Isolation and Characterization of PAGE-1 and GAGE-7 NEW GENES EXPRESSED IN THE LNCaP PROSTATE CANCER PROGRESSION MODEL THAT SHARE HOMOLOGY WITH MELANOMA-ASSOCIATED ANTIGENS*

The LNCaP progression model of human prostate cancer consists of lineage-related sublines that differ in their androgen sensitivity and metastatic potential. A differential display polymerase chain reaction was employed to evaluate mRNA expression differences between the LNCaP sublines in order to define the differences in gene expression between the androgensensitive, nontumorigenic LNCaP cell line and the androgen-insensitive, metastatic LNCaP sublines, C4-2 and C4-2B. An amplicon, BG16.21, was isolated that showed increased expression in the androgen-independent and metastatic LNCaP sublines, C4-2 and C4-2B. Hybridization screening of a l gt11 expression library with BG16.21 revealed two transcripts, both homologous to BG16.21 at the 3* end. A GenBank data base search using the GCG Wisconsin software package revealed the shorter ;600-bp transcript (designated GAGE-7) to be a new member of the GAGE family. The second ;700-bp transcript was a novel gene (designated PAGE-1, “prostate associated gene”) with only 45% homology to GAGE gene family members. RNA blot analysis demonstrated that GAGE-7 mRNA was expressed at equal levels in all lineage related prostate cancer cell sublines, while PAGE-1 mRNA levels were elevated 5-fold in C4-2 and C4-2B as compared with LNCaP cells. Neither GAGE-7 nor PAGE-1 demonstrated any regulation by androgens in the prostate cancer cell lines used in this study. PAGE-1 and GAGE-7 expression was found to be restricted to testes (high) and placenta (low) on human multiple tissue Northern blots. As GAGE/MAGE antigens were reported previously to be targets for tumor-specific cytotoxic lymphocytes in melanoma, these results suggest that PAGE-1 and GAGE-7 may be related to prostate cancer progression and may serve as potential targets for novel therapies.

An experimental model system to study advanced and me-tastatic prostate cancer has been developed based on the observation that nontumorigenic bone or prostate fibroblasts, when co-inoculated with the nontumorigenic, prostate-specific antigen (PSA) 1 -secreting cell line LNCaP, induced prostate adenocarcinoma growth in vivo in nude mice (1,2). By manipulating the androgen status of the hosts, the laboratory was able to generate various LNCaP sublines from the primary tumors (3,4). The direct lineage-related sublines C4, C4 -2, and C4 -2B4 share cytogenetic markers and HLA haplotype with parental LNCaP, but otherwise possess distinct biologic and biochemical profiles. Most importantly, the sublines exhibit progressive tumorigenicity, metastatic potential, and androgen independence. C4, the first subline derived from LNCaP, forms tumors in intact nude mice without the need for a co-inoculum of bone fibroblasts or Matrigel®; while growth in castrated hosts still requires a co-inducer. C4-2 is a subline derived from C4, which is highly tumorigenic in intact or castrated nude mice without a co-inducer. C4-2B4 is a bone metastatic subline derived from orthotopic injection of C4-2 and has demonstrated an increased aggressiveness as compared with C4-2.
The isolation of novel prostate cancer progression related genes may lead to new prognostic markers and/or therapeutic targets. Therefore, this lineage-related progression model represented an ideal system to screen for prostate cancer progression associated genes. Differential mRNA display polymerase chain reaction (DD-PCR) (5,6) was used to screened the LN-CaP sublines for differences in gene expression. Sixty-nine amplicons were cloned, sequenced, and screened by RNA blot hybridization. One amplicon demonstrated increased expression with increasing metastatic potential in the LNCaP sublines. Screening of a cDNA library from C4-2 cells, where the amplicon expression was elevated, resulted in the isolation of clones of two lengths, ϳ600 and ϳ700 bp. Sequence analysis and GenBank TM homology search revealed the two genes to be related to the GAGE family of melanoma associated antigens. Herein, we report the isolation and preliminary characterization of these GAGE-related genes: PAGE-1 (prostate-associated gene), and GAGE-7.

EXPERIMENTAL PROCEDURES
Cell Lines and Culture-The LNCaP, C4, C4-2, and C4-2B4 cell lines were cultured in T medium containing 5% fetal bovine serum as described previously (7,8). Cells were grown to confluence in 150-mm culture dishes in a 37°C incubator equilibrated with 5% CO 2 in humidified air.
Differential Display Polymerase Reaction-DD-PCR was performed * This work was supported by National Institutes of Health Grant CA 64863 (to L. W. K. C) and by an American Foundation for Urologic Disease Research Scholarship (to M. E. C.). Institutional GenBank TM search resources were supported by the Information Technology Center-Academic Computing Health Sciences at the University of Virginia and by National Institutes of Health NCI Grant CA-16672 to the M. D. Anderson Cancer Center. 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.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AFO58988 and AFO58989.
A 0.2-g aliquot of DNase I-treated RNA from each of the four cell lines was reverse transcribed. Briefly, 0.2 g of RNA was placed in solution containing 50 mM Tris-HCl, pH 8.3, 75 mM KCl, 3 mM MgCl 2 , 10 mM dithiothreitol, 20 units of RNase inhibitor (Pharmacia Biotech), 20 M dNTPs (Pharmacia Biotech), 1 M T12-MN downstream primer, where M ϭ A, G, or C bases and N ϭ A, G, C, and T bases (Operon Technologies, Alameda, CA). The reaction was heated to 65°C for 5 min and then allowed to cool slowly to room temperature. 30 units (1.5 ml) of moloney murine leukemia virus reverse transcriptase (Life Technologies, Inc.) was added to each sample and incubated at 37°C for 1 h.
DD-PCR reactions were then carried out. The downstream primer used was the same T12-MN primer used in the reverse transcription reaction. The upstream primer, OPA-n, was an arbitrary 10-mer with 60 -70% GC base content and no self-complementarity (OPA primer set, Operon Technologies, Alameda, CA). One tenth (2 l) of reverse transcription reaction was used for each DD-PCR reaction. Reactions were carried out in 10 mM Tris-HCl, pH 8.3, 1.5 mM MgCl 2 , 50 mM KCl, 1 M appropriate T12-MN downstream primer, 0.5 M OPA-n upstream primer, 0.50 M ␣-35 S-labeled dATP (Amersham Corp.), 0.0625 units of Taq polymerase (Boehringer Mannheim) in nuclease-free water. The samples were then placed in a thermal cycler (model 2800, Perkin-Elmer) for the following profile: denature at 94°C, 2 min; followed by 40 cycles of denature at 94°C, 30 s; anneal at 42°C, 2 min; and extension at 72°C, 30 s with a final extension at 72°C for 5 min.
4 l of sequencing stop buffer (95% formamide, 20 mM EDTA, 0.05% bromphenol blue, 0.05% xylene cyanol-ff) were added. The DD-PCR products were resolved on a 6% denaturing polyacrylamide gel. After drying the gel on 3-mm blotting paper, an autoradiograph was exposed for 48 h. Differentially expressed bands (amplicons) between the parental LNCaP and the C4-2 and C4-2B sublines were selected and excised. The DNA was eluted by boiling for 15 min and recovered by adjusting the sodium acetate concentration to 0.3 M followed by ethanol precipitation. The amplicons were reamplified by PCR using the same primers and conditions as described for DD-PCR above, except that no radioactive label was added and the dNTP concentration was 20 M. Successful reamplification was confirmed by running a small aliquot of the sample on a 1% agarose gel stained with ethidium bromide. The reamplified amplicons were then cloned by the TA system into the pCRII plasmid (Invitrogen, San Diego, CA).
Sequencing and Sequence Handling-Cloned amplicons and phage inserts were sequenced by dideoxy chain termination (10) using Sequenase version 2.0 (Amersham Corp.). Initial cDNA characterization and translation was performed using MacVector DNA analysis software (Oxford Molecular Group, Campbell, CA). Homology search in the Gen-Bank TM data base was carried out using the FASTA algorithm as implemented in the GCG Wisconsin version 8.01 software package (Genetics Computer Group, Madison, WI). Multiple sequence alignments, DNA, and protein were made using Pileup and edited using Lineup features of the GCG Wisconsin version 8.01 software package with a weighted representation of the first methionine codon. Publication output for multiple sequence alignments was generated using ALSCRIPT (11).
Northern Blot-Expression of novel amplicons (no match in Gen-Bank TM ) was evaluated by formaldehyde denatured RNA blotting essentially as described (12). Because DD-PCR performed as described generates a cloned cDNA fragment representing the 3Ј-polyadenylated end of the mRNA, the sense strand is known. Therefore, [ 32 P]CTPlabeled antisense riboprobes of the cloned amplicons were generated from either the Sp6 or T7 transcription sites on the pCRII plasmid as appropriate. Riboprobes were generated as recommended by the manufacturer in the SP6/T7 transcription kit (Boehringer Mannheim). 5 l of CTP (800 Ci/mmol, NEN Life Science Products) was used in the labeling reaction. For Northern blots, 30 g of total RNA from each of the four cell lines were loaded per lane. The RNA was transferred to ZetaProbe®-charged nylon membrane (Bio-Rad) by electrotransfer for 16 h at 15 mA and 4°C in a Hoefer TE-42 transphor tank system (Pharmacia Biotech Inc.). Blots were hybridized at 65°C in Rapid Hyb® solution (Amersham Corp.). 1 ϫ 10 6 dpm of riboprobe were added per ml of hybridization solution. Blots were hybridized overnight and washed in SSC/SDS solutions of increasing stringency as appropriate. RNase digestion (riboprobes only) of the blots was performed in order to confirm hybridization specificity and differential mRNA expression. The membranes were placed in 2ϫ SSC containing RNase at a concentration 0.5 g/ml. After autoradiography, expression levels were evaluated by semiquantitative densitometry and normalized to GAPDH expression. Expression of sequences recovered by library screening was evaluated as described above, except that cDNA probes were used.
Expression Library Screening-In the course of DD-PCR, a novel amplicon, BG16.21, demonstrated greatly increased expression in C4-2 and C4-2B4. It was therefore decided to recover the full gene sequence from a gt11 expression library using BG16.21 as a probe. A custom C4-2 cell line gt11 phage library was constructed from poly(A) ϩselected RNA using Invitrogen Custom Services (Invitrogen Corp., San Diego, CA). Briefly, poly(A) ϩ -selected RNA was primed using oligo(dT) primers. cDNAs were made double-stranded and were ligated to NotI/ EcoRI open reading frame adapters. The cDNA smear started at 500 bp and was continous to 12 kb. Size selection was made for greater than 500 bp. Adapted cDNAs were ligated into gt11 arms and packaged. The final titer of the amplified library was 8.8 ϫ 10 8 plaque-forming units/ml. Nineteen independent phage were analyzed for the presence of inserts. All 19 phage had inserts with an average size of 1.1 kb.
A single colony of Y1090 bacteria was expanded in 15 ml of LB broth containing 10 mM MgSO 4 and 0.2% maltose. Host bacteria were diluted in 10 mM MgSO 4 to A 600 ϭ 0.5 absorbance units. Bacteria were infected with 10 l of a 5 ϫ 10 3 plague-forming units/l phage stock in SM buffer (13) plus 5% chloroform. Initial screening of the C4-2 library was accomplished by hybridizing 32 P-labeled double-stranded cDNA from the amplicon BG16.21 against 1 ϫ 10 6 phage plaques from 20 dishes. Two filter replicates were generated per plate using 150-mm BA-85 nitrocellulose.
The filters were denatured in 1.5 M NaCl, 0.5 M NaOH for 2 min at room temperature; neutralized in 1.5 M NaCl, 0.5 M Tris-Cl, pH 8 for 5 min at room temperature; and rinsed with 2ϫ standard saline citrate (SSC), 0.2 M Tris-Cl, pH 7.5, for 30 s. Filters were baked for 1.5 h at 80°C in vacuo. More than 50 plaques were positive. Hybridization was performed at 65°C using 1 ϫ 10 6 dpm denatured [ 32 P]dCTP-labeled cDNA probe/ml of RapidHyb® buffer (Amersham Corp.). Twenty plaques were pulled and stored. Eight plaques were chosen at random for further purification. These eight plaque cores were titered and replated at a density approaching 5000 plaque-forming units/150-mm dish. Replica filters were generated and hybridized to BG16.21 32 Plabeled cDNA as described. Pure plaques, approaching 100% hybridization, were obtained after three rounds of screening.
Preparation of phage DNA was accomplished by eluting phage from the purified plaques essentially as described previously (13) by placing the plaques into 500 l of SM buffer overnight at room temperature. 100 l of phage elution was then incubated with 100 l of saturated Y1090 bacteria. These were shaken at 37°C in 50 ml of NCZYM broth until lysis was observed (about 4 h). At this point 1.5 ml of chloroform was added to lyse the remaining intact bacteria. The cultures were cleared by centrifugation for 15 min at 4°C using 3600 rpm in a Dupont RT6000B tabletop centrifuge in an HB3600 rotor. Cleared phage lysates were treated for 2 h at 37°C with 50 g of DNase I and 250 g of RNase A. Lysates were transferred to 30-ml OakRidge centrifuge tubes and centrifuged for 3 h, 40 min at 30,000 rpm at 4°C to pellet the phage. Phage were resuspended in 200 l of Tris-Cl, pH 8.0. A PCR was performed on the purified phage to determine the insert size. Briefly, 2 l of purified phage was used for each 50-l PCR reaction. Reaction buffer consisted of 3 mM final MgCl 2 , 0.5 mM dNTP mixture, 200 ng of forward primer, 200 ng of reverse primer, 1ϫ Boehringer Mannheim PCR buffer, and 5 units of Taq DNA polymerase (Boehringer Mannheim). The thermal cycle conditions were denature 94°C, 40 s; annealing 56°C, 1 min 40 s; extension 72°C, 2 min for 35 cycles with a 7-min, 72°C final extension. 20% of the reaction was loaded onto 1% agarose, TAE (0.04 M Tris-acetate, 0.01 M EDTA) gel for examination. PCR products were TA-cloned into pCRII per the manufacturer's recommendations (Invitrogen Corp.) and were sequenced.
5Ј-RACE-5Ј-RACE was performed using the CLONTECH kit (CLONTECH, Palo Alto, CA). 1 mg of total RNA from the C4-2 cell line was reverse transcribed. The RNA was digested and a second strand was made. The 5Ј adapter was then ligated to the double-stranded cDNAs. A PCR was performed with a 5Ј adapter-specific upstream primer and a gene-specific downstream primer per the manufacturer's recommendations. The PCR products were evaluated with a 1% aga-rose, 0.5 ϫ TAE gel. PCR products were gel-purified, TA-cloned as described above, and sequenced.
Polyclonal Antibody Generation and Immunoblotting-An antigenic peptide AL5-30 from amino acid 21-41, EESSDEQPDEVESPTQS-QDST, was chosen based on the antigenicity of the translated PAGE-1(AL5) polypeptide (MacVector DNA analysis software, Oxford Molecular Group, Campbell CA) and was synthesized in the Baylor College of Medicine Peptide Synthesis Core Facility. Briefly, the peptide was synthesized using Fmoc (N-(9-fluorenyl)methoxycarbonyl) chemistry (14) on a Perkin-Elmer Applied Biosystems model 432 peptide synthesizer (Perkin-Elmer). The peptide was linked to mitogen-activated protein (MAP) in a four-branch procedure. MAP-conjugated peptide was used to immunize primed murine hosts for the generation of polyclonal antibodies according to established protocols (15)(16)(17)(18). Briefly, male Balb/c mice, age Ն90 days, were immunized with 25 g of purified peptide in 0.5 ml of phosphate-buffered saline in an equal volume of complete Freund's adjuvant. The immunizations were repeated for a total of three times on day 0 intradermally, day 28 subcutaneously, and day 50 subcutaneously. The last two injections used incomplete Freund's adjuvant. Mice were injected at day 70 intraperitoneally with 0.5 ml of pristane followed 3 days later with an additional challenge by the MAP-conjugated AL5 peptide (25 g of peptide in phosphate-buffered saline intraperitoneally. Four days later mice were injected with 6 ϫ 106 mouse myeloma cells per mouse intraperitoneal ascites was collected at days 15-20 after the last presentation of antigen. Ascites were screened for immunoreactivity using enzyme-linked immunosorbant assays in peptide-coated plates as described previously (15).
Immunoblotting-Cells were lysed on the dish in 3ϫ SDS sample buffer (3% SDS, 0.6% 2-mercaptoethanol, 15% glycerol, 0.035% bromphenol blue). For immunoblot analysis, 100 g of total cellular protein were separated electrophorectically in 12.5% polyacrylamide gels under reducing conditions and blotted onto nitrocellulose filters. The filters were blocked with 5% (w/v) nonfat dry milk at 4°C overnight. Murine serum was used at 1:250 dilution, and the secondary antibody were used at the dilution recommended by the manufacturer for enhanced chemiluminescence (ECL) antibody detection (Amersham Corp.).

RESULTS
Using DD-PCR, an amplicon, BG16.21, was observed that appeared to be differentially expressed, with expression observed in the C4-2 and C4-B4 lanes (Fig. 1). Reamplification, cloning, and sequencing of this amplicon revealed it to be a typical DD-PCR product that represented the 3Ј end of a mRNA transcript. A second amplicon, slightly larger than BG16.21 on the polyacrylamide gel was in fact shown by sequencing to be identical to BG16.21 (Fig. 1, arrows). Such doublets due to the slightly different electrophoretic mobility of the complementary strands of the same species of DNA under denaturing conditions have been observed previously (19).
Differential expression of the BG16.21 amplicon was confirmed by Northern blot using riboprobes under the conditions described above. It was observed that expression increased progressively from LNCaP through C4-2B4 (data not shown).
At the time of initial isolation, BG16.21 had no significant homology to sequences in the GenBank TM data base. It was therefore decided to recover the full-length cDNA sequence by screening a gt11 cDNA expression library of the C4-2 cell line. Of the eight individual phage chosen after three rounds of plaque purification, two insert sizes of ϳ600 (AL4) and ϳ700 bp (AL5) were observed (data not shown). Both phage inserts were PCR-amplified, TA-cloned into pCRII (Invitrogen Corp.), and sequenced (see below). The 3Ј end of the 524-bp (AL4) insert had approximately 80% homology at the nucleotide level to the DD-PCR amplicon, BG16.21, while the 3Ј end of the 676-bp (AL5) insert matched exactly the BG16.21 amplicon. On Northern blot, using the 524-bp (AL4) and 676-bp (AL5) inserts as cDNA probes, it was found that the shorter insert (AL4) had similar levels of expression in LNCaP and the three lineagerelated sublines ( Fig. 2A). The expression of the longer insert (AL5) increased progressively in the LNCaP sublines as had been observed with the shorter BG16.21 riboprobe (Fig. 2B). By semiquantitative densitometry, the up-regulation of the AL5 transcript was 5-fold. The estimated transcript sizes on Northern blot corresponded to the size of the phage inserts observed (Fig. 2).
Both strands of the cDNAs for AL4 and AL5 were sequenced (Fig. 3, A and B). 5Ј-RACE (CLONTECH) was used to confirm the 5Ј ends of the cDNAs. The 5Ј-RACE reaction yielded an additional 21 bp of 5Ј-UTR for AL5 (Fig. 3B, underlined). FASTA alignment (GCG8, Madison, WI) revealed significant homology of these genes to the GAGE family of melanoma antigens (data not shown). Multiple sequence alignment found the highest homology, Ͼ99% at the DNA level in the coding sequence (Fig. 4B) and 99% at the polypeptide sequence (Fig.  4C), between AL4 and GAGE-4. The difference amounts to a single amino acid substitution in the coding sequence, Asp 3 His at position 118, Fig. 4C (amino acid 76, Fig. 3A), in (AL4)GAGE-7 as compared with GAGE-4. This amino acid substitution appears to be the result of a single G 3 C transversion at position 469, Fig. 4B (Fig. 3A, nucleotide 302). The same substitution of Asp 76 3 His 76 at position 118 was observed in GAGE-2 but not GAGE-4. The extended 5Ј-UTR of (AL4)GAGE-7 when compared with GAGE-4 may be due only to the isolation of a longer cDNA since the remaining 5Ј-UTR FIG. 1. Relevant area of a 6% polyacrylamide/urea/TBE gel from a reverse transcription DD-PCR comparing the LNCaP lineage-derived subline RNAs. The bands, BG16.21, that are differentially expressed between the cell sublines are indicated by arrows. This doublet, a product of denaturing conditions, was subsequently eluted, PCR-amplified, cloned, sequenced, and used as the source for antisense riboprobes to confirm differential expression. was perfectly homologous. Analysis of the predicted peptide using MacVector DNA analysis software (Oxford Molecular Group, Campbell, CA) predicted a peptide of 117 amino acids, having a molecular mass of 12,997 daltons and an isoelectric point of 4.10. A motif search of the predicted polypeptide showed three potential casein kinase II phosphorylation sites (data not shown), two potential protein kinase C phosphorylation sites (Fig. 4C, boxed), and one potential N-myristoylation site (Fig. 4C, blue shading). These features were found to be common to all GAGE family members and were maintained in the sequence alignment.
Within the coding sequence, the homology of AL5 to the GAGE family was 56% at the nucleotide level with two additional gaps of 24 and 12 bp in (AL5)PAGE-1 (addition of the gap lowers the nucleotide homology to 47%) and 48% at the protein level when compared with the most similar GAGE family member, GAGE-2, and clearly represents a divergent family member. Striking differences were noted in the 3Ј-UTR and a midsequence insertion of 126 bp (42 amino acids) that appears to be a duplication of the sequence that immediately follows with some divergence. (AL5)PAGE-1 also has an 8-amino acid gap followed by a 4-amino acid gap in the N-terminal area (amino acids 21-28 and 35-38 of GAGE-2) when compared with all other members of the GAGE family. Although no significance of this region has been elucidated, the gap has lost 3 Pro and 3 charged residues. Translation of the (AL5)PAGE-1 coding sequence predicted a peptide of 143 amino acids, having a molecular mass of 16,133 daltons and a pI ϭ 3.95. Based on the divergence of AL5 from the GAGE family it was renamed PAGE-1 (prostate associated gene-1). By comparison to the peptide motifs found for (AL4)GAGE-7 and the other GAGE family members, (AL5)PAGE-1 had several differences. There were found to be 6 potential casein kinase II phosphorylation sites (data not shown), 1 potential protein kinase C phosphorylation site (Fig. 4C, boxed) and 3 potential N-myristoylation sites (Fig. 4C, blue shading). Both of the consensus protein kinase C sites found in the GAGE family were not found in (AL5)PAGE-1. The loss of the protein kinase C motifs was due to the N-terminal divergence of PAGE-1 while a new protein kinase C motif was generated in the central, nonhomologous duplication in the PAGE-1 gene. The additional N-myristoylation sites were generated by the centrally located sequence duplication observed in the PAGE-1 gene.
The mRNA expression of (AL5)PAGE-1 and, albeit to a lesser degree, (AL4)GAGE-7 were examined by RNA blotting to determine their hormonal regulation, cancer cell line expression, and normal tissue distribution. Levels of PAGE-1 mRNA expressed were unaffected by androgen treatment (Fig. 5) following a 24-h exposure to 10 nM dihydrotestosterone. When compared with hybridization for GAPDH as a loading control, the difference in expression with dihydrotestosterone treatment was less than 2-fold. (AL5)PAGE-1 and (AL4)GAGE-7 mRNA expression were examined by RNA blot in a variety of cancer cell lines (Fig. 6). Discordant expression of (AL5)PAGE-1 and (AL4)GAGE-7 mRNAs was not observed. However, in the LN-CaP progression series (LNCaP, C4, C4-2, and C4-2B4), HeLa, HT144, and HT1080 cell lines the expression of (AL4)GAGE-7 greatly exceeded the expression of (AL5)PAGE-1. Neither of the androgen-independent, PSA negative prostate cancer cell lines, DU145, nor PC-3, expressed (AL5)PAGE-1 or (AL4)GAGE-7 mRNA. Furthermore, the other genitourinary tumor-derived cell lines, SKOV-3, UpS7, Hu609, and A431, did not express either (AL5)PAGE-1 or (AL4)GAGE-7 mRNAs. The highly aggressive cervical carcinoma, HeLa, did, however, express both (AL5)PAGE-1 and (Al4)GAGE-7 RNAs. Human multiple tissue mRNA blots, MTNI and II (CLONTECH) were used to determine the tissue specificity of PAGE-1 expression (Fig. 7). (AL5)PAGE-1 mRNA expression was detectable in the placenta, and high levels of mRNA were found expressed in the testes. The tissue-specific expression of (AL4)GAGE-7 was assumed to be the same as for the other GAGE family members and was not examined.  4. Comparison of the homology between (AL4)GAGE-7, (AL5)PAGE-1, and the other members of the GAGE melanomaassociated tumor antigen gene family. A, schematic representation of the homology alignment. (AL5)PAGE-1 has divergent 5Ј-UTR sequences Murine serum AL5-30 1:250 was used in an immunoblot to detect PAGE-1 in lysates of prostate cancer cell lines (Fig. 8). Consistent with the results obtained for RNA blotting, the detection of PAGE-1 protein in LNCaP and C4 was much lower than the expression observed for C4-2 or C4-2B4. Again, these data correlated with the increased tumorigenicity, androgen independence, and metastatic capacity of C4-2 and C4-2B4 as compared with LNCaP and C4. However, no protein expression was detected in the PSA-negative, androgen-independent, and highly tumorigenic prostate cancer cell lines, DU145 and PC-3, under normal growth conditions. DISCUSSION Prostate cancer is the most common noncutaneous malignancy in the United States, and second only to lung cancer in mortality (20). Recent epidemiological analysis anticipates 334,500 new cases and more than 41,000 deaths in 1997 (21,22). For metastatic disease, androgen ablation (surgical or medical) has been a long standing standard therapy (23,24). With time, however, prostate cancer progresses from an androgen-sensitive state to an androgen-insensitive state. The change in hormonal sensitivity is the hallmark of treatment and disease dissemination.
Until recently, there were no human-derived model systems to study PCa metastasis to bone and corresponding acquisition of androgen independence. This laboratory was able to manipulate the androgen-sensitive, PSA-secreting human PCa cell line LNCaP in vivo to produce lineage-related cell sublines, C4, C4-2, and C4-2B, that were progressively more tumorigenic, metastatic, and androgen-independent (3,4). In this report we have documented the cloning and initial characterization of a gene, (AL5)PAGE-1, that is differentially expressed as the cell lines acquire further metastatic and androgen independence. Furthermore, we have isolated a new member of the GAGE family of melanoma-associated tumor antigens, (AL4)GAGE-7, the expression of which does not correlate with tumor progression in this model system.
The regulation and function of the PAGE-1 is currently uncharacterized. The expression pattern of PAGE-1 in normal tissues is restricted similarly to that of the MAGE, BAGE, and GAGE gene families. Like PAGE-1, MAGE gene RNA expression was found to be very restricted in normal tissues; expression was limited to testicular and placental samples (25). A lack of MAGE expression in normal prostate was documented for MAGE-Xp using multiple human tissue poly(A) ϩ Northern blots, while a strong testicular expression was maintained (26). Gaugler et al. (27) showed that MAGE-3 was expressed in some prostate cancers. They did not, however, examine PCa cell lines or normal prostate tissue. Interestingly, human ovarian tissue as well as a large insertion in the coding region that appears to be a duplication of an exon followed by divergence. B, nucleotide pile-up of (AL4)GAGE-7, (AL5)PAGE-1, and the GAGE gene family. Sequences were weighted at the translation initiation to align the coding sequence. Conserved nucleotides are indicated by a dash, while nucleotide differences are indicated in lowercase letters and gaps are indicated by dots. C, amino acid pile-up of the translated sequences. Boxes represent putative protein kinase C phosphorylation sites, while blue shaded regions represent potential N-myristoylation sites. (AL5)PAGE-1 has lost two protein kinase C sites conserved in the other GAGE members but has gained a new protein kinase C site in the large insertion. Furthermore, (AL5)PAGE-1 has gained two potential N-myristoylation sites when compared with the other GAGEs. failed to express MAGE RNAs. Similar data was acquired for the GAGE antigens. Van den Eynde et al. (28) demonstrated GAGE expression in fetal and adult testes but in no other normal tissues. This report showed further that GAGE expression could be found in some prostate cancers but not in normal prostate tissue (28). These data establish that other members of the MAGE/GAGE gene families are expressed as cancer develops in the prostate.
The deregulated expression of MAGE/GAGE family members in many cancer specimens from varied tissues of origin appears to be a common event. Deregulated expression of GAGE or MAGE melanoma-derived tumor antigen family members in cancer has been documented in melanoma, lung, breast, bladder, stomach, and tumors of neuroectodermal origin (25, 29 -35) among others with some showing promise for predicting disease progression (32,34). The lack of PAGE-1 expression in MCF-7 breast cancer cells is similar to that described previously for MAGE (25,36), while other breast cancer cell lines were positive for MAGE expression. These data and others (25,28,37) reflect the heterogeneous expression of MAGE/GAGE family members in cancer specimens. The apparent lack of expression of PAGE-1 and GAGE-7 in the androgen-independent prostate cell lines DU145 and PC3 may either reflect heterogeneous expression in prostate cancer or may result from the highly aggressive nature or advanced progression of these tumor cell lines. Regardless, to determine the relevance of PAGE-1 expression in prostate cancer, further analysis of human prostate tissues will be required.
While not sharing apparent peptide sequence homology between families, MAGE, BAGE, and GAGE proteins have predicted pI values between 3.9 and 4.8. Of these three analogous families, the MAGE group has been the most extensively studied. MAGE proteins have been localized to the cytosol (38 -40). The MAGE promoter contains a CpG methylation site (41). In vitro expression of MAGE-3 is up-regulated by treatment with the demethylating agent 5-azadeoxycytidine (42). Demethylation of CpG islands has recently been shown to be one mechanism for expression of MAGE-1 in melanoma cells (43). In addition, the MAGE promoter region contains two consensus sites for the ets transcription factor, and one site for Sp1 (44). While conferring activity, the transcription factor sites do not appear to specifically regulate MAGE expression. Transfection of a MAGE-4 promoter-reporter construct in a sarcoma cell line not expressing MAGE-4 nevertheless resulted in generation of transcriptional activity from the construct (44).
The MAGE/GAGE gene families were cloned from peptides isolated from reactive cytolytic T lymphocytes against patient melanomas (45). As tumor-associated antigens, processed MAGE, BAGE, and GAGE peptide fragments are presented in the antigen-presenting site of the class I major histocompatibility complex, where they prime and signal a specific antitumor immune response by autologous cytotoxic T lymphocytes (46,47). The genes that were cloned, therefore, represent functional targets for immune cell presentation, which has been well documented (46,48). There has also been some success treating patients with adoptive immunotherapy following sensitization to MAGE antigens (49 -52). As an analogous protein, PAGE-1 has the potential for use as an immunotherapy target in prostate and other cancers.