Background: Antibodies isolated from cancer patients have been used to identify genes encoding tumor-associated antigen epitopes relevant to immune responses in cancer patients. In this report, we used an immunoglobulin G (IgG) purified from serum of a patient with breast cancer to identify its corresponding epitope, gene, and protein—retinoblastoma-binding protein-1-like protein-1 (RBP1L1)—and determined whether it is a potential molecular marker for various cancers. Methods: IgG purified from the serum of a patient with breast cancer was used to screen an MCF-7 breast cancer cell complementary DNA (cDNA) expression library for immunoreactive clones. The cDNAs identified were cloned and sequenced. Immunoreactivity of specific amino acids in the epitope was determined by western blot analysis and enzyme-linked immunosorbent assay. The cellular location of the antigen was determined by immunoperoxidase staining with purified RBP1L1-specific IgG. Gene expression in various human carcinomas and normal tissues was examined by northern blot analysis and the reverse transcription–polymerase chain reaction. Results: Our purified IgG recognized just one epitope on RBP1L1. The complete 5802-base-pair RBP1L1 cDNA encodes a 1226-amino acid protein containing the antigenic epitope IKPSLGSKK. The derived protein sequence of RBP1L1 shares 74% and 37% amino acid identity, respectively, with a partial sequence of the retinoblastoma-binding protein and the complete sequence of retinoblastoma-binding protein-1. The RBP1L1 epitope was localized to the cytoplasm of MCF-7 cells but was not detected in peripheral blood mononuclear cells. High expression of RBP1L1 messenger RNA was found in human breast, lung, colon, pancreatic, and ovarian cancers and in normal testis, but expression was limited in other normal tissues. Conclusions: RBP1L1 appears to be a molecular marker associated with a broad range of human malignancies.
Molecular identification of human tumor antigens that can elicit both antibody and cellular immune responses is a central objective in developing immunotherapies that are active and specific. Because of their antigen-binding specificity and human origin, human monoclonal antibodies specific for tumor-associated antigens are useful probes for the molecular identification of tumor-associated proteins. Using a human immunoglobulin M (IgM) monoclonal antibody, we previously identified an immunogenic peptide of a 43-kilodalton (kd) human cancer-associated antigen (1). The peptide induced both antibody and cytotoxic T-lymphocyte (CTL) responses in melanoma patients (2– 4), and high serum titers of an IgM antibody against this peptide antigen were associated with improved prognosis (4).
Autologous or allogeneic serum antibodies from cancer patients are an alternative to monoclonal antibodies. Although not as specific as monoclonal antibodies, serum antibodies can be useful probes to identify tumor-associated antigens if they can be isolated from the serum in high titers (5–7). One well-defined antigen identified with autologous serum antibodies is NY-ESO-1 (8). The transcript for NY-ESO-1 is highly expressed in various human cancers, including breast, prostate, and ovarian cancers and melanoma; however, among normal human tissues, it is restricted to the testis. Both human serum antibodies and CTLs are responsive to the antigen; the epitope recognized by CTLs is a nonameric peptide, SLLMWITQC, corresponding to NY-ESO-1 amino acids 157–165 (9). In a previous study (7), we used an allogeneic immunoglobulin G (IgG) antibody prepared from the serum of a patient with breast cancer to screen a complementary DNA (cDNA) expression library of MCF-7 breast cancer cells. We identified a tumor-associated antigen that corresponded to the retinoblastoma-binding protein-1 (RBP1). Although the RBP1 gene is present in normal human cells and in human cancer cells (10), expression of the antigenic epitope is restricted to certain types of cancer, such as breast, prostate, and renal cancers. Human RBP1 contains an heptameric antibody-binding peptide epitope, KASIFLK, corresponding to RBP1 amino acids 250–256, as well as two overlapping decameric peptide epitopes, GLQKASIFLK and KASIFLKTRV, respectively, corresponding to RBP1 amino acids 247–256 and 250–259. These epitopes were highly immunogenic, inducing both specific antibody and CTL responses. In fact, the CTL responses induced by the decameric peptide epitopes exhibited strong cytotoxic activity against HLA-A2- and HLA-A3-positive breast cancer cells (11). Thus, antigenic epitopes identified by human antibodies appear to induce T-cell responses in cancer patients and, therefore, may be useful targets for active and highly specific immunotherapy.
In our investigation of the epitope KASIFLK and its IgG antibody, we discovered a second IgG antibody in serum from the same patient (7). To identify additional tumor-associated genes containing epitopes that can elicit immune responses against cancer cells, we cleared anti-KASIFLK antibodies from an IgG preparation by affinity chromatography and used the resulting antibody preparation at a reduced dilution to rescreen the MCF-7 cDNA expression library.
Herein, we report the isolation and characterization of retinoblastoma-binding protein-1-like protein-1 (RBP1L1), a novel protein containing an antigenic nonameric peptide epitope that reacts specifically with the purified antibody preparation. The gene encoding this antigen was initially called breast cancer-associated antigen (BCAA; GenBank accession number 214114); we have since renamed this antigen retinoblastoma-binding protein-1-like protein-1 (RBP1L1; the accession number has remained the same) to reflect its homology with RBP1.
Materials and Methods
Human IgG Purification and MCF-7 cDNA Library Immunoscreening
An IgG fraction was prepared from the serum of a patient with breast cancer (7). Antibodies against Escherichia coli proteins and antibodies against the heptapeptide KASIFLK in the IgG preparation were, respectively, removed first by affinity chromatography with an immobilized E. coli Y1090 lysate affinity resin and then with a KASIFLK peptide affinity resin. A λgt11 cDNA expression library from MCF-7 cells was screened with the resulting purified IgG (at 20 μg/mL, an approximately 1 : 400 dilution of the original serum) by the method of Young and Davis (12).
Expression and Analysis of Fusion Proteins
E. coli Y1090 and Y1089 were used for screening and protein expression of λgt11 recombinants, respectively. Immunopositive recombinant clones 131 and 151 were detected, and plaques were purified. Lysogenic E. coli Y1089 cells, grown in Luria broth (LB) medium (pH 7.5) containing 0.2% maltose at 37 °C to late logarithmic phase, were diluted in LB medium containing 10 m M MgCl 2 and infected with λgt11 recombinant clones 151 and 131 at a multiplicity of infection of approximately 5 for 20 minutes at 32 °C. Infected cells were incubated on LB plates at 32 °C overnight. Twenty colonies from each clone were streaked onto two LB plates. The first plate was incubated at 42 °C, and the second plate was incubated at 32 °C. Colonies that grew at 32 °C but not at 42 °C were chosen for fusion protein expression in E. coli Y1089 as described previously (1). Protein samples were separated by sodium dodecyl sulfate (SDS)–polyacrylamide gel electrophoresis (PAGE) on 4%–20% gradient gels, and proteins were stained with Coomassie blue R-250 or electrotransferred to nitrocellulose membranes for western blot analysis. The western blot method of assessing IgG antibody reactivity against β-galactosidase–antigen fusion proteins has been described previously (13). Peroxidase-conjugated rabbit anti-human IgG (Pierce Chemical Co., Rockford, IL) was used as the secondary antibody.
Sequence Analysis of Immunoreactive Clones
DNA inserts from immunoreactive λgt11 clones 131 and 151 were amplified by polymerase chain reaction (PCR) with λgt11 forward and reverse primers and cloned into the PCRII plasmid (Invitrogen Corp., Carlsbad, CA). Total RNAs from MCF-7 cells and peripheral blood mononuclear cells (PBMCs) were amplified by reverse transcription (RT)–PCR, and cDNA from human ovarian cells (Marathon-Ready™ kit; Clontech Laboratories, Palo Alto, CA) was amplified by PCR. Products from all reactions were cloned into PCRII plasmids. Plasmid DNA, prepared as described by the manufacturer (Qiagen Inc., Valencia, CA), was completely sequenced with an ABI Prism automated sequencer (Perkin-Elmer Corp., Norwalk, CT) at the DNA-sequencing core facility at the University of California, Los Angeles, and the University of California, Irvine. DNA and derived amino acid sequences were compared with sequences in GenBank by use of the BLAST program at the National Center for Biotechnology Information (NCBI), Bethesda, MD. Publication output for RBP1 and RBP1L1 protein sequences alignment was generated with the Baylor College of Medicine (Houston, TX) search launcher BOXSHADE program.
Northern Blot Analysis
Northern blot analysis was performed on human multiple tissue northern (MTN II) and human endocrine system MTN blots (Clontech Laboratories). On these blots, each lane contains approximately 2 μg of poly(A) + RNA from human spleen, thymus, prostate, testis, ovary, small intestine, colon, peripheral blood leukocyte, pancreas, adrenal medulla, thyroid, adrenal cortex, or stomach. Membranes were hybridized with a 32 P-labeled 441-base-pair (bp) fragment of RBP1L1 cDNA that was amplified by PCR from immunopositive clone 151. Blots were washed twice in 0.1% standard saline citrate and 0.1% SDS for 20 minutes and then exposed to Kodak BIOMAX film at –80 °C with an intensifying screen for 36 hours. As a loading control, blots were stripped and reprobed with the control glyceraldehyde-3-phosphate dehydrogenase (G3PDH) gene to prove RNA integrity.
Reverse Transcription–Polymerase Chain Reaction
A stop codon at nucleotide 322 in clone 151 was investigated to determine whether it was the result of a somatic mutation in the gene or an artifact of the construction of the cDNA library. Total RNA (1 μg) from MCF-7 cells and PBMCs was primed with (dT) 12–18 and reverse transcribed with SuperScript RT (Life Technologies Inc. [GIBCO BRL], Rockville, MD). A 470-bp segment of RBP1L1 spanning the region that contained the stop codon in the clone 151 was amplified by PCR with sequence-specific primers (sense = 5′-ATGGAGGAGGAGAGGAATATAATACCAAG-3′; antisense = 5′-CTGAAATGGTGGTTTGGACAAGCGCCGA-3′) (Operon Technologies, Alameda, CA). PCR was performed as follows: denaturation at 94 °C for 2 minutes, followed by 35 amplification cycles of denaturation at 94 °C for 30 seconds, annealing at 68 °C for 30 seconds, and extension at 72 °C for 1 minute, with a final extension at 72 °C for 7 minutes in a thermal cycler (Perkin-Elmer Corp.). Several RT–PCR products were sequenced to determine whether they also encoded a stop codon at the same position.
To identify the 5′ untranslated region (UTR) and the remaining 5′ cDNA sequence of RBP1L1, we used a rapid amplification of 5′ cDNA ends system with cDNA from human ovarian cells (Marathon-Ready™; Clontech Laboratories). Because the complete cDNA sequence of the rat retinoblastoma-binding protein-1-related protein (rRBP1-R accession number AF245512) shares 84% identity to the partial RBP1L1 gene sequence, our initial amplification used a sense primer (5′-AGAGTCACCATGAAGGCCCTTGATGATGAGC-3′) corresponding to the 5′ end of rRBP1-R at 5′ nucleotides 30–57 and an antisense primer (5′-TGGGATTATATTCCTCTCCTCCTCCATC-3′) corresponding to the 5′ end of the incomplete human RBP1L1 (clone 151, 4032 bp) at 5′ nucleotides 12–39. By sequencing this reaction product, we identified the remaining 5′ cDNA sequence of human RBP1L1. To obtain the 5′ UTR sequence, we amplified this region by using adaptor primer 1 (AP1; Clontech Laboratories) and an antisense primer (5′-ATACGGCATCAGGCTTTGGTGCAGTGTCAC-3′) that corresponds to 5′ nucleotides 741–771 of rRBP1-R. PCR was performed as follows: denaturation at 94 °C for 2 minutes, followed by 35 amplification cycles of denaturation at 94 °C for 30 seconds, annealing at 68 °C for 30 seconds, and extension at 72 °C for 3 minutes, with a final extension at 72 °C for 7 minutes in a thermal cycler (Perkin-Elmer Corp.).
To examine the expression level of RBP1L1 messenger RNA (mRNA), quantitative PCR was performed on cDNA panels from normal and cancer tissues (MTC II and tumor MTC, respectively; Clontech Laboratories). These panels contain first-strand cDNA from human cancers of the breast, lung, colon, lung, prostate, colon, ovary, and pancreas and from normal human thymus, prostate, testis, ovary, small intestine, colon, and peripheral blood leukocytes. A set of RBP1L1-specific oligonucleotides (sense = 5′-AGGGAATAGCTCGCCAGCAGGTTTTGATG-3′; antisense = 5′-TCGGCACTTGTCATATTTTCCAGGTCCGAC-3′) spanning a 441-bp cDNA segment was used for PCR with 1 ng of cDNA from each tissue. PCR was performed as follows: denaturation at 94 °C for 2 minutes, followed by 25 amplification cycles of denaturation at 94 °C for 30 seconds, annealing at 68 °C for 30 seconds, and extension at 72 °C for 1 minute, with a final extension at 72 °C for 7 minutes. Reaction products (6 μL) were evaluated by gel electrophoresis on 2% gels, and bands were visualized with ethidium bromide. As a loading control, amplified products from a set of G3PDH control PCR primers (Clontech Laboratories) were included to ensure the integrity of the cDNA samples of each tissue.
Purification of Fusion Proteins and Antibody
β -Galactosidase fusion protein (1.5 mg) was purified from 50 mg of total protein extracted from immunopositive clone 151 with the ProtoSorb LAC Z immunoaffinity adsorbent system (Promega Corp., Pittsburgh, PA). The purified fusion protein (2 mg/mL) was coupled to cyanogen bromide-activated Sepharose 4B by the manufacturer's protocol (Amersham Pharmacia Biotech Inc., Alameda, CA). Three milliliters of the purified IgG (1.5 mg/mL) was applied to the fusion protein affinity column, and the column was washed extensively with phosphate-buffered saline (PBS) to remove nonspecifically adsorbed proteins. The IgG specifically bound to the RBP1L1 epitope peptide was next eluted with 0.1 M acid glycine at pH 3.0, and the pH was adjusted immediately with 1.0 M Tris buffer at pH 9.0. This IgG was named purified RBP1L1-specific IgG.
As a negative control antibody, we used a human IgG that was isolated from a human hybridoma cell line and was not immunoreactive to MCF-7 cells. This control IgG was prepared by growing human hybridoma cells in RPMI-1640 medium (Life Technologies Inc.) containing 10% fetal bovine serum until senescence. Conditioned culture medium was removed from the cultures, and debris was removed by centrifugation at 1000 g for 10 minutes at 4 °C. IgG was purified with a protein A–Sepharose 4B Fast Flow column (Amersham Pharmacia Biotech Inc.) and eluted with 0.1 M acid glycine at pH 3.0. The pH of the eluted IgG was adjusted immediately with 1 M Tris buffer at pH 9.0, and the purified IgG was dialyzed against PBS.
Carboxyl-Terminal Truncation of Glutathione S-Transferase (GST)–RBP1L1 and Expression of GST Fusion Proteins
To identify the site of the immunogenic epitope within the first 107 residues of clone 151, a series of carboxyl-terminal deletion constructs was generated by amplifying overlapping 5′ DNA fragments of clone 151 (nucleotides 1–105, 1–114, 1–142, 1–154, 1–208, 1–235, and 1–279) and cloning them into the GST expression vector pGEX-2T. The sense PCR primer contained a Bam HI, and all antisense PCR primers contained an Eco RI site (indicated in boldface type below), allowing amplified products to be cloned into the respective sites of pGEX-2T. The primers used to generate each of the GST–RBP1L1 constructs were as follows: nucleotides 1–105 (sense = 5′-TGAC GGATCC TGCGGCCGCAAAG-3′; antisense = 5′-TCTG GAATTC CTTCCCAGAGAGAGGGC-3′); nucleotides 1–114 (antisense = 5′-TCTG GAATTC ATTCTTTTTACTTCC-3′); nucleotides 1–142 (antisense = 5′-TCTG GAATTC CCTGATCAGAATGTGTAGG-3′); nucleotides 1–154 (antisense = 5′-TCTG GAATTC CCCAGATTTTCATTGTCTTC-3′); nucleotides 1–208 (antisense = 5′-TCTG GAATTC CCTACCCTAGTTGTGTC-3′); nucleotides 1–235 (antisense = 5′-TCTG GAATTC GCTTTAATCCATTCATC-3′); nucleotides 1–279 (antisense = 5′-TCTG GAATTC GCTTTTTCTTCCTCAGC-3′). Expression of GST–RBP1L1 fusion proteins was induced with 0.1 m M isopropyl β- d -thiogalactopyranoside (IPTG). Bacterial lysates were subjected to SDS–PAGE on 10% gels and subsequently transferred to nitrocellulose membranes. The reactivity of the purified IgG against the GST fusion proteins was assessed by western blotting as described previously (13).
Peptides were synthesized on the basis of the deduced amino acid sequence of the immunoreactive clone and enriched by high-pressure liquid chromatography to greater than 80% purity (Research Genetics Inc., Huntsville, AL).
Inhibition of Antibody Activity by Synthetic Peptides
To determine the minimal number of amino acids necessary for antibody binding, we tested synthetic peptides of various lengths for their ability to inhibit the binding of purified RBP1L1-specific IgG to GST–RBP1L1 fusion protein by western blot analysis. The synthetic peptides tested were IKPSLGSKKN, IKPSLGSKK, KPSLGSKKN, IKPSLGSK, KPSLGSK, PSLGSKK, IKPSLGS, KPSLGSK, and SLGSKKN. The final concentration of each peptide was 100 μg/mL, and the concentration of purified RBP1L1-specific IgG was 1.5 μg/mL. The mixture of antigen and purified RBP1L1-specific IgG was agitated at 4 °C overnight, and the inhibition of antibody binding was assessed by western blotting with the GST–RBP1L1 fusion protein as the target antigen.
Enzyme-Linked Immunosorbent Assay With Synthetic Peptide Antigens
The binding of purified RBP1L1-specific IgG to synthetic peptides was assessed by enzyme-linked immunosorbent assay (ELISA). Wells of 96-well polystyrene plates (Reacti-Bind maleic anhydride-activated plates; Pierce Chemical Co.) were coated with each of the peptides (100 μL at 5 μg/mL in PBS; IKPSLGSKKN, IKPSLGSKK, KPSLGSKKN, IKPSLGSK, KPSLGSK, and PSLGSKK), incubated overnight at 4 °C, and then blocked with 1% bovine serum albumin. Purified RBP1L1-specific IgG (200 μg/mL) was serially diluted 1 : 250, 1 : 500, 1 : 1000, 1 : 2000, and 1 : 4000 times, then added to triplicate peptide-coated wells (500 ng per well), and incubated for 3 hours at room temperature. Peroxidase-conjugated goat anti-human IgG (Pierce Chemical Co.) was added at room temperature for 1 hour, followed by the addition of o -phenylenediamine dihydrochloride (400 μg/mL) in 0.03% hydrogen peroxide. Reactivity was measured at 490 nm. Background absorbance without the primary antibody was subtracted from each sample's absorbance.
Immunoperoxidase Staining of MCF-7 Cells and PBMCs
PBMCs were centrifuged (1500 rpm for 5 minutes at 4 °C) onto glass slides by use of a cytocentrifuge, and MCF-7 cells were plated on slide chambers and incubated in the CO 2 incubator for 2 days. The cells were fixed in 2% paraformaldehyde. The slides were dipped in PBS for 5 minutes and then treated sequentially with 0.1% Triton X-100 for 10 minutes to permeabilize cell membranes and with 3% hydrogen peroxide for 10 minutes to quench endogenous peroxidase activity. The slides were then washed in running water for 5 minutes, and the cells were stained immunocytochemically by use of the VECTASTAIN ABC system (Vector laboratories Inc.). Purified, RBP1L1-specific IgG (10 μg/mL), a human monoclonal IgG (10 μg/mL; MCF-7 cell negative control), and PBS (PBMC negative control) were incubated overnight with slides that had been blocked with goat serum. After being washed with PBS, the slides were incubated with biotin-labeled goat anti-human IgG at room temperature for 1 hour. After another wash with PBS, the slides were incubated with ABC reagent for 30 minutes and then with the horseradish peroxidase substrate for color development. The slides were counterstained with hematoxylin–eosin.
Identification and Expression Analysis of Immunopositive Clones
We screened 4 × 10 6 plaques of a λgt11 MCF-7 cDNA expression library with human IgG purified from the serum of a patient with breast cancer and detected two positive clones, clones 131 and 151. PCR analysis of both clones with the use of λgt11 forward and reverse primers identified a 4.0-kilobase (kb) insert. Lysogenic E. coli Y1089 was infected with λgt11 DNA from clones 131 and 151, and expression of the β-galactosidase fusion protein was induced with IPTG. Western blot analysis of the whole-cell lysates with the purified IgG detected a fusion protein of about 128 kd for each clone (Fig. 1). Given that β-galactosidase is 116 kd, the size of the open reading frame of each clone would be 12 kd. Therefore, only a small portion of the approximately 4.0-kb insert appeared to encode the protein detected by the purified IgG.
Sequence Analysis of Immunopositive Clones
cDNA inserts of immunopositive clones 131 and 151 were amplified by PCR, cloned into the PCRII vector, and sequenced completely on both strands. The 4032-bp sequence of the two clones was identical. Beyond the β-galactosidase sequence, each clone contained an open reading frame of 107 amino acids ending at a stop codon (TAA) at nucleotides 322–324. We investigated this premature stop codon by sequencing RT–PCR products, amplified from total RNA from MCF-7 cells and PBMCs, around nucleotide 322. None of the 10 cDNA clones examined had a TAA stop codon at nucleotides 322–324; all had the sequence AAA. Consequently, we believe that the stop codon resulted from a mutation that was introduced during the synthesis of MCF-7 cDNA. Correction of this mutation extended the open reading frame an additional 683 nucleotides and predicted that the full-length protein was 790 amino acids long, consistent with the length of other retinoblastoma-binding proteins.
To obtain the full-length cDNA sequence of the RBP1L1 gene, we used rapid amplification of 5′ cDNA ends and identified additional 1770 nucleotides at the 5′ end of clone 151 that contained the 5′ UTR and the rest of the 5′ cDNA sequence of RBP1L1. The complete cDNA sequence (5802 bp) and the predicted amino acid sequence (1226 amino acids) are shown in supplemental Fig. 1 on the Journal's Web site at http://jnci.oupjournals.org . The GenBank accession number of the partial (BCAA) and complete RBP1L1 cDNA sequence is AF214114.
NCBI BLAST search results revealed that the following four protein sequences are similar to the sequence of RBP1L1: 1) human retinoblastoma-binding protein (hRBP; GenBank accession number AF083249), 2) human RBP1-like protein (hRBP1-L; GenBank accession number NP057485), 3) rat RBP1-related protein (rRBP1-R; GenBank accession number AF245512), and 4) human RBP1 (hRBP1; GenBank accession number NM002892). Only partial sequences of hRBP and hRBP1-L genes are available. Alignment of the derived amino acid sequences of RBP1L1 and human RBP1 is shown in supplementary Fig. 2 on the Journal's Web site (see above). RBP1L1 shares 74%, 72%, 86%, and 37% amino acid sequence identity with human RBP, human RBP1-L, rat RBP1-R, and human RBP1, respectively.
Expression of RBP1L1 in Normal Human Testis and Human Cancer Tissues
Expression of RBP1L1 mRNA in normal human tissues was analyzed by northern blotting with RBP1L1-specific probes. Abundant expression of a 7.5-kb transcript was detected in the testis, with less expression detected in the thymus, prostate, and ovary (Fig. 2, A). Expression was very low or absent in the nine other adult tissues examined. To study the expression of RBP1L1 mRNA in a variety of cancer tissues, we used RT–PCR with a set of RBP1L1-specific primers. Of the six histologic types of cancer examined, five (breast, ovary, lung, colon, and pancreatic cancers) expressed substantially higher levels of RBP1L1 mRNA than the six normal tissues examined (thymus, prostate, ovary, small intestine, colon, and peripheral blood leukocyte; Fig. 2 , B). Expression in the testis and expression in the cancers examined were comparable. Essentially no expression was detected in prostate cancer cells. Differential expression was detected by PCR under nonsaturating conditions with restricted PCR cycles. Thus, RBP1L1 has a very restricted tissue distribution, being expressed predominantly in cancer tissues where it may contribute to the pathophysiology.
Antigen Epitope Mapping in Recombinant Protein From Clone 151
Peptide sequences that contain the antigenic epitope were identified by testing the following GST–RBP1L1 fusion constructs: construct 1, nucleotides 1–105; construct 2, nucleotides 1–114; construct 3, nucleotides 1–142; construct 4, nucleotides 1–154; construct 5, nucleotides 1–208; construct 6, nucleotides 1–235; and construct 7, nucleotides 1–279. Western blot analysis revealed that the purified IgG recognized GST–RBP1L1 fusion constructs 2–7 but not GST–RBP1L1 fusion construct 1, suggesting that fusion constructs 2–7 contained the antigen epitope. Because fusion constructs 1 and 2 differed by only three residues at the carboxyl-terminal end, we expected the epitope to be located between nucleotides 84 and 114 in clone 151 within a 10-amino acid sequence.
The minimum number of amino acids required for the antibody binding was determined by testing various truncated synthetic peptides in ELISAs and on western blots. The following set of peptides was generated by sequentially reducing the 10-amino acid peptide by one amino acid from either the amino terminus or the carboxyl terminus: IKPSLGSKKN, IKPSLGSKK, IKPSLGSK, IKPSLGS, KPSLGSKKN, KPSLGSKK, KPSLGSK, and PSLGSKKN. The antigenic activity of each synthetic peptide (5 μg/mL) was tested for its ability to competitively inhibit the binding of the purified RBP1L1-specific IgG (0.5 μg/mL) to the GST–RBP1L1 fusion protein on western blots. The peptides IKPSLGSKKN and IKPSLGSKK completely inhibited the purified RBP1L1-specific IgG from binding to the GST–RBP1L1 fusion protein from clone 151, indicating that the carboxyl-terminal asparagine (N) is not essential for the antibody binding. However, deletion of the carboxyl-terminal lysine (K) from IKPSLGSKK substantially reduced the inhibition, and deletion of amino-terminal isoleucine (I) from IKPSLGSKK eliminated the inhibition. We verified this result with an ELISA. Thus, IKPSLGSKK, amino acid residues 465–473 of the full-length RBP1L1 protein sequence, was the minimal epitope recognized by the purified RBP1L1-specific human IgG (Fig. 3 ).
Cytologic and Immunocytochemical Analyses of Antigen
The purified RBP1L1-specific IgG that had been affinity purified by affinity chromatography on RBP1L1 fusion-protein affinity resin was used to determine the cellular location of the IKPSLGSKK epitope antigen in MCF-7 cells and in PBMCs. Immunocytochemical staining with the purified RBP1L1-specific IgG was strong in the cytoplasm, with little or no staining of other cellular components (Fig. 4, A and B). Antigen-specific staining was not observed with PBMCs obtained from healthy donors ( Fig. 4 , C and D).
In this report, we have identified a novel retinoblastoma-binding proteinrelated gene, RBP1L1, isolated from an MCF-7 cDNA expression library screened with a purified human IgG from a patient with breast cancer. We detected little or no expression of RBP1L1 transcripts in normal adult pancreas, prostate, ovary, adrenal medulla, thyroid, adrenal cortex, spleen, thymus, colon, stomach, and PBMCs by northern blotting, although expression was detected in normal testis. Consistent with this observation, quantitative RT–PCR analysis of normal tissues did not detect RBP1L1 mRNA, except in the testis. However, in cancers of the breast, lung, colon, ovaries, and pancreas, high levels of RBP1L1 mRNA were detected. Thus, RBP1L1 mRNA is expressed abundantly in cancer cells and in normal testicular cells. The mRNA differential expression detected by PCR was performed under nonsaturating conditions with restricted PCR cycles. This restricted pattern of expression suggests that RBP1L1 may be a diagnostic molecular marker for a broad range of human cancers.
The purified human IgG binds to a nonameric peptide antigen, IKPSLGSKK, encoded by the RBP1L1 gene. Immunohistochemical and cytologic studies with this antibody show that the antigen is located in the cytoplasm. Although T-cell recognition of antigen was not investigated in this study, previous studies (2, 3 , 7 , 8 , 10 ) have shown that serologically identified antigens are recognized by human leukocyte antigen (HLA) class I-restricted CTLs. We have reported previously (7) that RBP1 encodes a heptameric peptide antigen, KASIFLK (RBP1 amino acids 250–256), that was detected by a human IgG purified from the serum of a patient with breast cancer. In vitro stimulation of HLA-A2- and HLA-A3-positive PBMCs with decameric peptides containing the KASIFLK antigen generated peptide-specific CTLs that were highly cytotoxic to HLA-A2- and HLA-A3-positive breast cancer cells but not to normal cells (11). Similarly, tumor antigens recognized by HLA class I-restricted CTLs are also recognized by antibodies and can induce an antibody response in patients with tumors. MAGE, TRP-2, and gp100 tumor-associated antigens, which were initially characterized for their HLA-restricted CTL reactivity, elicited strong antibody responses to their respective recombinant antigens in melanoma patients immunized with an antigen-containing melanoma cell vaccine (14, 15 ). Such immunologically dominant peptide antigens may be potential targets for cancer cell destruction via a dual-effector immune system.
Because of the extensive sequence conservation between RBP1L1 and RBP1 (64% identity in the N-terminal 450 amino acids and 42% identity in the C-terminal 300 amino acids), one can speculate that the functional pathway of RBP1L1 in part mirrors that of RBP1. Like RBP1, RBP1L1 also contains an ARID and a BRIGHT DNA-binding domain (NCBI; search the Conserved Domain Database). In addition, RBP1L1 and RBP1 share certain immunologic features. Both antigens are overexpressed in human cancer cells and contain antigenic peptide sequences (a heptamer and a nonamer, respectively) that are detected by human antibodies. Both antigens are localized mainly to the cytoplasm of cancer cells when detected by the human IgG antibodies, although RBP1 has been reported (16) to be primarily in the cell nucleus when a rabbit antibody against C-terminal 15 amino acids peptide of RBP1 was used. The details of the cell-regulatory functions and pathways of retinoblastoma-binding proteins have not been well defined and are currently under extensive investigation (10, 16 – 18 ). The retinoblastoma tumor suppressor protein (19) plays a critical role in controlling the cell cycle and cell proliferation (20), and loss of the retinoblastoma gene is associated with malignant transformation of cells. The retinoblastoma gene product (pRB) and related proteins p130 and p107 (21) regulate cell cycle progression through interactions with transcription factors belonging to the E2F family (22). RBP1 binds to the pocket of pRB and is thus termed a pocket protein (10); RBP1 represses transcriptional activity by interacting with p130–E2F and pRB–E2F complexes during cell cycle arrest. Overexpression of RBP1 inhibits E2F-dependent gene expression and suppresses cell growth (18). RBP1 represses E2F-dependent transcription by recruiting histone deacetylase complexes, which do not interact directly with retinoblastoma family proteins but rather use RBP1 to target the pocket of these proteins (23). Because of the high protein sequence identity between RBP1L1 and RBP1 and their common immunologic features, RBP1L1 may also be able to associate with the pRB pocket and to regulate the transcription of genes that control the cell cycle, differentiation, proliferation, and apoptosis.
Genes with high mRNA expression in human testis and tumor cells are not unique. MAGE (24), BAGE (25), and GAGE (26), initially discovered as CTL-responding antigens, fall into this category. In addition, mRNAs of many of the genes for human cancer-associated antigens, identified by use of human antibodies to screen recombinant cDNA expression libraries, are also overexpressed in the testis (8, 27 , 28 ). It is interesting that some of the genes identified by antibody screening have been shown to belong to the MAGE family (29). However, to our knowledge, our study is the first to show that a retinoblastoma-binding protein-related gene encodes a cancer-associated antigen that is highly expressed by various cancer types and normal testicular cells. As yet, RBP1L1 shows no homology to any reported cancer/testis genes that might be targets for cancer immunotherapy (30, 31 ). Because the testis is an immunologically privileged site, testicular cells expressing tumor antigens should escape direct contact by antigen-presenting cells. Consequently, CTLs and antibody responses induced by peptide antigens derived from cancer/testis antigen genes might attack autologous cancer cells without killing normal testicular cells. The lack of HLA class I expression on the surface of testicular cells also favors their ability to escape from CTL recognition and killing. Thus, IKPSLGSKK, or a peptide containing this epitope, may have potential as a cancer vaccine, a possibility that merits further investigation.
We thank Dr. Sanjiv Ghanshani (Department of Physiology and Biophysics, College of Medicine, University of California, Irvine) for critical reading of the manuscript. We also thank Drs. Duan-ren Wen (John Wayne Cancer Institute, Santa Monica, CA) and Ning Ru (Department of Microbiology and Molecular Genetics, College of Medicine, University of California, Irvine) for their expertise in the immunocytochemical analysis of the antibody and Mr. Eddy Suruki (Department of Microbiology and Molecular Genetics, College of Medicine, University of California, Irvine) for his excellent technical assistance.
- northern blotting
- polymerase chain reaction
- gene expression
- western blotting
- enzyme-linked immunosorbent assay
- amino acids
- immune response
- amino acid sequence
- fusion protein
- clone cells
- dna, complementary
- rna, messenger
- immunoglobulin g
- breast cancer
- ovarian cancer
- tumor antigens
- breast cancer cell
- immunoperoxidase stain
- peripheral blood mononuclear cell
- mcf-7 cells