screening and identification of a novel target specific ... · three-dimensiona stl ructurapr l...

Acta Biochim Biophys Sin (2008) | Volume 40 | Issue 5 | Page 443

Screening and identification of a novel target specific for HepG2Acta Biochim Biophys Sin (2008): 443-451 | © 2008 Institute of Biochemistry and Cell Biology, SIBS, CAS | All Rights Reserved 1672-9145http://www.abbs.info; www.blackwellpublishing.com/abbs | DOI: 10.1111/j.1745-7270.2008.00412.x

Screening and identification of a novel target specific for hepatoma cell lineHepG2 from the FliTrx bacterial peptide library

Wenhan Li1,2#, Ping Lei1#, Bing Yu1, Sha Wu1, Jilin Peng1, Xiaoping Zhao1, Huifen Zhu1, Michael Kirschfink2, andGuanxin Shen1*1 Department of Immunology, Tongji Medical College, Huazhong University of Science and Technology, Wuhan 430030, China2 Institute of Immunology, University of Heidelberg, 69120 Heidelberg, Germany

Received: December 15, 2007 Accepted: March 5, 2008This work was supported by the grants from the Hi-tech Research andDevelopment Program of China (No. 2006AA02Z158) and the ScienceFoundation of the Ministry of Education of China (No. 20060487024)# These authors contributed equally to this work*Corresponding author: Tel, 86-27-83692611; E-mail, [email protected]

To explore new targets for hepatoma research, we used asurface display library to screen novel tumor cell-specificpeptides. The bacterial FliTrx system was screened with livingnormal liver cell line L02 and hepatoma cell line HepG2successively to search for hepatoma-specific peptides. Threeclones (Hep1, Hep2, and Hep3) were identified to be specificto HepG2 compared with L02 and other cancer cell lines.Three-dimensional structural prediction proved that peptidesinserted into the active site of Escherichia coli thioredoxin(TrxA) formed certain loop structures protruding out of thesurface. Western blot analysis showed that FliC/TrxA-peptidefusion proteins could be directly used to detect HepG2 cells.Three different FliC/TrxA-peptide fusion proteins targetedthe same molecule, at approximately 140 kDa, on HepG2 cells.This work presented for the first time the application of theFliTrx library in screening living cells. Three peptides wereobtained that could be potential candidates for targeted livercancer therapy.

Keywords tumor target; hepatoma; bacterial display;FliTrx system; counter-screening; living-cell panning; 3-Dmodeling

Estimates from the year 2000 indicate that liver cancerremains the fifth most common malignancy in men andthe eighth most common in women worldwide, burdenedby a constantly increasing frequency. This tumoraccounted for 5.6% of all human cancers (7.5% among

men and 3.5% among women) [1,2]. The main obstaclesto improving control and treatment of hepatoma are thelack of biomarkers for early diagnosis [3] and selectivedelivery of chemotherapeutic drugs into the tumor cells invivo [4].

Phage display technology is a very powerful tool forthe identification of critical amino acids responsible forprotein-protein interaction and the discovery of newtherapeutic targets [5,6]. In the current study, we chose abacterial peptide library, the FliTrx Escherichia colithioredoxin (TrxA) scaffold random peptide library, whichhas been widely used to identify conformationallyconstrained dodecamer peptide sequences specificallyrecognized by soluble proteins, such as monoclonalantibodies [7,8]. The FliTrx peptide library has an estimateddiversity of 1.77×108 different random dodecamer loopsequences and does not contain any predefined structuralmotif. Thus, novel peptide sequences could be identifiedin this library that mimic the functions of protein domainsfound in nature [9]. In this flagella display library, peptidesare directly displayed on the surface of E. coli fused withtwo proteins, the major bacterial flagellar protein (FliC)and TrxA. Various dodecapeptides are inserted within theactive loop of TrxA, and the fusion protein forms a stableprotrusion from the bacterial cell surface with the help ofbacterial flagella [10,11]. We wanted to identify FliTrxpeptides specifically targeting tumor cells, not only toinvestigate target molecules, but also to deliver drugs intomalignant tissues.

We identified three FliTrx peptides from the bacterialrandom library that are specific for hepatoma cell lineHepG2 and do not bind to other cancer cell lines or normalliver cells. Results of structure prediction suggested thatTrxA-peptide fusion proteins based on the TrxA activesite might be more valuable than linear peptides themselves.Immune blotting results further proved that FliC/TrxA-


Screening and identification of a novel target specific for HepG2

peptide fusion proteins could be used to detect HepG2cells and indicate a novel target molecule on these cells.

Materials and Methods

Cell cultureHuman hepatoma cell line HepG2 and human liver cell lineL02 (both cell lines were maintained in our laboratory)were cultured in RPMI 1640 (Gibco, Carlsbad, USA)supplemented with 10% (V/V) fetal bovine serum (Gibco)at 37 ºC in a 5% CO2 incubator.

Selection of HepG2 cell-binding peptidesThe FliTrx random peptide display library (Invitrogen,Carlsbad, USA) containing 1.77×108 primary clones wasused to identify specific HepG2 cell-binding peptidesincorporating negative selection using the normal liver cellline L02.

Bacteria were treated according to the manufacturer’sinstructions. The bacterial cells were diluted 10-fold inIMC medium (1× M9 salts, 0.2% casamino acids, 0.5%glucose, 1 mM MgCl2) (Invitrogen) with 100 μg/mlampicillin and cultured with shaking (225−250 rpm) at 25ºC for 15 h.

The expression of the bacteria was induced by adding 3ml of the overnight culture (approximately 1×1010 cells) to50 ml IMC medium containing 100 μg/ml ampicillin and100 μg/ml tryptophan (Invitrogen) and the mixture wascultured with shaking for 6 h. Biopanning was carried outafter induction of the bacterial library with tryptophan,which activated the transcription of the FliC/TrxA peptidefusion proteins.

A confluent dish with L02 cells was first blocked withIMC medium containing 1% bovine serum albumin (BSA)and 1% α-methyl mannoside (Sigma-Aldrich, St. Louis,USA) at 37 ºC for 1 h. Blocked cells were incubated with10 ml induced bacterial suspension in IMC mediumcontaining 1% BSA and 1% α-methyl mannoside at roomtemperature for 1 h. Unbound bacteria were transferredand incubated in another dish with blocked L02 cells atroom temperature for 1 h. Then the unbound bacteria weretransferred to the blocked HepG2 cells and incubated foranother 1 h. Dishes with HepG2 cells were washed withIMC medium containing 1% α-methyl mannoside for 5min and the wash was repeated four more times. Bindingbacteria were obtained by stirring in a vortex for 30 s andsubsequently amplified in IMC medium containingampicillin (100 μg/ml) overnight for the next selectionround. Recovered bacteria were counted by plating onRMG plates (1× M9 salts, 2% casamino acids, 0.5%

glucose, 1 mM MgCl2, and 1.5% agar) (Invitrogen)supplemented with 100 μg/ml ampicillin. After five roundsof selection, individual clones were isolated and analyzed.

Clone polymerase chain reaction (PCR)Individual clones were picked and lysed in 20 μl of 0.1%Triton X-100 and the lysate was used as a template.Reactions were carried out with FliTrx forward sequencingprimer (5'-ATTCACCTGACTGACGAC-3'), FliTrx reversesequencing primer (5'-CCCTGATATTCGTCAGCG-3')(synthesized by BioAsia, Shanghai, China), and Taq DNApolymerase (MBI Fermentas, St. Leon-Roth, Germany) ina volume of 20 μl for 2 min predenaturing at 94 ºC and 30cycles of 1 min denaturing at 94 ºC, 1 min annealing at 55ºC, and 1 min extension at 72 ºC. PCR products wereanalyzed by electrophoresis on 1% agarose gel. Clonescontaining inserts of 170 bp were selected for furtheranalysis.

Flow cytometryIndividual clones were further screened for binding withHepG2 by flow cytometry, using L02 as negative control.Cells (1×105 cells/well) were removed from the dish withtrypsin, washed once with phosphate-buffered saline(PBS)/1% BSA, and incubated with 50 μl induced bacteria(1012 c.f.u./ml in PBS/1% BSA and 1% α-methylmannoside) for 1 h on ice. Cells were washed twice withPBS/1% BSA and incubated with mouse Anti-Thio antibody(1:500 diluted) (TrxA; Invitrogen) for 30 min on ice. Cellswere washed twice with PBS/1% BSA and incubated withfluorescein-isothiocyanate-conjugated anti-mouse immuno-globulin G (1:500 diluted) (Invitrogen) for 30 min on ice.Then the cells were washed twice with PBS/1% BSA andanalyzed using FACSCalibur (Becton Dickinson, Oxford,UK). Clones that only bound HepG2, not L02, were selectedfor sequencing.

DNA sequencing and homologyIndividual clones were isolated and sequenced by BioAsia(Shanghai) with forward sequencing primer 5'-ATTCACC-TGACTGACGAC-3' and reverse sequencing primer 5'-CCCTGATATTCGTCAGCG-3'. Sequences of dodecapep-tides were submitted to the PubMed (http://www.ncbi.nlm.nih.gov/blast/Blast.cgi), SwissProt (http://expasy.org/tools/blast/), and European Molecular Biology Laboratory(EMBL) (http://www.ebi.ac.uk/blast2/index.html)databases for homology analysis.

Characterization of specificity with flow cytometryTo characterize specificity, individual clones from the last



screening round were analyzed using flow cytometry forbinding with the following cancer cell lines: SLC (squamouslung carcinoma), MKN28 (gastric adenocarcinoma),K562 (chronic myelogenous leukemia), HeLa (cervixadenocarcinoma), MCF7 (mammary gland adenocarcinoma),HL60 (peripheral blood acute promyelocytic leukemia), andSMMC7721 (hepatocellular carcinoma). Similar techniqueswere used as above described.

Structure predictionThe sequences of selected peptides were analyzed withExPASy proteomics and sequence analysis tools (http://expasy.org/tools/). Secondary structure prediction wascarried out by AGADIR (http://www.embl-heidelberg.de/Services/serrano/agadir/agadir-start.html). Information onTrxA-peptide fusion proteins was sent to the 3D-JIGSAWcompa r a t i ve mod e l i n g s e r ve r ( h t t p : / / b mm.cancerresearchuk.org/~3djigsaw/). VMD 1.8.5 (http://www.ks.uiuc.edu/Research/vmd/vmd-1.8.5/) was used todisplay and analyze the results of the modeling of fusionregions.

Expression analysis of FliC/TrxA-peptide fusionprotein and Western blot analysisAfter being induced with 100 μg/ml tryptophan underoptimized conditions for 6 h, the cultured bacteria werecollected and the flagellin extracted as described by Ibrahimet al [12], with some modifications. Briefly, the flagellinprotein was extracted by exposure of the bacterial cells to1 N hydrochloric acid at pH 2.0 for 20 min. Cellular debriswas then separated by centrifugation at 1500 g for 15 minand the flagellin protein in the supernatant was collected.The pH of this supernatant solution was adjusted to 7.2with 1 N sodium hydroxide before incubation with thepolyvinylidene fluoride membrane.

HepG2, L02, and SLC cells were lysed in the sodiumdodecyl sulfate sample buffer; proteins were separatedon 8% polyacrylamide gels and transferred topolyvinylidene fluoride membranes. Blots were incubatedwith 10 μg/ml FliC/TrxA-peptide fusion proteins(supernatant as mentioned above in PBS/0.2% Tween-20/5% skimmed milk) overnight at 4 ºC and washedwith PBS/0.2% Tween-20. Membranes were detectedusing mouse Anti-Thio antibody (1:5000 diluted) andhorseradish peroxidase-conjugated goat anti-mouseimmunoglobulin G (1:2000 diluted). The final detectionmethod is to add the horseradish peroxidase substratesolution containing DAB (3,3'-diaminobenzidine) (1.5mg/ml) and watch for the appearance of coloured bandon the nitrocellulose.

Results

BioscreeningHepG2 binding-specific peptides were screened from theFliTrx random peptide library containing 1.77×108 primaryclones. Induced bacterial display peptides were counterselected with normal liver cell line L02 to deplete healthytissue-binding peptides. After incubation with L02 cells,unbound bacteria were obtained and panned with HepG2cells. Five rounds of selection were carried out and theefficiency of selection was monitored by comparing thenumber of recovered bacteria in each round. An ever-in-creasing ratio of output/input (Table 1) indicated that thepositive clones specific for HepG2 were fully enriched,especially in the last round.

Selection Bacteria input Bacteria output Recovery rateround (c.f.u.) (c.f.u.) (output/input)

1 1.5×1010 1.8×104 1.2×10−6

2 5.0×1010 1.2×105 2.4×10−6

3 5.8×1010 1.0×105 1.7×10−6

4 8.9×1010 1.8×105 2.0×10−6

5 4.0×1010 1.9×106 4.8×10−5

Table 1 Recovery rate (output/input) of peptides during biopanning

After five rounds of biopanning, the ratio of output/input increasedfrom 1.2×10−6 to 4.8×10−5. The specific clones were successfully en-riched during subtract bioscreening.

After five subtract-screening rounds, 700 clones werepicked from the recovered bacteria in the last round forbacterial PCR to select clones containing whole fusionprotein. Approximately 28% (200/700) of the selectedclones containing the full-sized insert of 170 bp (Fig. 1)

Fig. 1 Selection of clones by polymerase chain reaction (PCR)1, sequence-known clone from peptide library as the positive control;2, HB2151 Escherichia coli as the negative control; 3−7, PCR productsof individual clones; M, DNA marker.

Acta Biochim Biophys Sin (2008) | Volume 40 | Issue 5 | 46


were selected for further investigation. Each of the clonesand the corresponding exogenous sequence was given asequential name from 1 to 200.

Confirmation of in vitro binding by flow cytometryThe binding ability of individual clones was verified byflow cytometry. To identify the specificity of selectedclones for HepG2, the cell line L02 served as the control.Ten of the 200 peptides had clearly higher binding activitywith HepG2 than with L02 (Fig. 2). They were chosen tobe sequenced.

No. Sequences

11/16/76 IAVAPGWLWEEE12 KELCELDSLLRI19 TRGRPRDVANGH34/60 IRELYSYDDDFG106 ESLSVDFMGERA109 QELAPYSWSEKD173 ARRILKGGGVHT

Table 2 Sequences and frequency of peptide inserts isolatedfrom biopanning

The inserts of clones 11, 16, and 76 are identical. Clones 34 and 60 arerepeated sequences. The amino acids highlighted in grey are high-fre-quency motifs.

Fig. 3 Binding of positive clones to different carcinoma celllines HeLa, cervix adenocarcinoma; HL60, peripheral blood acutepromyelocytic leukemia; K562, chronic myelogenous leukemia; MCF7,mammary gland adenocarcinoma; MKN28, gastric adenocarcinoma;and SLC, squamous lung carcinoma.

Fig. 2 Binding of positive clones to hepatoma HepG2 cellscompared with normal liver L02 cells Black lines are the isotypecontrol. Overlayed colorful lines show the different affinities of 10selected individual clones to HepG2 (left) and L02 (right). FITC,fluorescein-isothiocyanate.

Sequence and homology analysisAmong the 10 selected clones, seven peptide sequenceswere identified due to repeated events (Table 2). Threeclones, 11, 16, and 76, included the same dodecamerpeptide sequence. Two clones, 34 and 60, were totallyrepeated sequences. Analysis of the amino acid sequencessuggested that individual clones might share someconsensus groups at an identical position, such as VA, EL,RI, AP, S-S, L-E, and W-E. In other words, there were

some high-frequency amino acids. But these sequenceswere different from hepatocellular antigens or any otherpeptide or protein sequence available, as confirmed byBLAST in various protein databases such as SwissProt,PubMed, and the EMBL databases.

Characterization of specificity of peptidesAdditional flow cytometry analysis was carried out tofurther determine the binding specificity of these sevenpeptides with tumor cell lines SLC, MKN28, K562, Hela,MCF7, and HL60. Three clones (11, 12, and 34) showedobviously lower binding activity with other tumor cell linesthan with HepG2 (Fig. 3). We named these three peptidesHep1 (11), Hep2 (12), and Hep3 (34). Furthermore, weidentified their binding specificity with hepatocellularcarcinoma cell SMMC7721. Results showed that they



Fig. 4 Binding of clones Hep1, Hep2, and Hep3 to hepatocellular carcinoma SMMC7721 cells compared with HepG2

could also bind to SMMC7221 cells, but the affinity wassignificantly lower compared with HepG2 (Fig. 4).

Prediction of the structure of TrxA-peptide fusionproteinThe amino acid sequences of Hep1, Hep2, and Hep3 were

submitted to AGADIR and the results revealed that bothHep1 and Hep2 peptides had a very high helical potential[Fig. 5(A−C)]. This, in turn, implies that the dodecamerpeptides might adopt a coiled-coil structure. As Hep3peptide showed a very low helical potential, it was selectedas a control for further structural and recognition

Fig. 5 Structure prediction of peptides and Escherichia coli thioredoxin (TrxA)-peptide fusion proteins Secondary structure predictionof peptides Hep1 (A), Hep2 (B), and Hep3 (C) using the AGADIR method (http://www.embl-heidelberg.de/Services/serrano/agadir/agadir-start.html).The AGADIR prediction is expressed as helicity per residue along the peptide sequence. Three-dimensional structural models of TrxA-peptide fusionproteins TrxA-Hep1 (D), TrxA-Hep2 (E), and TrxA-Hep3 (F), developed by the 3D-JIGSAW server (http://bmm.cancerresearchuk.org/~3djigsaw/).The extended peptides are circled in the models.

(A) (B) (C)

(D) (E) (F)



investigation.As dodecapeptides were inserted within the active loop

of TrxA, we submitted the sequences of TrxA-peptidefusion proteins to the 3D-JIGSAW server to furtherdevelop the TrxA-peptide structural model. The overallstructure is shown in Fig. 5(D,E). The fusion proteinconsisted of two regions; TrxA had a central pleated β-sheet with five parallel and antiparallel β-strands,surrounded by three α-helices. Dodecapeptides formed astable protrusion from the active site of TrxA, located atthe beginning of a long α-helix. Hep1 formed a big loop[Fig. 5(D)], Hep2 was displayed as a β-sheet combinedwith one β-fold of TrxA overhanging from the TrxA region[Fig. 5(E)], and Hep3 stuck out from the TrxA region[Fig. 5(F)], forming two β-sheets by itself. The modelshowed that even Hep3 might not be a simple linearstructure in the fusion protein.

Comparison with the secondary structure analysisshowed that only the structure of Hep3 in the fusion pro-tein was affected by TrxA. Hep1 and Hep2 were alsoconstrained by TrxA, but the tertiary structure was simi-lar to the secondary structure and not influenced by thefusion protein.

Identification of a novel target on HepG2 cellsThe flagellin protein was extracted for Western blot assay.The Western blot analysis revealed that: (1) FliC/TrxA-peptide fusion proteins were approximately 65 kDa; (2)no binding occured on L02 cell lysate or SLC cell lysatewith FliC/TrxA-peptide fusion proteins; and (3) threeFliC/TrxA-peptide fusion proteins bound to the same

fragment of approximately 140 kDa of the HepG2 celllysate (Fig. 6).

Discussion

Targeting specific binding ligands on tumor cells is anefficient way to improve early cancer diagnosis schemesand therapy. It has been reported that cancer cells oftendisplay high levels of certain cell surface molecules, suchas tumor-associated antigens or tumor-specific antigens,that are sparse on normal tissues, representing potentialsites for delivery of toxic agents [13,14]. Therefore, wecarried out subtract-screening based on the FliTrx bacterialpeptide library to search for HepG2 cell-specific peptides.As HepG2 is a human hepatoma cell line derived fromhepatic neoplasia, which still retains numerous cellularfunctions typical of differentiated, normal hepatocytes(such as synthesis of albumin, transferring, α1-antitrypsin,fibrinogen, and certain other coagulation factors), it iswidely accepted as a valuable and informative modelsystem for studying human hepatocyte function [15,16].

Phage display libraries have been applied in ligandselections using cultured cells, animals [17], or humanpatients [18], yielding peptides specific for cell surfacemarkers [19]. The principal advantage of display librarymethodologies is that selection can be carried out in a native-like, membrane-bound environment without priorknowledge of the target cell receptors [20]. Peptide ligandsgenerated using this method have been proved useful intumor diagnosis and target therapy [21,22]. The FliTrxsystem has been widely used as a novel bacterial displayed

Fig. 6 Western blot analysis of TrxA-peptide fusion protein (A) The membrane was incubated with lysate of one sequence-knownEscherichia coli thioredoxin (TrxA)-fusion protein. 1, induced bacteria lysate (TrxA proteins are approximately 65 kDa); 2, squamous lungcarcinoma cell line SLC cell lysate; 3, protein marker; 4, normal liver cell line L02 cell lysate; 5, hepatoma cell line HepG2 cell lysate. (B−D)Membranes were incubated with lysate of three TrxA proteins. A 140 kDa band was detected by all three FliC/TrxA-peptide fusion proteins.



peptide library for studying protein-protein interactionsinvolved in receptor-ligand binding, enzyme-substratespecificity, and antibody-antigen recognition [8,23].

Cell surface proteins are frequently post-translationallymodified, including malignancy-specific modifications.Thus, peptides selected against purified recombinant proteinmight not be able to access their targets on living cells[24,25]. Here, for the first time, we described successfulliving cell biopanning with this particular bacterial library.The data proved that positive clones specific for HepG2were fully enriched after subtract screening and threeHepG2-specific clones were selected from the peptidelibrary.

Although it is possible to screen peptide libraries to selectpeptidic ligands that bind to previously uncharacterizedmolecules on the cell surface, a problem in this approachis the risk of selecting clones that bind a whole populationof cell surface molecules, not the target molecule alone.To enhance the selection of specific ligands from the phage-displayed peptide library, subtractive approaches have beenapplied to eliminate potential selection of clones that bindto irrelevant surface molecules [26−28]. In this case, wechose the normal liver cell line L02 as the negative controland after five subtract-screening rounds we obtained 200clones containing the full-sized fusion from 700 randomclones. Of these, 10 clones showed obviously higheraffinity with HepG2 than with L02. Finally, seven differentsequences were identified on account of repeated cases.We carried out not first sequencing of all random clonespicked up from recovered bacteria, but bacterial PCR foridentification of correct construction of fusion regions afterbioscreening. The results proved that, with even the obviousenrichment of bacteria specifically binding to HepG2, only28% of clones containing full-size inserts could be usedfor further binding ability assay. And only 5% of the clonesappeared to bind more effectively than the others, and wereable to bind specifically to hepatoma cells. Such a lowpositive rate was not what we aimed for and the evidencerevealed the meaning of identification of binding ability ofselected clones for deletion of non-specific peptides andavoidance of blind random sequencing.

The analysis of amino acid sequences in clones initiallyindicated some consensus groups, such as VA, EL, RI,AP, S-S, L-E, and W-E, shared by individual clones.However, after subsequent sequenced BLASTS in variousprotein databases, including SwissProt, PubMed, andEMBL databases, these peptides turned out to be differentfrom any hepatocellular-specific antigens or any otherpeptide or protein sequences available. Further flowcytometry assays distinguished three clones that showed

obviously higher affinity with HepG2 cells compared withother tumor cell lines. They were named Hep1, Hep2, andHep3, and they could also bind to another liver carcinomacell line SMMC7721, although the affinity was lowercompared with HepG2.

Screening of protein databases allows determination ofproteins with homologies to selected peptide sequences.In that event, the matched proteins are biologically relevant(can theoretically serve as ligands for the target molecules).But, in our experiment, the consensus sequence of theselected peptides did not bear any obvious similarity to thesequence of any characterized protein in the databases. Itis very possible that the selected peptides mimic a complexbinding epitope of potential ligands and, thus, cannot befound in databases [29]. Therefore, we tried to investigatethe characteristics and gather biological information aboutthese HepG2-specific peptides.

Variation in the FliTrx loop epitope sequence, the locationof the linear epitope sequence, and the non-epitopesequence composition indicated that some different aminoacid sequences are recognized by the same HepG2 cells.Bioinformatics analysis and prediction of the secondarystructures of peptides and tertiary structures of TrxA-peptide fusion proteins suggested that: (1) three peptidesisolated from the FliTrx library were constrained in thefusion region, and formed different 3-D structures stickingout of the fusion protein, which benefit their interactionwith molecules; (2) the structure of Hep3 was influencedby TrxA, however, the secondary and tertiary structuresof Hep1 and Hep2 showed no significant difference; and(3) the structure character might be more important forpeptide function and recognition than the sequence ofpeptides. We propose, therefore, that the 3-D structuralmodel of TrxA confirmed and guaranteed the structure ofpeptides and affected their affinity.

As TrxA-peptide fusion proteins are more available thanthe respective peptides, we attempted to verify the recog-nition between fusion proteins and tumor cells usingWestern blot analysis. We used the flagellin protein, so-called FliC/TrxA-peptide fusion proteins, to identify thetumor cell lysate. Induced bacteria were incubated withcell lysate blot and Anti-Thio positive bands were observedonly on the HepG2 cell lysate. Selective binding to thesame band of HepG2 cell lysate occurred by differentFliC/TrxA-peptide fusion proteins but not by the negativecontrol. These results revealed that FliC/TrxA-peptidefusion proteins recognize a certain molecule or subunit onHepG2 cells, but not TrxA or the peptide itself.Furthermore, the fragment of approximately 140 kDa onHepG2 cells, which binds to fusion proteins, was also



different from the other known tumor antigens, such astransferrin 95 kDa, fibrinogen 340 kDa, and α1-antitrypsin54 kDa. The fact that several peptide consensus groupswere apparent among the sequences isolated using thebacterial library suggests that these clones recognizedifferent cell surface receptors or epitopes on whole cells[30]. However, the results suggested that different clonesdetected the same target molecule, implying the existenceof a possible protein or subunit of approximately 140 kDaon HepG2 cells. These three peptides are not repeatedsequences and their 3-D structures are not identical, butthey can bind the same target molecule on tumor cells.

A protein epitope can either be continuous ordiscontinuous. While a continuous (also called sequentialor linear) epitope is a sequential fragment from the proteinsequence, a discontinuous (also called conformational)epitope is composed of several fragments scattered alongthe protein sequence and brought together in spatialproximity when the protein is folded [31]. Most epitopesare discontinuous, although they are often composed ofsmall continuous elements of the sequence. Most linearepitopes of approximately five residues are involved, fourresidues being involved in the binding [32], and the otherreplaceable residues essentially fulfilling a spacer function.E (glutamin) and L (leucine) show a high frequency amongthe amino acid sequences of Hep1, Hep2, and Hep3.Peptides could recognize and bind the same epitope basedon the folding and reversal of the 3-D structure. Peptidesselected using this methodology have moderate biologicalactivity towards their receptor molecules on cell surfaces.This indicates that the peptides can serve as potential leadsfor the development of diagnostic and therapeutic agents[33].

In summary, the development of new biologicaltechnology provides powerful methods to identifybiomarkers on tumor cells or tissues, to consummate thetumor therapy methods and drugs, by specific delivery ofdrugs in tumor cells. A major advantage of these techniques,especially the combination-interdisciplinary method, is notaffecting on innocent bystander cells. The successencountered with the bacterial peptide library with TrxA-dodecapeptides suggests that the approach described inthis study might be used to develop powerful and morespecific peptides. Three FliC/TrxA-peptide fusion proteinsselected in this study (FliC/TrxA-Hep1, FliC/TrxA-Hep2,and FliC/TrxA-Hep3) showed remarkable cell specificityof HepG2 cells. And all the facts surely pointed toward theimprovement of a novel target in HepG2 cells, as therewas a 140 kDa band detected by the FliC/TrxA-peptidefusion proteins. Structural bioinformatic approaches could

be a helpful methodology for further investigation of targetmolecules on HepG2 cells and improvement of the affinityof HepG2-specific peptides.

Acknowledgements

We thank Jingfang Shao, Yue Zhang, Jing Yang, ZhihuiLiang, Wei Feng, Xiaodan Jiang, Ping Xiong, and YongXu (Department of Immunology, Tongji Medical College,Huazhong University of Science and Technology, Wuhan,China) for their help.

References

1 Bosch FX, Ribes J , Díaz M, Cléries R. Primary liver cancer:worldwide incidence and trends. Gastroenterology 2004, 127: S5−S16

2 Ferlay J, Bray F, Pisani P, Parkin DM. GLOBOCAN 2000: CancerIncidence, Mortality and Prevalence Worldwide. Version 1.0. Lyon:IARC Press 2001

3 El-Houseini ME, Mohammed SM, Wael ME, Tarek DH, Omar SD,Anwar AE. Enhanced detection of hepato-cellular carcinoma.Cancer Control 2005, 12: 248−253

4 Zhang B, Zhang Y, Wang J, Zhang Y, Chen J, Pan Y, Ren L et al.Scr eening a nd i dent i f i ca t io n of a t a rge t ing p ept ide tohepatocarcinoma from a phage display peptide library. Mol Med2007, 13: 246−254

5 Pan B, Li B, Russell SJ, Tom JY, Cochran AG, Fairbrother WJ.Solution structure of a phage-derived peptide antagonist in com-plex with vascular endothelial growth factor. J Mol Biol 2002,316: 769−787

6 Su n LY, Chu T W, Wa ng Y, Wa ng XY. Ra diola bel ing a ndbiodistribution of a nasopharyngeal carcinoma-targeting peptideidentified by in vivo phage display. Acta Biochim Biophys Sin2007, 39: 624−632

7 Lu Z, Murray KS, Van Cleave V, LaVallie ER, Stahl ML, McCoyJM. Expression of thioredoxin random peptide libraries on theEscherichia coli cell surface as functional fusions to flagellin: asystem designed for exploring protein-protein interactions.Biotechnology 1995, 13: 366−372

8 Lu Z, LaVallie ER, McCoy JM. Using bio-panning of FLITRXpeptide libraries displayed on E. coli cell surface to study protein-protein interactions. Methods Mol Biol 2003, 205: 267−280

9 Tripp BC, Lu Z, Bourque K, Sookdeo H, McCoy JM. Investigationof the “switch-epitope” concept with random peptide librariesdisplayed as thioredoxin loop fusions. Protein Eng 2001, 14: 367−377

1 0 Craik DJ, Simonsen S, Daly NL The cyclotides: novel macrocyclicpeptides as scaffolds in drug design. Curr Opin Drug Discov Devel2002, 251−260

1 1 Li P, Roller PP. Cyclization strategies in peptide derived drugdesign. Curr Top Med Chem 2002, 2: 325−341

1 2 Ibrahim GF, Fleet GH, Lyons MJ, Walker RA. Method for theisolation of highly purified Salmonella flagellins. J Clin Microbiol1985, 22: 1040−1044

1 3 Nilsson F, Tarli L, Viti F, Neri D. The use of phage display for thedevelopment of tumor targeting agents. Adv Drug Deliv Rev 2000,



43: 165−1961 4 Bussolati B, Grange C, Tei L, Deregibus MC, Ercolani M, Aime S,

Camussi G. Targeting of human renal tumor-derived endothelialcells with peptides obtained by phage display. J Mol Med 2007,85: 897−906

1 5 Javitt NB. HepG2 cells as a resource for metabolic studies:lipoprotein, cholesterol, and bile acids. FASEB J 1990, 4: 161−168

1 6 She YM, Narindrasorasak S, Yang S, Spitale N, Roberts EA, SarkarB. Identification of metal-binding proteins in human hepatomalines by immobilized metal affinity chromatography and massspectrometry. Mol Cell Proteomics 2003, 2: 1306−1318

1 7 Pasqualini R, Ruoslahti E. Organ targeting in vivo using phagedisplay peptide libraries. Nature 1996, 380: 364−366

1 8 Zurita AJ, Troncoso P, Cardó-Vila M, Logothetis CJ, PasqualiniR, Arap W. Combinatorial screenings in patients: the interleukin-11 receptor α as a candidate target in the progression of humanprostate cancer. Cancer Res 2004, 64: 435−439

1 9 Rasmussen UB, Schreiber V, Schultz H, Mischler F, Schughart K.Tumor cell-targeting by phage-displayed peptides. Cancer GeneTher 2002, 9: 606−612

2 0 Dane KY, Chan LA, Rice JJ, Daugherty PS. Isolation of cell specificpeptide ligands using fluorescent bacterial display libraries. JImmunol Methods 2006, 309: 120−129

2 1 Ladner RC. Polypeptides from phage display. A superior source ofin vivo imaging agents. Q J Nucl Med 1999, 43: 119−124

2 2 Park JW, Hong K, Kirpotin DB, Colbern G, Shalaby R, Bseiga J,Shao Y et al. Anti-Her2 immunoliposomes: enhanced efficacyattributable to targeted delivery. Clin Cancer Res 2002, 8: 1172−1181

2 3 Zitzmann S, Krämer S, Mier W, Mahmut M, Fleig J, Altmann A,Eisenhut M et al. Identification of a new prostate-specific cyclicpeptide with the bacterial FliTrx system. J Nucl Med 2005, 46:

782−7852 4 Barry MA, Dower WJ, Johnston SA. Toward cell-targeting gene

therapy vectors: selection of cell-binding peptides from randompeptide-presenting phage libraries. Nat Med 1996, 2: 299−305

2 5 Zhang J, Spring H, Schwab M. Neuroblastoma tumor cell-bindingpeptides identified through random peptide phage display. CancerLett 2001, 171: 153−164

2 6 Giordano RJ, Cardo-Vila M, Lahdenranta J, Pasqualini R, Arap W.Biopanning and rapid analysis of selective interactive ligands. NatMed 2001, 7: 1249−1253

2 7 Romanov VI, Durand DB, Petrenko VA. Phage display selectionof peptides that affect prostate carcinoma cells attachment andinvasion. Prostate 2001, 47: 239−251

2 8 Binetruy-Tournaire R, Demangel C, Malavaud B, Vassy R, RouyreS, Kraemer M. Identification of a peptide blocking vascularendothelial growth factor (VEGF)-mediated angiogenesis. EMBOJ 2000, 19: 1525−1533

2 9 Folgori A, Tafi R, Meola A, Felici F, Galfré G, Cortese R, MonaciP et al. A general strategy to identify mimotopes of pathologicalantigens using only random peptide libraries and human sera. EMBOJ 1994, 13: 2236−2243

3 0 Daugherty PS, Olsen MJ, Iverson BL, Georgiou G. Developmentof an optimized expression system for the screening of antibodylibraries displayed on the Escherichia coli surface. Protein Eng1999, 12: 613−621

3 1 Van Regenmortel MH. Antigenicity and immunogenicity ofsynthetic peptides. Biologicals 2001, 29: 209−213

3 2 Geysen HM, Mason TJ, Rodda SJ. Cognitive features of continuousantigenic determinants. J Mol Recognit 1988, 1: 32−41

3 3 Desha yes K, Scha ffer ML, Sk el ton NJ , Na ka mu ra GR,Kadkhodayan S, Sidhu SS. Rapid identification of small bindingmotifs with high-throughput phage display: discovery of peptidicantagonists of IGF-1 function. Chem Biol 2002, 9: 495−505

screening and identification of a novel target specific ... · three-dimensiona stl ructurapr l...

Documents