Fingerprinting the circulating repertoire of antibodies from cancer patients

Figure 1. Strategy and screening of index patients. (A) Scheme of screening and validation algorithm. (B) Selection of a phage-display random-peptide library on immunoglobulins purified from the serum of four prostate cancer index patients. Each successive round of panning shows an increase in selectivity relative to the serum of age-matched control men. At the end of the third round of selection, a marked enrichment (up to 1,000-fold) was obtained in three patients; in the fourth patient (lower right panel) no enrichment was observed. Bars represent mean values for phage transducing units (TU) recovered from immunoglobulins s.e.m. from triplicates. (C) Binding inhibition of the antibodies isolated from the serum of patients to immobilized GST-fusion peptide-coated wells is specific. Bars represent mean s.e.m. from triplicates. Full Figure and legend (63K)

Table 1. Peptide sequences displayed by phage binding to purified IgGs Full Table

Figure 2. Reactivity between the serum from prostate cancer patients or control men and the selected peptide is stage-specific. (A−C) Serum samples derived from a large panel of prostate cancer patients (n = 108) were divided into four groups: organ-confined (n = 17), locally advanced (n = 31), metastatic androgen-dependent (n = 31), and metastatic androgen-independent (n = 29). Serum samples derived from 71 age-matched volunteer blood-donor men served as a negative control group. For each serum sample, serial dilutions determined optimal reactivity by ELISA (data not shown). Distribution of reactivity is shown as a ratio of GST-peptide to GST alone in (A) and as percentages of positive reactivity in (B). Positive reactivity was defined by a ratio of GST peptide to GST alone 2. (C) Correlation between overall survival and serological reactivity against the peptide CNVSDKSC selected in the screening. The same population of prostate cancer patients was used to generate the Kaplan-Meier survival curves shown. The difference in overall survival between prostate cancer patients reacting positively versus not reacting to CNVSDKSC was assessed and a correlation between poor survival outcome and positive serum reactivity against the peptide CNVSDKSC was statistically significant (log-rank test, P = 0.02). GST, glutathione S-transferase; A.D., androgen-dependent; A.I., androgen-independent. Full Figure and legend (48K)

Figure 3. Immunohistological analyses of tumors from individuals with prostate cancer. (A−F) Immunostaining of sections from prostate cancer metastatic to the bone marrow of the patient whose screening yielded CNVSDKSC and from normal prostate are shown. Strong staining was observed on metastatic tumor with the autologous immunopurified IgGs (A) and with a rabbit polyclonal antibody raised against the synthetic form of the CNVSDKSC soluble peptide (B). No staining was observed with rabbit pre-immune serum (C) or with secondary antibody alone (data not shown). A recombinant CNVSDKSC peptide (D) inhibited the staining shown in (A). Similarly, the recombinant CNVSDKSC peptide (E) also inhibited the staining shown in (B). (F) A weak staining was observed with the same rabbit polyclonal antibody used in (B) on normal prostate. Scale bar (A−F), 50 m. Full Figure and legend (51K)

Figure 4. Identification of the corresponding antigen. (A) Prostate cancer cells were harvested and sheared, and the fraction containing cell membrane and cytosolic contents was semi-purified and separated on a 4%−20% gradient SDS-PAGE (Coomassie blue staining, left panel). Western blot analysis (right panel) was performed with anti-peptide CNVSDKSC antibody at 1:100. An ECL system was used to detect the antigen. A single 80 kDa protein was immunodetected (right panel). (B) The corresponding band was excised from the Coomassie blue−stained gel (A, arrow), reprocessed in an 8% gradient SDS-PAGE, and used for protein sequencing. The partially purified sample resolved in the 8% SDS-PAGE (left panel) was also used for a western blot performed with the anti-peptide CNVSDKSC antibody used in (A) (middle panel). Pre-immune serum served as a negative control (right panel). (C) Amino acid sequence of glucose regulated protein−78 kDa (GRP78) is shown. Five peptides (sequences marked in red) were obtained by mass spectrometry and matched to GRP78 by BLAST homology search with 100% identity. (D) Identity of the purified protein (B, left panel) was confirmed as GRP78 by western blot analysis with an anti-GRP78 antibody as probe. (E) Reciprocal co-immunoprecipitations with either anti-GRP78 antibody or anti-peptide CNVSDKSC antibody are shown; protein G−conjugated beads served as a negative control. (F) Whole lysates were prepared from frozen tissue samples of normal prostate and bone marrow metastases. Equivalent amounts of protein (20 g) were resolved on 4%−20% SDS-PAGE and probed with an anti-GRP78 antibody on western blots. GRP78 was weakly detected in normal prostate tissue but was strongly expressed in bone marrow metastases (left panel). Recombinant GRP78 or the peptide-GST fusion protein GST-CNVSDKSC can inhibit GRP78 (middle and right panels). Full Figure and legend (74K)

Figure 5. Cross-inhibition of binding activity of the patient-derived serum by GRP78 or CNVSDKSC. Microtiter plate wells were coated with (A) recombinant GRP78 or (B) CNVSDKSC, and various concentrations of patient serum, anti-GRP78 antibody, and anti-CNVSDKSC antibodies were added and analyzed by ELISA. Bars, mean s.e.m. from triplicates. Full Figure and legend (16K)

Figure 6. Reactivity against GRP78 is a candidate marker of prostate cancer. Microtiter plate wells were coated with recombinant GRP78 and triplicates of serum samples were added at a 1:50 dilution, as described. (A) Serum samples from the same prostate cancer population presented in Figure 2 were examined. Male (n = 155) and female (n = 48) donors served as negative controls; positive reactivity by ELISA was defined as an absorbance 0.95 by a standard statistical classification and regression tree49. To test whether reactivity against GRP78 was restricted to prostate cancer, three additional non-prostate cancer tumor types are shown as controls: metastatic non−small cell (N.S.C.) lung cancer (n = 31), metastatic breast cancer (n = 32), and advanced ovarian cancer (n = 32). Percentages of positive reactivity are presented. (B) Correlation between overall survival and serological reactivity against GRP78. The prostate cancer patient population (n = 91) was used to generate the Kaplan-Meier survival curve. A log-rank test was used to detect significant differences in survival time between patients reacting positively versus those not reacting to GRP78. A significant correlation was observed between shorter overall survival and positive reactivity against GRP78 (log-rank test, P = 0.07). Full Figure and legend (37K)

Figure 7. Expression pattern of the protein GRP78 and specific immunostaining inhibition. (A−D) Immunohistochemical analyses of normal prostate and bone marrow metastases from prostate cancer with anti-GRP78 antibody and anti-CNVSDKSC antibody. Strong staining was observed in bone marrow metastases (A), whereas weak staining was observed in the normal prostate (B). Pre-incubation with either the recombinant GRP78 (C) or with CNVSDKSC (D) inhibits staining. Scale bar, 100 m. Full Figure and legend (67K)

	Top

	Top

Protein G−agarose beads (Gibco-BRL, Rockville, MA) were used to purify IgGs from the serum of prostate cancer patients. To enhance the selection of peptides binding to IgGs specifically associated with prostate cancer, we designed and used a dual procedure. First, a pre-clearing step was used to remove nonspecific clones by pre-absorbing the phage peptide library onto purified IgGs from normal serum pooled from volunteer age-matched blood-donor control men. Next, the pre-cleared phage library was selected onto the pool of IgGs purified from the serum of prostate cancer patients. For the pre-clearing step, 10⁹ transducing units of a CX₆C phage library were incubated with IgGs purified from 50 l of serum pooled from healthy control men and immobilized on 50 l of protein G for 1 h at 4 °C. For the screening step, the pre-cleared phage library was selected onto the pool of IgGs purified from the serum of prostate cancer patients; affinity selection on immobilized IgGs from the serum of each patient for 2 h at 4 °C was used for screening. Phage clones bound to patient-derived IgGs were eluted with glycine buffer (pH 2.2) and neutralized with Tris buffer (pH 9.0). Phages were eluted from infected host Escherichia coli strain K91kan in log phase, as described^{3, 5, 6, 7}. Serial dilutions of the bacterial culture were spread onto Luria-Bertani (LB) agar plates containing 40 g/ml of tetracycline and grown overnight at 37 °C. At least 200 individual bacterial colonies were picked, amplified, and precipitated for subsequent rounds of selection; three rounds of selection were performed per patient. For binding assays, solutions of purified GST or GST-fusion peptides in 100 mM NaHCO₃ (each at 20 g/ml) were used to coat multiwell plates (Nalge-Nunc, Naperville, IL) and incubated overnight at 4 °C. Plates were covered with blocking buffer containing 4% milk (wt/vol), 2% casein (wt/in/vol), and 0.05% Tween-20 (vol/vol) for 4 h. Serial dilutions (from 1:100 to 1:1,200) of serum from prostate cancer patients or control men were added and incubated for 90 min, rinsed with wash buffer containing 1% milk (wt/vol), 0.5% casein (wt/vol), and 0.025% Tween-20 (vol/vol), and then incubated at 4 °C with alkaline phosphate−conjugated anti-human antibodies. For peptide inhibition assays, peptides CHQKPWEC, CKDRFERC, and CNVSDKSC were synthesized as biotinylated cyclic forms; control biotinylated cyclic peptides with unrelated sequences (CNGRC and CTFAGSSC) were synthesized by Merrifield synthesis and cyclized (AnaSpec, San Jose, CA). Increasing concentrations of cyclic synthetic peptides were pre-incubated with the corresponding serum antibodies from the index prostate cancer patients (patients 1−3) for 30 min and then incubated on the immobilized GST fusion peptide−coated wells. For each sample, serial dilutions were performed. Little or no reactivity occurred with the controls, whereas strong reactivity occurred with the selected peptides (data not shown). To determine the inhibitory activity of the peptides, serum from each patient was pre-incubated for 30 min at room temperature with the peptides at various concentrations and then incubated for 1 h on immobilized peptides. Finally, the substrate p-nitrophenyl phosphate (p-NPP; Sigma, St. Louis, MO) was added and the reaction was incubated for 30 min at room temperature. Absorbance at 405 nm was determined in an automated ELISA reader (Bio-Tek, Winooski, VT).

Written informed consent was obtained from subjects and approved by the Institutional Review Board (IRB) of the University of Texas M. D. Anderson Cancer Center (UTMDACC). Experimental samples of patients with prostate cancer were obtained from the UTMDACC Specialized Program of Research Excellence (SPORE) Serum Bank; control samples were obtained from age-matched blood-donor volunteers. In each case, the patient was evaluated in reference to tumor staging (locally advanced or metastatic disease) and hormone responsiveness of the disease (androgen-dependent or androgen-independent). Criteria for enrollment consisted of a combination of the Tumor-Nodal-Metastases (TNM) classification⁴⁷ and histological grading⁴⁸. Patients diagnosed with adenocarcinoma of the prostate with stage T1c or T2 with Gleason score 7 and serum PSA <10 ng/ml were considered to have clinically organ-confined prostate cancer. Study entry in the locally advanced group required appropriate primary tumor staging (stage T1c or T2 with Gleason score >7; or clinical stage T2b−2c with Gleason score 7 and serum PSA >10 ng/ml; or clinical stage T3) and no regional (N0) or distant (M0) metastases. Study entry in the metastatic group required evidence of regional (N1) or distant (M1) metastases or both on radionuclide bone scan, chest radiography, or computed tomography of the abdomen and pelvis. Androgen-independence was defined as serum testosterone lower than 50 ng/ml and serially rising serum PSA. The index patients used in the initial selection had metastatic androgen-independent disease, but we had no other knowledge of serum reactivity levels at the time their serum samples were obtained.

Probabilities of survival for each group were analyzed by the Kaplan-Meier method¹¹. Reactivity to selected peptides was considered positive if the ratio of GST peptide to GST alone was 2 by ELISA. Appropriate statistical tests (log-rank, -square, or t-tests) were used to assess statistical differences between groups as indicated. To identify the cutoff point for reactivity, a standard statistical classification and regression tree (termed CART) methodology⁴⁹ was used. Censored survival data were transformed into a single uncensored data value (the null Martingale residual), which is used as input into a standard regression tree algorithm.

Archived, formalin-fixed, paraffin-embedded tissue sections were obtained from the UTMDACC Prostate Specialized Program of Research Excellence (SPORE Tumor Bank). Sections (4 m) were stained with purified biotinylated antibodies or peptide antibodies by immunoperoxidase detection with an antigen retrieval kit (Dako, Carpinteria, CA) and diaminobenzidine (DAB); the slides were counter-stained with hematoxylin. Purified IgGs were coupled to biotin and resolved by SDS-PAGE. Biotinylated serum-immunopurified IgGs were used at a 1:60 dilution. Polyclonal antibodies against the displayed peptides were raised in rabbits. Pre-immune and immune serum were purified with a T-gel immunoglobulin purification kit and protein G columns (Pierce, Rockford, IL). The purified antibodies were used at 0.01 mg/ml. For immunostaining inhibition, antibodies against the displayed peptides were pre-incubated for 30 min at room temperature with the corresponding fusion GST-peptide fusion (500 g each). For the GRP78 protein immunostaining, anti-GRP78 antibody (C-20) at 1:350 was used (Santa Cruz Biotechnology, Santa Cruz, CA).

DU-145 prostate cancer cells (American Type Culture Collection, Manassas, VA), which express the antigen (data not shown), were used for protein purification. Cells were grown to 70% confluence, harvested in PBS, and treated with 100 mM Tris-Cl, 2 mM MgCl₂, and 1% (vol/vol) Triton X-100 (TM buffer). Cells were sheared to separate nuclei from cytoplasm and other organelles. The cytosolic/membrane fraction was centrifuged; the supernatant was collected, resolved on a 4%−20% gradient SDS-PAGE, probed with rabbit anti-peptide antibodies on western blots, and detected by enhanced chemiluminescence (ECL; Pharmacia, Piscataway, NJ). The band containing the protein recognized by the antiserum was excised and used for protein sequencing. Results from mass spectrometry analysis were compared to databases containing protein sequences by BLAST homology search. For immunoprecipitation, 200 l of protein G−agarose beads (Pierce) was coupled to anti-GRP78 or rabbit anti-peptide antibodies, and 150 g of the recombinant GRP78 (Stressgen, Victoria, British Columbia, Canada) was added and incubated for 4 h. As a negative control, protein G−agarose beads alone were used. The immunoprecipitates were recovered by centrifugation, rinsed with wash buffer containing 0.05% (vol/vol) Tween-20 in PBS, and resolved by SDS-PAGE. A western blot was probed with either anti-CNVSDKSC or anti-GRP78 antibodies (each at 1:200 dilution) and detected by ECL. For detection of GRP78 in the normal prostate and bone marrow metastases, whole lysates from frozen tissue samples were prepared by grinding the tissue in a Dounce homogenizer in a 2 ml of a tissue protein extraction reagent (Pierce) per sample with protease inhibitors (10 g/ml of leupeptin and aprotinin). The homogenate was incubated on ice for 10 min and subjected to grinding again. The homogenate was spun at 610 g for 5 min and the supernatant was removed; protein concentration was measured using the Protein DC Assay (BIO-RAD, Hercules, CA). A total of 20 g of protein from normal prostate and bone marrow metastasis lysates were resolved on a 4%−20% SDS-PAGE, western blotted using an anti-GRP78 antibody probe, and detected by ECL.

Ninety-six-well microtiter plates were coated with 10 g/ml recombinant GRP78 (Stressgen) or GST-CNVSDKSC in 100 mM NaHCO₃ overnight at 4 °C, washed, and then blocked with blocking buffer containing 2% (wt/vol) milk, 1% (wt/vol) casein, and 0.05% (vol/vol) Tween-20 in PBS for 2 h at 37 °C. To determine the inhibitory activities of GRP78 and GST-CNVSDKSC, patient serum (1:50), anti-GRP78 (1:1,000), and anti-GST−CNVSDKSC (1:20) were incubated with GRP78 (50-100 g) or GST-CNVSDKSC (100-300 g). The mixtures were incubated for 1 h at 37 °C and then added to the coated wells. After 1 h of incubation at room temperature, the wells were washed several times with 0.05% (vol/vol) Tween-20 in PBS (PBST buffer). Secondary antibodies conjugated to horseradish peroxidase were added at 1:5,000 dilution, incubated for 30 min at room temperature, and washed five times with PBST buffer. Finally, the substrate 3,3',5,5'-tetramethylbenzidine (TMB; Calbiochem, San Diego, CA) was added and incubated for 15 min at room temperature, after which the reaction was stopped by addition of 0.5 M H₂SO₄. Absorbance at 450 nm was determined in an ELISA reader (Bio-Tek).

	Top

1. Smith, G.P. Filamentous fusion phage: novel expression vectors that display cloned antigens on the virion surface. Science 228, 1315–1317 (1985). | PubMed  | ISI | ChemPort |
2. Brown, K.C. New approaches for cell-specific targeting: identification of cell-selective peptides from combinatorial libraries. Curr. Opin. Chem. Biol. 4, 16–20 (2000). | Article | PubMed  | ISI | ChemPort |
3. Phage Display: A Laboratory Manual (eds. Barbas, C.F. III, Burton, D.R., Scott, J.K. & Silverman, G.J.; Cold Spring Harbor Laboratory Press, New York, NY, 2000).
4. Kolonin, M., Pasqualini, R. & Arap, W. Molecular addresses in blood vessels as targets for therapy. Curr. Opin. Chem. Biol. 5, 308–313 (2001). | Article | PubMed  | ISI | ChemPort |
5. Giordano, R.J., Cardó-Vila, M., Lahdenranta, J., Pasqualini, R. & Arap, W. Biopanning and rapid analysis of selective interactive ligands. Nat. Med. 7, 1249–1253 (2001). | Article | PubMed  | ISI | ChemPort |
6. Scott, J.K. & Smith, G.P. Searching for peptide ligands with an epitope library. Science 249, 386–390 (1990). | PubMed  | ISI | ChemPort |
7. Smith, G.P. & Scott, J.K. Libraries of peptides and proteins displayed on filamentous phage. Meth. Enzymol. 217, 228–257 (1993). | Article | PubMed  | ISI | ChemPort |
8. Boon, T. & Old, L.J. Cancer tumor antigens. Curr. Opin. Immunol. 9, 681–683 (1997). | Article | PubMed  | ISI | ChemPort |
9. Rosenberg, S.A. Progress in human tumour immunology and immunotherapy. Nature 411, 380–384 (2001). | Article | PubMed  | ISI | ChemPort |
10. Smyth, M.J., Godfrey, D.I. & Trapani, J.A. A fresh look at tumor immunosurveillance and immunotherapy. Nat. Immunol. 2, 293–299 (2001). | Article | PubMed  | ISI | ChemPort |
11. Kaplan, E.L. & Meier, P. Nonparametric estimation from incomplete observations. J. Am. Statist. Assoc. 53, 457–481 (1958). | ISI |
12. Ganglberger, E. et al. Allergen mimotopes for 3-dimensional epitope search and induction of antibodies inhibiting human IgE. FASEB J. 14, 2177–2184 (2000). | Article | PubMed  | ISI | ChemPort |
13. Buhl, L., Szecsi, P.B., Gisselo, G.G. & Schafer-Nielsen, C. Surface immunoglobulin on B lymphocytes as a potential target for specific peptide ligands in chronic lymphocytic leukaemia. Br. J. Haematol. 116, 549–554 (2002). | Article | PubMed  | ISI | ChemPort |
14. Suphioglu C. et al. A novel grass pollen allergen mimotope identified by phage display peptide library inhibits allergen-human IgE antibody interaction. FEBS Lett. 502, 46–52 (2001). | Article | PubMed  | ISI | ChemPort |
15. Suphioglu C. et al. A novel grass pollen allergen mimotope identified by phage display peptide library inhibits allergen-human IgE antibody interaction. FEBS Lett. 502, 46–52 (2001). | Article | PubMed  | ISI | ChemPort |
16. Feige, U. & Polla, B.S. Hsp70—a multi-gene, multi-structure, multi-function family with potential clinical applications. Experientia 50, 979–986 (1994). | PubMed  | ISI | ChemPort |
17. Melnick, J. & Argon, Y. Molecular chaperones and the biosynthesis of antigen receptors. Immunol. Today 16, 243–250 (1995). | Article | PubMed  | ISI | ChemPort |
18. Lee, A.S. The glucose-regulated proteins: stress induction and clinical applications. Trends Biochem. Sci. 26, 504–510 (2001). | Article | PubMed  | ISI | ChemPort |
19. Brown, J.M. & Giaccia, A.J. The unique physiology of solid tumors: opportunities (and problems) for cancer therapy. Cancer Res. 58, 1408–1416 (1998). | PubMed  | ISI | ChemPort |
20. Bicknell, R., Lewis, C.E. & Ferrara, N. (eds.) Tumour Angiogenesis (Oxford University Press, Oxford, 1997).
21. Alaiya, A.A. et al. Identification of proteins in human prostate tumor material by two-dimensional gel electrophoresis and mass spectrometry. Cell Mol. Life Sci. 58, 307–311 (2001). | PubMed  | ISI | ChemPort |
22. Thomas, S.A. et al. Detection and distribution of heat shock proteins 27 and 90 in human benign and malignant prostatic tissue. Br. J. Urol. 77, 367–372 (1996). | Article | PubMed  | ISI | ChemPort |
23. Triantafilou, M., Fradelizi, D. & Triantafilou, K. Major histocompatibility class one molecule associates with glucose regulated protein (GRP) 78 on the cell surface. Hum. Immunol. 62, 764–770 (2001). | Article | PubMed  | ISI | ChemPort |
24. Triantafilou, K., Fradelizi, D., Wilson, K. & Triantafilou, M. GRP78, a coreceptor for coxsackievirus A9, interacts with major histocompatibility complex class I molecules which mediate virus internalization. J. Virol. 76, 633–643 (2002). | Article | PubMed  | ISI | ChemPort |
25. Rao, R.V. et al. Coupling endoplasmic reticulum stress to the cell death program: role of the ER chaperone GRP78. FEBS Lett. 514, 122–128 (2002). | PubMed  | ISI | ChemPort |
26. Tamura, Y., Peng, P., Liu, K., Daou, M. & Srivastava, P.K. Immunotherapy of tumors with autologous tumor-derived heat shock protein preparations. Science 278, 117–120 (1997). | Article | PubMed  | ISI | ChemPort |
27. Srivastava, P.K., Kumar, S. & Mendonca, C. in Principles and Practice of Biologic Therapy of Cancer (ed. Rosenberg, S.A.) 1–13 (Lippincott Williams & Wilkins, New York, 2001).
28. Srivastava, P.K., Menoret, A., Basu, S., Binder, R.J. & McQuade, K.L. Heat shock proteins come of age: primitive functions acquire new roles in an adaptive world. Immunity 8, 657–665 (1998). | Article | PubMed  | ISI | ChemPort |
29. Golub, T.R. et al. Molecular classification of cancer: class discovery and class prediction by gene expression monitoring. Science 286, 531–537 (1999). | Article | PubMed  | ISI | ChemPort |
30. Dhanasekaran, S.M. et al. Delineation of prognostic biomarkers in prostate cancer. Nature 412, 822–826 (2001). | Article | PubMed  | ISI | ChemPort |
31. Chen, Y.T. Cancer vaccine: identification of human tumor antigens by SEREX. Cancer J. Sci. Am. 6, S208–S217 (2000). | PubMed  |
32. Tureci, O., Sahin, U. & Pfreundschuh, M. Serological analysis of human tumor antigens: molecular definition and implications. Mol. Med. Today 8, 342–9 (1997). | Article |
33. Sahin, U. et al. Human neoplasms elicit multiple specific immune responses in the autologous host. Proc. Natl. Acad. Sci. USA 92, 11810–11813 (1995). | PubMed  | ChemPort |
34. Sioud, M. & Hansen, M.H. Profiling the immune response in patients with breast cancer by phage-displayed cDNA libraries. Eur. J. Immunol. 3, 716–725 (2001). | Article |
35. Crameri, R., Achatz, G., Weichel, M. & Rhyner, C. Direct selection of cDNAs by phage display. Methods Mol. Biol. 185, 461–469 (2002). | PubMed  | ChemPort |
36. Nakatsura, T. et al. Gene cloning of immunogenic antigens overexpressed in pancreatic cancer. Biochem. Biophys. Res. Commun. 281, 936–944 (2001). | Article | PubMed  | ISI | ChemPort |
37. Kouzmitcheva, G.A., Petrenko, V.A. & Smith, G.P. Identifying diagnostic peptides for Lyme disease through epitope discovery. Clin. Diagn. Lab. Immunol. 8, 150–160 (2001). | Article | PubMed  | ISI | ChemPort |
38. Motti, C. et al. Recognition by human sera and immunogenicity of HBsAg mimotopes selected from an M13 phage display library. Gene 146, 191–198 (1994). | Article | PubMed  | ISI | ChemPort |
39. Puntoriero, G. et al. Towards a solution for hepatitis C virus hypervariability: mimotopes of the hypervariable region 1 can induce antibodies cross-reacting with a large number of viral variants. EMBO J. 17, 3521–3533 (1998). | Article | PubMed  | ISI | ChemPort |
40. Prezzi, C. et al. Selection of antigenic and immunogenic mimics of hepatitis C virus using sera from patients. J. Immunol. 156, 4504–4513 (1996). | PubMed  | ISI | ChemPort |
41. Scala, G. et al. Selection of HIV-specific immunogenic epitopes by screening random peptide libraries with HIV-1-positive sera. J. Immunol. 162, 6155–6161 (1999). | PubMed  | ISI | ChemPort |
42. Dybwad, A., Forre, O., Kjeldsen-Kragh, J., Natvig, J.B. & Sioud, M. Identification of new B cell epitopes in the sera of rheumatoid arthritis patients using a random nanopeptide phage library. Eur. J. Immunol. 23, 3189–3193 (1993). | PubMed  | ISI | ChemPort |
43. Lunardi, C. et al. Systemic sclerosis immunoglobulin G autoantibodies bind the human cytomegalovirus late protein UL94 and induce apoptosis in human endothelial cells. Nat. Med. 6, 1183–1186 (2000). | Article | PubMed  | ISI | ChemPort |
44. Rowley, M.J. et al. Prediction of the immunodominant epitope of the pyruvate dehydrogenase complex E2 in primary biliary cirrhosis using phage display. J. Immunol. 164, 3413–3419 (2000). | PubMed  | ISI | ChemPort |
45. Cortese, I. et al. Identification of peptides specific for cerebrospinal fluid antibodies in multiple sclerosis by using phage libraries. Proc. Natl. Acad. Sci. USA 93, 11063–11067 (1996). | Article | PubMed  | ChemPort |
46. Mennuni, C. et al. Selection of phage-displayed peptides mimicking type 1 diabetes-specific epitopes. J. Autoimmun. 9, 431–436 (1996). | Article | PubMed  | ISI | ChemPort |
47. Schroder, F.H. et al. The TNM classification of prostate cancer. Prostate 4, 129–138 (1992). | ChemPort |
48. Gleason, D.F. in Urologic Pathology: The Prostate (ed. Tannenbaum, M.) 171–197 (Lea & Febiger, Philadelphia, PA, 1977).
49. Breiman, L., Friedman, J.H., Olshen, R.A. & Stone, C. in Classification and Regression Trees pp. 1–336 (Wadsworth International Group, Belmont, California, 1984).

	Top

Competing interests statement:  The authors declare competing financial interests.