ORIGINAL ARTICLE

A novel melanoma-targeting peptide screened by phagedisplay exhibits antitumor activity

Alisson L. Matsuo & Aparecida S. Tanaka &

Maria A. Juliano & Elaine G. Rodrigues &

Luiz R. Travassos

Received: 5 April 2010 /Revised: 22 July 2010 /Accepted: 11 August 2010 /Published online: 28 August 2010# Springer-Verlag 2010

Abstract Peptide display on the phage surface has beenwidely used to identify specific peptides targeting several invivo and in vitro tumor cells and the tumor vasculature,playing a role in the discovery of bioactive antitumor agents.Bioactive peptides have been selected to target importanttumor receptors or apoptosis-associated molecules such asp53. Presently, we attempted to identify potentially anti-tumor bioactive molecules using the whole cell surface as therecognizable static matrix. Such methodology could beadvantageous in cancer therapy because it does not requireprevious characterization of target molecules. Using a C7Cphage display library, we screened for peptides binding to theB16F10-Nex2 melanoma cell surface after pre-absorption onmelan-A lineage. After a few rounds of enrichment, 50

phages were randomly selected, amplified, and tested forinhibition of tumor cell proliferation. Seven were active, andthe corresponding peptide of each phage was chemicallysynthesized in the cyclic form and tested in vitro. Threepeptides were able to preferentially inhibit the melanomalineage. A unique peptide, [-CSSRTMHHC-], exhibited invivo antitumor inhibitory activity against a subcutaneousmelanoma challenge, rendering 60% of mice without tumorgrowth. Further, this peptide also markedly inhibited in vitroand in vivo the tumor cell invasion and cell-to-cell adhesive-ness in vitro. This is the first report on a bioactive peptidederived from a C7C library active against whole melanomacells in vitro and in vivo.

Keywords Melanoma . Phage display . Bioactive peptide .

Antitumor activity

Introduction

Melanoma is a highly aggressive cancer with poor prognosisin the metastatic stage. It is responsible for 80% of total skincancer deaths despite the fact that it accounts for only 4% ofdermatologic cancers [1]. Many studies have shown thattumor and associated cells may over express surfaceproteins, biochemically altered components, and even neo-antigens [2, 3]. Since the main goal in cancer treatment is totarget cancer cells and preserve normal tissues, newmethodologies have been explored to identify moleculartargets and specific ligands [4].

Phage display peptide libraries have been widely used toidentify specific peptides and proteins targeting cancer cellsin vitro and in vivo [5–8] and/or the tumor vasculature [9].

A. L. Matsuo (*) : E. G. Rodrigues : L. R. TravassosExperimental Oncology Unit (UNONEX),Department of Microbiology,Immunology and Parasitology,Federal University of São Paulo,Rua Botucatu 862, oitavo andar,São Paulo, São Paulo 04023-062, Brazile-mail: almatsuo@unifesp.br

A. S. TanakaDepartment of Biochemistry,Federal University of São Paulo,Rua Três de Maio 100,São Paulo, São Paulo 04044-020, Brazil

M. A. JulianoDepartment of Biophysics,Federal University of São Paulo,Rua Três de Maio 100,São Paulo, São Paulo 04044-020, Brazil

J Mol Med (2010) 88:1255–1264DOI 10.1007/s00109-010-0671-9

The procedure showed promising results in the discovery ofbioactive antitumor and internalizing agents [4, 10, 11]. Itbasically consists in the selection of peptides or proteinsdisplayed on phage libraries that recognize a specific ligand,followed by phage DNA sequencing to determine the aminoacid sequence. Most peptides selected for cell surface proteinbinding may target receptors or adhesion molecules [12, 13].The use of synthetic peptides, based on phage selectivebinding, is valuable in cancer therapy because of their lowinteraction with the immune system, potential large scaleproduction, reproducibility, and good tumor and tissuepenetration [14].

In the melanoma model, there are few studies using thePhage Display methodology, including the treatment withwhole phages expressing a specific melanoma peptide [15],a human melanin binding peptide to address melanoma cells[16], and more recently, a peptide targeting a disintegrinwhich is important for melanoma adhesion [17]. In thepresent work, our aim was to identify specific bioactivepeptides with antitumor activity using the Phage Displaytechnique directly on whole tumor cells. After three roundsof subtractive/selection protocol at 4°C, we have individu-ally tested 50 phages. Only seven peptides showed anti-proliferative effects on B16F10-Nex2 melanoma cells. Threepeptides displayed high inhibitory effect on melanomalineage. The selection of the phage peptide library in vivo,using animals challenged with B16F10-Nex2 cells, led to asingle peptide that was able to strongly inhibit subcutaneoustumor development and markedly reduced the number oflung nodules in a metastatic colonization model. This is anovel report on a C7C peptide displaying anti-melanomaactivity in vitro and in vivo.

Materials and methods

Mice and cultured cell lines

Eight-week-old male C57BL/6 mice were obtained fromthe Center for Development of Experimental Models(CEDEME), Federal University of São Paulo (UNIFESP).Animal experiments were carried out according to theguidelines provided by the UNIFESP Ethics Committee forAnimal Experimentation.

The B16F10 murine melanoma cell line is syngeneic inC57Bl/6 mice and was originally obtained from the LudwigInstitute for Cancer Research, São Paulo branch. Themelanotic subline Nex2 (B16F10-Nex2) was isolated atthe Experimental Oncology Unit (UNONEX). It forms lethalsubcutaneous tumors, with no spontaneous metastasis to thelung unless injected intravenously. Human leukemia (HL-60) and breast cancer (MDA-MB-231) were also obtainedfrom the Ludwig Institute for Cancer Research, São Paulo

branch. The human melanoma SKmel28 and SKmel25 celllines were originally obtained from the Memorial SloanKettering Cancer Center, New York. Human cervicalcarcinoma (HeLa) was acquired from Dr. Hugo PequenoMonteiro, UNIFESP. Tumoral cell lineages and humanumbilical vein endothelial cells (HUVEC) were maintainedin complete medium consisting of RPMI 1640, pH 7.2,supplemented with 10 mM N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid (HEPES), 24 mM sodium bicarbon-ate, 10% heat-inactivated fetal calf serum (FCS) fromGibco (Minneapolis, MN, USA), and 40 μg/ml gentamicinsulfate (Hipolabor Farmacêutica, Sabará, MG, Brazil).NIH-3T3 cells were provided by Mary CS Armelin(Department of Biochemistry, IQ-USP) and were grown inDMEM supplemented with 10 mM HEPES, 24 mMsodium bicarbonate, 10% heat-inactivated FCS, and40 μg/ml gentamicin sulfate. The non-tumorigenic murinemelanocytic melan-A cell line was provided by MichelRabinovitch (Department of Parasitology, UNIFESP, SãoPaulo, Brazil) and was cultured in RPMI 1640 (pH 6.9),supplemented with 5% fetal calf serum and 200 nM 12-o-tetradecanoyl phorbol-13-acetate (PMA; Tocris, Ellisville,MO) at 37°C in a humidified atmosphere with 5%CO2.

Phage library

Peptide phage library, PhD C7C, was purchased from NewEngland Biolabs. This library consists of 7-random aminoacid peptides flanked by cysteine residues from the N-terminal sequences of the M13 PIII coat protein. All phageprotocols, including phage amplification, titration, andDNA isolation, were performed as recommended by themanufacturer. The control bio-panning provided with thekit was used.

In vitro bio-panning

Bio-panning was performed with three rounds of a subtrac-tion/selection protocol as described elsewhere [8] with a fewmodifications. Briefly, a melan-A cell monolayer grown insix-well plates was incubated in RPMI with phage library(1011 plaque forming units (pfu)) for 1 h at 4°C forunspecific phage depletion. Unbound phages were trans-ferred to a B16F10-Nex2 cell monolayer and incubatedagain for 1 h at 4°C. Cells were gently washed five timeswith PBS containing 0.02% Triton X-100 followed by phageelution with 1 ml of glycine 0.1 M, pH 2.2, for 10 min. Afterneutralization with 150 μl of Tris 1 M, pH 9.1, eluted phageswere titrated and amplified as described in the manufacturer’sinstructions. This cycle was repeated three times, andindividual phages were amplified after the third selectionround.

1256 J Mol Med (2010) 88:1255–1264

Cell viability assay

Cell viability was measured as described elsewhere [18].Briefly, 2×103 plated cells were incubated for 48 h in thepresence of the selected peptides. For phage experiments,5×102 B16F10-Nex2 cells were incubated for 72 h with108 pfu of each isolated phage. Proliferation was measuredusing the Cell Proliferation Kit I (MTT; BoehringerMannheim), an MTT-based colorimetric assay for quantifi-cation of cell proliferation and viability. Readings wereperformed in a plate reader at 570 nm. All experiments weremade in quadruplicates at least twice.

Peptide synthesis

Peptides were synthesized by the solid phase and classicalsolution methods of peptide synthesis. Fluorescent peptideswere labeled using o-aminobenzoic acid (Abz) as thefluorescent group [19]. All the obtained peptides werepurified by semi-preparative HPLC on an Econosil C-18column. The molecular masses of synthesized peptides wereconfirmed by MALDI-TOF mass spectrometry, using aTofSpec-E from Micromass, Manchester, UK.

Treatment of subcutaneous murine melanoma

The protocol used was the same as described before [20].Basically, 5×104 B16F10-Nex2 cell suspension was pre-incubated for 10 min at room temperature with 50 μg of theselected peptides and inoculated subcutaneously on theright flank of C57Bl/6 mice. The tumor volume in cubicmillimeter was measured by the equation V=(0.52×D12×D2), where D1 and D2 correspond to the smaller and greaterdiameter in millimeters, respectively. Animals were sacri-ficed by cervical displacement with maximal tumor volumesof 3,000 mm3.

To mimic a therapeutic protocol, 5×104 melanoma cellswere injected subcutaneously in the mouse right flank.Mice treatment was carried out on days 1, 3, 5, and 7 afterinoculation via i.p. with doses of 50 μg of peptide on day 1,followed by 25 μg in the subsequent days. Tumor volumedetermination and animal sacrifice were performed asdescribed above.

In vivo bio-panning

In vivo bio-panning was carried out according to Pasqualiniet al. [21], Kennel et al. [22], and Eriksson et al. [15].Briefly, 5×104 melanoma cell suspension was injectedsubcutaneously in the right flank of mice. When tumorsreached approximately 1,000 mm3, mice were injectedintravenously with 1011 pfu of the random 7C7 library orphage-20 peptide in the tail vein. After 60 min, mice were

sacrificed by cervical displacement, and the kidney, spleen,and tumor mass were excised for phage titration. Since thetumor forms a soft tissue, cells were resuspended in PBS andpassed through a 70-μM cell strainer (Becton–Dickinson). Toremove blood contaminants, cells were resuspended in ice-cold 0.2% NaCl and hemolyzed with one volume of ice-cold1.6% NaCl. After several washes with PBS, tumor cells,kidney, and spleen tissues were weighed, and phages wereeluted with 0.1 M of glycine, pH 2.2, for 10 min andneutralized with 1 M of Tris, pH9.1, and titrated out asrecommended by the manufacturer. The ratio of pfu per gramof tissue was used to analyze the fold-increase of phage-20peptide in comparison with the C7C peptide library.

Fluorescent peptide binding to tumor cells

To test the binding of labeled peptide to B16F10-Nex2cells, 1×104 cells were plated on milk-white 96-well platesby centrifugation (1,500×g for 10 min) and fixed with 0.5%of glutaraldehyde overnight at 4°C. After neutralizationwith 0.5 M of glycine for 1 h at room temperature, cellswere blocked with PBS containing 1% of BSA for 3 h at37°C followed by incubation with the fluorophore Abz-labeled peptide 20 or the labeled scramble peptide in PBScontaining 0.1% BSA overnight at 4°C. Cells wereextensively washed and read on a plate reader at 320 nmand emission at 420 nm. To visualize the cell-boundpeptide by fluorescence microscopy, 1×104 B16F10-Nex2cells were cultivated on 13-mm cover slips, fixed andtreated as described previously, and analyzed at 350 nm andemission at 400 nm using 400× magnification. Thisfluorophore has been used in internally quenched peptidesto determine the specificity of proteolytic enzymes [23].Presently, we used the bound fluorophore to determinepeptide localization.

Invasion assay

Matrigel invasion by tumor cells was performed asdescribed elsewhere [24]. Briefly, a suspension of 2×105

B16F10-Nex2 cells in 500 μl of RPMI was incubated with20 μM of peptide 20 or the scramble peptide and was addedto the upper chamber of an 8-μm Transwell (Becton–Dickinson) pre-coated with Matrigel (Becton–Dickinson).As chemoattractant, we used 10% FCS in RPMI in thetranswell lower chamber. After 24 h at 37°C in humidifiedatmosphere with 5% CO2, non-invasive cells were excludedwith a swab, and cells that crossed the membrane werefixed with methanol and stained with Giemsa. Cells werecounted in three randomly chosen fields for each transwellmembrane, and the numbers of invasive cells are expressedas percentages. This experiment was performed three timesin duplicates.

J Mol Med (2010) 88:1255–1264 1257

Clonogenic assay

Clonogenic assays were performed as described by Frankenet al. [25]. Basically, 1% soft agar in RPMI supplementedwith 10% FCS was added on 6-well plates to create thelower layer. About 1×103 tumor cells in suspension weretreated or not with 100 μM of the selected peptides for 1 h atroom temperature. Cells were then mixed with 1% of softagar in RPMI and 10% of FCS and added on the previoussoft agar layer. After 25 days at 37°C in a humidifiedatmosphere with 5% CO2, colonies were counted asvisualized with a magnifying glass.

Metastatic in vivo tumor protection assays using peptide 20

In vivo metastatic assays were performed as described inGuimarães-Ferreira et al. [24]. Male, 7- to 8-week-old,C57Bl/6 mice were injected i.v. with 5×105 syngeneicB16F10-Nex2 viable cells in 0.1 ml RPMI in each mouse.For antitumor protection experiments, three groups of micewere challenged with tumor cells and treated with i.p. dosesof peptide 20 at 100 μg, the scramble peptide at 100 μg, orPBS at the same time. After 13 days, mice lungs werecollected and inspected for metastatic colonization and themelanotic nodules counted at 2× magnification.

Adhesion assays

Adhesion inhibition assays were performed as described byHu et al. [8] with modifications. Briefly, 96-well microtiterplates were plated with 2×104 B16F10-Nex2 cells over-night to build a cell monolayer. To test intercellularadhesion interference, 5×104 B16F10-Nex2 cells wereincubated for 30 min at room temperature with the peptide20 (100 μM) and the necessary controls. The peptide-treated and untreated cell suspensions were added to atumor cell monolayer and incubated for 1 h at 37°C inhumidified atmosphere with 5% CO2. Non-adherent cellswere gently washed with PBS, and attached cells werecounted in a Neubauer chamber.

Cadherin immunoprecipitation

Confluent B16F10-Nex2 cells (∼3×107) grown in RPMIsupplemented with 10% FCS were washed with ice-coldPBS and lysed in 3 mL of non-denaturing lysis buffer(20 mM Tris pH 8.0, 137 mM NaCl, 1% Triton X-100,10% glycerol, 2 mM EDTA) plus protease inhibitorcocktail. The cell lysate was maintained under constantagitation for 30 min at 4°C and centrifuged at 12,000×g for20 min. The supernatant was incubated with mouse antipan-cadherin mAb (Santa Cruz Biotechnology, CA) over-night at 4°C on a rocking shaker, and then 50 μl of Protein-

A Agarose was added and incubated overnight at 4°C. Theimmunoprecipitated protein complex was collected bycentrifugation at 2,500 rpm at 4°C for 5 min, washed fourtimes with PBS, eluted with 50 mM glycine pH 3, andneutralized with Tris.

Binding of fluorescent peptide 20 to cadherin

A white 96-well plate was coated overnight at 4°C with theimmunoprecipitated cadherin (10 μg/mL) from the tumorcell lysate in 100 mM of carbonate buffer (pH 8.6). Wellswere blocked with PBS containing 1% BSA for 4 h at roomtemperature and incubated with 100 μM of the fluorescentpeptide 20 or the scramble peptide overnight at 4°C. Afterwashing with PBS–Tween 0.1%, the amount of fluores-cence was quantified in a plate reader with excitation at320 nm and emission at 420 nm. Specific arbitraryfluorescent units for peptide 20 and the scramble peptidewere determined by subtracting the peptide fluorescencereading in the immunoprecipitated cadherin from that of theBSA wells.

Statistical analyses

The Kaplan–Meier log rank test was applied to the survivaldata. The CFU data in infected tissues were analyzed byStudent’s t test. p values ≤0.05 were considered significant.

Results

Bio-panning and identification of bioactive peptides

Melanoma B16F10-Nex2 cells were used as an affinitymatrix for selection of C7C random peptide phage library,and the melanocytic melan-A cell line was used forunspecific phage depletion as described in the “Materialsand methods” section. After three rounds of phagesubtraction/selection, a significant enrichment of boundphages was evident. Fifty individual phages obtained fromthe last bio-panning were amplified and titrated. Toevaluate the peptide bioactivity against melanoma cells,individual phages (108 pfu) were incubated with 5×102

B16F10-Nex2 cells in 96-well plates for 72 h at 37°C.Quantification of viable cells for each phage was performedusing MTT colorimetric assay. As shown in Fig. 1, seven ofthe 50 initially selected phages were able to significantlydecrease the number of viable cells. Peptide amino acidsequences of these seven phages were deduced after DNAsequencing and are shown on Table 1. All sequences exceptfor phage peptide 20 had one to three proline residues, andpeptides 38 and 44 showed the Pro-Asn-Met three-peptidemotif.

1258 J Mol Med (2010) 88:1255–1264

Screening of bioactive peptides for melanoma cell lineage

To investigate the peptide bioactivity to B16F10-Nex2cells, phage peptides were synthesized and tested in vitro ina concentration of 100 μM against the melanoma cells incomparison with the melan-A cell line. As shown in Fig. 2,all peptides, except peptide 34, were able to reduce thenumber of viable cells, but among them, only peptides 20,38, and 44 significantly decreased the viability of tumorcells when compared with the melan-A control.

In vivo antitumor activity of selected peptides

B16F10-Nex2 cells (5×104) were incubated for 10 min atroom temperature with peptides 20 and 38 at 50 μg, or thecorresponding scrambles peptides 20Sc, 38Sc, respectively,and were injected subcutaneously in the right flank ofC57bl/6 mice. Unexpectedly, peptide 38 did not affecttumor development. On the other hand, peptide 20 wassignificantly protective (p<0.05) with only one mouse in agroup of five developing tumors (Fig. 3a).

Using a therapeutic protocol, mice were treated with50 μg peptide 1 day after tumor challenge, and on days 3,5, and 7 with 25 μg of peptides 20 and 38. Results were

similar to the previous protocol in that peptide 38 did notaffect the tumor growth and peptide 20 caused a delay intumor development and protected 60% mice (*p<0.05),which remained tumor free (Fig. 3b). In both experiments,

Fig. 2 Selection of bioactive peptides against B16F10-Nex2 cells.Synthesized peptides (100 μM) were incubated for 48 h with themelanoma cells. The melanocytic cell line melan A was used as acontrol. From the seven previously selected peptides, three peptidesshowed inhibitory activity against murine melanoma cells. Cellviability was measured by MTT assay, and experiments wereperformed in quadruplicate and repeated at least three times (*p<0.05)

Fig. 3 In vivo activity of selected peptides in the subcutaneousmelanoma model. a About 5×104 melanoma cells were pre-incubatedfor 10 min with 50 μg of peptides and inoculated in the right flank ofeach mouse. b Mice challenged with B16F10-Nex 2 cells were treatedon the next day with 50 μg of selected peptides followed by threetreatments in alternating days with 25 μg of peptides by intraperito-neal injections. In both experiments, animals were sacrificed whentumors reached 3,000 mm3 or after 150 days after inoculation. Foreach experiment, five animals were used per group

Fig. 1 B16F10-Nex2 cell growth inhibition by individually selectedphage peptides. A 7C7 phage peptide library was used as control.Cells were incubated with the phage library (108 pfu) for 72 h. Thecell viability was measured by MTT assay. Experiments wereperformed in quintuplicate and repeated at least twice (*p<0.05)

Phage Sequence

5 CRPDPGSPC

19 CLPLWPGAC

20 CSSRTMHHC

34 CTRTPAKVC

38 CPNMFSTSC

44 CMSGPPNMC

47 CPNIRAPLC

Table 1 Selected phage librarycyclic peptide sequences (C-7X-C)

J Mol Med (2010) 88:1255–1264 1259

tumor-free mice were observed and sacrificed after 150 dayspost inoculation.

Predominant binding of peptide 20 as determinedby in vivo bio-panning

To determine the interaction of peptide 20 with melanomadeveloping in vivo, a protocol of in vivo bio-panning wasused. We performed the bio-panning in the mouse withapproximately 1 cm3-sized tumor using 1011 pfu of the C7Crandom phage library or the isolated phage expressing peptide20. In both experiments, we titered the number of elutedphages from lung, heart, liver, kidney, spleen, and tumor mass;the results were analyzed by the number of pfu by gram oftissue. Figure 4 shows the ratio of eluted phage 20 to the elutedrandom phages from each tissue. A discrete increase wasfound in the elution ratio from kidney, lung, liver, and spleen,while in the tumor mass, a threefold increase was observed. Atumor-binding preference for peptide 20 was therefore shown.

In vitro B16F10-Nex2 cell binding using labeled peptide 20

To confirm the binding of peptide 20 directly to B16F10-Nex2 melanoma cells, the peptide 20 was labeled with thefluorophore Abz. To visualize peptide cell binding byfluorescence microscopy, B16F10-Nex2 cells were culti-vated on 13-mm cover slips, fixed and treated as describedin the “Materials and methods” section (Fig. 5a–d).Glutaraldehyde-fixed cells (1×104/well) were incubatedovernight at 4°C with 100 μM of the labeled peptide 20or 20Sc, respectively. Fluorescence was read at 320 nm andemission at 420 nm (Fig. 5e). The labeled peptide 20 boundto melanoma cells while the labeled scramble peptide 20Scshowed only background fluorescence.

In vitro effect of peptide 20 against normal and tumor cell lines

We further evaluated the specificity of the peptide 20against normal cell lineages, melan A, HUVEC and 3T3,and tumoral cell lines HL-60, MDA, HeLa, B16-F10Nex2,A2058, SKmel25, and SKmel28. In Fig. 6, 100 μM ofpeptide 20 showed high inhibitory effect on all tumoral celllines in comparison with melan A (*p<0.05) except forHL-60. Human endothelial (HUVEC) and fibroblast (3T3)cells were poorly affected.

In vitro and in vivo invasion inhibition assays

To examine whether peptide 20 could inhibit melanoma cellinvasiveness, the in vitro test using a Matrigel-coated

Fig. 5 Labeled peptide 20 binding to B16F10-Nex2 cells. a, c Phasecontrast microscopy of tumor cells treated with labeled 20Sc orlabeled peptide 20, respectively; b no reaction with peptide 20Sc(100 μM); d binding of fluorescent peptide 20 (100 μM) tomelanoma. e Total arbitrary fluorescent units provided by theinteraction of labeled peptide with the cell measured by a fluorometerin triplicate (*p<0.05), 400× magnification. Scale bars 20 μm

Fig. 4 Prevalence of peptide 20 binding in developing tumors. Mice withtumors of approximately 1,000 mm3 were endovenously injected with1011 pfu of 7C7 phage library or phage peptide 20. Kidney, lung, heart,liver, and spleen organs were used as controls. The graphic represents thefold-increase of eluted phage 20 compared to the naive library. Elutedphages were titrated in duplicate with three serial dilutions

1260 J Mol Med (2010) 88:1255–1264

transwell was used. As observed in Fig. 7a, peptide 20 wasable to markedly inhibit cell invasion in vitro.

A metastatic model in vivo was then tested to evaluatethe activity of peptide 20. For this experiment, 5×105

viable cells were injected i.v. in each mouse which wastreated i.p. on the same day with 100 μg of peptide 20,20Sc, or PBS. After 13 days, animals were sacrificed, andthe number of metastatic colonies was counted. Peptide 20treatment significantly decreased the number of metastaticnodules whereas that with the 20Sc peptide had no effect(Fig. 7b). Figure 7c illustrates the recovered lungs frommice treated with peptide 20Sc or 20, respectively.

Clonogenic and cell adhesion assays

Clonogenic assays can identify substrate-independent anti-proliferative agents. The effect of peptide 20 at 100 μM wastested on 1×103 B16F10-Nex2 cells for 1 h followed byplating on soft agar. After 25 days, the number of colonieswhere counted comparing peptide 20 and 20Sc-treated cells.No differences were observed (data not shown) betweencontrol and treated cells. Action of peptide 20 in vivo couldthen involve a decrease in the tumor cell proliferation rateand/or interference in cell adherence and invasion capacity.We therefore investigated the possible inhibition of intercel-lular adhesion. Tumor cells (5×104) were incubated with100 μM of peptide 20 for 30 min at room temperature andplated on B16F10-Nex2 cell monolayer. After incubation at37°C, unbound cells were gently washed with PBS, and theremaining cells were counted in a Neubauer chamber.Results are expressed in percentage of adherent cells, the

control without peptides taken as 100%. As observed inFig. 8, peptide 20 highly inhibited intercellular adherence.

Peptide 20 binds to cadherin

Since peptide 20 inhibited cell-to-cell contact, we furtherinvestigated its binding to cadherin. To address this possibil-ity, total cadherin was immunoprecipitated from B16F10-Nex2 lysates with anti pan-cadherin mAb (that recognizesP-, N-, E- K-, M, and R-cadherins). Immunoprecipitatedcadherins were distributed on a white 96-well plate, blockedwith BSA, and incubated overnight with labeled peptide 20 orits scramble peptide. After several washes, arbitrary units

Fig. 7 Inhibition of tumor invasion by peptide 20. a Evaluation ofinhibition of tumor invasion by peptide 20, performed in vitro using atranswell coated with Matrigel. Values are percentages of cell invasionwhen incubated with 20 μM of peptide 20 or the scramble 20Scpeptide. The experiment was performed three times in duplicate. bNumbers of metastatic melanoma nodules in the lung in animalstreated with 100 μg of the peptides injected i.p. c Lungs treated withpeptide 20 or with the scramble 20Sc peptide. For each experiment,six animals were used per group (*p<0.05)

Fig. 6 Specificity of bioactive peptides for different cell lines. Normalcell lines melan A, HUVEC, fibroblast 3T3, and tumoral lineages HL-60, MDA, HeLa, B16F10-Nex 2, A2058, SKmel25, and SKmel28were incubated with 100 μM of the peptide 20 for 48 h. Experimentswere performed in quadruplicate at least twice; cell viability wasdetected by MTT assay (*p<0.05)

J Mol Med (2010) 88:1255–1264 1261

were measured in a fluorescent plate reader. Fluorescenceobtained with labeled peptide 20 was almost seven timesgreater than that with the labeled scramble peptide (Fig. 9),suggesting the specific binding of peptide 20 to cadherins.

Discussion

Phage display of peptides and proteins is a very powerfultool to identify specific ligands of cellular targets [26–28].Only a limited number of bioactive peptides obtained fromphage display libraries were found to be active againstmelanoma cells. Recently, a peptide selected from a 12-merphage library targeting ADAM15 disintegrin was shown to

inhibit melanoma cell adhesion [17], whereas anotherpeptide inhibited B16F10 cell adhesion to hyaluran [29].Another group focused on a heptapeptide obtained from aphage display library that bound to human tumor melaninand could act as a potential vehicle for targeting radionu-clide therapy to metastatic melanoma [16].

In the present work, we describe the isolation of abioactive peptide with antitumor activity after three roundsof a subtraction/selection protocol. A C7C phage librarywas chosen because short peptides are expected to haveenhanced diffusion in solid tumors and little chance ofeliciting an immune response in the cyclic form which alsohelps them to have increased stability in vivo, withoutconformational changes and with increased resistance toproteolysis [4, 14]. On starting with 50 isolated phages,only seven showed antiproliferative effects on B16F10-Nex2 cells. The corresponding peptides were synthesizedand tested in vitro against murine melanoma cells. Allpeptide sequences except that of peptide 20 had one to threeproline residues, and peptides 38 and 44 contained the tri-peptide motif, PNM. The three peptide sequences, phage 20[-CSSRTMHHC-], phage 38 [-CPNMFSTSC-], and phage44 [-CMSGPPNMC-], showed high activity against mela-noma cells. As peptide 38 and 44 share the same tri-peptidemotif and sequence 38 was more effective in vitro againstB16F10-Nex2 cells, we choose to continue our experimentswith peptides 20 and 38.

Peptide 38 showed a good antitumor activity in vitro butwas inactive in vivo. That could be explained by rapiddegradation or inability to reach the tumor site. Peptide 20not only delayed tumor growth but caused total inhibitionof tumor development in most animals challenged withmelanoma cells. This result was unexpected because thepeptide was not cytotoxic in clonogenic assays. In vivo,however, peptide 20 could inhibit the tumor cell-to-cellcontact, hence interfering in tumor cell establishment. It isimportant to note that HL-60 lineage was insensitive topeptide 20; this cell line grows in suspension and may lackadhesion molecules. The peptide was also able to markedlyinhibit tumor cell invasion in vitro and in vivo. A peptidewith similar property which partially inhibited gastriccancer cells adherence to collagen type IV was able toabolish liver metastasis in vivo [8].

As peptide 20 inhibits cell-to-cell adhesion, we furtherinvestigated that it could bind to a cadherin molecule.Studies of cadherin distribution in cancer cell metastasissuggest different roles for different molecular species. Lossof E-cadherin is important for metastasis progression [30],whereas overexpression of other cadherin types is importantfor tumor development. Alterations, as by ligand bindingand molecular switching of cadherins, have been associatedwith invasion and metastasis in cancer [31]. Recently, itwas shown in hepatoma cells that cadherin-17 is upregu-

Fig. 9 Binding of labeled peptide 20 to cadherin. White 96-well platewas coated overnight at 4°C with the immunoprecipited cadherin(104 g/mL), blocked with PBS containing 1% BSA for 4 h at roomtemperature, and incubated with 100 μM of the fluorescent peptide 20or its scramble overnight at 4°C. After several washes, arbitrary unitswere measured in a fluorescent plate reader. Specific arbitraryfluorescent units for peptide 20 and its scramble were determined bysubtracting the fluorescence obtained in coated cadherin immunopre-cipited from coated BSA wells. The experiment was performed intriplicates (*p<0.05)

Fig. 8 Cell adhesion assay. About 5×104 cells incubated with100 μM of peptides were added on a B16F10-Nex 2 monolayer for1 h at 37°C. After gently washing, attached cells were counted on aNeubauer chamber. Values are percentages of attached cells where100% attachment was considered for untreated cells. The experimentwas repeated twice in triplicates (*p<0.05)

1262 J Mol Med (2010) 88:1255–1264

lated in human liver cancers and that downregulation byinterference RNA was able to partially inhibit tumor cellproliferation through Wnt signaling and activation of tumorsuppressor genes in vitro and in vivo [32]. In another work,inhibition of cadherin 22 also partially blocked proliferationof colorectal cancer cells in vitro and in vivo and inhibitedtumor metastasis [33]. All these properties, antiprolifera-tive, antitumor, and anti-metastatic, were observed in cellstreated with peptide 20, seemingly in support of a cadherinbinding hypothesis.

In conclusion, we identified an anti-melanoma bioactivepeptide by phage display using whole tumor cells as matrix.The short cyclic peptide [-CSSRTMHHC-] was preferentiallyactive in vitro against tumoral lineages, including human celllines except for HL-60. Peptide 20 was highly effective inprotecting against subcutaneous melanoma challenge with60% survival and markedly reduced the number of melanomanodules in a lung metastatic model using a single dose. Aspeptide 20 inhibits cell-to-cell adhesion, we showed that itcould bind to a cadherin constituent at the tumor cell surface.Here, we have shown that phage display peptide selection is avery fast tool to identify bioactive ligands, particularly againstfast growing tumors.

Acknowledgments The present work was supported by Fundação deAmparo à Pesquisa do Estado de São Paulo (FAPESP) and the ConselhoNacional de Desenvolvimento Científico e Tecnológico (CNPq).

Disclosure of potential conflict of interests The authors declare noconflict of interests related to this study.

References

1. Miller AJ, Mihm MC (2006) Mechanisms of disease—melanoma.N Engl J Med 355:51–65

2. Ruoslahti E (2002) Drug targeting to specific vascular sites. DrugDiscov Today 7:1138–1143

3. St Croix B, Rago C, Velculescu V, Traverso G, Romans KE,Montgomery E, Lal A, Riggins GJ, Lengauer C, Vogelstein B,Kinzler KW (2000) Genes expressed in human tumor endotheli-um. Science 289:1197–1202

4. Shadidi M, Sioud M (2002) Identification of novel carrierpeptides for the specific delivery of therapeutics into cancer cells.Faseb J 16:256–258

5. Zhang BH, Zhang YQ, Wang JW, Zhang YD, Chen JJ, Pan YF,Ren LF, Hu ZY, Zhao JF, Liao MM, Wang SW (2007) Screeningand identification of a targeting peptide to hepatocarcinoma froma phage display peptide library. Mol Med 13:246–254

6. Kolonin MG, Bover L, Sun J, Zurita AJ, Do KA, Lahdenranta J,Cardo-Vila M, Giordano RJ, Jaalouk DE, Ozawa MG, Moya CA,Souza GR, Staquicini FI, Kunyiasu A, Scudiero DA, Holbeck SL,Sausville EA, Arap W, Pasqualini R (2006) Ligand-directedsurface profiling of human cancer cells with combinatorial peptidelibraries. Cancer Res 66:34–40

7. Li ZH, Zhao RJ, Wu XH, Sun Y, Yao M, Li JJ, Xu YH, Gu JR(2005) Identification and characterization of a novel peptideligand of epidermal growth factor receptor for targeted delivery oftherapeutics. Faseb J 19:1978–1985

8. Hu SJ, Guo XN, Xie HH, Du YL, Pan YL, Shi YQ, Wang J, HongL, Han S, Zhang DT, Huang DW, Zhang KD, Bai FH, Jiang HP,Zhai HH, Nie YZ, Wu KC, Fan DM (2006) Phage displayselection of peptides that inhibit metastasis ability of gastriccancer cells with high liver-metastatic potential. Biochem BiophysRes Commun 341:964–972

9. Pasqualini R, Arap W (2002) Translation of vascular diversity intotargeted therapeutics. Ann Hematol 81:S66–S67

10. Zhang JB, Spring H, Schwab M (2001) Neuroblastoma tumorcell-binding peptides identified through random peptide phagedisplay. Cancer Lett 171:153–164

11. Hong FD, Clayman GL (2000) Isolation of a peptide for targeteddrug delivery into human head and neck solid tumors. Cancer Res60:6551–6556

12. Kay BK, Kurakin AV, Hyde-DeRuyscher R (1998) From peptidesto drugs via phage display. Drug Discov Today 3:370–378

13. Lee JH, Engler JA, Collawn JF, Moore BA (2001) Receptormediated uptake of peptides that bind the human transferrinreceptor. Eur J Biochem 268:2004–2012

14. Ladner RC, Sato AK, Gorzelany J, de Souza M (2004) Phagedisplay-derived peptides as a therapeutic alternatives to anti-bodies. Drug Discov Today 9:525–529

15. Eriksson F, Culp WD, Massey R, Egevad L, Garland D, PerssonMAA, Pisa P (2007) Tumor specific phage particles promotetumor regression in a mouse melanoma model. Cancer ImmunolImmunother 56:677–687

16. Howell RC, Revskaya E, Pazo V, Nosanchuk JD, Casadevall A,Dadachova E (2007) Phage display library derived peptides that bindto human tumormelanin as potential vehicles for targeted radionuclidetherapy of metastatic melanoma. Bioconjug Chem 18:1739–1748

17. Wu J, Wu MC, Zhang LF, Lei JY, Feng L, Jin J (2009)Identification of binding peptides of the ADAM15 disintegrindomain using phage display. J Biosci 34:213–220

18. Paschoalin T, Carmona AK, Rodrigues EG, Oliveira V, MonteiroHP, Juliano MA, Juliano L, Travassos LR (2007) Characterizationof thimet oligopeptidase and neurolysin activities in B16F10-Nex2 tumor cells and their involvement in angiogenesis and tumorgrowth. Mol Cancer 6:44–58

19. Hirata IY, Cezari MHS, Nakaie C (1994) Internally quenchedfluorogenic peptide substrates: solid phase, synthesis and fluores-cence spectroscopy of peptides containing orthoaminobenzoil/dinitrophenyl groups as donor acceptor-pairs. Lett Peptide Sci1:299–308

20. Dobroff AS, Rodrigues EG, Moraes JZ, Travassos LR (2002)Protective, anti-tumor monoclonal antibody recognizes a confor-mational epitope similar to melibiose at the surface of invasivemurine melanoma cells. Hybrid Hybridomics 21:321–331

21. Pasqualini R, Ruoslahti E (1996) Organ targeting in vivo usingphage display peptide libraries. Nature 380:364–366

22. Kennel SJ, Mirzadeh S, Hurst GB, Foote LJ, Lankford TK,Glowienka KA, Chappell LL, Kelso JR, Davern SM, Safavy A,Brechbiel MW (2000) Labeling and distribution of linear peptidesidentified using in vivo phage display selection for tumors. NuclMed Biol 27:815–825

23. Oliveira V, Campos M, Hemerly JP, Ferro ES, Camargo ACM,Juliano MA, Juliano L (2001) Selective neurotensin-derivedinternally quenched fluorogenic substrates for neurolysin (EC3.4.24.16): comparison with thimet oligopeptidase (EC 3.4.24.15)and neprilysin (EC 3.4.24.11). Anal Biochem 292:257–265

24. Guimaraes-Ferreira CA, Rodrigues EG, Mortaray RA, Cabralz H,Serrano FA, Ribeiro-dos-Santos R, Travassos LR (2007) Anti-tumor effects in vitro and in vivo and mechanisms of protectionagainst melanoma B16F10-Nex2 cells by fastuosain, a cysteineproteinase from Bromelia fastuosa. Neoplasia 9:723–733

25. Franken NAP, Rodermond HM, Stap J, Haveman J, van Bree C(2006) Clonogenic assay of cells in vitro. Nat Protoc 1:2315–2319

J Mol Med (2010) 88:1255–1264 1263

26. Scott JK, Smith GP (1990) Searching for peptide ligands with anepitope library. Science 249:386–390

27. Sioud M, Forre O, Dybwad A (1996) Selection of ligands forpolyclonal antibodies from random peptide libraries: potentialidentification of (auto)antigens that may trigger B and T cellresponses in autoimmune diseases. Clin Immunol and Immuno-pathol 79:105–114

28. Kolonin M, Pasqualini R, Arap W (2001) Molecular addresses inblood vessels as targets for therapy. Curr Opin ChemBiol 5:308–313

29. Mummert ME, Mummert DI, Ellinger L, Takashima A (2003)Functional roles of hyaluronan in B16-F10 melanoma growth andexperimental metastasis in mice. Mol Cancer Ther 2:295–300

30. Agiostratidou G, Li MM, Suyama K, Badano I, Keren R, ChungS, Anzovino A, Hulit J, Qian BZ, Bouzahzah B, Eugenin E,

Loudig O, Phillips GR, Locker J, Hazan RB (2009) Loss of retinalcadherin facilitates mammary tumor progression and metastasis.Cancer Res 69:5030–5038

31. Takeichi M, Watabe M, Shibamoto S, Ito F, Oda H, Uemura T,Shimamura K (1993) Dynamic control of cell–cell adhesion formulticellular organization. C R Acad Sci Serie III 316:818–821

32. Liu LX, Lee NP, Chan VW, Xue W, Zender L, Zhang CS, Mao M,Dai HY, Wang XL, Xu MZ, Lee TK, Ng IO, Chen YC, Kung HF,Lowe SW, Poon RTP, Wang JH, Luk JM (2009) Targetingcadherin-17 inactivates Wnt signaling and inhibits tumor growthin liver carcinoma. Hepatology 50:1453–1463

33. Zhou J, Li JM, Chen JZ, Liu YH, Gao WZ, Ding YQ (2009)Over-expression of CDH22 is associated with tumor progressionin colorectal cancer. Tumor Biol 30:130–140

1264 J Mol Med (2010) 88:1255–1264