This Article Abstract Full Text (PDF) Supplementary Data Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Kolonin, M. G. Articles by Pasqualini, R. Search for Related Content PubMed PubMed Citation Articles by Kolonin, M. G. Articles by Pasqualini, R.

Ligand-Directed Surface Profiling of Human Cancer Cells with Combinatorial Peptide Libraries

¹ The University of Texas M.D. Anderson Cancer Center, Houston, Texas; and ² Developmental Therapeutics Program, Division of Cancer Treatment and Diagnosis, National Cancer Institute, NIH, Bethesda, Maryland

Requests for reprints: Renata Pasqualini or Wadih Arap, The University of Texas M.D. Anderson Cancer Center, 1515 Holcombe Boulevard, Houston, TX 77030. Phone: 713-792-3873; Fax: 713-745-2999; E-mail: rpasqual{at}mdanderson.org; warap{at}mdanderson.org.

	Abstract

Top Abstract Introduction Materials and Methods Results Discussion References

 
A collection of 60 cell lines derived from human tumors (NCI-60) has been widely explored as a tool for anticancer drug discovery. Here, we profiled the cell surface of the NCI-60 by high-throughput screening of a phage-displayed random peptide library and classified the cell lines according to the binding selectivity of 26,031 recovered tripeptide motifs. By analyzing selected cell-homing peptide motifs and their NCI-60 recognition patterns, we established that some of these motifs (a) are similar to domains of human proteins known as ligands for tumor cell receptors and (b) segregate among the NCI-60 in a pattern correlating with expression profiles of the corresponding receptors. We biochemically validated some of the motifs as mimic peptides of native ligands for the epidermal growth factor receptor. Our results indicate that ligand-directed profiling of tumor cell lines can select functional peptides from combinatorial libraries based on the expression of tumor cell surface molecules, which in turn could be exploited as "druggable" receptors in specific types of cancer. (Cancer Res 2006; 66(1): 34-40)

	Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

 
The National Cancer Institute panel of human cancer cell lines from different histologic origins and grades (NCI-60) has been extensively used to screen compounds for anticancer activity (1, 2). The NCI-60 includes carcinomas of several origins (kidney, breast, colon, lung, prostate, and ovarian), tumors of the central nervous system, malignant melanomas, leukemias, and lymphomas. Gene expression determined by high-throughput microarrays has been used to survey the variation in abundance of thousands of distinct transcripts in the NCI-60; such data provided functional insights about the corresponding gene products in tumor cell transformation (2–4). This information-intensive genomic approach has yielded candidate diagnostic tumor markers to be validated at the protein level in prospective studies (4). Moreover, systematic proteomic studies based on two-dimensional PAGE (5) and protein microarrays (4) have also been implemented. Finally, in parallel with the NCI-60 transcriptome and proteome initiatives, pharmacologic sensitivity of the cells to >10⁵ different chemical compounds has been registered (1, 2). Indeed, for some genes, correlation of expression data to drug sensitivity profiles has uncovered the mechanistic basis for the drug activity (3, 6–10). Thus, conventional genomic and proteomic approaches have identified several potential tumor markers and drug targets. However, despite such advances, correlation between drug activity and gene expression profiles has not as yet been established for most of the compounds tested (9, 11, 12). This suggests the likely existence of unknown factors and the need to develop alternative methodology to discover "druggable" molecular targets.

Over the past few years, it has been proposed that (a) characterization of molecular diversity at the tumor cell surface level (represented primarily by membrane-associated proteins that are often modified by lipids and carbohydrates) is required for the development of ligand-directed anticancer therapies, and that (b) peptides binding to surface receptors preferentially expressed on tumor cells may be used to ligand-direct therapeutics to sites of disease with potential for increased therapeutic windows (13, 14). It has become increasingly clear that selective cell surface features can be mapped by screening libraries of peptides (14–17). In fact, combinatorial peptide libraries displayed from pIII protein of an M13-derived phage have now been successfully screened on intact cells and in vivo (13–15). Peptide ligands selected from unbiased screens without any predetermined notions about the nature of the cellular receptor repertoire have been used for the subsequent identification of the corresponding target cell surface receptors (16–21). In addition, novel techniques, such as the biopanning and rapid analysis of selective interactive ligands (BRASIL), have enabled high-throughput phage library screening on cells (16). Here, we used the BRASIL method to systematically screen combinatorial libraries on tumor cells of the NCI-60 panel. Results of this feasibility study suggest that tumor cells can be grouped by profiles of their peptide ligands directed to differentially expressed cell surface receptors. Our data support the notion that many tumor cell surface-exposed receptors are expressed irrespective of tumor origin, thus suggesting they could be developed as broad tumor targets. Integration of ligand-directed surface profiling with other approaches related to the NCI-60 may uncover functional ligand-receptor pairs for the targeted drug delivery.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

 
Combinatorial library screening on cells. All the NCI-60 cell lines (1), except MDA-N (unavailable), were grown in RPMI 1640 supplemented with 5% fetal bovine serum (FBS) and 5 mmol/L L-glutamine. A phage display random peptide library based on the vector fUSE5 displaying the insert CX₇C was screened by using BRASIL as described (16). Exponentially growing cells were harvested with 0.5 mmol/L EDTA, 0.4 g/L KCl, 8 g/L NaCl, and 1 g/L dextrose, washed once with phosphate buffer saline (PBS), and resuspended in RPMI containing 1% bovine serum albumin (BSA) and 1 mmol/L HEPES. Cells ( 10⁶) were incubated for 2 hours on ice with 10⁹ transduction units of CX₇C phage in 200-µL suspension, transferred to the top of a nonmiscible organic lower phase (dibutyl phtalate/cyclohexane, 9:1), and centrifuged at 10,000 x g for 10 minutes. The phage-bound cell pellet was incubated with 200 µL of K91 bacterial culture, and the bound phages were amplified and used in the following round. To prevent preferential isolation of peptides containing the RGD motif, which is selected on tissue-cultured cells due to expression of cell adhesion molecules binding to vitronectin, library screening was done in the presence of 1 mg/mL of the synthetic peptide RGD-4C (AnaSpec, San Diego, CA) in each round. After three rounds of selection, phage peptide-encoding inserts were sequenced as described (15, 17, 21).

Hierarchical cluster analysis of peptide motif/cell line association. We created an interactive sequence management database of all peptide sequences isolated in the screen. Calculation of tripeptide motif frequencies in CX₇C peptides (in both directions) was done by using a character pattern recognition program based on SAS (version 8.1.2, SAS Institute, Cary, NC) and Perl (version 5.6.1) as described (17). To identify the most closely related tripeptides and cell lines, we generated clustered image maps (CIM) by using online software CIMminer available at http://discover.nci.nih.gov/tools.jsp. Data were centered (mean subtracted and divided by SD) on both cell lines and tripeptide motifs; correlation coefficient metric with average linkage algorithm was used as distance measurement. The tripeptide motif frequencies across the NCI-60 cell lines formed a two-dimensional data matrix that was used to correlate motif enrichment with groups of cell lines. To evaluate whether CIMMiner algorithm is appropriate for clustering analysis of peptide frequency data, we devised a simulation test assuming that the frequencies of tripeptide motifs in a given data set follow an independent Poisson distribution. We simulated a random 3,280 x 59 data matrix of the dimension identical to that of tripeptide motif frequency data matrix (corresponding to the set of 3,280 tripeptides and 59 cell lines). These simulated data were centered the same way as the experimental data by transforming to mean of 0, variance of 1. For CIM in Fig. 1, tripeptides selected on all but one cell line of common origin (17) were used. Specificity of five tripeptides selectively overrepresented or underrepresented in lung tumor cell binding peptides for the 11 boxed cell lines (against the other 48 cell lines) was evaluated by using the R Package, version 2.0.0 (http://www.r-project.org) by performing two-sample t test (one tailed), as well as using Wilcoxon rank sum test (one tailed) and Fisher exact test (one tailed) as described (17).

View larger version (75K): [in this window] [in a new window]   Figure 1. Selectivity of broad-specificity tripeptides for clusters of NCI-60 cell lines. Two-dimensional hierarchical clustering was applied to the frequencies of 38 tripeptides (rows) encountered in CX7C peptides selected on NCI-60 cell lines (columns). Tripeptides selected on all but one cell line of common origin were clustered based on their correlations with cell lines; cell lines were clustered based on their correlations with the tripeptides. Tripeptide frequencies were mean subtracted and average linkage clustered with correlation metric. Amino acid color code: red, hydrophobic; green, neutral and polar; purple, basic. The color in each CIM segment ranges from blue (negative correlation) to red (positive correlation), as indicated by the scale bar. Cell lines are color-coded based on previously defined histologic tumor origin (1, 2). Bars underneath dendrogram, clusters of cells of similar tumor tissue origin (one exception allowed). Boxed, cluster of lung cancer–derived cell lines and associated/dissociated tripeptides.

View larger version (39K): [in this window] [in a new window]   Figure 2. Identification of peptides mimicking EGFR ligands. A, EGFR-binding peptide sequences isolated from the SKOV3-selected phage pool were matched in each orientation to protein sequences of biological human EGFR ligands (leader peptide sequence underlined). Matches displayed are peptides with three or more amino acids being identical (red) and one or more being from the same class (green) as the correspondingly positioned protein amino acids. Tripeptides listed in Table 1 (yellow). B, isolation of peptides targeting EGFR. Binding of SKOV3-selected phage pool to immobilized EGFR compared with BSA in rounds 1 and 2 of biopanning of SKOV3-selected phage pool on immobilized human EGFR.

Validation of epidermal growth factor receptor as one of the peptide targets. To isolate peptides binding to epidermal growth factor receptor (EGFR), phage clones selected on SKOV3 in rounds 2 and 3 of the screening were individually amplified and pooled, and 10⁹ transduction units of the mixed phage were incubated overnight at 4°C with 10 µg of purified human EGFR (Sigma, St. Louis, MO), or BSA control immobilized on plastic. Unbound phages were extensively washed off with PBS, and then the bound phages were recovered by infecting host K91 Escherichia coli directly on the plate, and tetracycline-resistant clones were selected/quantified and sequenced. To identify EGFR ligand-matching motifs among phage-displayed SKOV3-binding peptides, custom-designed Perl 5.8.1–based software was used to run peptide sequences against biological EGFR ligand sequences. Each 7-mer peptide sequence was aligned in each orientation against the EGFR ligand sequences from the NH₂ to COOH terminus in one-amino-acid shifts. The peptide/protein similarity scores for each residue were calculated based on a BLOSUM62 matrix modified to identify peptide matches of at least three amino acids in any position being identical and one being similar to the corresponding amino acid positions in the EGFR ligands (Fig. 2A).

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

 
Isolation of peptides binding to surface of the NCI-60 cancer cells. As an initial attempt to profile cell surface of the tumor cell panel, we screened a large (2 x 10⁸ unique sequences) cyclic random peptide library with the basic structure CX₇C (C, cysteine; X, any residue) on every cell line of the NCI-60. Phage selection was done in the excess of a competing Arg-Gly-Asp (RGD) synthetic integrin-binding peptide (13) to minimize the recovery of RGD-containing peptides. This strategy was designed to facilitate the recovery of ligands binding to nonintegrin families of cell surface receptors because RGD tends to become dominant in the screening due to the high levels of integrin expression in adherent cells.³ Preferential cell binding of specific cell-targeting peptides results in enrichment, defined by the increased recovery frequency of these peptide motifs in each subsequent round of the screen (14, 21). Thus, we set out to profile the expression of nonintegrin cell surface molecules among the cell lines of the NCI-60 according to the differential selection of motifs enriched in the screen.

Hierarchical cluster analysis of peptides binding to the NCI-60 cells. To analyze the spectrum of the peptides resulting from the screening and compare those among different cell lines of the panel, we adopted a combinatorial statistical approach based on the premise that three residue motifs (tripeptides) provide a sufficient structure for protein-peptide interactions in the context of phage display (17). For each NCI-60 cell line, we sequenced CX₇C peptide-encoding DNA inserts from 96 phage clones recovered after three rounds of selection. We did a computer-assisted survey of all tripeptides within the library-derived sequences selected on each cell line by analyzing a database of 26,031 tripeptides contained within the 5,270 CX₇C-encoded 7-mer peptides isolated (an average of eighty-nine 7-mer peptide sequences analyzed per each NCI-60 cell line). Thus, each cell line was assigned a unique set of tripeptides that was identified during the selection for cell surface binders, and the frequencies of each motif among all peptides for a given cell line were calculated.

To classify cell lines according to their association with particular motifs, which might provide inference on the targeted surface molecules, we did a hierarchical clustering analysis of the 3,280 nonredundant tripeptides based on the frequency of association with the NCI-60 cell lines. For the construction of a CIM, we adapted a hierarchical clustering algorithm and a pseudo-color visualization matrix initially designed to address differential gene expression among the cells of the panel (3, 6–8). CIMMiner (2) was used for inference of the variation in peptide binding specificity across the cell lines by comparing relative frequencies of tripeptides found in 7-mer peptides binding to each cell (see Materials and Methods). Clustering of peptide motifs with similar cell selectivity revealed that the peptide distribution of the combinatorial library within the NCI-60 set was nonrandom (Supplementary Fig. S1A). Computer simulations of the permutated data set show that the observed pattern could not be generated by random chance (Supplementary Fig. S1B), thus indicating that the discontinuous tripeptide frequency data is applicable for cluster analysis.

The selective spectra of peptide motifs interacting with the clustered cell lines suggest the existence of shared targeted surface receptor(s) expressed in these lines. In this study, we chose to focus on putative peptide-targeted receptors with broad cell line specificity, which would be more informative for an initial peptide binding/receptor expression correlation analysis. We therefore excluded from the data set motifs selected only on a single or few cell lines. Instead, we focused on 38 tripeptides that showed a semiubiquitous distribution among the NCI-60 lines (Fig. 1). A CIM constructed according to the isolation frequency of these broader-specificity tripeptides from each cell line revealed several apparent clusters of cell lines that displayed distinct profiles of association with certain classes of peptide motifs. For example, the majority of lung cancer–derived cell lines segregated as a separate group, suggesting that some of the receptors targeted may be conserved among cell lines derived from a common origin (Fig. 1). Thus, although we severely restricted our analysis by limiting it to semiubiquitous tripeptides, clustering of some of them (predominantly with cell lines derived from the same tumor type) is consistent with their relative tissue specificity. To evaluate individual motifs for selectivity, we identified a distinct cluster of five tripeptides associated with lung tumor–derived cell lines (Fig. 1, boxed). We compared tripeptide frequencies for the 11 cell lines within this cluster with their frequencies for the rest of NCI-60 lines by using statistical tests (Fisher exact, Wilcoxon rank-sum, and t test). Consistently, we observed that motif GGS was isolated for the clustered lines significantly (P < 0.05) more frequently than for the other NCI-60 cell lines (Supplementary Table S1).

Notably, the distribution of cell lines in the dendrogram (Fig. 1) was partially consistent with the reported association of cells derived from tumors with common tissue origin (3, 4). This suggests that some of the receptors, such as the one presumably recognized by the lung tumor–specific tripeptide GGS (Fig. 1; Supplementary Table S1), may be up-regulated only in certain cancer origins. However, the tumor cell phylogeny was recapitulated only to an extent; the majority of the observed clusters contained cell lines derived from unrelated tumor types (Fig. 1). The limited grouping of lines derived from tumors of common origin is perhaps not surprising: the relationship between different cell lines in our study is based on peptide binding to putative cell surface molecules, many of which may be tumor induced rather than characteristic of the tissue of origin. If so, our analysis of broad-specificity motif distribution may be well suitable for identification of specific surface molecules that are generally up-regulated by tumors and thus may constitute broad drug targets against cancer.

Identification of candidate receptor targets for peptide motifs. We next attempted to identify the targets for the 38 broad-specificity tripeptides, most of which presumably bind to receptors expressed by multiple NCI-60 cell lines. We used the NCI Molecular Targets Database that contains detailed information on the expression and activity of 1,218 human proteins measured by nonarray methods (22). By using the COMPARE algorithm (6), we correlated the selectivity profiles of the 38 tripeptide motifs with the expression profiles of the characterized molecular targets. We observed that several of the qualifying proteins, expression of which correlated with enrichment profiles of certain motifs, represented tyrosine kinase receptors, such as those for ligands belonging to families of EGFs, fibroblast growth factors (FGF), nerve growth factors (NGF), and ephrins (Table 1). When transferred to molecular target correlation data, the order of the 38-tripeptide motif set in the dendrogram (Fig. 1) revealed clusters of tripeptides for which cell line association profile correlated with expression profiles of EGF, FGF, NGF, or ephrin receptors (Table 1).

The peptide distribution-correlating tyrosine kinase receptors, belonging to EGFR, FGFR, NGFR, and ephrin receptor families (Table 1), are often up-regulated in many types of cancer (23). To determine if the cell-binding peptides may target these tyrosine kinases, we employed the notion that receptor-binding peptide motifs often mimic natural ligands for these receptors (16, 17, 19). Thus, we tested whether the selected motifs mimic ligands for the candidate tyrosine kinases by determining whether tripeptides listed in Table 1 are embedded into longer peptides that may be responsible for cell surface binding. We analyzed the CX₇C phage inserts containing the 38 tripeptides by using the ClustalW software and compiled extended motifs containing the tripeptides shared among multiple peptides selected during the screen (data not shown). To identify candidate prototype human ligands, epitopes of which could be mimicked, we screened each of the ClustalW-extended motifs against the nonredundant database of human proteins by using the BLAST software (National Center for Biotechnology Information). As a result of this analysis, we found the motifs containing 34 of 38 tripeptides (89%) to be identical or very similar to segments of proven or putative ligands for the tyrosine kinase receptors listed (Table 1).

Validation of EGFR as a targeted receptor. To show that the approach taken can lead to actual targetable tumor cell surface proteins, we chose to test if the EGFR is bound by any of the tripeptide motifs distributed in the panel in a profile correlating with EGFR expression. Consistently, 24 of 38 tripeptides surveyed displayed NCI-60 cell line association pattern consistent with that of EGFR expression (Table 1). Of these tripeptides, 22 were isolated in the screens on ovarian cancer cell lines SKOV3 and OVCAR4 (data not shown). Because EGFR is well known to be associated with ovarian cancer (23), we deemed these cell lines to be likely expressers of targetable EGFR, which would account for the selection of EGFR ligand-mimicking motifs. To validate EGFR binding by the selected motifs, we screened the SKOV3-binding phage sublibrary (pooled clones recovered in rounds 2 and 3) on immobilized human EGFR. After two rounds of selection, we analyzed phage displaying the EGFR-binding peptides: the majority were comprised by different 7-mer peptides (Fig. 2A) that contained 17 of 22 SKOV3-selected tripeptide motifs distributed in the panel in a profile correlating with EGFR expression (Table 1). Phage displaying these peptides had specific affinity to EGFR, as determined by subjecting the same sublibrary to immobilized BSA control binding (Fig. 2B). Remarkably, computer-assisted analysis of sequences (Fig. 2A) revealed that 12 of the 7-mer EGFR-binding peptides contained amino acid motifs similar to those present in some of the biological EGFR ligands (23). These peptides, containing eight of the candidate tripeptides (RVS, AGS, AGL, GVR, GGR, GGL, GSV, and GVS), were found highly similar to fragments of EGF, amphiregulin, heparin-binding EGF-like growth factor, and epiregulin (Fig. 2A). Similarity search using the same algorithm on the same twelve 7-mers did not reveal any matches to two other EGFR ligands, transforming growth factor- and ß-cellulin, or randomly chosen control ligands of tyrosine kinase receptors from the three other candidate families listed in Table 1: ephrin A, NGF-ß, and FGF6 (data not shown). Taken together, these data suggest that at least some of the peptides selected on the NCI-60 cells target EGFR, whereas others may bind to different tyrosine kinases, possibly including those from TRK, ephrin, or FGF receptor families.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

 
Expression profiles of the candidate receptor targets for peptides identified in the screen illustrate the concept that in cancer, at least some tumor-associated cell surface molecules seem up-regulated regardless of cancer tissue origin. As such, this is the case for the EGFR and other tyrosine kinases possibly targeted by peptide ligands selected on the NCI-60 cell panel. This may also be the case for many other receptors with a role in tumorigenesis, expression profiles of which may not correlate with the overall proteomic profile of the original tumor tissue. In fact, these observations may account for the relatively limited success in correlating drug toxicity profiles with the genomic and/or proteomic profiles of the NCI-60 panel (12). On the other hand, some of the receptors, such as EphA5 presumably targeted by GGS tripeptide and its derivatives predominantly selective for lung tumor-derived cell lines (Fig. 1), seem to be at least partially specific for the progenitor cancer type.

The candidate ligand-receptor leads identified in this study can be characterized further for the development of targeted agents selective for tumors. Moreover, the peptides identified by the approach described here may map receptor interaction domains of biological (native) ligands. Similarity of peptides to the corresponding receptor-binding ligands has already been used for validation of the IL-11R receptor as a target of an interleukin-11 mimic peptide homing to blood vessels in the prostate (17, 24). We and others have modeled the usage of peptides homing to receptors expressed by tumors (18) or nonmalignant tissues (19, 20) for directing the delivery of cytotoxics, proapoptotic peptides, metalloprotease inhibitors, cytokines, fluorophores, and genes (13, 14). Thus, our approach provides a straightforward way to identify drug-accessible tumor cell surface receptors and to discover peptide ligands that can serve as mimetic prototype drugs. Unlike genomic or proteomic-based approaches that rely on differential expression levels of transcripts or protein products, this discovery platform directly addresses functional protein-protein interactions at the level of physical binding. In contrast to protein array systems, it is possible to select binding peptides even if the ligand-receptor interaction is mediated by conformational (rather than linear) epitopes. In summary, ligand-directed screening of combinatorial libraries on tumor cell surfaces may lead to improved selection of functionally relevant peptides that can be developed for targeting "druggable" molecular targets.

	Acknowledgments

 
Grant support: NIH (R. Pasqualini, W. Arap, M.G. Kolonin, and K-A. Do), SPORE Programs (Prostate Cancer and Leukemia, R. Pasqualini and W. Arap), Department of Defense (M.G. Kolonin) IMPACT (W. Ki Hong) and grant DAMD17-03-1-0638 (M.G. Kolonin), Lung Cancer Program (W. Ki Hong), A.J. Zurita was supported by a BEFI fellowship.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

	Footnotes

 
Note: M.G. Kolonin and L. Bover contributed equally to the work.

Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).

³ Unpublished observations.

Received 8/ 3/05. Revised 10/12/05. Accepted 11/ 2/05.

	References

Top Abstract Introduction Materials and Methods Results Discussion References

 

1. Monks A, Scudiero D, Skehan P, et al. Feasibility of a high-flux anticancer drug screen using a diverse panel of cultured human tumor cell lines. J Natl Cancer Inst 1991;83:757–66.[Abstract/Free Full Text]
2. Weinstein JN, Myers TG, O'Connor PM, et al. An information-intensive approach to the molecular pharmacology of cancer. Science 1997;275:343–9.[Abstract/Free Full Text]
3. Scherf U, Ross DT, Waltham M, et al. A gene expression database for the molecular pharmacology of cancer. Nat Genet 2000;24:236–44.[CrossRef][Medline]
4. Nishizuka S, Charboneau L, Young L, et al. Proteomic profiling of the NCI-60 cancer cell lines using new high-density reverse-phase lysate microarrays. Proc Natl Acad Sci U S A 2003;100:14229–34.[Abstract/Free Full Text]
5. Myers TG, Anderson NL, Waltham M, et al. A protein expression database for the molecular pharmacology of cancer. Electrophoresis 1997;18:647–53.[CrossRef][Medline]
6. Zaharevitz DW, Holbeck SL, Bowerman C, Svetlik PA. COMPARE: a web accessible tool for investigating mechanisms of cell growth inhibition. J Mol Graph Model 2002;20:297–303.[CrossRef][Medline]
7. Blower PE, Yang C, Fligner MA, et al. Pharmacogenomic analysis: correlating molecular substructure classes with microarray gene expression data. Pharmacogenomics J 2002;2:259–71.[CrossRef][Medline]
8. Rabow AA, Shoemaker RH, Sausville EA, Covell DG. Mining the National Cancer Institute's tumor-screening database: identification of compounds with similar cellular activities. J Med Chem 2002;45:818–40.[CrossRef][Medline]
9. Wallqvist A, Rabow AA, Shoemaker RH, Sausville EA, Covell DG. Establishing connections between microarray expression data and chemotherapeutic cancer pharmacology. Mol Cancer Ther 2002;1:311–20.[Abstract/Free Full Text]
10. Szakacs G, Annereau JP, Lababidi S, et al. Predicting drug sensitivity and resistance: profiling ABC transporter genes in cancer cells. Cancer Cell 2004;6:129–37.[CrossRef][Medline]
11. Brown JM. NCI's anticancer drug screening program may not be selecting for clinically active compounds. Oncol Res 1997;9:213–5.[Medline]
12. Wallqvist A, Rabow AA, Shoemaker RH, Sausville EA, Covell DG. Linking the growth inhibition response from the National Cancer Institute's anticancer screen to gene expression levels and other molecular target data. Bioinformatics 2003;19:2212–24.[Abstract/Free Full Text]
13. Arap W, Pasqualini R, Ruoslahti E. Cancer treatment by targeted drug delivery to tumor vasculature in a mouse model. Science 1998;279:377–80.[Abstract/Free Full Text]
14. Kolonin MG, Pasqualini R, Arap W. Molecular addresses in blood vessels as targets for therapy. Curr Opin Chem Biol 2001;5:308–13.[CrossRef][Medline]
15. Pasqualini R, Ruoslahti E. Organ targeting in vivo using phage display peptide libraries. Nature 1996;380:364–6.[CrossRef][Medline]
16. Giordano RJ, Cardo-Vila M, Lahdenranta J, Pasqualini R, Arap W. Biopanning and rapid analysis of selective interactive ligands. Nat Med 2001;7:1249–53.[CrossRef][Medline]
17. Arap W, Kolonin MG, Trepel M, et al. Steps toward mapping the human vasculature by phage display. Nat Med 2002;8:121–7.[CrossRef][Medline]
18. Pasqualini R, Koivunen E, Kain R, et al. Aminopeptidase N is a receptor for tumor-homing peptides and a target for inhibiting angiogenesis. Cancer Res 2000;60:722–7.[Abstract/Free Full Text]
19. Kolonin MG, Pasqualini R, Arap W. Teratogenicity induced by targeting a placental immunoglobulin transporter. Proc Natl Acad Sci U S A 2002;99:13055–60.[Abstract/Free Full Text]
20. Kolonin MG, Saha PK, Chan L, Pasqualini R, Arap W. Reversal of obesity by targeted ablation of adipose tissue. Nat Med 2004;10:625–32.[CrossRef][Medline]
21. Pasqualini R, Arap W, Rajotte D, Ruoslahti E. In vivo selection of phage-display libraries. In: Barbas C, Burton D, Silverman G, Scott J, editors. Phage display: a laboratory manual. New York (NY): Cold Spring Harbor Laboratory Press; 2001. p. 22.1–4.
22. Holbeck SL. Update on NCI in vitro drug screen utilities. Eur J Cancer 2004;40:785–93.[CrossRef][Medline]
23. Vogelstein B, Kinzler KW. Cancer genes and the pathways they control. Nat Med 2004;10:789–99.[CrossRef][Medline]
24. Zurita AJ, Troncoso P, Cardo-Vila M, et al. Combinatorial screenings in patients: the interleukin-11 receptor as a candidate target in the progression of human prostate cancer. Cancer Res 2004;64:435–9.[Abstract/Free Full Text]

This article has been cited by other articles:

				S. Nishimura, S. Takahashi, H. Kamikatahira, Y. Kuroki, D. E. Jaalouk, S. O'Brien, E. Koivunen, W. Arap, R. Pasqualini, H. Nakayama, et al. Combinatorial Targeting of the Macropinocytotic Pathway in Leukemia and Lymphoma Cells J. Biol. Chem., April 25, 2008; 283(17): 11752 - 11762. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) Supplementary Data Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Kolonin, M. G. Articles by Pasqualini, R. Search for Related Content PubMed PubMed Citation Articles by Kolonin, M. G. Articles by Pasqualini, R.