High JC virus load in gastric cancer and adjacent non‐cancerous mucosa

Abstract

The JC virus (JCV) infects a large proportion of the worldwide population and approximately 90% of adults are seropositive. Recent reports have described the possibility of its oncogenetic role in several malignancies. The aim of the present study was to assess the oncogenetic significance of JCV for gastric cancer. Twenty‐two sample pairs of fresh tumor and adjacent non‐cancerous tissue (ANCT) as well as 10 normal gastric mucosa specimens were investigated on the basis of nested polymerase chain reaction (PCR) followed by Southern blotting, DNA direct sequencing, real‐time PCR, in situ PCR and immunohistochemistry. The T antigen sequence was detected in 86.4% of gastric cancers and ANCT, and in 100% of the normal mucosa samples, as for virus capsid protein, 54.1%, 68.1% and 70%, respectively. A generally low incidence was noted for agnoprotein. The JCV DNA load was approximately 10‐fold higher in both gastric cancers and paired ANCT (4784 ± 759 and 5394 ± 1466 copies/µg DNA, respectively) than in normal gastric tissue (542.4 ± 476.0 copies/µg DNA, P < 0.0001). In situ PCR revealed sporadic JCV genome‐positive cancer cells and foveolar epithelial cells. T antigen protein expression assessed by immunohistochemistry was detected only in one case (1/22; 4.5%), probably because the half life of T antigen might be short. It was concluded that the gastric epithelium in most Japanese people is infected with JCV at a low rate but levels of infection are increased markedly in both cancer cells and ANCT, indicating that multiplication of JCV copies might be a risk factor and a background for gastric carcinogenesis. (Cancer Sci 2007; 98: 25–31)

Serological studies have indicated that approximately 90% of the world's adult population is asymptomatic for infection with the human polyomavirus, JC virus (JCV). The virus enters through tonsillar stromal tissue and persists quiescent in the kidney and lymphoid tissue during latency, but may be activated under immunosuppressive conditions, leading to the lethal demyelinating disease, progressive multifocal leukoencephalopathy (PML).^{( 1 )} The JCV genome is double‐stranded and characterized by negatively supercoiled circular DNA, 5130 bp in length, with early and late coding regions separated by a transcription control region. The early region is alternatively spliced to produce two mRNAs that encode the large T antigen and the small t antigen. The T antigen, a nuclear phosphoprotein active in viral DNA replication, binds to the viral replication region to promote unwinding of the double helix and recruitment of cell proteins that are required for DNA synthesis. The late region encodes the capsid structural proteins virus capsid protein (VP)1, VP2 and VP3 due to alternative splicing, and a small regulatory protein known as agnoprotein.^{( 2 )}

JC virus infection initiates binding to JCV‐sensitive cell surfaces and JCV caspids undergo endocytosis and are transported to the nucleus where the viral DNA is uncoated and the early region begins to be transcribed by a clathrin‐dependent mechanism. In permissive infections, viral DNA can replicate, resulting in lytic infection (PML) with viral amplification. When this is blocked, abortive infection or cell transformation may result.^{( 3 )} JCV can transform cells, as manifested by distinct morphological changes such as rapid division, prolonged life span, enhanced production of plasminogen activator, anchorage‐dependent growth, unstable multicentric chromosomes, centric and acentric rings, and the ability to form dense foci in culture,^{( 4 )} and intravenous or intracranial inoculation of JCV into experimental animals has been shown to cause astrocytomas, glioblastomas, neuroblastomas and medulloblastomas.^{( 1 )} The JCV T antigen has a multifunctional modular structure, acting as an ATPase, a helicase, a polymerase and in DNA binding; all of which are essential for DNA replication.^{( 5} , ^{6 )} It can also inactivate p53 and members of the retinoblastoma protein (pRb) family to promote uncontrolled proliferation and immortal survival.^{( 7 )} Additionally, the T antigen exerts oncogenic activity partially through deregulation of the Wnt signaling pathway, associated with increased levels of β‐catenin in the nucleus and c‐myc expression.^{( 8 )} A transgenic mouse strain expressing the JCV T antigen is reported to exhibit pituitary adenomas in approximately 50% of animals by 1 year of age, some transforming to malignant peripheral nerve sheath tumors.^{( 9 )} In recent years, the presence of JCV has also been suggested to correlate with the development of various types of human cancer, including esophageal, colorectal, prostatic and bronchopulmonary carcinomas, brain tumors and B cell lymphomas.^{( 10} , ¹¹ , ¹² , ¹³ , ¹⁴ , ^{15 )} Thus there is abundant evidence that JCV can act as an oncovirus.

Gastric carcinogenesis is a multistage process with a multifactorial etiology. Exposure to biological and chemical carcinogens, such as Helicobacter pylori, the Epstein–Barr virus, nitrosamines and oxidants, can lead to DNA damage and mutation of gastric epithelial cells and increase the likelihood of neoplasia through sequential steps: chronic gastritis, multifocal atrophy, intestinal metaplasia and intraepithelial neoplasia, with accumulation of genetic alterations. H. pylori has a strong oncogenic role.^{( 16 )} With regard to gastric carcinogenic oncoviruses, Epstein–Barr virus is the most obvious candidate,^{( 17} , ^{18 )} but one investigation showed that JCV can also live and gradually evolve in the human stomach.^{( 19 )} Therefore we hypothesized that JCV could also be involved in gastric carcinogenesis as an oncovirus. In the present study, we tested this by looking for the presence of the JCV genome in gastric cancers, paired adjacent non‐cancerous mucosa and normal gastric epithelium, and calculating copy numbers.

Materials and Methods

Cases.  Twenty‐two fresh sample pairs of gastric cancer and adjacent non‐cancerous tissue (ANCT) were obtained from surgical material at the University of Toyama Hospital, with informed consent of the patients (17 men and five women with an average age of 71.3 years). Tissue samples were frozen in liquid nitrogen and stored at −80°C until use for DNA extraction. Residual specimens were fixed with 10% neutralized formaldehyde, embedded in paraffin, sectioned at 4 µm and stained with hematoxylin and eosin (HE) for confirmation of the histological diagnosis and definition of other histopathological characteristics. Ten normal gastric mucosa samples, five from men and five from women (average age 69.4 years), were obtained from endoscopic biopsies for the medical examination of health and were treated as for the biopsy material. All patients and healthy volunteers examined in the present study had no immunodeficiency diseases. The Ethics Committee of the hospital gave approval for genetic experiments restricted to JCV.

DNA extraction.  DNA from 100‐mg aliquots of frozen paired samples, cancer and ANCT was extracted using a standard method with proteinase K digestion and phenol–chloroform. Paraffin‐embedded tissues were also sectioned at 10 µm for microdissection of areas defined according to HE staining of serial sections. After deparaffinization, DNA was extracted with a QIAamp DNA mini kit. Before polymerase chain reaction (PCR) and real‐time PCR, DNA was reacted with 0.5 U topoisomerase I for 30 min at room temperature.

Nested PCR for the JCV genome.  Polymerase chain reaction amplification was carried out using three individual sets of primers for T antigen, agnoprotein and VP. For the JCV T antigen, PEP1 and PEP2 (nucleotides 4255–4272 of the Mad‐1 strain, 5′‐AGT CTT TAG GGT CTT CTA‐3′; and nucleotides 4408–4427, 5′‐GGT GCC AAC CTA TGG AAC AG‐3′, respectively), which amplify sequences in the NH₂‐terminal region of the JCV T antigen, were used for the first PCR, and PEP1 and PEP3 (nucleotides 4364–4345, 5′‐TGA AGA CCT GTT TTG CCA TG‐3′) were used for the second PCR (110 bp). For the VP capsid gene sequence, VP2 and VP3 (nucleotides 1828–1848, 5′‐TGT GCA CTC TAA TGG GCA AGC‐3′; and 2019–39, 5′‐CTA GGT ACG CCT TGT GCT CTG 5′, respectively) were used for the first PCR, followed by VP2 and VP4 (nucleotides 2004–1982, 5′‐GAT TGC ACT GTG GCA TTC TTT GG‐3′) for the second PCR (177 bp). Lastly, for JCV agnoprotein, AGNO1 and AGNO2 (nucleotides 280–298, 5′‐GTC TGC TCA GTC AAA CCA CTG‐3′; and 458–438, 5′‐GTT CTT CGC CAG CTG TCA C‐3′, respectively), which amplify a region within the coding region of JCV agnoprotein, were used for the first PCR and AGNO1 and AGNO3 (nucleotides 395–415, 5′‐GCA CAG GTG AAG ACA GTG TAG‐3′) were used for the second PCR (64 bp). The primer sequences for PEP1, PEP2, VP2, VP3, AGNO1 and AGNO2 were described in an earlier report.^{( 10 )} Twenty‐five‐microliter reaction mixtures contained 0.125 µL Takara Ex Taq HS (Takara, Tokyo, Japan) with 2.0 mM MgCl₂, 2.5 µL 10× PCR buffer, 2.5 µL dNTP mixture, 1 µM of each primer set (external primers) and 250 ng template DNA. The PCR conditions were: denaturation at 95°C for 10 min, followed by 30 cycles of denaturation at 95°C for 15 s, annealing for 30 s, and extension at 72°C for 30 s. The annealing steps were carried out at temperatures of 55°C for the T antigen primers, 57°C for the agnoprotein primers, and 54°C for the VP primers. As a termination step, the extension time of the last cycle was increased to 7 min. DNA extracted from JCI cells (JCV cultured neuroblastoma cell line, kindly provided by Associate Professor H. Sawa, Department of Neuropathology, Hokkaido University, Sapporo, Japan) served as a positive control. Samples amplified in the absence of template DNA were used as negative controls. Nested PCR was carried out as for the first PCR cycles, using 0.08% (volume) of the first PCR product with the internal primers in each case. These procedures were conducted in triplicate for confirmation of the results. Before investigating formalin‐fixed and paraffin‐embedded samples, the DNA was amplified using β‐globin primers (sense, 5′‐ACA CAA CTG TGT TCA CTA GC‐3′; and antisense, 5′‐GTC TCC TTA AAC CTG TCT TG‐3′) (175 bp) and by 30 cycles of denaturation at 95°C for 30 s, annealing at 55°C for 30 s, and extension at 72°C for 30 s to confirm the integrity of the DNA.

Southern blotting.  Southern blots were carried out after resolving 10 µL of each nested PCR product on 2% agarose gels stained with ethidium bromide. After gels were denatured, neutralized and Southern blot transferred onto nylon membranes (Hybond N1; Amersham, USA), hybridization was performed using 10 pmol/mL of digoxygenin‐labeled oligonucleotide probes (nucleotide 4303–4327 for T antigen, 5′‐GTT GGG ATC CTG TGT TTT CAT CAT C‐3′; nucleotide 1872–1891 for VP 5′‐AGC CAG TGC AGG GCA CCA GC‐3′ and nucleotide 395–415 for agnoprotein, 5′‐AAA GAC AGA GAC ACA GTG GTT‐3′) at 48°C for 3 h. After washing the membranes with 2× SSC and 0.1% sodium dodecyl sulfate (SDS), and 0.1× SSC and 0.1% SDS at the same temperature as for the hybridization, luminescence was detected with X‐ray film (Fuji RX‐U, Kanagawa, Japan) using a Dig luminescent detection kit for nucleic acids (Boehringer Mannheim, MA, USA).

DNA direct sequencing for PCR products.  The presence of JCV was further confirmed by direct sequencing of nested PCR products. Amplified DNA fragments initially identified by Southern blot hybridization were excised from preparatory agarose gels stained with ethidium bromide, and DNA was purified using a QIAEX II PCR purification kit according to the manufacturer's instructions (QIAGEN). After extraction, the DNA was sequenced using a BigDye Terminator v3.1 cycler sequencing kit and an ABI prism 3100 genetic analyzer (Applied Biosystems, Foster City, CA, USA).

Quantitation of JCV DNA by real‐time PCR.  A real‐time, fluorescence probe‐based PCR method was used for quantitative JCV of samples with the Mx3000P Real‐Time PCR system (Stratagene, La Jolla, CA, USA). For establishment of the quantitative technique for estimating the JCV copy number, a JCV‐containing plasmid (pBS‐JCVMad1, kindly provided by Associate Professor H. Sawa, Department of Neuropathology, Hokkaido University) was serially diluted and served as a standard reference. This standard and DNA of gastric cancer and normal gastric tissues were subjected to PCR amplification of the 188‐bp sequence of the T antigen gene using RT‐JCV1 (nucleotides 3492–3511, 5′‐GCC ACC CCA GCC ATA TAT TG‐3′) and RT‐JCV‐2 (nucleotides 3619–3595, 5′‐GTT GAC AGT ATC CAT ATG ACC AGA GAA‐3′) as the forward and reverse primers, respectively. Amplicon development was monitored using a double dye probe. RT‐JCV‐3 (nucleotides 3515–3539, 5′‐TAA AAC AGC ATT GCC ATG TGC CCC A‐3′) was labeled with FAM at the 5′ end and TAMRA at the 3′ end. The reactions used the TaqMan Universal PCR master mix (Applied Biosystems). Twenty‐five‐microliter reaction mixtures contained 12.5 µL TaqMan (×2) with 2.25 µL (10 µM) of each primer, 2.5 µL (2.5 µM) of double dye probe and 100 ng of template DNA. The protocol included the following parameters: an initial 10 min of incubation at 95°C for TaqMan DNA polymerase activation followed by 60 cycles of denaturation at 95°C for 30 s, annealing at 55°C for 1 min, and extension at 72°C for 30 s.

In situ PCR.  Sections (10 µm thick) were digested with proteinase K (20 µg/mL) for 15 m at 37°C. After rinsing with phosphate‐buffered saline (PBS), the tissue was fixed in 4% neutralized paraformadehyde and subjected to washing with 2× SSC. Then, 125 µL of PCR solution (1 µm primers of JCT‐1A [nucleotides 3009–3027, 5′‐AGGTAGGCCTTTGGTCTAA‐3′] and JCT‐1AS [nucleotides 3069–3050, 5′‐TGCCTAGAACTTTACAGG‐3′], 200 µM DIG‐11‐dUTP, 4.5 mM MgCl₂, PCR buffer and 2 IU Taq polymerase) were placed on the tissue under membrane sealing and PCR was carried out on the slide griddle of a programmable thermal controller: 94°C for 3 min, followed by 15 cycles of 92°C for 15 s, 55°C for 15 s and 72°C for 30 s, and finally 72°C for 5 min. These primers result in amplification of a 100‐bp fragment of JCV. The tissue was then washed with 2× SSC and incubated with blocking solution (100 µg/mL Salmon testis DNA, 100 µg/mL yeast tRNA and 5% bovine serum albumin in PBS) for 1 h. Finally, the specimens were incubated with an antidigoxigenin antibody coupled to alkaline phosphatase overnight, followed by fuchsin as a chromogen, and counterstained with methyl green. JCI cells were used as the positive control in this experiment.

Immunohistochemistry.  For immunohistochemistry, 4 µm‐thick sections of formalin‐fixed and paraffin‐embedded gastric cancers, ANCT, normal gastric mucosa and JCI cells were deparaffinized with xylene, dehydrated through an alcohol gradient and immersed in heated target retrieval buffered solution (TRS; DAKO, Denmark) with intermittent microwave irradiation for 15 min.^{( 20 )} Methanol solution with H₂O₂ was applied for 5 min to block endogenous peroxidase. The primary antibody used was a mouse monoclonal antisimian virus T antigen that crossreacts with JCV T antigen (1:100 dilution; clone Pab 101; Santa Cruz Biotechnology, Santa Cruz, CA, USA). Incubation was with the primary antibody overnight, followed by exposure to EnVision Labeled Polymer (DAKO) for 30 min. Staining was developed by reaction with diaminobenzidine chromogen, and counterstaining for 1 min with hematoxylin was then carried out.

Statistical analysis.  Statistical evaluation was carried out using Mann–Whitney tests to differentiate non‐parametric means. SPSS 10.0 software was used to analyze all data and P < 0.05 was considered to be statistically significant.

Results

Detection of the JCV genome.  In fresh frozen tissue, 19 out of 22 gastric cancer cases were positive for T antigen (86.3%), as were 19 of 22 ANCT cases (86.3%) and all cases of normal gastric mucosa after the triplicate experiments. The Southern blotting results are demonstrated in Fig. 1 and Table 1. In formalin‐fixed and paraffin‐embedded samples, 12 (54.5%) and 19 (86.3%) cases of gastric cancer and ANCT, respectively, demonstrated positivity for T antigen, as opposed to only 30% for normal gastric mucosa. With regard to VP in frozen tissues, 12 (54.5%) and 15 (68.1%) gastric cancers and ANCT, respectively, were positive and the normal gastric mucosa demonstrated a similar incidence (70%). In formalin‐fixed and paraffin‐embedded samples, similar values were found for gastric cancer and ANCT but the figure for normal gastric mucosa was lower (30%). As for agnoprotein, generally a low incidence was noted in this detection system. In total, all samples examined had some evidence of the JCV genome, either T antigen, VP or agnoprotein.

Figure 1.

Southern blot for T antigen, virus capsid protein (VP) and agnoprotein. Clear single bands are seen in all cases except case 2 for T antigen, cases 1, 5, 6 and 9 for VP and cases 1 and 3 for agnoprotein. NC, negative control; PC, positive control. The presence of amplifiable DNA was confirmed by the amplification of β‐globin.

Table 1.

Detection rates of JC virus genomes in gastric cancers, adjacent non‐cancerous tissue (ANCT) and normal gastric mucosa of fresh frozen and paraffin embedded tissues

Tissue	T antigen		VP		Agnoprotein
	n	%	n	%	n	%
Fresh frozen
Gastric cancer (22)	19	86.3	12	54.5	2	9.0
ANCT (22)	19	86.3	15	68.1	1	4.5
Normal gastric mucosa (10)	10	100	7	70	1	10
Paraffin‐embedded
Gastric cancer (22)	12	54.5	13	59.0	2	9.0
ANCT (22)	19	86.3	17	72.7	1	4.5
Normal gastric mucosa (10)	3	30	3	30	1	10

VP, virus capsid protein.

Direct DNA sequencing of PCR products.  The products of nested PCR were judged as authentic JCV (and not the BK or SV40 species of polyoma virus) genome by DNA sequencing. No differences in the DNA sequences were observed between the examined gastric cancers, ANCT and normal gastric tissues. Although there were several point mismatches, we judged these to be due to Taq polymerase error after repeated examination.

Quantitation of JCV DNA by real‐time PCR.  Real‐time PCR for T antigen in fresh samples was successful with all cases examined (Table 2). The viral DNA loads were 4784 ± 759 copies/µg DNA (mean ± SD; range 3539–6304) in gastric cancers, 5394 ± 1466 copies/µg DNA (range 3239–9006) in paired ANCT versus 542.4 ± 476.0 copies/µg DNA (range 28.8–1513.6) in normal tissue. There were significant differences between normal mucosa and gastric cancer or paired ANCT (P < 0.0001, Table 3). The value for control of JCI cells was 4.39 × 10⁶ copies/µg DNA. When fresh frozen samples were processed by formalin fixation for paraffin embedding, viral DNA load was reduced to one‐sixth in gastric cancers, 1/20 in ANCT and one‐fifth in normal gastric mucosa, the differences being significant (P < 0.05).

Table 2.

Results of the triplicate examination for T antigen and real‐time polymerase chain reaction for JC virus DNA load of fresh frozen and paraffin‐embedded tissues in individual cases

Case	Age (years)	Sex	WHOHist.	Depth	LN	UICC stage	T antigen		Fresh tissue		Fixed tissue
							Cancer	ANCT	Cancer	ANCT	Cancer	ANCT
1	77	M	tub	ss	+	?	+	+	3808	5672	1025.4	1407.8
2	78	M	tub	sei	+	?	–	+	3539	8121	1428.3	3.16
3	68	M	tub	si	+	?	+	+	3997	6130	595.7	1.16
4	63	M	tub	ss	+	?	+	+	4693	6557	1156.6	143.9
5	77	M	tub	se	+	?	+	+	5230	9006	1384.1	47.6
6	60	F	tub	ss	+	?	+	+	4866	5056	7.02	559.3
7	83	M	tub	sm	+	?B	+	+	4550	7189	921.1	432.9
8	65	M	tub	mp	+	?	+	–	5846	4945	103.6	ND
9	84	F	tub	ss	+	?B	+	+	4914	3982	309.7	1.83
10	78	M	muc	se	+	?	+	+	3681	4977	159.6	47.2
11	70	M	pap	m	–	IA	+	+	5467	6020	1535.8	ND
12	61	M	tub	ss	+	?	+	–	4329	5656	159.6	ND
13	74	M	tub	ss	+	?	–	+	4266	6431	ND	3.27
14	50	M	tub	si	+	IV	+	–	6304	3239	709.4	ND
15	54	F	tub	ss	+	?A	+	+	5151	5467	0.68	1469.4
16	78	M	tub	ss	–	?B	+	+	6146	4677	325.5	ND
17	78	M	tub	ss	–	?B	+	+	4645	3239	1249.8	81.2
18	80	F	tub	ss	–	?B	+	+	4313	4566	1545.2	26.9
19	67	F	tub	se	+	?A	+	+	4329	5277	1537.3	0.088
20	83	M	tub	ss	+	?A	+	+	5103	4614	1568.9	10.55
21	53	M	tub	se	+	?A	–	+	4582	4045	4.41	0.85
22	87	M	tub	se	+	IV	+	+	5498	3808	1313	2.2

Depth, depth of cancer invasion; Fixed tissue, formalin‐fixed and paraffin‐embedded tissue; Fresh tissue, fresh frozen tissue; LN, positive for lymph node‐metastasis; mp, muscularis propria; muc, mucinous adenocarcinoma; ND, not detected; pap, papillary adenocarcinoma; se, serosa exposure; sei, serosa exposure and invasive; si, serosa invasive; sm, submucosa; ss, subserosa; tub, tubular adenocarcinoma; UICC, International Union Against Cancer; WHO Hist., WHO histological classification.

Table 3.

JC virus DNA load (JC virus copies/1 µg DNA, (mean ± SD) of gastric cancers, adjacent non‐cancerous tissue (ANCT) and normal gastric mucosa of fresh frozen and paraffin‐embedded tissues

Tissue	Fresh tissue					Fixed tissue
Gastric cancer	4784 ± 759			]	**	811.5 ± 606.0	]	***
ANCT	5394 ± 1466	]	*			249.4 ± 475.4
Normal gastric mucosa	542.4 ± 476.0					114.6 ± 136.4

^* P < 0.0001,

^** P < 0.0001,

^*** P < 0.05, comparison with fresh frozen samples and corresponding paraffin‐embedded samples, P < 0.05.

In situ PCR.  In situ PCR demonstrated clear positivity in 20–30% of JCI cells (Fig. 2a). In some cases, positive cells were found sporadically in areas of gastric cancer (Fig. 2b) and ANCT. In ANCT, positivity was restricted to foveolar cells (Fig. 2c), and lymphocytes were totally negative. In normal gastric mucosa, no signals were found in any of the cases.

Figure 2.

(a) In situ polymerase chain reaction demonstrated clear positivity in 20–30% of JCI cells (JC virus cultured neuroblastoma cell line) (×400). (b) In some cases, positive cells were found sporadically in areas of gastric cancer (×400) and adjacent non‐cancerous tissue (ANCT). (c) In ANCT, positivity was restricted to foveolar cells (×1000).

Immunohistochemical staining for T antigen.  With regard to immunohistochemistry of T antigen, clot section of JCI cells infected by JCV was used as a positive control (Fig. 3a). We found positive nuclear reaction in one case of gastric cancer (1/22, 4.5%; Fig. 3b), and no nuclear detection of T antigen was noted in ANCT and normal gastric mucosa.

Figure 3.

Immunohistochemistry of T antigen. (a) T antigen‐positive JCI cells (JC virus cultured neuroblastoma cell line) infected by JC virus was used as a positive control, of which 20–30% cells were positive. (b) T antigen‐positive cells in a well‐differentiated adenocarcinoma case.

Discussion

In the present study, T antigen was detected at an incidence of 86.4% in both cancer and ANCT specimens and in 100% of normal gastric mucosa samples when fresh tissues were used, but relatively low frequencies in formalin‐fixed and paraffin‐embedded samples in line with a previous report.^{( 19 )} Kato et al. reported that in formalin‐fixed, paraffin‐embedded samples, amplification of the JCV regulatory region of tonsillar tissues by nested PCR was observed in eight of 10 trials, with 25 genome equivalents of JCV DNA per 1–2 µg DNA, but only 1 of 10 trials had 2.5 genome equivalents.^{( 21 )} Although targeted JCV regions and procedures differed between our experiments, it seems likely that copies are not normally abundant in gastric mucosa. The detection rates for agnoprotein and VP were lower than that of T antigen in the present study and also in an earlier report.^{( 10 )} Some medulloblastoma samples positive for agnoprotein were found to lack the T antigen,^{( 13 )} but this may be explained by JCV integration at random into the chromosomal DNA of cells on non‐permissive infection.^{( 22 )} In laboratory animal studies, JCV‐related tumor tissue exhibited the T antigen protein in the nuclei of tumor cells, but attempts to detect VP protein were unsuccessful.^{( 22 )} Detection rates for T antigen and VP in the present study were severely reduced when we used formalin‐fixed and paraffin‐embedded tissue rather than fresh tissue. For estimating the quality of DNA extracted from paraffin‐embedded blocks it is essential to evaluate PCR products. Inoue et al. found 100% of samples to be eligible after fixation in non‐buffered 10% formalin for 1 day, but only 44% after 2–3 days and 14% after 4–6 days, using PCR amplification with primers producing a 190‐bp targeted DNA segment of the p53 exon.^{( 23 )} Here we used samples fixed in formalin for 3 days and could amplify the β‐globin house‐keeping gene (175 bp) in all samples. The cause of this reduction in T antigen and VP detection rates after fixation is unclear but, as noted previously, JCV DNA research should only be conducted with fresh frozen specimens or after brief formalin fixation when at all possible.^{( 24 )}

The results of real‐time PCR for determining JCV DNA load revealed almost equal levels in gastric cancer, which was 10‐fold higher than that in normal gastric mucosa. It was earlier reported that colon cancers and adenomas demonstrated a higher JCV DNA load calculated by real‐time PCR than background tissue or normal colonic mucosa.^{( 25 )} From the data available, the JCV copy number appears to be two‐ to four‐fold higher in colon than gastric cancers, but almost equal in background ANCT in the two organs.

In general, standard DNA in situ hybridization (ISH) requires approximately 1000 viral copies per cell for detection.^{( 26 )} We could not find any signals by ISH, except in the positive control (data not shown), which might be due to a low copy number. Whereas Zanbrano et al. reported that ISH revealed the sporadic presence of JCV in normal prostatic glandular epithelium using a JCV large T antigen probe, data on actual infection levels were not available.^{( 27 )} In contrast, in situ PCR can detect even a few viral copies per cell,^{( 28 )} and we could find sporadic nests of positive cells in cancers and surrounding areas but not in normal mucosa, in line with the general PCR results.

With regard to immunohistochemical T antigen expression, we observed its expression only in one case (1/22, 4.5%), which differed from the report by Shin et al. (9/23, 39%).^{( 29 )} In our experiments, it was hard to prove the clear positivity of T antigen, but the antibody used in the present study carried limited success. It is speculated that the half life of T antigen might be very short and it is not easy to show evidence of T antigen positivity when formalin‐fixed and paraffin‐embedded tissues are used.

JC virus T antigen shares more than 70% homology with SV40 T antigen, particularly in the N‐terminus. Based on SV40 studies, the N‐terminal domain serves multiple functions, including regulation of DNA replication, protein stability and cell immortalization.^{( 30 )} The protein region encoded by the large T/small t common exon plays a critical role in controlling the cell cycle by interacting with key cellular proteins such as the pRb and Rb family members p107 and p130.^{( 31 )} Various point mutations, missense mutations, deletions and insertions have been detected in this region (Mad 1; 4521–4949, 429 bp, encoding small t antigen and N‐terminus of the translation product) of JCV in gastric cancers.^{( 19 )} However, we could not find any consistent mutation in the T antigen region examined (Mad 1; 4255–4364, 110 bp). Further studies are needed for confirmation.

Earlier studies have revealed the presence of JCV DNA sequences in the respiratory tract, as well as the upper and lower human gastrointestinal tract.^{( 1 )} Previous work has demonstrated the presence of replicating JCV DNA in B lymphocytes from peripheral blood, tonsil and spleen, and it has been hypothesized that lymphocytes may be one site of JCV persistence.^{( 32 )} Detection of viral gene products in renal tubules and excretion of JC virions in the urine suggests JCV persistence in the kidney.^{( 33 )} In the present study, however, JCV DNA contamination from JCV persistent B lymphocytes in the gastric mucosa, which contain abundant B lymphocytes, could not be ruled out. However, it was suggested that the majority of JCV‐infected cells were tonsil stromal cells, not B lymphocytes, on the basis of our results of in situ PCR using tonsil tissues (data not shown), in line with the report by Monaco et al.^{( 32 )}

According to the proposed genetic model for gastric carcinogenesis, accumulation of a series of genetic variations results in the activation of oncoproteins and inactivation of tumor suppressor genes.^{( 34 )} Although the significance of the presence of JCV DNA for the genesis of human epithelial malignant tumors remains unclear, the detection of viral proteins, including the T antigen, indicates a potential involvement in pathways leading to the development or progression of cancer.^{( 35 )} Transcriptional activation of the JCV early promoter is initiated by interleukin‐1α,^{( 36 )} for which there is evidence of a contribution to gastric cancer.^{( 37 )} If JCV presence was a cause of cancer, one might expect to find the viral genome in cancer tissue but not in normal tissue. In fresh samples of the present study, JCV copy number was calculated to be approximately 5 copies/100 cells in both gastric cancer and ANCT on the basis of the results of real‐time PCR and our DNA extraction from gastric epithelial cells (∼100 000 cells/µg DNA, data not shown). However, the ‘hit and run’ hypothesis claims that viruses can mediate cellular transformation through an initial ‘hit’, maintenance of the transformed state then being compatible with the loss or ‘run’ of viral molecules.^{( 38 )} Such a ‘hit and run’ or multistep transformation strategy has been suggested for some JCV‐induced cancers, such as colorectal carcinomas and neural‐origin tumors.^{( 2} , ^{3 )} If this is the case, one virus genome inserted into the human genome in cancer tissue would not be essential for viral carcinogenesis. The equivalent presence of JCV in ANCT may be implicated by the presence of high JCV load, which may itself be a risk factor for gastric carcinogenesis.

It is concluded that the gastric epithelium in most Japanese people is infected with JCV at a low level but higher levels of infection may be associated with gastric cancer development. Further investigations to explore this possibility appear warranted.