Tumor vaccines represent one type of molecularly targeted therapy being investigated for the treatment of prostate cancer. Although many prostate-specific proteins are being tested as target antigens for prostate cancer vaccines, most are not natural targets of an immune response in patients with cancer. Using sera from cancer patients, several research groups have identified a large family of immunologically recognized proteins whose expression is normally confined to immune-privileged testis tissue but which may be expressed in cancers of different histological origins. These proteins, so-called cancer-testis (CT) antigens, are appealing targets for immune-based therapies because they are essentially tumor-restricted antigens and there is less risk of preexisting immune tolerance. In addition, specifically targeting these proteins by means of vaccines should reduce the risk of potential autoimmune reactions to normal tissues. In the current study, we hypothesize that prostate CT antigens can be identified using a SEREX screening method with sera from patients with prostate cancer and probing with a human testis cDNA expression library. We have identified several potential prostate cancer antigens with predominantly testis-specific expression in normal tissues, including MAD-CT-1 (protamine 2) and MAD-CT-2. Each was independently identified from different subjects with prostate cancer. Antigens identified by these studies can be investigated further as potential prostate cancer tumor antigens.
This article was published in Cancer Immunity, a Cancer Research Institute journal that ceased publication in 2013 and is now provided online in association with Cancer Immunology Research.
Prostate cancer is a significant health problem that is common in men above the age of 50. It is the most commonly diagnosed malignancy, and the second leading cause of cancer-related death, in men in the United States (1). At present, there is no cure for metastatic prostate cancer, and approximately 12% of patients diagnosed with prostate cancer ultimately die from the disease. There is considerable interest in developing molecularly targeted therapies for prostate cancer, including tumor vaccines (2). The results of several clinical vaccine trials suggest a clinical benefit to patients with prostate cancer, and phase III trials are currently underway in patients with androgen-independent prostate cancer (3, 4). The antigens chosen for inclusion in antigen-specific prostate cancer vaccines have predominantly been proteins with known prostate-restricted expression, including prostate-specific antigen, prostate-specific membrane antigen, and prostatic acid phosphatase. There is little evidence that any one of these antigens is superior to another as a vaccine target. To date, there has been little evaluation of prostate cancer-associated proteins that may be natural targets of CTLs. In contrast, the identification of antigens recognized by melanoma TILs has guided the evaluation of specific antigens for melanoma tumor vaccines (5, 6). Identifying immunologically recognized proteins in patients with prostate cancer could lead to the development of other potential vaccine target antigens. Unfortunately, the paucity of prostate cancer tissue obtained from individual subjects, and the difficulties associated with propagating autologous prostate cell lines compared with melanoma cell lines, has for the most part prohibited an evaluation of TILs analogous to those conducted with melanoma (7).
It has already been demonstrated that IgG responses to tumor-associated proteins exist in cancer patients, making it possible to use human sera to identify immunologically-recognized potential tumor antigens (8). SEREX (serological analysis of recombinant cDNA expression libraries of human tumors), is a robust methodology that uses IgG antibody-based screening to identify potential antigens of a tissue-specific cDNA expression library using human sera (8). Many antigens discovered through SEREX are known targets of CTLs (9, 10). Given the limitations described above, SEREX has become an attractive method for identifying tumor-associated antigens of solid tumors, including prostate cancer (11, 12, 13, 14, 15). The use of SEREX in other solid tumor systems has led to the identification of a large family of proteins whose expression is normally confined to testis tissue (16). Antigens restricted to the testis are considered immunologically privileged, given the absence of MHC molecule expression on germ cells (17). The aberrant expression in cancers of proteins that are normally restricted to germ cells may make them available for immune recognition (11). Members of this family of CT antigens, which includes MAGE, BAGE, GAGE-1, HOM-MEL-40, and NY-ESO-1, are expressed in a variety of solid tumors and have been found to be immunologically recognized in the sera of patients with cancers of diverse types (16, 18). Given their restricted tissue expression, in many cases, cancer-specific expression, CT antigens are promising candidates for antitumor vaccines (18). Moreover, in the case of prostate cancer, many patients with metastatic disease are treated with surgical castration. In this context, CT antigens expressed in prostate cancer cells should be unique tumor antigens.
Several members of the CT antigen family are expressed in prostate cancer tissues, including NY-ESO-1, LAGE-1, and XAGE-1 (19, 20, 21). This suggests that several known CT antigens warrant investigation as potential prostate tumor vaccine antigens. Other research groups have identified CT antigens in patients with breast cancer (22, 23) and with renal cell cancer (24) by using sera from patients to screen a testis cDNA expression library. To our knowledge, there have been no published reports of the direct identification of prostate CT antigens. In this report, we discuss whether prostate CT antigens can be identified prospectively using sera obtained from patients with prostate cancer. We used SEREX to screen for antigenic proteins, expressed from a human testis-tissue cDNA library, that were recognized by high-titer IgG antibodies in prostate cancer patients’ sera. We identified several potential prostate cancer antigens. Two of the proteins were identified independently in sera from separate individuals. These proteins were most highly expressed in testis tissue, but demonstrated low-level expression in some other, normal tissues. One of these, MAD-CT-2, was expressed in both normal and malignant prostate tissues. Expression of MAD-CT-1 was not detectable in normal prostate tissue, but was detectable in 2 of 10 metastatic prostate cancer specimens.
Testis protein-specific IgG in the sera of patients with prostate cancer
To identify possible CT antigens from patients with prostate cancer, we used the SEREX methodology to screen a normal testis cDNA phage expression library with sera from prostate cancer patients. Because the identification of antigens is essentially dependent on the choice of sera for screening, this analysis was conducted with sera from prostate cancer patients from a variety of different clinical scenarios. Many of these patients had atypical histories of prostate cancer regression without specific treatment or had evidence of antibody responses to other prostate cancer antigens (25), as described in Table 1. SEREX screening was conducted with paired sera to speed the analysis, and immunoreactive plaques were isolated and rescreened individually with the unpaired sera used for initial detection. A strongly immunoreactive plaque identified from this initial screen is shown in Figure 1. From this process, 122 potentially immunoreactive phage plaques were identified. These phages were then rescreened with the reactive sera at lower plaque density to isolate purified phages and to eliminate false positives and phages with weak immunoreactivity from further evaluation. Following this, 27 immunoreactive phages were identified and ultimately sequenced. Seven of the twenty-seven sequenced phages were found to be unique (shown in Table 2). It is noteworthy that two of these phages, encoding MAD-CT-1 or MAD-CT-2, were each isolated from the sera of two different patients.
Antigen-specific IgG in the sera of other patients with prostate cancer
To determine if any of the 7 gene products identified and shown in Table 2 were recognized by IgG in sera from other patients with prostate cancer, a larger analysis was performed with phage immunoscreening. For this experiment, purified phages encoding each antigen were directly spotted to bacterial lawns in multiple replicates and evaluated using sera from 109 patients in various stages of prostate cancer and from 52 male control blood donors without histories of prostate cancer. Examples are shown in Figure 2, and the findings are summarized in Table 3. None of the antigens were recognized significantly more frequently in the sera of patients versus controls (P > 0.05). Consequently, in order to prioritize these proteins for further evaluation, the three antigens recognized by the greatest number of patients relative to controls, MAD-CT-1, MAD-CT-2, and MAD-CT5, were chosen.
MAD-CT-5-specific IgG antibodies
The protein encoded by MAD-CT-5, the Sjögren’s syndrome B antigen (SSB, autoantigen La), is a ubiquitously expressed protein associated with RNA polymerase III (26, 27). It is also a known autoantigen typically recognized in patients with Sjögren’s syndrome. A commercially available ELISA method was used to assay for the presence of MAD-CT-5 (SSB)-specific antibodies (Figure 3A). This confirmed the presence of MAD-CT-5-specific antibodies in the sera of patient # 20, from which the antigen was identified, thereby validating our SEREX immunoscreening methodology. This ELISA methodology was used in a larger panel consisting of normal controls (n = 47) and subjects with prostate cancer (n = 95) to determine if low-titer antibodies to SSB can be detected, perhaps below the level of detection of our immunoblot methodology. As shown in Figure 3B, antibody responses were detectable in a few individuals, consistent with our phage immunoscreening findings. IgG responses to SSB, however, were not detected more frequently in the sera of patients with early (n = 37) or late stage (n = 58) prostate cancer as compared with control male blood donors (n = 47).
MAD-CT-1 and MAD-CT-2 expression
MAD-CT-1 (protamine-2) is a known spermatid protein with essentially testis-specific tissue expression (28), whereas expression of MAD-CT-2 has not been reported previously. In order to determine the expression of MAD-CT-2 in normal tissues, PCR was performed with cDNA from several normal human tissues (Figure 4A). As is shown, expression of MAD-CT-1 and MAD-CT-2 was highest in normal testis tissue. MAD-CT-1 expression was also detectable in brain tissue, and lower expression was detectable in bladder tissue. MAD-CT-2 was also found to have low-level expression in normal bladder and prostate tissue. Similar results were found by RT-PCR using total tissue RNA (data not shown). To determine if either of these antigens might be expressed in prostate cancer, a similar analysis was conducted with cDNA specimens obtained from 10 separate prostate cancer metastases (Figure 4B). By this analysis, we found low but detectable levels of MAD-CT-1 transcript in 2 of 10 prostate cancer specimens (patients # 8 and # 9, Figure 4B). MAD-CT-2 transcript was detectable in 9 of the 10 prostate cancer specimens.
The group of antigens known as CT antigens, or germ cell antigens, is a large family of proteins expressed primarily in adult, male germ cell tissues. These proteins are also expressed in cancers of different origins. It is believed that these antigens may escape normal immune recognition, given their restricted expression in MHC class I-negative germ cells (29). Thus, there has been considerable interest in identifying CT antigens, as they may serve essentially as tumor-restricted antigens for the development of vaccines that are at lower risk for preexisting immune tolerance and for creating potential autoimmune reactions to normal tissues. Several members of the CT antigen family are expressed in prostate cancer, including PAGE-4 (30), NY-ESO-1 (19), LAGE-1 (21), XAGE-1 (31), and members of the MAGE-A family (32). To date, we are unaware of any published reports that directly identify prostate cancer CT antigens using sera from patients with prostate cancer. In this study, we used SEREX to screen a normal testis cDNA library using sera from patients with prostate cancer, including sera from patients with early stage, nonmetastatic prostate cancer (n = 10) as well as from patients with metastatic prostate cancer (n = 12). Within this population, we included sera from patients with clinical histories suggesting disease response or with previous evidence of immune response to other prostate cancer antigens. We identified several proteins by this method, of which MAD-CT-1 (PRM2) and MAD-CT-2 are likely prostate cancer antigens. Their expression, however, is not entirely germ cell-specific, as has been reported for some CT antigens such as NY-ESO-1 (11).
Northern blot analysis and immunohistochemical analysis (28), as well as evaluation of EST databases (33, 34), had previously suggested that, of the 7 proteins found in this study, PRM2 is testis-specific in terms of tissue expression. It had also been previously identified as an autoantigen in infertile men without cancer (35) and in patients with sarcoma (36). Curiously, in the latter study the authors suggested that the protein is ubiquitously expressed based on their quantitative RT-PCR analysis, and they did not explore its potential as a CT antigen (36). We suspect that the primers used in their study were not entirely specific for PRM2, as we have performed similar quantitative RT-PCR analysis demonstrating essentially tissue-specific expression (data not shown). Our data are also in concordance with the gene expression predicted by EST analysis (37), in which expression was predicted to be predominantly in testis, but with low-level expression in brain and placenta (34). The identification of PRM2 from a sarcoma library, and its detection in prostate cancer but not normal prostate tissue, suggests that it is indeed a tumor antigen. Furthermore, the earlier identification of PRM2 as an autoantigen in infertile men, as well as in subjects with various cancers, also suggests that it is an immunogenic protein. Strictly speaking, however, the detection of PRM2 expression in other normal tissues, including brain, suggests it is not a "pure" CT antigen.
Little is known of the gene product encoded by MAD-CT-2, and it has not been previously named or characterized. The cDNA had been identified earlier by sequencing a testis cDNA library, but at present there is no other information about tissue expression, nor is there any information in the NIH EST database (34). From our RT-PCR analysis of RNA from normal human tissues, it does appear to be predominantly expressed in testis, with lower levels of expression in normal bladder and prostate tissues. The predicted gene product has a predominance of charged residues (38.5% aspartate, glutamate, lysine, or arginine residues), and 24.2% of the amino acids are glycine residues. The presence of many charged residues, and the fact that it exhibits some homology to known DNA-binding proteins, suggest that it may be a DNA-binding protein. It also, however, demonstrates homology to several members of a large family of Golgi proteins, many of which are being characterized as autoantigens (38). The testis-predominant expression of MAD-CT-2, its expression in prostate cancer tissues, and the presence of antibodies to this protein in the sera of several patients with prostate cancer, suggest that it is also a prostate cancer antigen. Like MAD-CT-1, however, it cannot be classified as a "pure" CT antigen, given that its expression is not entirely restricted to germ cells and tumor cells. Given its restricted expression to testis, bladder, and prostate, however, this gene product could potentially be investigated as a prostate or bladder tumor antigen.
The autoantigen La (MAD-CT-5) was recognized in the sera of one individual. Complete medical histories were not available for the patients from whom the sera was taken, thus we do not know whether this patient had symptoms of Sjögren’s syndrome. The known, generally ubiquitous, expression of this protein suggests that it is not a CT antigen. The identification of this protein in this study, however, did permit us to validate our SEREX methodology using clinically standardized reagents in an ELISA. As indicated, we saw few significant IgG responses to this protein by immunoscreening with phage or by ELISA in a larger panel of patients versus controls.
The other proteins identified have not specifically been identified as autoantigens in other studies, to our knowledge. There have, however, been multiple reports of autoantibody responses to other ribosomal proteins, similar to MAD-CT-7, in other cancers and autoimmune diseases, suggesting that these are common autoantigens (39, 40). Curiously, we did not identify other commonly recognized CT antigens such as NY-ESO-1, LAGE-1, or members of the MAGE-A family. In particular, we expected to detect NY-ESO-1 in our analysis, as a report by Nakada and colleagues suggested that NY-ESO-1 is expressed in 38% of prostate tumor specimens (19). The detection of antibodies to a particular protein in our analysis, however, was strictly dependent on the choice of sera used. We specifically chose sera for our primary SEREX screening to include sera from patients with early stage, nonmetastatic prostate cancer, as well as patients with metastatic prostate cancer, for the broadest possible sampling. In the earlier report of NY-ESO-1 expression in prostate cancer, antibodies to NY-ESO-1 were only detectable in the sera of 8% of patients with advanced-stage prostate cancer (19). Consequently, our population of sera samples for screening, of which only 12 had metastatic disease, might not have been large enough to identify NY-ESO-1 in particular. In related studies, we are currently assembling a large panel of phages encoding previously identified CT antigens in order to identify immunologically recognized CT antigens by a similar methodology, using sera from many patients with prostate cancer. Together with the current study, these studies should further define known CT antigens that could be explored as therapeutic or diagnostic target antigens for prostate cancer.
Materials and methods
Sera were obtained from patients with various stages of prostate cancer who gave written Internal Review Board-approved informed consent for their blood products to be used for immunological research. Blood was collected at the University of Washington Medical Center (Seattle, WA, USA) between 1997 and 2001, and at the University of Wisconsin Hospital and Clinics (Madison, WI, USA) between 2001 and 2004. Sera were stored in aliquots at -80˚C until used. Control sera were obtained from age-matched male volunteer blood donors who also gave written informed consent.
SEREX analysis was conducted in a manner similar to what we described previously (14). A phage cDNA expression library made from normal human testis tissue mRNA (TriplEx, BD Biosciences Clontech, Palo Alto, CA, USA) was used to transduce the E. coli strain XL-1 Blue (Stratagene, La Jolla, CA, USA). Transduced bacteria were then plated in top agarose in a lawn at a phage density of approximately 6500 pfu per 150-mm LB agar plate. After plaques developed, nitrocellulose filters (Millipore, Bedford, MA, USA) presoaked with 10 mM IPTG were overlaid to induce expression of recombinant proteins. After 12-16 h, the membranes were removed, washed twice with 10 mM Tris 8.0, 150 mM NaCl, 0.5% Tween-20 (TBST), and once with 10 mM Tris 8.0, 150 mM NaCl (TBS), blocked with TBST plus 1% BSA, and then probed with human sera diluted 1:100 in TBST plus 1% BSA overnight. Sera from 22 prostate cancer patients were arbitrarily paired to speed screening, and a mixture of 2 sera was used in each of 11 screens. Membranes were then washed and blocked again, and human IgG was detected with an alkaline phosphatase-conjugated monoclonal antihuman IgG antiserum (Sigma, St. Louis, MO, USA) diluted in TBST plus 1% BSA. After repeated washing, the membranes were developed with 0.3 mg/mL nitro blue tetrazolium chloride (NBT) (Fisher Biotech, Pittsburgh, PA, USA) and 0.15 mg/mL 5-bromo 4-chloro 3-indoylphosphate (BCIP) (Fisher Biotech) in 100 mM Tris 9.5, 100 mM NaCl, 5mM MgCl2. Immunoreactive plaques were then removed from the corresponding agar plates and the phages were eluted overnight in SM buffer (100 mM NaCl, 10 mM MgSO4, 35 µM Tris 7.5, and 0.01% gelatin). This process was repeated with multiple agar plates, such that 75,000 to 250,000 pfu were ultimately screened with each sera pair. To eliminate false positives prior to plaque purification, an aliquot of each eluted heterogeneous plaque was individually spotted in replicate fashion onto lawns of E. coli growing on LB agar plates. Plaques were allowed to establish as before, and were then transferred to IPTG-impregnated nitrocellulose membranes. The membranes were then probed individually with each of the two sera of the original pairing to determine which of the two sera contained phage-specific IgG. Plaques were then rescued to pBluescript-derived phagemids by transduction of the recombinase-positive E. coli strain BM25.8 (BD Biosciences Clontech, Palo Alto, CA, USA). Recombinant bacteria were selected on ampicillin-containing media, and purified plasmids were then sequenced using plasmid-specific primers and ABI BigDye terminator sequencing reactions (Perkin-Elmer, Foster City, CA, USA). The GenBank database was searched to identify the gene products encoded by the recombinant phagemids. Gene products were labeled according to the standard SEREX database nomenclature (for example, MAD for "Madison," CT for "cancer-testis").
Purified phages corresponding to each of the sequenced plasmids were analyzed for immunoreactivity using a panel of sera from 109 prostate cancer patients and 52 normal controls. Lawns of XL-1 Blue E. coli were grown on LB agar plates, and purified plaques were individually spotted onto the lawns in replicate fashion. A phage encoding human IgG was included on each membrane as a positive control, and irrelevant purified phages were included on each membrane as negative controls. Protein expression was induced with IPTG, and phages were transferred to nitrocellulose membranes, as described above. Serum from each of the subjects and controls was used to screen the membranes. Immunoreactive plaques were recorded for each of the subjects, and the controls were observed and compared to the internal positive and negative control plaques. Four "blinded" individuals judged the immunoreactivity of plaques, and those plaques determined to be positive by at least three individuals were recorded as immunoreactive with the given sera, to reduce the subjectivity of some weakly immunoreactive plaques. A chi-square analysis was used to compare the number of immunoreactive plaques in subject and control populations.
Anti-SSB antibodies were detected using a commercial kit (Kallestad Anti-La Test Kit, Bio-Rad Laboratories, Hercules, CA, USA) according to the manufacturer’s instructions. Sera were diluted 1:100 in sample diluent prior to analysis. IgG reactivity is reported in units per mL by comparison with a standardized reference.
Tissue expression analysis by PCR
Gene-specific oligonucleotide primers (Proligo, Boulder, CO, USA) were designed to amplify MAD-CT-1 (Forward: 5’-GGGGTACCAATGGTCCGATACCGCGTGAGGAG-3’, Reverse: 5’-GGAATTCTTAGTGCCTTCTGCATGTTCTCTT-3’), MAD-CT-2 (Forward: 5’-GAGGATATGAGATCAGAAAGA-3’, Reverse: 5’-TCGGTAGAAGAATGGGATGC-3’), and actin (Forward: 5’-TCATGAAGTGTGACGTTGACATCCGT-3’, Reverse: 5’-CTTAGAAGCATTTGCGGTGCACGATG-3’) from cDNA specimens by PCR. PCR cycling and detection was conducted in a MyCycler thermocycler (Bio-Rad Laboratories) with 35 cycles of amplification under the following conditions: 60 s at 94˚C, 60 s at annealing temperature (65˚C for MAD-CT-1, 58.7˚C for MAD-CT-2, or 55˚C for actin), and 3 min at 72˚C. Plasmid template DNA encoding MAD-CT-1 or MAD-CT-2 was used as a positive control. In preliminary experiments with serial dilutions of plasmid DNA, the threshold level of detection was found to be less than or equal to 48 copies of MAD-CT-1 template and less than or equal to 880 copies of MAD-CT-2 template. Template DNAs for this analysis included cDNA specimens from normal human tissues (Biochain, Hayward, CA, USA), and cDNA derived from 10 metastatic prostate specimens (5 soft tissue or lymph node metastases and 5 bone metastases) from 10 separate patients. The latter cDNA specimens were graciously provided by the laboratory of Dr. Robert Vessella (University of Washington, Seattle, WA, USA).
This work is supported by NIH (K23 RR16489), the Howard Hughes Medical Institute at the University of Wisconsin, and the Prostate Cancer Foundation (CaPCURE).
- Received July 6, 2005.
- Accepted December 1, 2005.
- Copyright © 2006 by Douglas McNeel