novel target for peptide-based imaging and treatment of ......progression. even with the most...

Models and Technologies

Novel Target for Peptide-Based Imaging and Treatment ofBrain Tumors

Maija Hyv€onen1, Juulia Enb€ack1, Tuulia Huhtala5, Johanna Lammi1, Harri Sihto2, Janne Weisell5,Heikki Joensuu2,3, Katri Rosenthal-Aizman6, Samir El-Andaloussi6, Ulo Langel6, Ale N€arv€anen5,Gabriele Bergers7, and Pirjo Laakkonen1,4

AbstractMalignant gliomas are associated with high mortality due to infiltrative growth, recurrence, and malignant

progression. Even with the most efficient therapy combinations, median survival of the glioblastoma

multiforme (grade 4) patients is less than 15 months. Therefore, new treatment approaches are urgently

needed. We describe here identification of a novel homing peptide that recognizes tumor vessels and

invasive tumor satellites in glioblastomas. We demonstrate successful brain tumor imaging using radi-

olabeled peptide in whole-body SPECT/CT imaging. Peptide-targeted delivery of chemotherapeutics

prolonged the lifespan of mice bearing invasive brain tumors and significantly reduced the number of

tumor satellites compared with the free drug. Moreover, we identified mammary-derived growth inhibitor

(MDGI/H-FABP/FABP3) as the interacting partner for our peptide on brain tumor tissue. MDGI was

expressed in human brain tumor specimens in a grade-dependent manner and its expression positively

correlated with the histologic grade of the tumor, suggesting MDGI as a novel marker for malignant gliomas.

Mol Cancer Ther; 13(4); 996–1007. �2014 AACR.

IntroductionGlioblastoma multiforme, the most aggressive and

most common primary brain tumor in adults, comprisesapproximately 60% to 70% of all gliomas. Tumors ofglial cell origin are graded on a World Health Organi-zation scale as grade 1 to 4. Only grade 1 tumors areconsidered benign; they grow slowly with a cleartumor–brain interphase enabling the surgical removalof the tumor. Importantly, nearly all low-grade tumorseventually progress to high-grade malignancies (1).Owing to their infiltrative growth, gliomas are associ-ated with high morbidity and mortality (2). Even withthe most efficient therapy combinations, median sur-vival of glioblastoma multiforme patients (grade 4) isless than 15 months and most patients die within 2 yearswith 5-year survival less than 3% (3, 4). In addition,

malignant gliomas almost always recur. The recurrentgliomas are very aggressive with one-year survival rateless than 25%.

Therefore, new therapeutic approaches are urgentlyneeded. One of the main challenges is to identify meansbywhich block invasion and recurrence by destroying theinvasive tumor satellites that have left the main tumormass and that cannot be surgically removed. One attrac-tive approach would be to develop drugs/inhibitors thatspecifically target these satellite tumor cells and the bloodvessels they co-opt for growth.

The distinct molecular signatures in the normal andneoplastic tissueshavebeenefficiently exploitedusing thein vivo phage display screening to identify tissue-specificmarkers and to survey disease-specific differences (5).Using this technology, several peptides that specificallytarget the blood vessels in various normal tissues and intumors (6–8) have been isolated. Tumor lymphatic vesselshave also been targeted in this manner (9–11). Such pep-tides have been used for targeted delivery of drugs,oligonucleotides, imaging agents, liposomes, nanoparti-cles, and viruses (12, 13).

To identify peptides selectively targeting invasive satel-lites of malignant gliomas, we performed an in vivo phagedisplay screen using a tumor model that reproduces theinfiltrating glioblastomaphenotype (14, 15). Because thesetumor cells lack HIF-1, they are incapable of inducingtumor angiogenesis and rely on co-opted normal vesselsfor growth. We describe a peptide named "CooP" thatspecifically homes to invasive tumor satellites and thevessels in these tumors and other glioblastomas.

Authors' Affiliations: 1Research Programs Unit, Translational CancerBiology and Institute of Biomedicine and 2Laboratory of Molecular Oncol-ogy, Biomedicum Helsinki, University of Helsinki; 3Department of Oncol-ogy, Helsinki University Central Hospital; 4Finnish Cancer Institute, Hel-sinki; 5A.I. Virtanen Institute for Molecular Medicine, University of EasternFinland, Kuopio, Finland; 6Department of Neurochemistry, StockholmUniversity, Stockholm, Sweden; and 7UCSF Helen Diller Family CancerCenter, San Francisco, California

Note: Supplementary data for this article are available at Molecular CancerTherapeutics Online (http://mct.aacrjournals.org/).

Corresponding Author: Pirjo Laakkonen, University of Helsinki, Haartma-ninkatu 8, 00029, Helsinki, Finland. Phone: 358-9191-25537; Fax: 358-9191-25510; E-mail: [email protected]

doi: 10.1158/1535-7163.MCT-13-0684

�2014 American Association for Cancer Research.

MolecularCancer

Therapeutics

Mol Cancer Ther; 13(4) April 2014996

on August 13, 2021. © 2014 American Association for Cancer Research. mct.aacrjournals.org Downloaded from

Published OnlineFirst February 3, 2014; DOI: 10.1158/1535-7163.MCT-13-0684

http://mct.aacrjournals.org/

Materials and MethodsCellsHif-1a–deficient (HIFko) and VEGF-overexpressing

mouse astrocytes were propagated as described (15).U87MG human and BT4C rat glioma cells (gift fromSeppo Yl€a-Herttuala, University of Eastern Finland,Kuopio, Finland) were maintained in Dulbecco’s mod-ified Eagle medium (DMEM), 10% fetal calf serum(FCS), 2 mmol/L L-glutamine, 100 U/mL penicillin, and100 mg/mL streptomycin. MDA-MB-231 cells (gift fromJorma Keskioja, University of Helsinki, Helsinki, Fin-land) were cultured in the RPMI-1640, 10% FCS, 2mmol/L L-glutamine, 100 U/mL penicillin, 100 mg/mLstreptomycin, and U2-OS (gift from Marikki Laiho,University of Helsinki, Helsinki, Finland) cells inDMEM and 15% FCS. Human embryonic kidney cells(293FT, Functional Genomics Unit, University of Hel-sinki, Helsinki, Finland) used for lentivirus productionwere maintained in high-glucose (4.5 g/L) DMEM. Celllines were obtained during years 2006–2009. No authen-tication has been done by the authors.

Generation of stable cell linesTo establish U87MG cells stably expressing MDGI-

GFP fusion protein, MDGI gene was cloned to a lenti-viral expression vector (pBOB\cag\GFP) containingGFP (gift from Seppo Yl€a-Herttuala, University of East-ern Finland, Kuopio, Finland). Lipofectamine 2000(Invitrogen) was used to complex the GFP/MDGI orGFP–containing expression vector and lentiviral pack-aging plasmids pLP1, pLP2 and pLP-VSVG (Invitrogen)according to manufacturer’s instructions. Virus-con-taining supernatants of transfected 293FT cells werecollected at 72 hours and concentrated at 50,000 � gwith Optima L-80 XP ultracentrifuge (SW28 swingingbucket rotor, Beckmann Coulter). U87MG cells weretransduced with the concentrated MDGI-GFP or GFPviruses. To increase transgene-positive cells, transducedcells were sorted with a BD LSR II fluorescence-activat-ed cell sorter (BD Biosciences).

In vivo phage displayWe used an NNK-encoded CX7C peptide library (gift

from Erkki Ruoslahti, Sanford-Burnham Institute, LaJolla, CA) on the T7Select415-1-phage (Novagen). Phageselections were performed as previously described (16).For the first ex vivo round, tumor-derived cell suspensionwas incubated overnight at 4�C with 5 � 109 plaque-forming units (PFU) of the library. Bound phage wasrescued and amplified in E. coli (BLT5615, Novagen) andused for the second round of ex vivo selection. Ex vivoenriched phage pool was injected into the tail vein ofintracranial HIFko tumor-bearing mice, and allowed tocirculate for 15 minutes. Brain, including tumor, wasexcised, and the recovered and amplified phagewas usedfor the next rounds of in vivo panning. In each round,nonrecombinant control phage was injected to a separatemouse to assess the background.

Peptide synthesisCooP (NH2-ACGLSGLGVA-CONH2) and its control

peptide (NH2-ACVAALNADG-CONH2) were synthe-sized using an Apex 396 DCmultiple peptide synthesizer(Advanced ChemTech) with Fmoc strategy and O-Benzo-triazole-N,N,N 0 ,N0 -tetramethyl-uronium-hexafluoro-phosphate (HBTU, GLS Biochem) and N,N-diisopropy-lethylamine (DIPEA, Fluka) as coupling reagents, andrink resin as solid support (Novabiochem). DTPA wasconjugated directly to the alpha-amino group of the pep-tide. For radiolabeling, DTPA-conjugated peptides (20 mgper animal) were mixed with 0.2 mol/L NaAc (pH 5)followedby additionof 5MBqof 111Indiumchloride (111In,Mallinckrodt) per animal. Reaction mixture was incubat-ed for an hour at room temperature and radiochemicalpurity was measured using an instant thin-layer chroma-tography (ITLC-SG, Pall Corporation). Radiochemicalpurity of the peptides was 99% to 100%.

Isothiocynate-activated fluorescein (FITC) was conju-gated to peptides on resin via an additional N-terminallysine. After removal of Fmoc group using 20% piperdinein dimethyl formamide (DMF)-free N-terminus was acet-ylated with 20% acetic anhydride in DMF. Acid labilemethyltrityl protection group of the additional lysine wasremovedwith 2% trifluoroacetic acid inDCM.FITC (2mg)was added into reaction mixture following addition of 15mL triethanolamine. Conjugated peptides were cleavedfrom the resin using a mixture of 95% trifluoroacetic acid(TFA), 3% ethanedithiol (EDT), 1% triisopropylsilane, and1% H20. Peptides were purified with reverse-phase high-performance liquid chromatography (RP-HPLC) usingC18 reverse phase column (xTERRA, 250 � 10 mm,Waters) and an acetonitrile gradient (ACN, 0%–95%, 45minutes). Purity of the peptides was determined by ananalytical HPLC.

CooP-CPP-Cbl (NH2-CGLSGLGVAGGMVRRFLVTL-RIRRACGPPRVRVK(eCbl)-CONH2) peptide was synthe-sized on an ABI433A peptide synthesizer (Applied Bio-systems) by t-Boc chemistry using 4-methylbenzhydry-lamine-polystyrene resin to generate C-terminally ami-dated peptides. First amino acid Boc-Lys(eFmoc)OH wascoupled manually. All subsequent couplings used HOAt(1-hydroxy-7-azabenzotriazole) and N,N0-dicyclohexyl-carbodiimide to couple standard Boc protected aminoacids. Fully protected peptidyl-resin was removed fromsynthesizer and treated manually to generate freee-amine of C-terminal lysine using 20 vol% piperidinein DMF for 20 minutes followed by additional 40 minutesin 20% piperidine in 1 vol% Triton X-100 DCM/DMF/NMP 1:1:1 v/v/v. Chlorambucil (10 equivalent) wascoupled by 2-(1H-benzotriazol-1-yl)-1,1,3,3-tetramethy-luronium tetrafluoroborate, 1-hydroxybenzotriazole andDIPEA in DMF/NMP (1:1, v/v) for 1 hour at roomtemperature. The freeze-dried peptidyl resin was treatedwith anhydrous hydrogen fluoride containing 10 vol%scavengers (p-cresol and p-thiocresol, 1:1) at 0�C for 1hour. After evaporation of hydrogen fluoride, residuewas washed with diethyl ether and extracted with 10

Targeting Malignant Brain Tumors

www.aacrjournals.org Mol Cancer Ther; 13(4) April 2014 997




vol% acetic acid, and lyophilized. Peptide was purifiedby RP-HPLC (Discovery C-18 column, Supelco, Sigma-Aldrich) using a linear gradient of ACN/water contain-ing 0.1 vol% TFA (20%–60% ACN, 50 minutes). Identityof the purified product was verified by analytical RP-HPLC and by Perkin Elmer prOTOF 2000 matrix-assistedlaser desorption ionization time-of-flight mass-spectrom-eter. Mass spectra were acquired in positive ion reflectormode using a-cyano-4-hydroxycinnamic acid as matrix(10 mg/mL, 3/7 acetonitrile:water, 0.1% TFA).

SPECT/CT imagingFor a combined small-animal SPECT/CT (single pho-

ton emission computed tomography-computed tomogra-phy; Gamma Medica) imaging, radiolabeled peptide (20mg in 100 mL) was injected into the tail vein of intracranialU87MG tumor-bearing mice. Medium energy parallelhole collimators were used with a 171/245/416 keV ener-gy window for 111In. Planar two-dimensional (2D) imag-ing was performed at earliest possible time point afterinjection with 60 s acquisition time/image up to 30 min-utes. For three-dimensional (3D) imaging total of 64 pro-jections, (120 s/projection) were used. Matrix size was80� 80 in 125� 125mm2 field of view (FOV). CT imagingwas performedwith same coordinates as SPECTwith 512projections, 1024� 1024 projection matrix size, and 60 kVvoltage.

Planar SPECT images were combined to the X-rayimage using Matlab software (The MathWorks). Three-dimensional SPECT reconstruction was carried out usingthe LumGem software (Gamma Medica) with Butter-worth six-order low-pass post-filtering with 0.3 cut-offfrequency to obtain 3D images with dimensions of 512 �512 � 512 matrix size in 87 � 87 � 87 mm3 FOV. Both CTand SPECT images were interpolated into final 256 � 256� 256 size by using commercial software (Gamma Med-ica). Images were combined using IDL software (ITTVisual Information Solutions) and monitored with theAmira software (Mercury).

Analysis of radioactivity in organsAfter the last imaging session, animals were sacrificed

using CO2 and tissue samples were collected, weighted,and radioactivity was measured in an automated gammacounter. Obtained counts were corrected for backgroundradiation and physical decay. Tissue radioactivity wasexpressed as percentage of the injected dose per gram oftissue (% ID/g � SD).

Detection of peptides and antibodies in tissuesIntravenously injected fluorescein-labeled peptides

(100 nmoles) or antibodies (20 mg/animal) were allowedto circulate for 1 hour. To detect in vivo distribution of thefluorescein-labeled peptides, frozen sections (5–10 mm)were stained with rabbit anti-FITC antibodies. To visual-ize blood vessels, MDGI, and tumor cells, sections werestained with rat anti-mouse PECAM-1/CD31 (BD Phar-mingen), goat or rabbit anti-MDGI (Santa Cruz Biotech-

nology) or rabbit anti-SV40 large T antigen [gift fromDr. Hanahan, Swiss Institute for Experimental CancerResearch (ISREC), Lausanne, Switzerland] antibodies fol-lowed by Alexa-594 or Alexa-488 conjugated secondaryantibodies (Molecular Probes/Invitrogen). Nuclei werevisualized with 40, 6–diamidino–2–phenylindole (DAPI;Vector Laboratories). Sections were examined under aninverted fluorescence microscope (Zeiss Axioplan).

Yeast-2-hybrid screenSense- and antisense oligonucleotides (AATTCTGCG-

GACTGAGCGGGTTAGGCGTTGCTG, GATCCAGCA-ACGCCTAACCCGCTCAGTCCGCAG) encoding theCooP sequence were cloned into a pGBKT7 bait vector(Clontech Laboratories). Mouse embryonic (E12.5) libraryin the Pad-GAL4-2.1 prey vector was transformed intoyeast cells expressing the baits. Clones producing b-galac-tosidase were isolated and the prey cDNAs weresequenced. Screen was performed according to the Fieldsmethod (17) at the Yeast-Two-Hybrid Core Facility (Uni-versity of Helsinki, Helsinki, Finland).

Cloning of the MDGI from brain tumor cDNARNA was isolated from brains of intracranial HIFko

tumor-bearingmice using the Rneasy kit (QIAGEN). RNAfrom two brains (1 mg)was used for cDNAsynthesis by theQuantiTech RT Kit (QIAGEN). Full-length MGDI wascloned to the pcDNA3-9E10 plasmid using the followingprimers: forward: GGAATTCGCGGACGCCTTTGTCGG-TACCTGGAAG; reverse: CCTCGAGTCACGCCTCCTT-CTCATAAGTCCGAGTGCTC.

Tumor xenografts and treatment with targeted CblAthymic female Balb/c nu-nu mice (4–6 weeks) were

used in tumor experiments. Establishment of intracranialtumors has been described (15). Mice were used forexperiments 10 to 21 days after tumor cell inoculation.To establish subcutaneous tumors, 5 � 106 cells weregrafted on the abdominal side of the mice.

For treatment study, intracranial HIFko tumors wereestablished and treatmentwas started5days after implan-tation. Mice were randomly divided in groups (4 mice/group) and injected intravenously with PBS, CooP-CPP-Cbl (5 mg/kg), or free Cbl (same molar amount as CooP-CPP-Cbl) every second day. Weight of the animals wasmonitored during the study. Animal studies were con-ducted according to guidelines of the Provincial Govern-ment of Southern Finland and protocol was approved bythe Experimental Animal Committee.

Murine and human tumor materialFormalin-fixed paraffin-embedded tumor samples

obtained from patients diagnosed with brain tumor wereretrieved from archives of the Department of Pathology,Helsinki University Central Hospital, Finland. Study wasapproved by an Ethics Committee of Helsinki UniversityCentral Hospital, Finland. A permission to use tumortissue for the study was obtained from the Ministry of

Hyv€onen et al.

Mol Cancer Ther; 13(4) April 2014 Molecular Cancer Therapeutics998




Social Affairs and Health, Finland. In addition, formalin-fixed paraffin-embedded sections of murine xenograftsderived from EGFRvIII transformed, INK4/Arf �/� neu-ral stem cells (NSCG) and fromaglioblastomamultiformepatient–derived tumor (GBM43; ref. 18) were used for theimmunohistologic studies. Antigen retrieval was per-formed by heating the samples in 10 mmol/L sodiumcitrate buffer (pH 6.0) using a microwave oven (780W 5minutes, 380W 10 minutes). MDGI was detected from thesections using TSA Indirect Kit (PerkinElmer).To better visualize the localization of MDGI in gliomas,

double immunofluorescence staining of human gliomasamples using antibodies against MDGI and PECAM-1(endothelial marker) was performed. After deparaffiniza-tion, sections were blocked with blocking solution (5%BSA, 0.1%Triton X-100 in PBS) and incubated with theappropriate primary antibodies at þ4�C overnight. Afterthe PBS washes, sections were incubated for 1 hour atroom temperature with fluorescently labeled secondaryantibodies (Molecular Probes/Invitrogen) and Hoechst(1 mg/mL) in PBS to visualize the nuclei.

ResultsIdentification of peptides homing to malignant braintumorsBecause of the infiltrative growth, glioblastomas contain

invasive tumor satellites that cannotbe surgically removedand that contribute to recurrence of the tumors. We per-formed a combined ex vivo/in vivo phage display screen(16) to isolate peptides capable of homing to these inva-sive satellites that are nurtured by co-opted vasculature.As a model, we used the hypoxia-inducible factor–defi-cient transformedmurine astrocytes (HIFko) that are inca-pable of angiogenesis and adapt to that inability by co-optingexistingnormal vessels,which leads toperivascularinvasion and growth as invasive tumor satellites (14, 15).Two rounds of ex vivo selection using the CX7C phage-displayed peptide library on cell suspensions from intra-cranialHIFko tumors yielded a 5-fold enrichment of phagebinding to the tumor-derived cells compared with thecontrol (Fig. 1A).We subjected this preselected phage poolto in vivo selection to screen for peptides capable of homingto the tumors after intravenous injection. We observedan increase in tumor homing over the control phage from95-fold in the first in vivo round to approximately 200-foldin the third round (Fig. 1A). After the third in vivo round,we sequenced approximately 60 randomly selected indi-vidual clones and searched for enriched peptide sequen-ces. On the basis of the sequence data, we chose five clonestovalidate their ability tobindandhome to theHIFkobraintumors. Phage 11, 34, 36, 50, and 59 all showed high, up to1,700-fold, ex vivo binding to tumor cell suspensions overthe control phage. However, phage 11, 50, and 59 showedeven higher affinity for the normal brain (Fig. 1B) suggest-ing that thesephage/peptideswere rather brain thanbraintumor specific. The brain tumor to normal brain ratio wasabout 13 for phage 34 and over 80 for phage 36 demon-strating preference for brain tumor tissue (Fig. 1B). Pep-

tides encoded by phage 34 (CSESGLGVA) and 36(CGLSGLGVA) shared 7 of 9 amino acids.Whenwe testedthe in vivo homing of these phage to the intracranial HIFkotumors, phage 36, hereafter referred to as CooP, showednearly 20-fold homing to the tumor, while only negligibleaccumulation in other organs, including liver, kidney,lung, and normal brain, was detected (Fig. 1C). Phage34 showed only modest tumor homing, about 3-fold overthe control phage (data not shown).

To confirm that the homing was mediated by the CooPpeptide and to study the tissue localization of the peptidemore precisely, we conjugated the synthetic peptide toFITC. We injected the FITC-CooP intravenously to micebearing intracranial HIFko tumors and allowed it tocirculate for 60 minutes. The peptide accumulated in thetumor satellites in the brain (Fig. 1D and E). Identificationof the tumor satellites was confirmed by staining adjacenttissue sections for SV40 large T antigen to visualize thetumor cells (Fig. 1F). We detected no CooP peptide in thesurrounding histologically normal brain (Fig. 1E) or othertissues examined (data not shown). Somepeptide-derivedfluorescence was observed in the kidneys and the choroidplexus, a cerebrospinal fluid secreting epithelial structurelocated in brain ventricles. Since a control peptide (E3;ref. 19), and the previously reported tumor-homing pep-tides F3 (19), and LyP-1 (10) used as additional controls,also produced a similar fluorescent pattern at these sites(data not shown), the signals in these organs were mostlikely produced by nonspecific peptide or fluoresceinuptake related to excretion of the peptides.

To assess tumor-type specificity of the CooP peptide,we tested its homing to other types of brain tumors.Intravenously injectedCooP peptide homed to intracranialU87MG human glioma (Fig. 1G) and rat BT4C gliomaxenografts (Fig. 1H). No CooP peptide was detected inorthotopic MDA-MB-231 human breast cancer xeno-grafts (Fig. 1I). Interestingly, CooP peptide did not hometo subcutaneous tumors established from the HIFko cells(Fig. 1J). CooP peptide also did not home to tumors deriv-ed from VEGF-expressing HIFko cells (15), which contain-ed amassive network of angiogenic blood vessels (Fig. 1K).

Mammary-derived growth inhibitor is the bindingpartner for the CooP peptide

To identify protein(s) to which the CooP peptide bindsin the target tissue, we performed a yeast-two-hybridscreen (17). The peptide sequence was introduced intothe bait plasmid and cotransformed to bacteria with thetarget plasmids encoding a murine embryonic (E12.5)cDNA library. An embryonic library was chosen becausethe CooP peptide did not home to the normal brain, andmany embryonic proteins become upregulated in tumortissue but are absent in normal adult tissues (e.g., c-Myc,the EDB domain of oncofetal fibronectin; ref. 20, 21) andthe large isoform of tenascin C (22). The yeast-two-hybridscreen yielded ten in-frame hits, which were used forsequence identification from the BLAST database(http://blast.ncbi.nlm.nih.gov/Blast.cgi). These cDNAs






represented four different proteins: (i) an eukaryotictranslational initiation factor (EIF3S7), (ii) a neural pepti-dase (CPE), (iii) a mammary-derived growth inhibitor(MDGI), and (iv) an expressed sequence tag (EST). Threeof the hits represented two different cDNAs for MDGI.Only MDGI expression increased binding of CooP dis-playing phage to the transiently transfected U2-OS cells10-fold compared with the original library (Supplemen-

tary Fig. S1A, for methods see Supplementary Informa-tion).MDGI expression in transfected cellswas confirmedby Western blot analysis (Supplementary Fig. S1B, formethods see Supplementary Information). After this ini-tial validation, MDGI, also known as the fatty-acid bind-ing protein 3 (FABP3) or the heart-type fatty acid-bindingprotein (H-FABP) was further investigated as a potentialinteracting partner for the CooP peptide.

Figure 1. Identification of the CooP peptide that homes to a subset of brain tumors. A, phage screen comprising two ex vivo (Ev) rounds performed byincubating phage library with intracranial HIFko tumor-derived cell suspension followed by three in vivo (Iv) rounds with intravenously injected phagepool into the mice bearing intracranial HIFko tumors. Phage enrichment is shown as fold increase over the control phage. B, individual phage from the thirdin vivo selection round were tested for the ex vivo binding to the cell suspension derived from tumor containing brain and normal brain. Graphshowsphage binding to the brain tumor relative to the normal brain. C, homing of the CooP phage to brain tumor, histologically normal brain, liver, kidney, andlung tissue compared with the control phage. Values represent a mean of three experiments. Error bars, � SEM. D–F, fluorescein-conjugatedCooPpeptide (100 mmol/L, 100 mL/animal) was injected into the tail vein ofmice bearing intracranial tumors and allowed to circulate for 60minutes followed byperfusion of the mice through the right ventricle of the heart and excision of the tumor. Staining of the tumor sections with an anti-fluorescein antibody(red in E) shows accumulation of fluorescein-conjugated CooP in HIFko clusters identified by staining of an adjacent tissue section with an antibodyagainst theSV40 large Tantibody (green in F), amarker for tumor cells. Homing of the fluorescein-conjugatedCooPpeptide (red) to intracranial U87MG (G) andBT4C (H) gliomas. Absence of CooP homing to subcutaneous MDA-MB-231 breast tumors (I), to subcutaneous HIFko tumors (J), and to theVEGF-overexpressing intracranial tumors (K). G–K, sectionswere stainedwith an anti-PECAM1antibody (green) to visualize tumor vasculature. D, F–K, nucleiwere visualized with DAPI (blue). Magnification, �200. Images were digitally cropped in Photoshop CS6.

Hyv€onen et al.





To act as a receptor for CooP, MDGI should beexpressed either by the tumor cells and/or tumor-asso-ciated stromal cells. First, we investigated whetherMDGIwas expressed in those tumors towhich the CooP peptidehomed. FITC-labeled CooP peptide was injected intrave-nously into themicebearing intracranialHIFkoorU87MGgliomas. Tumor sections were then studied for the pres-ence of the peptide aswell as forMDGI expression.MDGIwas detected both in the U87MG and HIFko xenografts(Fig. 2B and D). Furthermore, partial colocalization ofMGDI and the peptide was observed in the tumor tissue(arrows in Fig. 2A–D). Importantly, MDGI was expressedin all tumors the CooP peptide homed to (intracranialHIFko, U87MG, and BT4C tumors) andMDGI expressionwas absent in tumors to which the CooP peptide did nothome (intracranial VEGF-overexpressing tumor, orthoto-pic MDA-MD-231 breast cancer xenograft, and subcuta-neous HIFko tumor). Moreover, we stained two addition-al xenograft models of glioblastoma multiforme for thepresence ofMDGI. Intracranial tumors derived both from

the murine EGFRvIII transformed, INK4/Arf�/� NSCGsand from a patient tumor (GBM43; ref. 18) expressedmoderate levels of MDGI (Supplementary Fig. S2).

To further validate MDGI as a receptor for the CooPpeptide, we established an MDGI expressing U87MGglioma cell line and monitored whether the presence ofMDGI would affect CooP homing to the intracranialtumors derived fromMDGI expressing or control U87MGcells. MDGI expression substantially increased the pep-tide homing to these tumors (Fig. 2E and F).

Circulating anti-MDGI antibody localizes to tumor-associated blood vessels

To act as a receptor for a systemically administeredpeptide, MDGI should be accessible via the circulation.Therefore, we next injected goat anti-MDGI antibodiesintravenously to U87MG tumor mice and investigatedwhether the antibodies could be detected in the tumortissue 60 minutes after injection. Blood vasculature wasvisualized with anti-PECAM-1 staining. Staining of tissuesections from various organs with secondary antibodiesshowed accumulation of the anti-MDGI antibody in thetumor-associated vasculature (Fig. 3A), whereas bloodvessels in other tissues such as muscle (Fig. 3B), heart (Fig.3C), normal brain (Fig. 3D), kidney, liver, and lung (datanot shown) showed no detectable accumulation of theantibody.Controlgoat IgGdidnotaccumulate in the tumoror any other tissue examined (Fig. 3E–H). We also stainedtissue sections forMDGI todetect expressionof thisproteinin sites not accessible to the circulating antibody. In addi-tion to the tumor (Fig. 3I), muscle (Fig. 3J) and heart (Fig.3K) were also positive for MDGI expression. No MDGIexpressionwas detected in normal brainwith the antibody(Fig. 3L). Thus, only in tumor MDGI was accessible viathe blood circulation and could therefore act as a specificreceptor molecule for the intravenously administeredhoming peptide and anti-MDGI antibody.

Peptide-based SPECT/CT imaging of brain tumorsNext, we studied the potential of the CooP peptide in

brain tumor imaging using SPECT/CT. We labeled thediethylene-triamine-penta-acetic acid conjugated CooPwith 111Indium (111In) and injected it intravenously intomice bearing intracranial U87MG tumors. We collectedtwo-dimensional images from all animals up to 15 or 30minutes after injection. A 3D image was reconstructed60 minutes, 120 minutes, or 24 hours after injection.Plasma clearance of the peptide was rapid because mostof the activity was excreted in the urine within the first 20minutes (Fig. 4A). Despite the clearance, uptake of 111In-CooP in the U87MG glioma was evident in 30% of mice(5.41 � 0.07% ID g�1) at 60 minutes after injection and in85% of mice (2.43 � 0.85% ID g�1) at 120 minutes afterinjection (Fig. 4B and D). The tumor-to-blood ratioincreased from 1.38 � 0.38 at 60 minutes to 2.49 � 0.02at 120 minutes resulting in improved tumor resolutionand detection in SPECT. No accumulation of an 111In-labeled control peptide was detected in tumor tissue

Figure 2. CooP peptide colocalizes with MDGI in gliomas and MDGIoverexpression increases CooP homing. Fluorescein-conjugatedCooP peptide (100 mmol/L, 100 mL/animal) was injected intravenouslyinto mice bearing intracranial HIFko (A–B) or U87MG (C–F) gliomas.CooP peptide and MDGI were visualized using antibodies againstfluorescein (red in A, C, E, and F) and MDGI (green in B and D).Overexpression of MDGI in the U87MG gliomas (F) increased homing ofthe CooP peptide compared with the control U87MG tumors (E).Nuclei were visualized with DAPI. Magnification, 400 �. Images weredigitally cropped in Photoshop CS6.






2 hours after injection (Fig. 4C and D). Tumor site radio-activity was approximately 20-fold higher at 15 minutesafter injection and about 12-fold higher at 2 hours afterinjection compared with the contralateral brain hemi-sphere, indicating marked CooP peptide accumulationin the malignant tissue (Fig. 4E). Importantly, only thetumors showed increased accumulation of the CooP pep-tide compared with the control peptide at 2 hours afterinjection (Fig. 4F).

CooP-targeted delivery of chemotherapeuticsSince CooP peptide homed to the brain tumor satellites,

we wanted to study the potential of CooP-targeted ther-apy to treat invasive brain tumors. As a proof-of-concept,we covalently conjugated clorambucil (Cbl) to the CooP

peptide. Since CooP was not taken up by the cells, a cellpenetrating peptide sequence (CPP; ref. 23) was addedbetween CooP and Cbl to facilitate conjugate internaliza-tion (Fig. 5A). On the basis of their infiltrative growthpattern, intracranial HIFko tumors were chosen as amodel. We treated the intracranial HIFko tumor-bearingmice with intravenous injections of either saline, free Cbl,or CooP-CPP-Cbl (5 mg/kg, 4 mice/group) every secondday starting on day 5 after implantation and monitoredthe weight of the animals during the study. At day 14,mice in the saline and free Cbl groups had lost more than10% of their starting weight and had to be sacrificed (Fig.5B). In the targeted therapy group, mice gained weightand did not show any signs of tumor burden indicatingthat the CooP-targeted drug treatment prolonged the life

Figure 3. MDGI is accessible via circulation in the tumor vasculature. To assess whether MDGI was accessible via circulation, we injected the goat anti-MDGI(A–D) or control goat IgG (E–H) antibodies into the tail vein of intracranial U87MG tumor-bearing mice. Tumors and organs were removed 1 hourafter injection, prepared for histologic analyses, and examined for the presence of antibodies using anti-goat Alexa 594 (red in A–H). Blood vessels werestained with anti-PECAM-1 antibody (green in A–H). Injected anti-MDGI antibody colocalized with blood vessels only in tumors (A), whereas nocontrol IgGcould bedetected in tumors (E). No anti-MDGI antibody or control IgGwasdetected in other tissues examined:muscle (B and F), heart (C andG), orbrain (D and H). To study the in situ expression of MDGI in tumors and other organs, we stained tissue sections with the goat anti-MDGI antibody followedby the anti-goat Alexa 594 (red in I–L). Fluorescein-conjugated tomato lectin (green in I–L) was injected intravenously to visualize the vasculature.MDGI colocalized with the vascular staining only in tumors (I) but not inmuscle (J), heart (K), or brain (L). Nuclei were visualizedwith DAPI (blue). Magnification,� 400. Images were digitally cropped in Photoshop CS6.

Hyv€onen et al.





span of the animals (Fig. 5B). Quantification of tumorsatellites in the brain showed significantly (P ¼ 0.045)fewer satellites in the targeted drug treated animals thanin the control groups (Fig. 5C), whereas no difference inthe size of the primary tumor was observed (Fig. 5D).

MDGI expression in human brain tumorsOur results suggest that MDGI acts as an interacting

partner for the CooP peptide in the experimental gliomamodels. Since MDGI expression in human brain tumorshas not been reported earlier, we investigated the pres-ence and localization of MDGI in clinical brain tumorsamples by immunohistochemistry. First, we analyzed apanel of astrocyte-derived tumor samples from cranioto-

my patients. Normal human brain (Fig. 6A) and grade 1astrocytomas (Fig. 6B) did not expressMDGI at detectablelevels. Interestingly, the majority of the grade 2 (60%; Fig.6C) and grade 3 astrocytomas (80%; Fig. 6D) showedmoderate levels of MDGI mostly in the vasculature andin perivascular compartment. Of glioblastomas, 17 of 18(94%) expressed veryhigh levels ofMDGI. The expressionwas concentrated on perinecrotic areas of the tumors (Fig.6E and F). Sixty-seven percent of ependymomas (Fig. 6G)were moderately positive while medulloblastomas (Fig.6H) showed noMDGI expression.We also had access to asmall tissue array of human glioblastomas (n¼ 46). In thisanalysis, 70% (32/46) of gliomas stained positive forMDGI. To investigate MDGI localization in gliomas more

Figure 4. Peptide-targeted SPECT/CT imaging of gliomas. A, majorityof the intravenously injectedpeptide (20 mg in 100 mL) wasrapidly excreted through thekidneys into urine. Two hours afterinjection, accumulation of the111In-conjugated CooP peptidewas detectable in brain tumors (Band D), whereas no control peptidecould be detected (C and D). E,accumulation of 5 MBq 111In-CooPin the tumor half and thehealthy half of the brain in sameanimals. F, biodistribution of 111In-CooP or control peptide 2 hoursafter injection. Error bars, � SD.






carefully, we stained paraffin-embedded sections withantibodies against MDGI and an endothelial marker,PECAM-1. MDGI was detected in tumor cells (Fig. 6I–K) and in vascular structures, which were positive forPECAM-1 staining in subsequent sections (Fig. 6L–P).Double-labeling with fluorescently labeled anti-MDGIand PECAM-1 antibodies showed that both proteinswerepresent in the same cells (Fig. 6Q). Thus, our resultsdemonstrate thatMDGI is present in human brain tumorsin a grade-dependent manner, its expression positivelycorrelates with the histologic grade of the tumor and it isexpressed both in tumor cells and in blood endothelium.

DiscussionHere we describe identification of a novel homing

peptide, CooP, which recognizes an epitope in certainbrain tumor cells and tumor-associated endothelium. Weprovide evidence to the effect that mammary-derivedgrowth inhibitor (MDGI), acts as an interacting partnerfor CooP in brain tumors, and show that the CooP peptidecan be used in selective delivery of an imaging agent andchemotherapeutic into glioblastomas. Finally, we showthat human glioblastomas, but not the human normalbrain, express MDGI, the CooP receptor.

The CooP peptide appears to have a highly selectivespecificity as a homing peptide. Binding of the CooPphage to cells isolated from glioblastomas, tumor hom-ing of this phage in vivo, and the accumulation of theCooP peptide in glioblastomas, but not in the normalbrain or other normal tissues demonstrate tumor spec-ificity of the peptide. Our results also suggest furtherlayers of specificity. CooP did not recognize the vesselsin HIFko tumors that had been engineered to expressVEGF, and unlike the original tumors, were able toinduce angiogenesis to provide the vasculature to sus-tain tumor growth. Also, an orthotopically grown breast

cancer was negative for CooP homing. These resultsshow that CooP does not recognize an angiogenesis-specific marker, like most of the currently knowntumor-homing peptides do (reviewed in ref. 12).Tumor-homing peptides and vascular tumor markersthat are tumor type–specific have been described(10, 24, 25). Moreover, CooP did not recognize HIFkotumors grown subcutaneously, suggesting a specificityfor brain-derived tumor vasculature. Other than theirproximity to tumor cells, there are no markers forvessels co-opted by tumors (as opposed to arising asa result of tumor-induced angiogenesis). Hence, it isdifficult to prove that CooP specifically recognizes co-opted tumor vessels, but the features discussed above,particularly the lack of binding to angiogenic tumorvessels, strongly suggest that this is the case. No peptidewith such specificity has been described before.

Several lines of evidence indicate that the MDGI/FABP3 is the receptor that is recognized by CooP and theselective expression of which endows CooP with thespecific homing properties. First, a yeast-2-hybrid screenusing the peptide sequence as bait yielded three differentMDGI cDNAs suggesting that it binds to the CooP pep-tide. Second, MDGI overexpression increased binding ofCooP to cells in vitro and homing to tumor xenografts invivo. Third, circulating antibodies against MDGI accumu-lated in tumor tissue and were nondetectable in othertissues, indicating that MDGI is accessible through thecirculation for intravenously injected ligands. It is note-worthy that MDGI is also expressed in heart and skeletalmuscle as previously described (26) but, importantly, thevascular expression is restricted to the tumor tissue,allowing the tumor-specific homing of the peptide. Stain-ing of human glioma samples with anti-MDGI andPECAM-1 antibodies confirmed the vascular expressionof MDGI.

CooPCPP-CPP

3,000

2,000

A

1,000Upt

ake

(pm

ol/m

g)

CooP

PBS Cbl CooP-CPP-Cbl

110

105

100

95

90

1412108642

B

Days after tumor implantation

Nor

mal

ized

wei

ght o

f the

ani

mal

s (%

)

Nor

mal

ized

are

a co

vere

d b

y sa

telli

tes

(%)

25

20

15

10

5

CblPBS CooP-Cbl

C

*

D

CblPBS CooP-Cbl

20

10

30

40

Nor

mal

ized

are

a co

vere

d b

y pr

imar

y tu

mor

(%

)

Figure 5. Peptide-targeted delivery of therapeutics prolongs the life-span of mice bearing intracranial HIFko tumors. A, internalization of fluorescein-labeledCooP, CPP, or Coop-CPP peptides. Fluorescence was measured at 494/518 nm after 1 hour of incubation of HIFko cells with 5 mmol/L fluorescein-labeledpeptides. Values represent the mean of 4 replicates in one representative experiment. Error bars, � SD. B, HIFko cells were injected intracraniallyinto nude mice. Treatment was started after tumor establishment at day 5 after implantation (arrow). Mice were treated with intravenous injections of eithersaline (PBS), free Cbl, or targeted drug (CooP-CPP-Cbl) every second day. Weight of the animals was monitored. Experiment was stopped whencriteria for sacrifice was met (>10% weight loss). Graph shows the normalized, mean body weight in each group during the experiment (� SEM). The meanweight on day 0 was set as 100%. C, brain sections of the treated mice were imaged and brain area occupied by tumor satellites was quantifiedfrom four sections/animal. Graph shows percentage of brain occupied by the tumor satellites (� SEM). D, brain sections of the treated mice were imagedand brain area occupied by the primary tumors was quantified from four sections/animal. Graph shows percentage of brain occupied by the primary tumors(� SEM). �, P � 0.05 (Student t test).

Hyv€onen et al.





Fatty acid binding proteins (FABP) are primarily intra-cellular proteins, although membrane-associated formshave also been reported (27). These FAPBs mediate thetransport of fatty acids and other hydrophobic ligands totissues (28). The extracellular/cell surface localization of

intracellular proteins is not unusual in tumors; cell surfaceor plasma membrane localization in tumors has beenreported previously for several intracellular proteins:e.g., GRP78 (29), NUCLEOLIN (30), ANNEXIN (31), andP32 (32).

Figure 6. MDGI is expressed in tumor cells and in blood endothelial cells, and its expression positively correlateswith tumor grade in clinical humanbrain tumorsamples. MDGI expression was studied in sections from normal brain (A), grade 1 (B), grade 2 (C), and grade 3 (D) astrocytomas, as well as in primary(E, I–P) and secondary (F) grade 4 glioblastomas, in ependymomas (G), and in medulloblastomas (H) with anti-MDGI antibodies using immunohistochemicalstainings.MDGI andPECAM-1 are shown as red color and nuclei in blue. I, MDGI expression in tumor cells. J, subsequent section stainedwith anti-PECAM-1antibodies. K and L, specificity of the MDGI staining was shown by using rabbit IgG staining as a control for panels I and M. M and P, MDGIexpression in PECAM-1–positive endothelial cells. N, higher magnification of the boxed area in M. O, higher magnification of the boxed area in P. Q, humanglioblastoma (grade 4) sections were stained with fluorescently labeled antibodies against MDGI (red) and PECAM-1 (green). Nuclei were visualizedwithDAPI (blue). Inset, highermagnification of the boxed area showing thatMDGI andPECAM-1 are present in the same cells.MagnificationA–K,�400;M–Q,�200. Images were digitally cropped in Photoshop CS6.






The role of MDGI in tumor progression is somewhatcontroversial and seems to vary among different cancertypes. MDGI appears to be the only FABP that affects cellproliferation and differentiation. This function may beseparate from its ligand-binding function since it can bemimicked by its C-terminal peptide, which cannot bindfatty acids (33, 34). In sporadic breast cancers, the genomicregion coding for MDGI is often deleted and MDGIappears to be downregulated in breast cancer cell linesvia hypermetylation (35, 36). Moreover, MDGI has beenreported to inhibit the growth ofMCF-7 breast cancer cellsand reduce their tumorigenicity (37) as well as inhibitinvasion andadhesionofMDA-MB-231breast cancer cells(36).

On the other hand, ectopic MDGI expression wasshown to render breast and lung cancer cells resistant totreatmentwith the anti-EGFRantibody, cetuximab (36). Insmall-cell lung cancer, MDGI expression is significantlyhigher in the highly aggressive cells than in their lessaggressive subtype (38). In gastric carcinoma, MDGIexpression is also associated with tumor aggressiveness,progression, and poor patient survival (39). Our datashowassociation of increasedMDGI expressionwith highgrade of brain tumors and suggest MDGI as a novelmarker for highly malignant brain tumors.

In contrast to antibodies, which have been traditionallyused for the specific delivery of therapeutic agents toseveral tumor types (40–43), homing peptides are fasterand less expensive to produce and nonimmunogenic(12, 44). Coupling of therapeutic agents with peptides hasalso been shown to enhance drug penetration into thetarget tissue (45). Because of their small size, peptidespenetrate the blood–brain barrier better, enabling thetargeted delivery of therapeutic agents directly into thetumor tissue. Therefore, utilization of homing peptidesoffers an attractive alternative for targeting solid tumorssuch as malignant gliomas, which adapt to novel anti-angiogenic therapies by increased invasion and co-optionof preexisting vasculature (45–50). Importantly, CooPpeptide-conjugated therapeutics reduced the number of

tumor satellites in the brain, suggesting that CooP-tar-geted therapies could be useful as drug carrier to ablatetumor cells in areas that cannot be surgically removed orreached by angiogenesis-based therapies. In addition,MDGI, the CooP receptor,may be a useful targetmoleculefor the development of new therapies against malignantgliomas.

Disclosure of Potential Conflicts of InterestNo potential conflicts of interest were disclosed.

Authors' ContributionsConception and design: J. Enb€ack, U. Langel, P.M. LaakkonenDevelopment ofmethodology: J. Enb€ack, T.Huhtala, J.Weisell, U. Langel,P.M. LaakkonenAcquisition of data (provided animals, acquired and managed patients,provided facilities, etc.): M. Hyv€onen, J. Enb€ack, T. Huhtala, J. Lammi,H. Sihto, J.Weisell, H. Joensuu, S. El-Andaloussi, A. N€arv€anen, G. Bergers,G. Bergers, P.M. LaakkonenAnalysis and interpretation of data (e.g., statistical analysis, biostatis-tics, computational analysis): M. Hyv€onen, J. Enb€ack, T. Huhtala,J. Weisell, H. Joensuu, A. N€arv€anen, P.M. LaakkonenWriting, review, and/or revision of the manuscript: M. Hyv€onen, J.Enb€ack, T. Huhtala, J. Lammi, H. Sihto, J. Weisell, H. Joensuu, U. Langel,A. N€arv€anen, P.M. LaakkonenAdministrative, technical, or material support (i.e., reporting or orga-nizing data, constructing databases): M. Hyv€onen, H. Sihto, U. LangelStudy supervision: U. Langel, P.M. LaakkonenSynthesis of peptides used for imaging and treatment: K. Rosenthal-Aizman

AcknowledgmentsThe authors thank Dr. Erkki Ruoslahti for the critical comments on the

manuscript, Anastasiya Chernenko and P€aivi Kivinen for excellent tech-nical assistance, and Molecular Imaging Unit (MIU) for help in imaging.

Grant SupportThis work was supported by the grants from the Academy of Finland

(No 107664, 124212 and 131732; to P. Laakkonen), Finnish Cancer Orga-nizations (to P. Laakkonen), and NIH (NIH-U54CA163155; to G. Bergers).M. Hyv€onen and J. Enb€ack have been supported by the Helsinki Biomed-ical Graduate Program.

The costs of publication of this article were defrayed in part by thepayment of page charges. This article must therefore be hereby markedadvertisement in accordance with 18 U.S.C. Section 1734 solely to indicatethis fact.

Received August 15, 2013; revised December 31, 2013; accepted January20, 2014; published OnlineFirst February 3, 2014.

References1. Louis DN, Ohgaki H, Wiestler OD, Cavenee WK, Burger PC, Jouvet A,

et al. The 2007 WHO classification of tumours of the central nervoussystem. Acta neuropathologica 2007;114:97–109.

2. Ekman M, Westphal M. Cost of brain tumour in Europe. Europeanjournal of neurology: the official journal of the European Federation ofNeurological Societies. 2005;12 Suppl 1:45–9.

3. Stupp R, Mason WP, van den Bent MJ, Weller M, Fisher B, TaphoornMJ, et al. Radiotherapy plus concomitant and adjuvant temozolomidefor glioblastoma. N Engl J Med 2005;352:987–96.

4. Erridge SC, Hart MG, Kerr GR, Smith C, McNamara S, Grant R, et al.Trends in classification, referral and treatment and the effect onoutcome of patients with glioma: a 20 year cohort. J Neurooncol 2011;104:789–800.

5. Ruoslahti E. Specialization of tumour vasculature. Nat Rev Cancer2002;2:8–90.

6. Arap W, Pasqualini R, Ruoslahti E. Cancer treatment by targeted drugdelivery to tumor vasculature in a mouse model. Science 1998;279:377–80.

7. Rajotte D, Arap W, Hagedorn M, Koivunen E, Pasqualini R,Ruoslahti E, et al. Molecular heterogeneity of the vascular endo-thelium revealed by in vivo phage display. J of Clin Invest 1998;102:430–7.

8. Zhang L, Hoffman JA, Ruoslahti E. Molecular profiling of heart endo-thelial cells. Circulation 2005;112:1601–11.

9. Laakkonen P, Akerman ME, Biliran H, Yang M, Ferrer F, Karpanen T,et al. Antitumor activity of a homing peptide that targets tumorlymphatics and tumor cells. Proc Natl Acad Sci U S A 2004;101:9381–6.

10. Laakkonen P, Porkka K, Hoffman JA, Ruoslahti E. A tumor-homingpeptide with a targeting specificity related to lymphatic vessels. NatMed 2002;8:751–5.

11. Zhang L, Giraudo E, Hoffman JA, Hanahan D, Ruoslahti E. Lymphaticzip codes in premalignant lesions and tumors. Cancer Res 2006;66:5696–706.

12. Laakkonen P, Vuorinen K. Homing peptides as targeted deliveryvehicles. Integr Biol 2010;2:326–37.

Hyv€onen et al.





13. Enback J, Laakkonen P. Tumour-homing peptides: tools for targeting,imaging and destruction. Biochem Soc Trans 2007;35:780–3.

14. Du R, Lu KV, Petritsch C, Liu P, Ganss R, Passegue E, et al. HIF1alphainduces the recruitment of bone marrow-derived vascular modulatorycells to regulate tumor angiogenesis and invasion. Cancer Cell 2008;13:206–20.

15. Blouw B, Song H, Tihan T, Bosze J, Ferrara N, Gerber HP, et al. Thehypoxic response of tumors is dependent on their microenvironment.Cancer Cell 2003;4:133–46.

16. Rivinoja A, Laakkonen P. Identification of homing peptides using the invivo phage display technology. Methods Mol Biol 2011;683:401–15.

17. Fields S, Song O. A novel genetic system to detect protein-proteininteractions. Nature 1989;340:245–6.

18. LuKV,Chang JP,ParachoniakCA,PandikaMM,AghiMK,MeyronetD,et al. VEGF inhibits tumor cell invasion and mesenchymal transitionthrough a MET/VEGFR2 complex. Cancer cell 2012;22:21–35.

19. Porkka K, Laakkonen P, Hoffman JA, Bernasconi M, Ruoslahti E. Afragment of the HMGN2 protein homes to the nuclei of tumor cells andtumor endothelial cells in vivo. Proc Natl Acad Sci U S A 2002;99:7444–9.

20. Carnemolla B, Balza E, Siri A, Zardi L, Nicotra MR, Bigotti A, et al. Atumor-associated fibronectin isoformgeneratedby alternative splicingof messenger RNA precursors. J Cell Biol 1989;108:1139–48.

21. Castellani P, Viale G, Dorcaratto A, Nicolo G, Kaczmarek J, Querze G,et al. The fibronectin isoform containing the ED-B oncofetal domain: amarker of angiogenesis. Int J Cancer 1994;59:612–8.

22. Borsi L, Carnemolla B, Nicolo G, Spina B, Tanara G, Zardi L, et al.Expression of different tenascin isoforms in normal, hyperplastic andneoplastic human breast tissues. Int J Cancer 1992;52:688–92.

23. Johansson HJ, El-Andaloussi S, Holm T, Mae M, Janes J, Maimets T,et al. Characterization of a novel cytotoxic cell-penetrating peptidederived from p14ARF protein. Mol Ther 2008;16:115–23.

24. Hoffman JA, Giraudo E, Singh M, Zhang L, Inoue M, Porkka K, et al.Progressive vascular changes in a transgenic mouse model of squa-mous cell carcinoma. Cancer Cell 2003;4:383–91.

25. Joyce JA, Laakkonen P, Bernasconi M, Bergers G, Ruoslahti E,Hanahan D, et al. Stage-specific vascular markers revealed by phagedisplay in a mouse model of pancreatic islet tumorigenesis. CancerCell 2003;4:393–403.

26. Zschiesche W, Kleine AH, Spitzer E, Veerkamp JH, Glatz JF. Histo-chemical localization of heart-type fatty-acid binding protein in humanand murine tissues. Histochem Cell Biol 1995;103:147–56.

27. Glatz JF, van der Vusse GJ. Cellular fatty acid-binding proteins: theirfunction and physiological significance. Prog Lipid Res 1996;35:243–82.

28. Veerkamp JH. Fatty acid transport and fatty acid-binding proteins.Proc Nutr Soc 1995;54:23–37.

29. Arap MA, Lahdenranta J, Mintz PJ, Hajitou A, Sarkis AS, Arap W, et al.Cell surface expression of the stress response chaperone GRP78enables tumor targeting by circulating ligands. Cancer Cell 2004;6:275–84.

30. Christian S, Pilch J, AkermanME, Porkka K, Laakkonen P, Ruoslahti E,et al. Nucleolin expressed at the cell surface is a marker of endothelialcells in angiogenic blood vessels. J Cell Biol 2003;163:871–8.

31. Oh P, Li Y, Yu J, Durr E, Krasinska KM, Carver LA, et al. Subtractiveproteomicmappingof the endothelial surface in lungandsolid tumoursfor tissue-specific therapy. Nature 2004;429:629–35.

32. Fogal V, Zhang L, Krajewski S, Ruoslahti E. Mitochondrial/cell-surfaceprotein p32/gC1qR as a molecular target in tumor cells and tumorstroma. Cancer Res 2008;68:7210–8.

33. Storch J, Corsico B. The emerging functions and mechanisms ofmammalian fatty acid-bindingproteins. AnnuRevNutr 2008;28:73–95.

34. Wang HL, Kurtz A. Breast cancer growth inhibition by delivery of theMDGI-derived peptide P108. Oncogene 2000;19:2455–60.

35. HuynhH, Alpert L, PollakM. Silencing of themammary-derived growthinhibitor (MDGI) gene in breast neoplasms is associated with epige-netic changes. Cancer Res 1996;56:4865–70.

36. Nevo J, Mattila E, Pellinen T, Yamamoto DL, Sara H, Iljin K, et al.Mammary-derived growth inhibitor alters traffic of EGFR and induces anovel form of cetuximab resistance. Clin Cancer Res 2009;15:6570–81.

37. HuynhH, PollakM. HH2A, an immortalized bovinemammary epithelialcell line, expresses the gene encoding mammary derived growthinhibitor (MDGI). In Vitro Cell Dev Biol Anim 1995;31:25–9.

38. Zhang L, Cilley RE, Chinoy MR. Suppression subtractive hybridizationto identify gene expressions in variant and classic small cell lungcancer cell lines. Journal of surgical research 2000;93:108–19.

39. Hashimoto T, Kusakabe T, Sugino T, Fukuda T, Watanabe K, Sato Y,et al. Expression of heart-type fatty acid-binding protein in humangastric carcinoma and its association with tumor aggressiveness,metastasis and poor prognosis. Pathobiology 2004;71:267–73.

40. Mancuso A, Sternberg CN. Colorectal cancer and antiangiogenictherapy: what can be expected in clinical practice? Crit Rev OncolHematol 2005;55:67–81.

41. Stevens PL, Oluwole O, Reddy N. Advances and application of radio-immunotherapy in non-Hodgkin lymphoma. Am J Blood Res 2012;2:86–97.

42. Spitaleri G, Berardi R, Pierantoni C, De Pas T, Noberasco C, Libbra C,et al. Phase I/II study of the tumour-targeting human monoclonalantibody-cytokine fusion protein L19-TNF in patients with advancedsolid tumours. J Cancer Res Clin Oncol 2013;139:447–55.

43. Spector NL, Blackwell KL. Understanding the mechanisms behindtrastuzumab therapy for human epidermal growth factor receptor 2-positive breast cancer. J Clin Oncol 2009;27:5838–47.

44. Yin H, Moulton HM, Betts C, Merritt T, Seow Y, Ashraf S, et al.Functional rescue of dystrophin-deficient mdx mice by a chimericpeptide-PMO. Mol Ther 2010;18:1822–9.

45. Sugahara KN, Teesalu T, Karmali PP, Kotamraju VR, Agemy L, GirardOM, et al. Tissue-penetrating delivery of compounds and nanoparti-cles into tumors. Cancer Cell 2009;16:510–20.

46. Rubenstein JL, Kim J, Ozawa T, ZhangM,Westphal M, Deen DF, et al.Anti-VEGF antibody treatment of glioblastoma prolongs survival butresults in increased vascular cooption. Neoplasia 2000;2:306–14.

47. Kunkel P, Ulbricht U, Bohlen P, Brockmann MA, Fillbrandt R, StavrouD, et al. Inhibition of glioma angiogenesis and growth in vivo bysystemic treatment with a monoclonal antibody against vascularendothelial growth factor receptor-2. Cancer Research 2001;61:6624–8.

48. di Tomaso E, Snuderl M, KamounWS, Duda DG, Auluck PK, FazlollahiL, et al. Glioblastoma recurrence after cediranib therapy in patients:lack of "rebound" revascularization as mode of escape. Cancer Res2011;71:19–28.

49. Leenders WP, Kusters B, Verrijp K, Maass C, Wesseling P, HeerschapA, et al. Antiangiogenic therapy of cerebral melanoma metastasesresults in sustained tumor progression via vessel co-option. ClinCancer Res 2004;10:6222–30.

50. Paez-RibesM,Allen E,Hudock J, Takeda T,OkuyamaH, Vinals F, et al.Antiangiogenic therapy elicits malignant progression of tumors toincreased local invasion and distant metastasis. Cancer Cell 2009;15:220–31.






2014;13:996-1007. Published OnlineFirst February 3, 2014.Mol Cancer Ther Maija Hyvönen, Juulia Enbäck, Tuulia Huhtala, et al. TumorsNovel Target for Peptide-Based Imaging and Treatment of Brain

Updated version

10.1158/1535-7163.MCT-13-0684doi:

Access the most recent version of this article at:

Material

Supplementary

http://mct.aacrjournals.org/content/suppl/2014/02/06/1535-7163.MCT-13-0684.DC1

Access the most recent supplemental material at:

Cited articles

http://mct.aacrjournals.org/content/13/4/996.full#ref-list-1

This article cites 49 articles, 14 of which you can access for free at:

Citing articles

http://mct.aacrjournals.org/content/13/4/996.full#related-urls

This article has been cited by 3 HighWire-hosted articles. Access the articles at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected]

To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://mct.aacrjournals.org/content/13/4/996To request permission to re-use all or part of this article, use this link



http://mct.aacrjournals.org/lookup/doi/10.1158/1535-7163.MCT-13-0684

http://mct.aacrjournals.org/content/suppl/2014/02/06/1535-7163.MCT-13-0684.DC1

http://mct.aacrjournals.org/content/13/4/996.full#ref-list-1

http://mct.aacrjournals.org/content/13/4/996.full#related-urls

http://mct.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://mct.aacrjournals.org/content/13/4/996


novel target for peptide-based imaging and treatment of ......progression. even with the most...

Documents