Main

Superficial bladder cancers often show multifocal occurrences or metachronous recurrence after transurethral resection (TUR), and eventually develop into invasive bladder cancer. Allelic loss on chromosome 9 is the most frequent genetic event in transitional cell carcinomas of the bladder,1, 2, 3, 4 that is observed in 70% of invasive bladder cancers and even in 50% of superficial bladder cancers at Stage G1.4 Whether or not loss on chromosome 9 arises in urothelial lesions such as dysplasia is crucial to the understanding of early genetic events in bladder carcinogenesis. Some authors have reported on the allelic loss of chromosome 9 that occurs in the small urothelial lesions and normal bladder urothelium in their attempts to trace genetic alterations using microsatellite markers.5, 6 However, it is still difficult to analyze allelic status in small epithelial regions obtained from formalin-fixed, paraffin-embedded tissues, and a few data have been reported regarding early genetic alterations in bladder dysplasia.3, 7 ABO (H) blood group antigens are constitutively expressed on epithelial cells such as those found in the gastrointestinal tract and urothelium. A reduction in blood-group A antigen (GalNAcα1-3[Fucα1-2]Galβ1-3GlcNAc-R) expression was reported in transitional cell carcinoma (TCC) of the bladder and showed significant correlation with an invasive phenotype.8, 9, 10, 11 Orntoft and Wolf12 examined the correlation between blood-group antigen expression and the activity of glycosyltransferases in TCC of the bladder and reported that the activity of A glycosyl transferase was severely reduced in tumors showing loss of A antigen expression. This phenomenon drew our attention, due to the fact that the determinant of the ABO blood-group antigen is synthesized by the action of the ABO gene encoding ABO glycosyltransferase assigned to chromosome 9q34.1, where loss of heterozygosity (LOH) was frequently reported in bladder cancer.1, 2, 3, 4 The ABO gene is composed of seven exons and six introns and encodes ABO glycosyltransferase, of which substrate specificity is determined by genetic polymorphisms in exons 6 and 7 (Figure 1).13, 14 Blood-group A antigen is synthesized by α-N-acetyl-galactosaminyltransferase (A-GalNAc transferase), which catalyzes the transfer of N-acetylgalactosamine to the subterminal β-galactosyl residue of the blood-group H carbohydrate chain. Blood-group B-antigen is synthesized by B-galactosyl transferase, which catalyzes the transfer of galactose to the subterminal β-galactosyl residue of the blood-group H carbohydrate chain. The ABO gene in blood-group O donors lacks glycosyltransferase activity, for it has a deletion on a guanine residue at the nucleotide position 261 in exon 6, causing protein truncation at codon 117.13, 14, 15, 16 Immunohistochemistry using anti-A monoclonal antibody in bladder cancer may be useful to evaluate the allelic status of the ABO gene locus at 9q34.1 in those who are heterozygous for ABO genotypes. Expression of blood-group A antigen is stable enough even in formalin-fixed paraffin-embedded specimens, and this could be applicable in the analysis of small lesions that are too small to be examined by genetic analysis. Two papers were so far reported as to the correlation between reduced expression of A antigen and A allelic loss in TCCs of the bladder.17, 18 Meldgaard et al17 analyzed 22 bladder tumors for LOH of the 9q allele by PCR-restriction fragment length polymorphism (RFLP) analysis of the ABO locus at 9q34. Seven tumors from heterozygous informative individuals were sorted by flowcytometry. LOHs were detected in the most aneuploid subpopulation of cells in two cases, but both cases were losing O-alleles. No LOHs were detected in analysis of the low aneuploid subpopulation. As all tumors showed loss of blood group ABH antigen expression, they concluded that LOH of the ABO locus on chromosome 9q34 is not the cause of loss of blood group ABH expression in human bladder cancer.17 Orlow et al18 analyzed 19 patients with bladder cancer serologically typed as blood group A. Expression of A antigen was maintained in 14 samples in normal urothelium, while it was reduced in nine tumors. PCR-RFLP analysis showed loss of the A allele in one tumor sample showing reduced expression of the A antigen. They indicated that the lack of the A antigen expression in certain bladder tumors is due to the allelic loss of the ABO gene and that in some of these tumors, the loss involved the surrounding chromosomal region at 9q34.1–4.18 These two reports did not support the correlation between A-allelic loss and the reduced expression of the A antigen in the majority of bladder cancers. Recent advance in cancer epigenetics shed light on the reduced expression of A antigen in malignant cells. Kominato et al19, 20 reported that hypermethylation of the promoter region of the ABO gene induced ABO gene silencing in their study using a human stomach carcinoma cell line. Iwamoto et al21 established subclones with positive or negative expression of the A antigen from parental colonic cancer cell lines and reported a distinct difference in the methylation pattern of the CpG island of the promoter region of the ABO glycosyltransferase, that is densely methylated in a subclone lacking the expression of the A antigen. Gao et al22 examined 30 oral squamous carcinomas for expression of the A and B antigens and A/B glycosyltransferase, together with LOH at the ABO locus and hypermethylation of the ABO gene promoters. Loss of A or B antigen expression was found in 21 of 25 tumors (84%), while the expression of the glycosyltransferase was absent in all of tumors showing negative expression of A or B antigens. Loss of the A or B allele was found in 3/20 tumors (15%) heterozygous for the ABO locus and hypermethylation of the promoter region in 10 of 30 tumors (33.3%).22 Furthermore, Habuchi et al23 reported that the region 9q32–9q33, which is in the vicinity of the ABO gene locus at 9q34.1, is a frequent target of LOH and methylation in bladder cancer. These findings prompted us to hypothesize that deletion of blood-group A antigen expression in TCC of the bladder might be regulated by a combination of genetic and epigenetic mechanisms, that is, an LOH of the ABO gene locus and hypermethylation of the ABO gene promoter region. The purpose of this study was to elucidate the relevant mechanisms underlying the loss of blood group A antigen expression in TCC of the bladder and whether it could be used as a phenotypic marker to estimate any underlying genetic and epigenetic abnormalities in normal urothelium and concomitant bladder dysplasia in patients with bladder cancer.

Figure 1
figure 1

Schema of single nucleotide polymorphisms (SNPs) in exons 6 of the ABO gene and electropherogram of the blunt-end SSCP analysis showing examples of normal DNA from blood group A or A/B donors. SNPs in nucleotide positions 261 and 297 were used for analysis in this study. DNA variants and their allelic frequencies reported in the Japanese are indicated.15 The blood group O gene has a single base deletion at position 261 resulting in a frame-shift mutation and causing protein termination at codon 117.

Full size image

Materials and methods

Samples and DNA Extraction

A total of 81 cases of TCC of the bladder were studied, of which 44 underwent TUR and 37 underwent radical cystectomy (Table 1). The histo-blood group for all cases was A (72 cases) or AB (nine cases) examined by routine hemagglutination tests at hospital. Tumors were graded and staged according to the WHO classification or the 1997 UICC TNM classification system. Based on patients’ history, the proportion of cases with advanced stage or high-grade tumors was significantly higher in those who underwent radical cystectomy than those who underwent TUR (P<0.01). In 44 patients who underwent TUR, DNA was extracted from fresh specimens and normal DNA was extracted from peripheral blood lymphocytes (PBL) by a standard procedure using proteinase K digestion followed by phenol–chloroform extraction. In 37 cases that underwent radical cystectomy, a total of 1130 paraffin-embedded specimens obtained from mapping study of the bladder were histologically confirmed by hematoxylin and eosin staining as being composed of tumor, dysplasia and normal tissues. DNA was extracted from manually dissected tumors and corresponding normal tissues using DEXPAT (TAKARASHUZO Co., Ltd, Shiga, Japan) according to the manufacturer's recommendation.

Table 1 Patient background
Full size table

Expression of Blood-Group A Antigen by Immunohistochemical Staining

In all, 4-μm-thick sections from formalin-fixed, paraffin-embedded specimens of resected tissues that underwent TUR or radical cystectomy were used for immunohistochemical staining (IHC). A mapping study of the bladder specimens revealed concomitant dysplastic lesions in 23 cases that underwent radical cystectomy, and they were then subjected to IHC performed as described previously.24 Mouse monoclonal antibody (mAb) directed against A antigen (clone 81FR2.2; DAKO, Carpinteria, CA, USA) was used as the primary antibody and the avidin–biotin-conjugated immunoperoxidase technique was performed with a DAKO LSAB2 Kit (DAKO, Carpinteria, CA, USA). Reportedly, the specificity of the mAb 81FR2.2 was characterized by transfection experiment of the A-glycosyl transferase gene to the HeLa cell (genotype OO).25 Erythrocytes, normal epithelium and vascular endothelium were used as internal positive controls, while muscle and connective tissues served as negative controls. To determine the specificity of A antigen, IHC was performed for normal urothelium of blood group B and O donors. Immunohistochemistry for A antigen was classified as follows: ‘negative’ if the section had no positively (0%) stained tumor cells, ‘positive’ if staining was seen across the section (>70% positively stained tumor cells), and ‘heterogenous’ if <70% of tumor cells stained positively. As to the correlation with A allelic loss or methylation status, cases showing positive or heterogenous expression were compared with those showing negative expression.

Allelic Status on 9q Loci Defined by Blunt-End Single-Strand DNA Conformation Polymorphism Analysis

LOH of the ABO gene locus was examined by blunt-end Single-strand DNA conformation polymorphism (SSCP) analysis,26 using genetic polymorphisms at nucleotide positions 261 and 297 in exon 6 of the ABO gene. Genotypes and their allelic frequencies in Japanese population were previously reported15 and shown in Figure 1. Four groups of alleles, A (A101, A102, A103), B (B101, B102, B103, A104), O1 (O101, O102, O202, O203) and O2 (O103, O201) were identified by the analysis of two genetic polymorphisms (nucleotides 261, 297) in exon 6 of the ABO gene. The 5′-terminus of the reverse primer was labeled with Cy5 flourescent dye. The nucleotide sequences of the forward and reverse primers were 5′-TCTCCATGTGCAGTAGGAAGGATG-3′ and 5′Cy5-ATGGCAAACACAGTTAACCCAATG-3′, respectively. PCR conditions were as follows: 0.5–1.0?μg of genomic DNA as a template, 0.2 μmol/l of each primer, 0.125?mmol/l deoxynucleoside triphosohate (dNTP), 0.25 units of AmpliTaq Gold DNA polymerase (Perkin Elmer-Cetus, Norwalk, CT, USA) in a total reaction volume of 25?μl. After the first denaturation step at 95°C for 12?min, 40 cycles were performed for amplification consisting of 30?s at 95°C, annealing for 30?s at 57°C, and extension for 30?s at 72°C followed by a final extension at 72°C for 7?min. PCR products were then treated with Klenow fragment (TAKARA SHUZO Co., Ltd, Shiga, Japan) to generate DNA fragments with blunt ends. To 1?μl of each PCR product, 0.5 units of Klenow fragment was added, and the mixture was incubated at 37°C for 30?min. One microliter of this reaction mixture was diluted with 10?μl of loading solution (90% deionized formamide, 20?mM EDTA, 0.05% bromophenol blue) and heat denatured at 95°C for 5?min. An ALF red automated DNA sequencer™ (Pharmacia, Tokyo, Japan) was used for blunt-end SSCP analysis. One microliter of the diluted mixture was applied onto a 15% polyacrylamide gel (30:1, acrylamide:bisacrylamide ratio) containing Tris/glycine buffer (25?mM Tris, 192?mM glycine). Electrophoresis was performed at 30?W for 16?h using a continuous buffer system consisting of 25?mM Tris and 192?mM glycine. During electrophoresis, the gel was maintained at a constant temperature of 18°C by a circulating water bath. The data were analyzed using the ALF Win Fragment analyzer 1.02™ software package (Pharmacia, Tokyo, Japan). LOH was determined by measuring the signal ratio between the opposing alleles and defined as tumor cellularity according to the equation that we previously reported.4, 26, 27 Supposing that the A1 allele is lost in a heterozygote carrying A1 and A2 alleles, T is the peak height of the signal from the tumor samples and N is the peak height of the signal from normal control. The tumor cellularity in the sample is thus given as follows:

Genomic DNA from normal PBL was analyzed to set the cutoff values for tumor cellularity. As previously reported, the mean+3s.d. values of the normal heterozygous DNA were used as a cutoff value for tumor cellularity, and tumor samples showing tumor cellularities above the cutoff level were considered to have LOHs.4 A104 allele was indistinguishable from B allele in this analysis, while the observed frequency of the A104 allele in the Japanese is reported to be as low as 1.3%. In fact, in all samples tested, the genotypes coincided with the patient's ABO isotypes. In addition, two single base nucleotide polymorphism markers (ALDOB, 9q21.3 and VAV2, 9q34.1) were used to assess the allelic status on 9q according to the method that we previously reported;4 the former is centromeric and the latter is telomeric to the ABO gene locus, respectively (Figure 4). Nucleotide sequences of the forward and reverse primers for ALDOB and VAV2 were as follows: 5′Cy5-GGGCTTGACTTTC CAACACG-3′ and 5′-TCTAGCCTCAATCCTCATAC-3′ (ALDOB), 5′-GTGTCTGCACTGGCCACACT-3′ and 5′Cy5-TCCAAAGGACCTTCTCCAAA-3′ (VAV2).

Figure 4
figure 4

A case of bladder cancer showing chimeric expression of the A antigen. DNA was extracted from areas showing positive or negative A antigen expression and subjected to blunt-end SSCP analysis using three single nucleotide polymorphic markers (ALDOB, ABO and VAV2) on 9q. The patient's genotype was A/O2. A allele was lost in the sample taken from the area showing negative A antigen expression, while O2 allele was lost in the sample taken from the area showing positive A antigen expression. Note that two polymorphic loci (ALDOB, 9q21.3 centromeric to the ABO locus and VAV2, 9q34.1 telomeric to the ABO locus) also showed LOH and suggested a large regional chromosome deletion, while the parental origin of the lost allele in these two loci was different between areas showing A-antigen positive or negative expression.

Full size image

Bisulfite PCR-SSCP Analysis and Methylation-Specific PCR

In cases that underwent TUR, methylation status in the promoter region of the ABO gene was analyzed by bisulfite PCR-SSCP (BiPS) and methylation specific PCR (MSP).24, 28, 29.Seven primer sets were designed to amplify seven overlapping regions spanning the CpG island located from −765 to +21 relative to the translation start site (Figure 2). Primer sets re 1 through re 6 were designed for BiPS analysis and RE7.M and RE7.UM were for MSP. Bisulfite treatment was performed using the CpGenome DNA Modification Kit (Intergen Co., New York, NY, USA). In all, 1?μg of tumor-derived DNA was treated with Na-bisulfite according to the manufacturer's recommendations. PCRs were performed in 25?μl reaction volumes containing 10 × buffer, 1.0?μl bisulfite-modified DNA corresponding to 50?ng of genomic DNA as a template, 0.2?μmol/l of each primer, 0.125?mmol/l dNTP and 0.25 units of AmpliTaq Gold DNA polymerase. PCR conditions were 95°C for 9?min for heat denaturation, 40 cycles of 94°C for 1?min, 1?min at the different annealing temperatures for each primer set (Table 2), 72°C for 2?min for amplification, followed by a final extension at 72°C for 10?min. The BiPS procedure was performed as previously described.28, 29 Nondenaturing polyacrylamide gels of 8% for re 2 and re 6, 10% for re 1, re 4 and re 5, and 15% for re 3 were used for the analysis. CpGenome™ Universal Methylated DNA (CHEMICON International, Temecula, CA, USA) was used as a positive control, and PBL obtained from healthy control donors were used as a negative control. When extra bands were observed, they were cut from the gels, reamplified and subjected to direct sequencing using ABI 3100 PRISM sequencer with a Big-Dye terminator sequencing kit (Perkin-Elmer). In analysis of cases that underwent radical cystectomy, BiPS analysis was not employed due to the technical difficulty for reliable amplification of relatively long sized DNA fragments from formalin-fixed paraffin-embedded sections. In cases that underwent radical cystectomy, methylation status was assessed by MSP of region 7, the most proximal to the translation start site. The size of the PCR product was as short as 96?bp and amplifiable from archival samples with relative ease; however, five out of 37 cases that underwent radical cystectomy failed in PCR amplification. Methylation status of region 7 was used as the surrogate indicator for extensive methylation of the CpG sites or full methylation.

Figure 2
figure 2

Map of the 5′ CpG island of the ABO gene and result of BiPS analysis. (Top) CpG sites in the promoter region of the ABO gene are indicated by vertical lines. (Middle) The amplified DNA fragments from regions 1 to 7 are indicated. PCR primer set of each region was indicated by arrows. (Bottom) BiPS analysis of the ABO gene. Extra bands are indicated by asterisks. After SSCP analysis, the extra bands were excised from gels, reamplified by PCR, and sequenced. Results of the direct sequencing of the case with full methylation were shown in the lower panel. U: unmethylated control, M: methylated control, N: no methylation, P: partial methylation, F: full methylation, *extra band showing mobility shift.

Full size image
Table 2 Primer sequences for BiPS analysis and MSP
Full size table

Statistical Analysis

Statistical analysis was performed using a likelihood χ2 analysis or Fisher's exact test. Probability (P) values of <0.05 were considered to be significant.

Results

Expression of the A Antigen in TCC of the Bladder by IHC

Expression of the A antigen in tumor and normal urothelium was examined by IHC (Figure 3A). The corresponding staining of A antigen on the normal urothelium from histo-blood-group B or O donors resulted in background levels only (data not shown). All of the normal urothelium from blood-group A individuals stained positively. The numbers of cases showing positive, heterogeneous and negative staining were 11 (25.0%), 11 (25.0%) and 22 (50.0%) in 44 tumor specimens that underwent TUR, while they were 14 (37.8%), 5 (13.5%) and 18 (48.6%) in 37 tumor specimens that underwent radical cystectomy. The overall frequencies of negative A antigen expression were 35.7% (5/14) for pTa, 58.3% (21/36) for pT1, 25.0% (3/12) for pT2, 40.0% (4/10) for pT3 and 77.8% (7/9) for pT4 stages, and 71.4% (5/7), 43.5% (10/23) and 49.0% (25/51) for Grade 1, 2 and 3 tumors, respectively. There were no significant differences between A antigen expression and tumor stages or histological grades.

Figure 3
figure 3

Expression of the blood-group A antigen, allelic status of the ABO gene and MSP of region 7 in cases that underwent radical cystectomy. (A) Immunostaining of A antigen in tumor (a, b, c), dysplasia (d, e, f), and corresponding normal urothelium (g, h, I) from cases A-10, AB-3 and A-15, respectively. A-10 showed positive staining in tumor (a), dysplasia (d) and normal urothelium (g), while the tumor section showed heterogeneous staining for the case AB-3 (b), and negative staining for the case A-15 (c). Normal urothelium from cases A-10 (g), AB-3 (h) and A-15 (i) stained positively. Reduced from × 100. High magnification view ( × 400) was shown as inset. (B, C) Analysis of LOH of the ABO gene locus using blunt-end SSCP and methylation status by MSP (RE 7). A-10 showed the expression of the A antigen in tumor tissue, no allelic loss and unmethylated CpG sites. AB-3 showed heterogenous expression of the A antigen and methylation of the ABO gene, while the A allele was retained. A-15 showed negative expression of the A antigen, loss of A alleles and methylation of the ABO gene.

Full size image

LOH on 9q in TCC of the Bladder

Allelic status of the ABO gene and neighboring loci were analyzed by blunt-end SSCP analysis using three polymorphic markers (ABO (9q34.1), ALDOB (9q21.3-22.2), VAV2 (9q34.1)) (Figure 4). Heterozygosity of each locus was 87.7% (71/81) for ABO, 52.6% (41/78) for ALDOB and 48.1% (38/79) for VAV2, respectively. As all samples were derived from patients with an A or AB blood group, heterozygosity at the ABO locus was highest of all the loci examined. Genotypes of the ABO gene were classified into four groups, that is, A/A (n=10), A/O1 (n=34), A/O2 (n=26) and A/B (n=9). The cutoff value for tumor cellularity in each genotype was defined as the mean+3s.d. of the normal DNA samples: 20% for A/O1, 22% for A/O2, 26% for A/B, respectively. In 44 cases that underwent TUR, frequencies of LOH were 53.7% (22/41) for ABO, 43.5% (10/23) for ALDOB and 50.0% (10/20) for VAV2, respectively. Frequencies of allelic loss at the ABO locus were 23.1% (9/39), 33.4% (6/18), 33.3% (5/15) and 33.3% (2/6) for A, O1, O2 and B allele, respectively. In 37 cases that underwent radical cystectomy, frequencies of LOH were 76.7% (23/30) for ABO, 77.8% (14/18) for ALDOB and 83.3% (15/18) for VAV2, respectively. Frequencies of allelic loss in the ABO locus was 23.3% (7/30), 50.0% (8/16), 54.5% (6/11) and 66.7% (2/3) for A, O1, O2 and B allele, respectively. There were no significant differences as to the frequencies of LOH between three markers and between four alleles of the ABO gene. Frequencies of LOH were higher in cases that underwent radical cystectomy as compared to the TUR cases, that is, 76.7% (23/30) vs 53.7% (22/41) for ABO (P=0.08), 77.8% (14/18) vs 43.5% (10/23) for ALDOB (P=0.054) and 83.3% (15/18) vs 50.0% (10/20) for VAV2 (P=0.043), among which VAV2 locus showed statistical significance.

Methylation Status of the ABO Gene Promoter Region

CpG island of the ABO gene extends from 0.7?kb upstream to 0.6kb downstream from the translation start site in exon 1. Reportedly, the promoter region of the ABO gene is located between −117 and +31 from the translation start site, of which hypermethylation regulates gene expression.19, 20 In the present study, we divided CpG island spanning −789 to +6 into six regions and examined the methylation status by BiPS analysis (Figure 2). In the preliminary experiment, methylated DNA could be identified as the extra band, if more than 25% of the template DNA was methylated (data not shown). Methylation patterns were defined as follows: full methylation if all regions showed methylation, partial methylation if at least one region showed methylation and no methylation. A total of 44 TUR cases were analyzed, and we assessed the correlation between methylation status and expression levels of the A antigen using a panel of 35 cases, for nine cases showing LOH of the A allele were not included in the first assessment (Tables 3 and 4). Frequencies of methylation in re 1 through re 6 were 17.1% (6/35), 28.6% (10/35), 34.3% (12/35), 11.4% (4/35), 14.3% (5/35) and 11.4% (4/35), respectively (Table 4). In re 4, re 5 and re 6, methylation was not detected in all cases showing positive or heterogenous expression and expression of the A antigen was negative in four cases showing full methylation. Frequencies of cases showing negative A antigen expression were 100% (4/4) in full methylation, 66.7% (6/9) in partial methylation and 27.3% (6/22) in no methylation and significant association was observed between methylation status (full, partial and no methylation) and expression of the A antigen (P=0.0093) (Table 4). In analysis using MSP, methylation of RE 7 was observed in nine cases, of which six cases showed full or partial methylation in BiPS analysis and the expression of the A antigen was negative in these six cases (Table 3). Discrepancies between MSP and BiPS analysis were shown in three cases, which showed methylation only in MSP and heterogeneous expression of the A antigen. Positive expression of the A antigen was found in 11 cases, in which two cases showed methylation of regions 1 through 3 by BiPS analysis and no cases showed methylation of RE 7 by MSP (Table 3).

Table 3 Methylation status in the ABO gene promoter region and expression of A antigen
Full size table
Table 4 Correlation of the expression of A antigen with methylation status in 35 cases underwent TUR
Full size table

Correlation of the Expression of A Antigen with A Allelic Loss and Hypermethylation of the ABO Gene Promoter Region

In analysis of 44 cases that underwent TUR, loss of the A allele was observed in nine cases, among which six cases showed negative and three cases showed heterogenous expression of the A antigen (Table 5). Cases homozygous for A allele were regarded as retaining at least one copy of the ABO gene. No statistical association was found between the expression level of the A antigen and A allelic loss (P=0.26). In BiPS analysis, expression of the A antigen was negative in all of the four cases with full methylation and statistical association was shown between the expression of the A antigen and methylation status (P=0.035). Taking A allelic loss or full methylation in combination, 76.9% (10/13) cases with A allelic loss and/or full methylation showed negative A antigen expression, while the expression of the A antigen was negative in 38.7% (12/31) of cases that retained A allele and showed partial or no methylation. Cases with A allelic loss and/or full methylation showed significant correlation with negative A antigen expression (P=0.02) (Table 5). In analysis of 37 cases that underwent radical cystectomy, A allelic loss was observed in seven cases and they all showed negative A antigen expression in the tumor (Table 6). Compared with 30 cases that retained the A allele (including A/A homozygotes), the frequency of A antigen expression was significantly low in those showing A allelic loss (P=0.003) (Table 6). MSP of RE 7 showed methylation in seven cases (18.9%) in which the expression of the A antigen was negative in six cases. Methylation status was significantly correlated with negative expression of the A antigen (P=0.03). Taking A allelic loss and methylation in combination, 91% (10/11) of cases with A allelic loss and/or methylation were negative for the A antigen expression, while the expression of the A antigen was negative in 23.8% (5/21) of cases showing retained A allele and no methylation. A allelic loss and methylation were significantly correlated with the expression level of the A antigen (P=0.0005) (Table 6). In one case, the expression of the A antigen was chimeric and the tumor was divided into areas showing positive or negative expression (Figure 4). This case was an A/O2 heterozygote, and the allelic status was determined from the dissected specimen. O2 allele was lost in the area showing positive staining, while the A allele was lost in the area showing negative staining. Allelic status was also examined in the ALDOB and VAV2 loci, where the parental origin of the lost allele was different between positively and negatively stained areas, indicating that allelic loss in the tumor involved large chromosomal region between 9q21.3 and 9q34.1.

Table 5 Correlation of the expression of A antigen with A alleleic loss and hypermethylation of the ABO gene promoter region in 44 cases that underwent TUR
Full size table
Table 6 Correlation of the expression of A antigen with A allele loss and/or hypermethylation of the ABO gene promoter region in 37 cases that underwent radical cystectomy
Full size table

Expression of the A Antigen in Dysplasia and Normal Urothelium

A total of 23 cases that underwent radical cystectomy were examined for expression of the A antigen in concomitant dysplastic lesions and normal urothelium (Table 7). In analysis of 13 cases showing positive A antigen expression in the tumor, A allele was retained in all cases and only one case showed hypermethylation together with normal expression of the A antigen in the dysplasia specimen. In analysis of 10 cases showing negative expression of the A antigen in the tumor, eight showed A allelic loss and/or methylation. Abnormal expression of the A antigen was observed only in one case (A-9), in which dysplasia specimen showed heterogeneous expression but A allelic loss and methylation were not observed in the tumor.

Table 7 Correlation of the expression of A antigen in the tumor, dysplasia and normal urothelium specimens with the genetic and epigenetic changes in the primary tumor
Full size table

Discussion

Previously, we reported that LOH on chromosome 9 was a frequent genetic event in TCCs of the bladder and its detection in urine samples would be an useful indicator for tumor recurrence in patients with TCC that underwent TUR.4 Frequencies of LOH of the ABO locus examined in this study seems higher than those reported previously.17, 18 In previous studies, allelic status of the ABO gene was examined by PCR/RFLP; however, LOH is barely detectable by PCR/RFLP if the proportion of tumor cells in the sample is below 60%, due to the formation of heteroduplex dimers that are resistant to the restriction enzyme digestion.30 Blunt-end SSCP analysis is a sensitive method to detect an LOH from clinical samples, of which the proportion of tumor cells is as low as 10–20%.26 However, LOH study from small lesions such as concomitant dysplasia was still difficult due to technical problems. Slebos et al31 reported that the lower the amount of DNA in the PCR, the greater the risk for allele ratios that were abnormal due to a chance distribution of alleles in the reaction and the DNA equivalent of a minimum of about 100 cells is required for a full representation of both alleles in the analysis. Furthermore, DNAs extracted from formalin-fixed paraffin-embedded sections often harbor degradation and fail in the PCR amplification, suggesting potential difficulty in assessing the allelic status of small lesions from archival materials. The aim of the present study was to elucidate the underlying mechanisms of reduced expression of the histo-blood group A antigen in bladder cancer, and to determine if IHC of the A antigen expression could be available as a hallmark to determine the allelic loss and/or epigenetic alterations of the ABO gene on a cell-to-cell basis.

In cases with radical cystectomy, allelic status was examined using DNAs extracted from histological slides and directly comparable with the A antigen expression in the same specimen and expression of the A antigen was negative in all cases showing A allelic loss. In cases that underwent TUR, three cases showed heterogenous expression of the A antigen, regardless of A allelic loss in the sample. In TUR cases, DNAs were extracted from fresh frozen samples obtained by cold-cup biopsies, while the expression of the A antigen was examined in formalin-fixed paraffin-embedded sections of the resected tumors. Discrepancies between A allelic loss and A antigen expression in TUR cases may be explained by the difference of materials subjected to analysis. As we indicated in Figure 4, some tumors show polyclonal development as to the allelic loss of chromosome 9 and direct comparison between biopsies and resected specimen may be difficult in such cases. In BiPS analysis, full methylation was observed in four cases and they all showed negative expression of the A antigen (Table 5). CpG islands were densely methylated in full methylation and they were closely correlated with the transcriptional silencing of the ABO gene. In cases with partial methylation, A antigen expression was also negative in 66.7% (6/9) of cases. Although partial methylation may play some role in transcriptional silencing, we used full methylation as an indicator of methylation in this study. As methylation extended to the most downstream of the ABO gene promoter region (re 6) in full methylation, we designed a primer set for MSP spanning region 7, which overlapped the downstream of region 6. The size of the amplified DNA fragment in MSP was as short as 96?bp and we used it as an indicator of full methylation in analysis of DNAs extracted from formalin-fixed paraffin-embedded sections. As MSP amplifies methylated DNA sequences selectively, its sensitivity is much higher than that of BiPS analysis and may have a risk of overestimation. In fact, MSP showed methylation in three cases that showed no methylation in BiPS analysis and the expression of the A antigen in these three cases were heterogenous. This may indicate the heterogeneity of the methylation status, suggesting only small number of cells harbored methylation (Table 3). In cases that underwent TUR, negative A antigen expression was significantly correlated with full methylation (P=0.035), but not with A allelic loss (P=0.26) (Table 5). In cases that underwent radical cystectomy, both methylation and A allelic loss were significantly correlated with the expression of the A antigen (P=0.003 for A allelic loss, P=0.03 for MSP, respectively). Using these two indices in combination, 29.5% (13/44) of the cases that underwent TUR and 29.7% (11/37) of cases that underwent radical cystectomy showed loss of the A allele and/or hypermethylation of the ABO gene. They were significantly correlated with the expression of the A antigen (P=0.02 for TUR cases, P=0.0005 for radical cystectomy cases) (Tables 5 and 6). Negative A antigen expression was observed in 50.0% (22/44) in TUR cases and 48.6% (18/37) in cases that underwent radical cystectomy, which was attributable to genomic deletion and/or hypermethylation of the ABO gene in at least 45% (10/22) of cases that underwent TUR and 66.7% (10/15) of cases that underwent radical cystectomy. It is apparent that A allelic loss and/or hypermethylation of the ABO gene could not be the sole cause for negative A antigen expression. As the antigenic determinant of the A antigen is the terminal structure of the carbohydrate chains, incomplete synthesis of carbohydrate chains associated with oncogenesis may also be concerned with the reduced expression of the A antigen. Methylation seems to be more predominant than loss of the A allele in cases that underwent TUR. This might be explained by the observation that superficial papillary tumors such as pTa or pT1 stages comprised most of the TUR cases, while more than 70% of them were invasive cancers above Stage pT2 in cases that underwent radical cystectomy. In our previous study, frequencies of LOH on chromosome 9 were 67% in pTa, 71% in pT1 and 80% in tumors ≧pT2 stages.4 As for the putative tumor suppressors found on chromosome 9, p16 and p14ARF are located on 9p21.32, 33 And an area on 9q31–34 is most prone to be deleted in TCC of the bladder,34, 35 which is also a candidate locus for a putative tumor suppressor gene. Reportedly, deletion of chromosome 9 is an early genetic event in the development of bladder cancers.1 However, there is not enough evidence to support this hypothesis regarding the occurrence of chromosome 9 deletion in preneoplastic lesions. In a few studies using microsatellite markers from microdissected specimens, allelic loss on chromosome 9 was observed in bladder dysplasia.3, 7 We studied the expression of the A antigen on the dysplasia specimens by IHC, aiming at screening genetic alterations in precancerous lesions of the bladder. Expression of the A antigen was examined in 23 cases of bladder cancer comprising dysplasia, among which the numbers of tumors showing positive or negative expression were 13 and 10, respectively. All of the cases showing positive expression retained the A allele in the tumor and only one case showed hypermethylation, while the expression of the A antigen was preserved in dysplasia and normal urothelium in all cases. In the analysis of 10 cases showing negative A antigen expression in the tumor, loss of the A allele and/or the hypermethylation was observed in eight cases. Expression of the A antigen was preserved in normal urothelium and dysplasia in all but one case showing heterogenous expression in the dysplasia. This case did not exhibit LOH or hypermethylation in the tumor. These results suggested that LOH and/or hypermethylation of the ABO gene were infrequent genetic and epigenetic alterations in dysplasia and normal urothelium of the bladder bearing TCC. Furthermore, one case showed chimeric expression of the A antigen in the tumor, among which the expression of the A antigen coincided with loss or retention of the A allele (Figure 4). Analysis of two polymorphic markers in the vicinity of ABO gene locus also showed LOHs and the parental origin of the lost allele in these two loci was opposite as was shown in analysis of the ABO gene locus. Previously, we reported loss of chromosome 9 was observed in 71% of TCCs of the bladder and nearly 50% of them involved both 9p and 9q, suggesting monosomy or uniparental aneusomy of chromosome 9.4 Thus, the deletion was considered to involve large chromosomal regions at least between 9q21.3 and 9q34.1 and possibly on the same allele. This finding may suggest the idea that the tumor showed polyclonal development as to the deletion of the 9q allele and that the loss of chromosome 9 might not be an early genetic event associated with tumorigenesis.

In conclusion, reduced expression of the A antigen in bladder cancer reflects allelic loss of the ABO gene assigned to 9q34.1 and/or hypermethylation of its promoter region, which is a specific marker for genetic and epigenetic alterations in bladder cancer but not in dysplasia.