnology 15 (4) (August): 305–13. doi:10.1016/j.copbio.2004.06.007. http://www.ncbi.nlm.nih.gov/pubmed/15358000.
James, J, and B K Simpson. 1996. “Application of Enzymes in Food Processing.” Critical Reviews in Food Science and Nutrition 36 (5): 437–463. http://www.mendeley.com/research/application‐enzymes‐food‐processing/.
Kaiser, C., S. Michaelis, and A. Mitchell. 1994. Methods in Yeast Genetics. CSHL Press, Cold Spring Harbor.
Kalisz, H M, H J Hecht, D Schomburg, and R D Schmid. 1991. “Effects of Carbohydrate Depletion on the Structure, Stability and Activity of Glucose Oxidase from Aspergillus Niger.” Biochimica Et Biophysica Acta 1080 (2) (October 25): 138–42. http://www.ncbi.nlm.nih.gov/pubmed/1932088.
Karst, S. M., N. Sadeghi, and T. M. Menees. 1999. “Cell Cycle Control of Reverse Transcriptase Activity for the Yeast Retrotransposon Ty3.” BioChemistry Biophysics Research Communication 254 (3): 679–84.
Kasai, Naoya, Ayako Konishi, Kazuya Iwai, and Gou Maeda. 2006. “Efficient Digestion and Structural Characteristics of Cell Walls of Coffee Beans.” Journal of Agricultural and Food Chemistry 54 (17) (August 23): 6336–42. doi:10.1021/jf0609072. http://www.ncbi.nlm.nih.gov/pubmed/16910728.
Kasana, Ramesh Chand, Richa Salwan, Hena Dhar, Som Dutt, and Arvind Gulati. 2008. “A Rapid and Easy Method for the Detection of Microbial Cellulases on Agar Plates Using Gram’s Iodine.” Current Microbiology 57 (5) (November 1): 503–7. doi:10.1007/s00284‐008‐9276‐8. http://www.springerlink.com/content/q7g54721205r26k3/.
Kawarasaki, Yasuaki, Karl E Griswold, James D Stevenson, Tzvia Selzer, Stephen J Benkovic, Brent L Iverson, and George Georgiou. 2003. “Enhanced Crossover SCRATCHY: Construction and High‐throughput Screening of a Combinatorial Library Containing Multiple Non‐homologous Crossovers.” Nucleic Acids Research 31 (21) (November 1): e126. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=275483&tool=pmcentrez&rendertype=abstract.
Keilin, D, and E F Hartree. 1952. “Specificity of Glucose Oxidase (notatin).” The Biochemical Journal 50 (3) (January): 331–41.
V I I . Literature P a g e | 90
http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=1197657&tool=pmcentrez&rendertype=abstract.
Khanna, Pyare L., Chiu C. Chang, and Edwin F. Ullman. 1989. “Method for the Determination of Peroxidase Using Fluorogenic Substrates.” http://www.freepatentsonline.com/4857455.html.
King, G.A., A.J. Daugulis, P. Faulkner, and M.F. Goosen. 1987. “Alginate‐polylysine Microcapsules of Controlled Membrane Molecular Weight Cutoff for Mammalian Cell Culture Engineering.” Biotechnology Progress 3: 231–240.
Klein, J., J. Stock, and K.D. Vorlop. 1983. “Pore Size and Properties of Spherical Ca‐Alginate Biocatalysts.” European Journal of Applied Microbioly and Biotechnology 18: 86–91.
Kriechbaum, M, H J Heilmann, F J Wientjes, M Hahn, K D Jany, H G Gassen, F Sharif, and G Alaeddinoglu. 1989. “Cloning and DNA Sequence Analysis of the Glucose Oxidase Gene from Aspergillus Niger NRRL‐3.” FEBS Letters 255 (1) (September 11): 63–6. http://www.ncbi.nlm.nih.gov/pubmed/2792372.
Kudriashova, E V, I S Vasileva, I N Zorov, A P Sinitsyn, and A V Levashov. 2009. “Stabilization of Alkaline Proteinase and Cellulases via Complex Formation with Chitosan for Use as Detergent Components.” Bioorganicheskaia Khimiia 35 (3): 368–75. http://www.ncbi.nlm.nih.gov/pubmed/19621052.
Kurisawa, Motoichi, Joo Eun Chung, Yi Yan Yang, Shu Jun Gao, and Hiroshi Uyama. 2005. “Injectable Biodegradable Hydrogels Composed of Hyaluronic Acid‐tyramine Conjugates for Drug Delivery and Tissue Engineering.” Chemical Communications (Cambridge, England) (34) (September 14): 4312–4. doi:10.1039/b506989k. http://www.ncbi.nlm.nih.gov/pubmed/16113732.
Kurtzman, A L, S Govindarajan, K Vahle, J T Jones, V Heinrichs, and P A Patten. 2001. “Advances in Directed Protein Evolution by Recursive Genetic Recombination: Applications to Therapeutic Proteins.” Current Opinion in Biotechnology 12 (4) (August): 361–70. http://www.ncbi.nlm.nih.gov/pubmed/11551464.
Labuza, T P, and W Breene. 1989. “Application of ‘Active Packaging’ Technologies for the Improvement of Shelf‐life and Nutritional Quality of Fresh and Extended Shelf‐life Foods.” Bibliotheca Nutritio Et Dieta (43) (January): 252–9. http://www.ncbi.nlm.nih.gov/pubmed/2658962.
Lee, N, Y Bessho, K Wei, J W Szostak, and H Suga. 2000. “Ribozyme‐catalyzed tRNA Aminoacylation.” Nature Structural Biology 7 (1) (January): 28–33. doi:10.1038/71225. http://www.ncbi.nlm.nih.gov/pubmed/10625423.
Lehmann, M, and M Wyss. 2001. “Engineering Proteins for Thermostability: The Use of Sequence Alignments Versus Rational Design and Directed Evolution.” Current Opinion in Biotechnology 12 (4) (August): 371–5. http://www.ncbi.nlm.nih.gov/pubmed/11551465.
Lehmann, Martin, Claudia Loch, Anke Middendorf, Dominik Studer, Søren F Lassen, Luis Pasamontes, Adolphus P G M van Loon, and Markus Wyss. 2002. “The Consensus Concept for Thermostability Engineering of Proteins: Further Proof of Concept.” Protein Engineering 15 (5) (May): 403–11. http://www.ncbi.nlm.nih.gov/pubmed/12034860.
Lenkei, R, J W Gratama, G Rothe, G Schmitz, J L D’hautcourt, A Arekrans, F Mandy, and G Marti. 1998. “Performance of Calibration Standards for Antigen Quantitation with Flow
V I I . Literature P a g e | 91
Cytometry.” Cytometry 33 (2) (October 1): 188–96. http://www.ncbi.nlm.nih.gov/pubmed/9773879.
Leskovac, V, S Trivić, G Wohlfahrt, J Kandrac, and D Pericin. 2005. “Glucose Oxidase from Aspergillus Niger: The Mechanism of Action with Molecular Oxygen, Quinones, and One‐electron Acceptors.” The International Journal of Biochemistry & Cell Biology 37 (4) (April): 731–50. doi:10.1016/j.biocel.2004.10.014. http://www.ncbi.nlm.nih.gov/pubmed/15694834.
Lin, Hening, and Virginia W Cornish. 2002. “Screening and Selection Methods for Large‐scale Analysis of Protein Function.” Angewandte Chemie (International Ed. in English) 41 (23) (December 2): 4402–25. doi:10.1002/1521‐3773(20021202)41:23<4402::AID‐ANIE4402>3.0.CO;2‐H. http://www.ncbi.nlm.nih.gov/pubmed/12458502.
Lindegren, C.C., and S.A Haddad. 1954. “Growth Rate of Individual Yeast Cells.” Genetica 27 (164): 45–53.
Locher, Christopher P, Nay Wei Soong, Robert G Whalen, and Juha Punnonen. 2004. “Development of Novel Vaccines Using DNA Shuffling and Screening Strategies.” Current Opinion in Molecular Therapeutics 6 (1) (March): 34–9. http://www.ncbi.nlm.nih.gov/pubmed/15011779.
Lu, C F, R C Montijn, J L Brown, F Klis, J Kurjan, H Bussey, and P N Lipke. 1995. “Glycosyl Phosphatidylinositol‐dependent Cross‐linking of Alpha‐agglutinin and Beta 1,6‐glucan in the Saccharomyces Cerevisiae Cell Wall.” The Journal of Cell Biology 128 (3) (February): 333–40. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2120349&tool=pmcentrez&rendertype=abstract.
M Rossi, Liane, Ashley D Quach, and Zeev Rosenzweig. 2004. “Glucose Oxidase‐magnetite Nanoparticle Bioconjugate for Glucose Sensing.” Analytical and Bioanalytical Chemistry 380 (4): 606–613. http://www.mendeley.com/research/glucose‐oxidasemagnetite‐nanoparticle‐bioconjugate‐glucose‐sensing/.
M145‐39. 2007. European Markets for Enzymes in Industrial Applications. Maki, Miranda, Kam Tin Leung, and Wensheng Qin. 2009. “The Prospects of Cellulase‐producing
Bacteria for the Bioconversion of Lignocellulosic Biomass.” International Journal of Biological Sciences 5 (5) (January): 500–16. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2726447&tool=pmcentrez&rendertype=abstract.
Mandels, M, R Andreotti, and C Roche. 1976. “Measurement of Saccharifying Cellulase.” Biotechnology and Bioengineering Symposium (6) (January): 21–33. http://www.ncbi.nlm.nih.gov/pubmed/1000065.
Mantyla, A., M. Paloheimo, and P. Suominen. 1998. “Industrial Mutants and Recombinant Strains of Trichoderma Reesei.” In Trichoderma & Gliocladium—Enzymes, Biological Control and Commercial Applications, ed. GF Harman and CP Kubicek, 291–309. London: Taylor & Francis.
Mastrobattista, Enrico, Valerie Taly, Estelle Chanudet, Patrick Treacy, Bernard T Kelly, and Andrew D Griffiths. 2005. “High‐throughput Screening of Enzyme Libraries: In Vitro Evolution of a Beta‐galactosidase by Fluorescence‐activated Sorting of Double Emulsions.”
V I I . Literature P a g e | 92
Chemistry & Biology 12 (12) (December): 1291–300. doi:10.1016/j.chembiol.2005.09.016. http://www.ncbi.nlm.nih.gov/pubmed/16356846.
Matsuura, Tomoaki, and Tetsuya Yomo. 2006. “In Vitro Evolution of Proteins.” Journal of Bioscience and Bioengineering 101 (6) (June): 449–56. doi:10.1263/jbb.101.449. http://www.ncbi.nlm.nih.gov/pubmed/16935245.
McMillan, James D, Edward W Jennings, Ali Mohagheghi, and Mildred Zuccarello. 2011. “Comparative Performance of Precommercial Cellulases Hydrolyzing Pretreated Corn Stover.” Biotechnology for Biofuels 4 (January): 29. doi:10.1186/1754‐6834‐4‐29. http://www.ncbi.nlm.nih.gov/pubmed/21899748.
Miller, G. L. 1959. “Use of Dinitrosalicylic Acid Reagent for Determination of Reducing Sugar.” Analytical Chemistry 31 (3) (March 1): 426–428. doi:10.1021/ac60147a030. http://dx.doi.org/10.1021/ac60147a030.
Mills, D R, R L Peterson, and S Spiegelman. 1967. “An Extracellular Darwinian Experiment with a Self‐duplicating Nucleic Acid Molecule.” Proceedings of the National Academy of Sciences of the United States of America 58 (1) (July): 217–24. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=335620&tool=pmcentrez&rendertype=abstract.
Muller, K. M., and K. M. Arndt. 2007. Protein Engineering Protocols (Methods in Molecular Biology). Ed. K. M. Muller and K. M. Arndt. Totowa: Humana Press. http://www.amazon.com/Protein‐Engineering‐Protocols‐Methods‐Molecular/dp/1617373389.
Murray, F R, D J Llewellyn, W J Peacock, and E S Dennis. 1997. “Isolation of the Glucose Oxidase Gene from Talaromyces Flavus and Characterisation of Its Role in the Biocontrol of Verticillium Dahliae.” Current Genetics 32 (5) (November): 367–75. http://www.ncbi.nlm.nih.gov/pubmed/9371889.
Musser, Richard O, Don F Cipollini, Sue M Hum‐Musser, Spencer A Williams, Judith K Brown, and Gary W Felton. 2005. “Evidence That the Caterpillar Salivary Enzyme Glucose Oxidase Provides Herbivore Offense in Solanaceous Plants.” Archives of Insect Biochemistry and Physiology 58 (2) (February): 128–37. doi:10.1002/arch.20039. http://www.ncbi.nlm.nih.gov/pubmed/15660363.
Nakazawa, Hikaru, Katsunori Okada, Ryota Kobayashi, Tetsuya Kubota, Tomoko Onodera, Nobuhiro Ochiai, Naoki Omata, Wataru Ogasawara, Hirofumi Okada, and Yasushi Morikawa. 2008. “Characterization of the Catalytic Domains of Trichoderma Reesei Endoglucanase I, II, and III, Expressed in Escherichia Coli.” Applied Microbiology and Biotechnology 81 (4) (December): 681–9. doi:10.1007/s00253‐008‐1667‐z. http://www.ncbi.nlm.nih.gov/pubmed/18762935.
Newman, Jeffrey D, and Anthony P F Turner. 2005. “Home Blood Glucose Biosensors: a Commercial Perspective.” Biosensors & Bioelectronics 20 (12) (June 15): 2435–53. doi:10.1016/j.bios.2004.11.012. http://www.ncbi.nlm.nih.gov/pubmed/15854818.
Olsen, M, B Iverson, and G Georgiou. 2000. “High‐throughput Screening of Enzyme Libraries.” Current Opinion in Biotechnology 11 (4) (August): 331–7. http://www.ncbi.nlm.nih.gov/pubmed/10975452.
V I I . Literature P a g e | 93
Olsen, M J, D Stephens, D Griffiths, P Daugherty, G Georgiou, and B L Iverson. 2000. “Function‐based Isolation of Novel Enzymes from a Large Library.” Nature Biotechnology 18 (10) (October): 1071–4. doi:10.1038/80267. http://www.ncbi.nlm.nih.gov/pubmed/11017045.
Patrick, Wayne M, Andrew E Firth, and Jonathan M Blackburn. 2003. “User‐friendly Algorithms for Estimating Completeness and Diversity in Randomized Protein‐encoding Libraries.” Protein Engineering 16 (6) (June): 451–7. http://www.ncbi.nlm.nih.gov/pubmed/12874379.
Pazur, J H, and K Kleppe. 1964. “The Oxidation of Glucose and Related Compounds by Glucose Oxidase from Aspergillus Niger.” Biochemistry 3 (April): 578–83. http://www.ncbi.nlm.nih.gov/pubmed/14188176.
Pizzariello, A, M Stred’ansky, and S Miertus. 2002. “A Glucose/hydrogen Peroxide Biofuel Cell That Uses Oxidase and Peroxidase as Catalysts by Composite Bulk‐modified Bioelectrodes Based on a Solid Binding Matrix.” Bioelectrochemistry Amsterdam Netherlands 56 (1‐2): 99–105. http://www.mendeley.com/research/glucosehydrogen‐peroxide‐biofuel‐cell‐uses‐oxidase‐peroxidase‐catalysts‐composite‐bulkmodified‐bioelectrodes‐based‐solid‐binding‐matrix/.
Powell, Keith A., Sandra W. Ramer, Stephen B. Del Cardayré, Willem P. C. Stemmer, Matthew B. Tobin, Pascal F. Longchamp, and Gjalt W. Huisman. 2001. “Directed Evolution and Biocatalysis.” Angewandte Chemie (International Ed. in English) 40 (21) (November 5): 3948–3959. http://www.ncbi.nlm.nih.gov/pubmed/12404465.
Prasetyo, Joni, Kazuya Naruse, Tatsuya Kato, Chuenchit Boonchird, Satoshi Harashima, and Enoch Y Park. 2011. “Bio‐conversion of Paper Sludge to Biofuel by Simultaneous Saccharification and Fermentation Using a Cellulase of Paper Sludge Origin and Thermotolerant Saccharomyces Cerevisiae TJ14.” Biotechnology for Biofuels 4 (1) (September 29): 35. doi:10.1186/1754‐6834‐4‐35. http://www.ncbi.nlm.nih.gov/pubmed/21958421.
Prodanovic, R., R. Ostafe, M. Blanusa, and U. Schwaneberg. 2012. “ViPer Assay for Flow Cytometry Sorting of Glucose Oxidase Gene Libraries in Double Emulsions.” Analytical and Bioanalytical Chemistry 404 (5): 1439–1474.
Prodanovic, Radivoje, Raluca Ostafe, Andreea Scacioc, and Ulrich Schwaneberg. 2011. “Ultrahigh‐throughput Screening System for Directed Glucose Oxidase Evolution in Yeast Cells.” Combinatorial Chemistry & High Throughput Screening 14 (1) (January): 55–60. http://www.ncbi.nlm.nih.gov/pubmed/20958255.
Promega. 2008. “TNT® T7 Insect Cell Extract Protein Expression System” Part# TM30. Qiagen. 2006. “June 2006 EasyXpress TM Insect Cell Protein Synthesis Handbook.” Quong, D, R J Neufeld, G Skjåk‐Braek, and D Poncelet. 1998. “External Versus Internal Source of
Calcium During the Gelation of Alginate Beads for DNA Encapsulation.” Biotechnology and Bioengineering 57 (4) (February 20): 438–46. http://www.ncbi.nlm.nih.gov/pubmed/10099220.
Rasiah, I.A., K.H. Sutton, F.L. Low, H.‐M. Lin, and J.A. Gerrard. 2005. “Crosslinking of Wheat Dough Proteins by Glucose Oxidase and the Resulting Effects on Bread and Croissants.” Food Chemistry 89 (3) (February): 325–332. doi:10.1016/j.foodchem.2004.02.052. http://dx.doi.org/10.1016/j.foodchem.2004.02.052.
V I I . Literature P a g e | 94
Reetz, M. T. 2008. “Asymmetric Organic Synthesis with Enzymes.” In Asymmetric Organic Synthesis with Enzymes, ed. Vicente Gotor, Ignacio Alfonso, and Eduardo Garca‐Urdiales. Weinheim, Germany: Wiley‐VCH Verlag GmbH & Co. KGaA. doi:10.1002/9783527622481. http://doi.wiley.com/10.1002/9783527622481.
Reiter, Birgit, Alexander Faschinger, Anton Glieder, and Helmut Schwab. 2007. “Random Strand Transfer Recombination (RSTR) for Homology‐independent Nucleic Acid Recombination.” Journal of Biotechnology 129 (1) (March 30): 39–49. doi:10.1016/j.jbiotec.2006.10.002. http://www.ncbi.nlm.nih.gov/pubmed/17175057.
Roberts, R W. 1999. “Totally in Vitro Protein Selection Using mRNA‐protein Fusions and Ribosome Display.” Current Opinion in Chemical Biology 3 (3) (June): 268–73. doi:10.1016/S1367‐5931(99)80042‐8. http://www.ncbi.nlm.nih.gov/pubmed/10359713.
Rogers, M J, and K G Brandt. 1971. “Multiple Inhibition Analysis of Aspergillus Niger Glucose Oxidase by D‐glucal and Halide Ions.” Biochemistry 10 (25) (December 7): 4636–41. http://www.ncbi.nlm.nih.gov/pubmed/5140182.
Ryu, Dewey D.Y., and Mary Mandels. 1980. “Cellulases: Biosynthesis and Applications.” Enzyme and Microbial Technology. doi:10.1016/0141‐0229(80)90063‐0. http://linkinghub.elsevier.com/retrieve/pii/0141022980900630.
Sabra, W, A P Zeng, and W D Deckwer. 2001. “Bacterial Alginate: Physiology, Product Quality and Process Aspects.” Applied Microbiology and Biotechnology 56 (3‐4) (August): 315–25. http://www.ncbi.nlm.nih.gov/pubmed/11548998.
Sakai, Shinji, and Koei Kawakami. 2007. “Synthesis and Characterization of Both Ionically and Enzymatically Cross‐linkable Alginate.” Acta Biomaterialia 3 (4) (July): 495–501. doi:10.1016/j.actbio.2006.12.002. http://www.ncbi.nlm.nih.gov/pubmed/17275429.
Saloheimo, Markku, Marja Paloheimo, Satu Hakola, Jaakko Pere, Barbara Swanson, Eini Nyyssönen, Amit Bhatia, Michael Ward, and Merja Penttilä. 2002. “Swollenin, a Trichoderma Reesei Protein with Sequence Similarity to the Plant Expansins, Exhibits Disruption Activity on Cellulosic Materials.” European Journal of Biochemistry / FEBS 269 (17) (September): 4202–11. http://www.ncbi.nlm.nih.gov/pubmed/12199698.
Setsukinai, Ken‐ichi, Yasuteru Urano, Kazuya Kikuchi, Tsunehiko Higuchi, and Tetsuo Nagano. 2000. “Fluorescence Switching by O‐dearylation of 7‐aryloxycoumarins. Development of Novel Fluorescence Probes to Detect Reactive Oxygen Species with High Selectivity.” Journal of the Chemical Society, Perkin Transactions 2 (12) (November 7): 2453–2457. doi:10.1039/b006449l. http://pubs.rsc.org/en/content/articlehtml/2000/p2/b006449l.
Shanmugam, S. 2009. Enzyme Technology. I. K. International Pvt Ltd. http://books.google.com/books?id=KrGn_obcy6QC&pgis=1.
Simoons, F.J. 1971. “The Antiquity of Dairying in Asia and Africa.” Geographical Review 61 (3). Singh, Madhu V, and P Anthony Weil. 2002. “A Method for Plasmid Purification Directly from
Yeast.” Analytical Biochemistry 307 (1) (August 1): 13–7. http://www.ncbi.nlm.nih.gov/pubmed/12137773.
Smidsrod, O, and G Skjak‐Braek. 1990. “Alginate as Immobilization Matrix for Cells.” Trends in Biotechnology 8 (3) (March): 71–8. http://www.ncbi.nlm.nih.gov/pubmed/1366500.
Steensma, David P, Michael Timm, and Thomas E Witzig. 2003. “Flow Cytometric Methods for Detection and Quantification of Apoptosis.” Methods in Molecular Medicine 85 (January):
V I I . Literature P a g e | 95
323–32. doi:10.1385/1‐59259‐380‐1:323. http://www.ncbi.nlm.nih.gov/pubmed/12710217.
Stemmer, Willem, and Brett Holland. 2003. “Survival of the Fittest Molecule.” American Scientist 91 (6): 526. doi:10.1511/2003.6.526. http://www.mendeley.com/research/survival‐fittest‐molecule/.
Sun, L., I. P. Petrounia, M. Yagasaki, G. Bandara, and F. H. Arnold. 2001. “Expression and Stabilization of Galactose Oxidase in Escherichia Coli by Directed Evolution.” Protein Engineering Design and Selection 14 (9) (September 1): 699–704. doi:10.1093/protein/14.9.699. http://peds.oxfordjournals.org/cgi/content/abstract/14/9/699.
Swoboda, B E. 1969. “The Mechanism of Binding of Flavin‐adenine Dinucleotide to the Apoenzyme of Glucose Oxidase and Evidence for the Involvement of Multiple Bonds.” Biochimica Et Biophysica Acta 175 (2) (March): 380–7. http://www.ncbi.nlm.nih.gov/pubmed/4305100.
Takegawa, K, K Fujiwara, S Iwahara, K Yamamoto, and T Tochikura. 1989. “Effect of Deglycosylation of N‐linked Sugar Chains on Glucose Oxidase from Aspergillus Niger.” Biochemistry and Cell Biology = Biochimie Et Biologie Cellulaire 67 (8) (August): 460–4. http://www.ncbi.nlm.nih.gov/pubmed/2511903.
Tang, Lian X., Frederick J. Rowell, and Robert H. Cumming. 1995. “Ultra‐rapid Fluorescence Assay for Cellulase Using a Substrate‐immobilized Mini‐bioreactor.” Analytical Proceedings Including Analytical Communications 32 (12) (January 1): 519. doi:10.1039/ai9953200519. http://pubs.rsc.org/en/content/articlehtml/1995/ai/ai9953200519.
Tawfik, D S, and A D Griffiths. 1998. “Man‐made Cell‐like Compartments for Molecular Evolution.” Nature Biotechnology 16 (7) (July): 652–6. doi:10.1038/nbt0798‐652. http://www.ncbi.nlm.nih.gov/pubmed/9661199.
Teather, R.M, and P.J. Wood. 1982. “Use of Congo Red Polysaccharide Interactions Complex Formation Between Congo Red and Polysaccharide in Detection and Assay of Polysaccharide Hydrolases.” Methods in Enzymology 160: 59–74.
Tonnesen, Hanne Hjorth, and Jan Karlsen. 2002. “Alginate in Drug Delivery Systems.” Drug Development and Industrial Pharmacy 28 (6) (July): 621–30. doi:10.1081/DDC‐120003853. http://www.ncbi.nlm.nih.gov/pubmed/12149954.
Trinder, P. 1969. “Determination of Blood Glucose Using 4‐amino Phenazone as Oxygen Acceptor.” Journal of Clinical Pathology 22 (2) (March): 246. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=474047&tool=pmcentrez&rendertype=abstract.
Tsuge, H, O Natsuaki, and K Ohashi. 1975. “Purification, Properties, and Molecular Features of Glucose Oxidase from Aspergillus Niger.” Journal of Biochemistry 78 (4) (October): 835–43. http://www.ncbi.nlm.nih.gov/pubmed/1213991.
Van de Vyver, Stijn, Jan Geboers, Pierre A. Jacobs, and Bert F. Sels. 2011. “Recent Advances in the Catalytic Conversion of Cellulose.” ChemCatChem 3 (1) (January 10): 82–94. doi:10.1002/cctc.201000302. http://doi.wiley.com/10.1002/cctc.201000302.
Wang, Q, A M Buckle, N W Foster, C M Johnson, and A R Fersht. 1999. “Design of Highly Stable Functional GroEL Minichaperones.” Protein Science: a Publication of the Protein Society 8 (10) (October): 2186–93. doi:10.1110/ps.8.10.2186.
V I I . Literature P a g e | 96
http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2144126&tool=pmcentrez&rendertype=abstract.
White, J W, M H Subers, and A I Schepartz. 1963. “The Identification of Inhibine, the Antibacterial Factor in Honey, as Hydrogen Peroxide and Its Origin in a Honey Glucose‐oxidase System.” Biochimica Et Biophysica Acta 73 (May 7): 57–70. http://www.ncbi.nlm.nih.gov/pubmed/14000328.
Witt, S, M Singh, and H M Kalisz. 1998. “Structural and Kinetic Properties of Nonglycosylated Recombinant Penicillium Amagasakiense Glucose Oxidase Expressed in Escherichia Coli.” Applied and Environmental Microbiology 64 (4) (April): 1405–11. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=106162&tool=pmcentrez&rendertype=abstract.
Witt, S, G Wohlfahrt, D Schomburg, H J Hecht, and H M Kalisz. 2000. “Conserved Arginine‐516 of Penicillium Amagasakiense Glucose Oxidase Is Essential for the Efficient Binding of beta‐D‐glucose.” The Biochemical Journal 347 (Pt 2) (April 15): 553–9. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=1220989&tool=pmcentrez&rendertype=abstract.
Witteveen, C F, M Veenhuis, and J Visser. 1992. “Localization of Glucose Oxidase and Catalase Activities in Aspergillus Niger.” Applied and Environmental Microbiology 58 (4) (April): 1190–4. http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=195573&tool=pmcentrez&rendertype=abstract.
Wong, Chun Ming, Kwun Hei Wong, and Xiao Dong Chen. 2008. “Glucose Oxidase: Natural Occurrence, Function, Properties and Industrial Applications.” Applied Microbiology and Biotechnology 78 (6) (April): 927–38. doi:10.1007/s00253‐008‐1407‐4. http://www.ncbi.nlm.nih.gov/pubmed/18330562.
Wong, T S, N Wu, D Roccatano, M Zacharias, and U Schwaneberg. 2005. “Sensitive Assay for Laboratory Evolution of Hydroxylases Toward Aromatic and Heterocyclic Compounds.” Journal of Biomolecular Screening 10 (3) (April): 246–52. doi:10.1177/1087057104273336. http://www.ncbi.nlm.nih.gov/pubmed/15809320.
Wong, Tuck Seng, Danilo Roccatano, and Ulrich Schwaneberg. 2007. “Are Transversion Mutations Better? A Mutagenesis Assistant Program Analysis on P450 BM‐3 Heme Domain.” Biotechnology Journal 2 (1) (January): 133–42. doi:10.1002/biot.200600201. http://www.ncbi.nlm.nih.gov/pubmed/17183510.
Wong, Tuck Seng, Daria Zhurina, and Ulrich Schwaneberg. 2006. “The Diversity Challenge in Directed Protein Evolution.” Combinatorial Chemistry & High Throughput Screening 9 (4) (May): 271–88. http://www.ncbi.nlm.nih.gov/pubmed/16724918.
Xiao, Zhizhuang, Reginald Storms, and Adrian Tsang. 2004. “Microplate‐based Filter Paper Assay to Measure Total Cellulase Activity.” Biotechnology and Bioengineering 88 (7) (December 30): 832–7. doi:10.1002/bit.20286. http://www.ncbi.nlm.nih.gov/pubmed/15459905.
Zhang, P. Y‐H, Michael E Himmel, Jonathan R Mielenz, and Y‐H Percival Zhang. 2006. “Outlook for Cellulase Improvement: Screening and Selection Strategies.” Biotechnology Advances 24 (5): 452–81. doi:10.1016/j.biotechadv.2006.03.003. http://www.ncbi.nlm.nih.gov/pubmed/16690241.
V I I . Literature P a g e | 97
Zhang, Y H Percival, Jiong Hong, and Xinhao Ye. 2009. “Cellulase Assays.” Methods in Molecular Biology (Clifton, N.J.) 581 (January): 213–31. doi:10.1007/978‐1‐60761‐214‐8_14. http://www.ncbi.nlm.nih.gov/pubmed/19768625.
Zhu, Ziwei, Carmen Momeu, Maxim Zakhartsev, and Ulrich Schwaneberg. 2006. “Making Glucose Oxidase Fit for Biofuel Cell Applications by Directed Protein Evolution.” Biosensors & Bioelectronics 21 (11) (May 15): 2046–51. doi:10.1016/j.bios.2005.11.018. http://dx.doi.org/10.1016/j.bios.2005.11.018.