Infectious Causes of Cancer
Human retinoblastoma is not caused by known pRb-inactivating human DNA tumor viruses
Article first published online: 4 JAN 2007
Copyright © 2006 Wiley-Liss, Inc.
International Journal of Cancer
Volume 120, Issue 7, pages 1482–1490, 1 April 2007
How to Cite
Gillison, M. L., Chen, R., Goshu, E., Rushlow, D., Chen, N., Banister, C., Creek, K. E. and Gallie, B. L. (2007), Human retinoblastoma is not caused by known pRb-inactivating human DNA tumor viruses. Int. J. Cancer, 120: 1482–1490. doi: 10.1002/ijc.22516
- Issue published online: 30 JAN 2007
- Article first published online: 4 JAN 2007
- Manuscript Accepted: 20 SEP 2006
- Manuscript Received: 5 JUL 2006
- Damon Runyon Cancer Research Foundation
- National Cancer Institute of Canada
- Canadian Genetic Diseases Network
- Canadian Institutes for Health Research
- Keene Retinoblastoma Perennial Plant Sale
- Royal Arch Masons of Canada
- Canadian Retinoblastoma Society
- human papillomavirus;
Retinoblastomas occur as the consequence of inactivation of the tumor suppressor retinoblastoma protein (pRb), classically upon biallelic inactivation of the RB1 gene locus. Recently, human papillomavirus (HPV) genomic DNA has been detected in retinoblastomas. To investigate the possibility that oncoproteins encoded by pRb-inactivating DNA tumor viruses play a role in the pathogenesis of human retinoblastoma, 40 fresh-frozen tumors were analyzed for the presence of HPV, adenovirus (HAdV) and polyomavirus (BKV, JCV and SV40) genomic DNA sequences by real-time polymerase chain reaction (PCR). Tumors were screened for genetic and epigenetic alterations in all 27 exons of the RB1 gene locus and promoter by exonic copy number detection, sequencing and methylation-specific PCR of the promoter region. Retinoblastoma tumors from children with bilateral familial (n = 1), bilateral nonfamilial (n = 1) and unilateral nonfamilial (n = 38) disease were analyzed. Inactivating modifications to the RB1 gene locus were identified on both the alleles in 27 tumors, one allele in 8, and neither allele in 5 cases. A median of over 107,000 tumor cells were analyzed for viral genomic DNA in each PCR reaction. All tumor samples were negative for 37 HPV types, 51 HAdV types, BKV and JCV genomic sequences. Very low copy number (0.2–260 copies per 100,000 tumor cells) SV40 genomic DNA detected in 8 of 39 samples was demonstrated to be consistent with an artifact of plasmid-derived SV40. In contrast to recent reports, we obtained substantial quantitative evidence indicating that neither HPV nor any other pRb-inactivating human DNA tumor viruses play a role in the development of retinoblastoma, regardless of RB1 genotype. © 2006 Wiley-Liss, Inc.
Retinoblastoma is a rare childhood cancer of the retina, classically initiated by loss or mutation of both alleles of the retinoblastoma gene (RB1) during retinal development. Children with an inherited or de novo germline mutation in the RB1 gene develop bilateral or unilateral retinoblastoma with high penetrance, accounting for ∼40% of cases. In the majority of remaining cases, unilateral tumors occur as a consequence of somatic mutations that inactivate both RB1 alleles. Upon detailed analysis, mutations of the RB1 promoter or coding region can be identified in 92% of suspected RB1 alleles either as heterozygous germline mutations in heritable cases or as biallelic mutations in retinoblastoma tumors.1 However, no RB1 inactivating event has been identified in ∼8% of RB1 alleles, leaving open the possibility that some retinoblastomas may be caused by alternative genetic or epigenetic events that modify retinoblastoma protein (pRB) function.
The RB1 gene on chromosome 13q14 encodes a nuclear phosphoprotein (the tumor suppressor pRb) that plays a key regulatory role in the cell cycle check point between G1 and entry into S-phase.2 In addition to mutations in the RB1 gene, pRb function can be abrogated by oncoproteins from several DNA tumor viruses that infect humans.3 HAdV E1A, the E7 protein of high-risk human papillomaviruses (HPV), and the large T-antigens from BKV, JCV and SV40 have all been demonstrated to bind to pRb through a homologous pRb binding domain4, 5, 6 and inactivate pRb by various mechanisms.3 Retinoblastomas have been observed in transgenic mice that express HPV16 E77 or SV40 T-antigen,2 in hamsters inoculated intraocularly with JCV and in primates infected with HAdV 12.8 Furthermore, human embryonic retinoblasts can be transformed by HAdV E1A and SV40.9 pRb inactivation by viral oncoproteins encoded by each of these viruses is essential for viral-mediated transformation of an infected cell.10, 11, 12
In several recent provocative studies, genomic DNA of high-risk HPV types was detected in 28–40% of retinoblastomas in patients from Mexico and South America.13, 14, 15 These data suggest that pRb inactivation by viral oncoproteins could play a role in the pathogenesis of sporadic, nonfamilial retinoblastoma.
We chose to investigate the hypothesis that human DNA viruses with viral oncogene products that functionally inactivate pRb, including HPV (E7), HAdV (E1A) and the polyomaviruses (JC and BK viruses and SV40 via large T-antigens), may play a role in the pathogenesis of human retinoblastoma in which both, one or no RB1 mutant alleles have been identified after detailed molecular analysis.
Material and methods
Research subjects were selected from among probands with a diagnosis of retinoblastoma from North America who consented to the use of tumor samples for retinoblastoma research. Cases were specifically chosen from among available retinoblastoma tumors by one of the investigators (B.G.) to purposefully enrich the study with tumors from patients with unilateral, nonfamilial retinoblastomas in which one or no mutations in the Rb1 gene or promoter were identified after detailed analysis. Pathological diagnosis was confirmed for all tumors.
Preparation of tumor DNA
Total genomic DNA was purified from fresh or fresh-frozen tumor samples by use of a Puregene kit for DNA purification (Gentra Systems, Minneapolis, MN). All preamplification procedures including DNA extraction were performed in an ultraviolet light-irradiated, laminar flow hood, and aliquots were made in a molecular biology laboratory. Purified retinoblastoma genomic DNA was shipped to the virology laboratory for further analysis, where stringent PCR contamination precautions were used for all experiments. Preamplification procedures were performed in a laboratory separate from the post-amplification laboratory. All SV40 control preparation and analysis was performed in a third laboratory. Aliquots of purified retinoblastoma DNA (100 ng/μl) were prepared on a single occasion from stock samples.
Screening for RB1 mutations
Retinoblastomas were screened for mutations in the RB1 gene or promoter, as well as RB1 promoter methylation, as previously described.1 Briefly, alterations in size or copy number of the RB1 exons were detected by quantitative, multiplex PCR amplification of all 27 exons of the gene by use of intronic primers designed to include splice sites (sequences available upon request). All 27 exons and the promoter region of RB1 were sequenced by the use of the Cy5/Cy5.5 dye primer cycle sequencing kit (Bayer/Visible Genetics, Walpole, MA) in 14 duplex sequencing reactions, encompassing recognized splice sites. In those cases where both RB1 mutant alleles were not identified, sodium bisulfite conversion and methylation-specific PCR was performed to detect methylation of the RB1 promoter.16 After the RB1 mutations were identified in tumors, peripheral blood lymphocytes were examined for the specific mutations identified in each case to evaluate the presence of a germline mutation. The virology lab was masked to the RB1 mutation data until all viral analysis was complete.
All retinoblastoma samples (2.5 μl, ∼250 ng) were analyzed for genomic sequences of 18 high-risk and 19 low-risk HPV types by multiplex PCR targeted to the conserved L1 region of the viral genome by use of PGMY09/11 L1 primer pools.17 HPV type specification was performed by hybridization to a probe array containing 37 HPV types and β-globin (Roche Molecular Systems, Alameda, CA).18 Positive controls, consisting of 10 and 100 HPV16 (SiHa, with ∼2 HPV copies per genome) or HPV18 (C4-2, with ∼1 HPV copy per genome)-positive cells diluted in a background of HPV-negative cells (K562), were included in each experiment. A negative control (K562 cells which lack HPV) was included in each row (1 in 12) of the PCR plate. Samples that were positive for β-globin were considered adequate for analysis. Samples were scored as negative or positive for HPV DNA.
Real-time TaqMan PCR amplification of viral genomic DNA and ERV-3
The sequence of the primers and probes, reagents used for standard curves for all assays and the optimized reaction conditions are provided in the Appendix Tables I and II. The primers and probes used for real-time TaqMan PCR amplification of viral target sequences were synthesized, labeled with 5′ FAM and 3′ Black Hole Quencher 1 and HPLC purified by Integrated DNA technologies (Granville, IA). Amplification conditions were optimized per the TaqMan Universal PCR master mix protocol (Applied Biosystems) or as previously described. Reactions were performed in a 96-well plate format in an Applied Biosystem 5700 or 7300 real-time PCR system (Foster City, CA). For all reactions, amplification efficiency was >90% and the correlation coefficient for the CT versus log concentration of the duplicate standard curve was ≥0.99. Unknown sample quantities were derived by interpolating threshold cycle (CT) values from a standard curve created by a duplicate 5- or 10-fold serial dilution series of control plasmids containing target sequences (Table II) in a background of human placental DNA (50 ng/μl), with the exception of the ERV-3 assay (see below). The threshold cycle (CT) was set at the midpoint of the linear range of amplification. Samples with a viral load ≥1 copy were considered positive. Water blanks were included in each row (1 in 10 samples) of the plate as negative controls for each assay. Two microliters (∼200 ng) of purified retinoblastoma DNA were added to each reaction. Because of limited quantities of purified retinoblastoma DNA, samples were not run in duplicate.
|Date of birth|
|Family history of retinoblastoma|
|Rb1 alteration identified in tumors|
|Germline mutation in PBL|
|No RB1 mutation in tumor||5|
|RB1 genetic alterations identified|
|Patient ID||Laterality||Family history||Mutation 1||Mutation 2||Germline|
|1||uni||no||no mutation found||no mutation found||ND|
|2||uni||no||no mutation found||no mutation found||ND|
|3||uni||no||no mutation found||no mutation found||ND|
|4||bi||no||no mutation found||no mutation found||ND|
|5||uni||no||no mutation found||no mutation found||ND|
|6||uni||no||delP->27||no mutation found||Normal|
|7||uni||no||delP->16||no mutation found||Normal|
|8||uni||no||delP->2||no mutation found||NA|
|9||uni||no||delP->27||no mutation found||Normal|
|10||uni||no||del2->20||no mutation found||Normal|
|11||uni||no||delP->6||no mutation found||Normal|
|12||uni||no||delP->1||no mutation found||Normal|
|13||uni||no||promoter methylation||no mutation found||Normal|
|14||uni||no||promoter methylation||promoter methylation||NA|
|15||uni||no||promoter methylation||promoter methylation||Normal|
|16||uni||no||promoter methylation||promoter methylation||Normal|
|17||uni||no||promoter methylation||promoter methylation||Normal|
|18||uni||no||promoter methylation||promoter methylation||Normal|
|19||uni||no||promoter methylation||promoter methylation||Normal|
|35||uni||no||g.162203 G->A||g. 162203 G->A||Normal|
|39||uni||no||c.1450delAT in exon 16||c.1450delAT in exon 16||Normal|
An estimate of the number of tumor cells analyzed in each reaction was made by TaqMan real-time PCR, targeting a single copy human gene on chromosome 7, human endogenous retrovirus 3 (ERV-3).19 Two microliters of stock purified genomic DNA from retinoblastomas were analyzed. A standard curve was generated from 10-fold dilutions (from 50,000 to 0.5) of a diploid human cell line, CCD-18LU (ATCC CCL-205, Manassas, VA). Results were reported as number of human diploid genome equivalents per microliter of purified genomic DNA from tumor samples.
HPV16 E6 and 18 E7 genomic sequences were identified in tumors by a validated real-time TaqMan PCR method.20
Genomic sequences of the hexon capsid gene of HAdV were assayed in tumors by use of a quantitative real-time PCR capable of detecting all 51 human adenovirus prototypes21 and for the E1A gene of Adenovirus subgroup A IV (serotypes 12, 18 and 31). pCMV-Ad12E1A was used as positive control for HAdV-12. Prototype virus stocks were obtained for HAdV-18 and 31 from the American Type Culture Collection (ATCC, Manassas, VA). The E1A genes for HAdV-18 E1A (bp 5-896 Genebank AY490822) and 31 E1A (bp 5-858 Genebank AY490825) were amplified and TopoTA-subcloned into pCRII plasmid (Invitrogen, Carlsbad, CA) for use as positive controls. Primers and probe were designed to amplify a target within the conserved CR1 Rb-binding domain of E1A.
Tumors were tested for human BK viral genomic DNA22 and for JC virus genomic DNA23 by real-time TaqMan PCR. Tumors were tested for SV40 large T-antigen exon by use of primers originally designed by Carbone et al.24 and an internal TaqMan probe. Additionally, SV40 large T-antigen intron genomic sequences were also assayed,25 using an internal TaqMan probe.
PCR for plasmid-specific SV40 sequences
All samples were subjected to PCR amplification by use of primers that bracketed an engineered junction of SV40 present in expression plasmids.25 This primer pair (SV2741-SV4279) amplifies a 210-bp product in plasmids, but a 1539-bp product from intact SV40 genome.25 pGL2 plasmid from Promega (Madison, WI) and tissue culture supernatant from SV40-inoculated BS-C-1 cells (ATCC 26) were used as positive controls. Two microliters of purified genomic DNA from retinoblastoma tumors were added to a 50-μl reaction containing 0.5 μM of each primer, 0.2 mM each dNTP, 3 mM MgCl, 1.25 units of TaqGold polymerase. PCR conditions were as follows: TaqGold activation at 95°C for 10 min, followed by 45 cycles at 94°C for 45 s, 60°C for 45 s, 72°C for 2 min, followed by a final extension of 72°C for 10 min. Samples were separated by agarose gel electrophoresis and bands of the appropriate size were scored as positives.
The characteristics of the study population and a summary of the RB1 mutation analysis are present in Table I. The majority of samples (38 of 40) included in the study were from probands with a diagnosis of unilateral, nonfamilial retinoblastoma. One of the patients with bilateral retinoblastoma had a positive family history.
After detailed screening for RB1 mutations, biallelic inactivating mutations or promoter methylation were identified in 27 tumors, only one mutant allele was identified in 8 tumors, and no mutations or methylation were detected in 5 tumors. Identified pRb inactivating mutations included small (n = 5) and large (11) deletions, nonsense mutations (20), insertions (4), and splice (7) mutations. Of the 18 cases in which biallelic mutations were identified, the same defined mutation for both first and second events was identified in 13 cases, suggesting somatic recombination as the mechanism for the second event. In 6 cases, the second event was a new mutation. RB1 promoter hypermethylation was detected as the inactivating event for 15 alleles. The absence of RB1 inactivating mutations or methylation in 18 (22.5%) of 80 alleles, as compared to an expected proportion of ∼8%, was the result of purposeful enrichment of tumors without such mutations in this study. We hypothesized that such tumors might be caused by viral protein inactivation of pRB.
Peripheral blood lymphocytes were available for 33 of the 35 cases with at least one RB1 mutation identified in the tumor. As expected, a germline mutation (c.1710insTT) was identified in the child with bilateral, familial retinoblastoma. One child with unilateral, isolated retinoblastoma was found to be mosaic (c.1330C->T), i.e., to have a mixture of wild-type and mutant genotypes in blood cells. Further details of the RB1 mutations identified are provided in Table II. The RB1 mutations detected in 19 of 40 tumors have been previously reported.1
To evaluate the quality of the extracted tumor DNA from fresh and fresh-frozen tumors, all samples were analyzed for the presence of a single copy human gene on chromosome 7, human endogenous retrovirus 3 (ERV-3), by quantitative TaqMan PCR. ERV-3 was successfully amplified from all 40 samples. By use of this assay, the median number of tumor cells analyzed for HPV genomic DNA was estimated to be 134,863 (Interquartile Range [IQR] 93,937–171,468) and for all other PCR assays was 107,890 (IQR 75,150–137,174). Purified genomic DNA from retinoblastoma for three of 40 samples was limited in quantity and, therefore, the number of samples analyzed ranged from 37–40.
Retinoblastoma tumors were analyzed for all 18 oncogenic HPV types epidemiologically associated with cervical carcinogenesis by use of multiplex PCR targeted to the conserved L1 capsid protein sequence followed by line blot hybridization. In stark contrast to previous reports, all retinoblastoma tumors from this North American population were negative for high-risk HPV genomic sequences (Table III). All samples were also negative for 19 low-risk HPV types (n = 39). Because previous studies had reported the presence of predominantly types HPV16 and 18, all tumors were tested for HPV16 E6 or 18 E7 by type-specific, quantitative TaqMan PCR (n= 38). All samples were negative by this analysis, thereby excluding the possibility that the negative results by consensus PCR were due to deletion of the L1 region during viral integration.
|Negative||Positive||Viral genomes/100,000 tumor cells Median (range)|
|L1 consensus PCR for 38 HPV types||39||0||NA|
|HPV16 E6 type-specific TaqMan PCR||38||0||NA|
|HPV18 E7 type-specific TaqMan PCR||38||0||NA|
|JC Virus large T antigen TaqMan PCR||39||0||NA|
|BK Virus VP1 gene TaqMan PCR||40||0||NA|
|Large T antigen interon TaqMan PCR||39||0||NA|
|Large T antigen exon TaqMan PCR||31||8||10.4 (0.20–260)|
|Hexon consensus PCR for 51 HAdV types||40||0||NA|
|HAdV12 E1A TaqMan PCR||37||0||NA|
|HAdV18 E1A TaqMan PCR||37||0||NA|
|HAdV31 E1A TaqMan PCR||37||0||NA|
Retinoblastomas were analyzed for the presence of HAdV and polyomavirus genomic DNA. It was considered more plausible that these pRb-inactivating human DNA tumor viruses could affect retinal development, because these viruses hematogenously disseminate and/or have been demonstrated to infect the retina. DNA sequences for the hexon gene of 51 HAdV types were not detected in retinoblastoma samples by a validated, consensus TaqMan PCR method (n = 40). To address the possibility that the HAdV E1A may be preferentially retained and the hexon gene deleted, tumors were also analyzed for the presence of the E1A gene of adenovirus subgroup A IV. This group (serotypes 12, 18 and 31) was chosen because these viruses have demonstrated high oncogenic potential in animal models.26 All samples were negative for HAdV-12, 18 and 31 E1A by type-specific TaqMan quantitative PCR assays designed to amplify a target within the conserved CR1 Rb-binding domain of E1A (n = 37).
None of the retinoblastoma tumors were positive for human BKV (n = 40) or JCV (n = 39) sequences (Table III). During optimization of the viral-specific TaqMan PCR, the BKV, JCV and SV40 primers and probes were demonstrated to have no cross-reactivity (data not shown). When tumors were evaluated for genomic sequences of the SV40 large T-antigen exon by use of the commonly utilized SV.for3 and SV.rev primer pair, 8 (∼21%) of 39 samples were positive. However, copy numbers of the virus were very low, ranging from 0.2 to 260 viral genomes per 100,000 tumor cells. This finding was confirmed by a second amplification of stored stock samples. No significant associations between the presence of SV40 genomic sequences and RB1 mutation status or promoter methylation status were observed.
The SV40 large T-antigen is present in many common laboratory plasmids and, therefore, the presence of SV40 sequences in tumors could be due to contamination of samples during laboratory processing. To determine the origin of SV40 exon sequences detected in retinoblastoma samples, all samples were tested for the presence of genomic sequences encoding the SV40 large T-antigen intron not commonly found in engineered plasmids. All tumors were negative for SV40 when evaluated by the use of the SVINTfor and SVINTrev primer pair targeted to a segment of the large T-antigen intron just 213 base pairs downstream from the SV.for3-SV.rev target site in native SV40. Given that the SV40 large T-antigen intron is present in native SV40, but deleted in the majority of SV40 containing plasmids, this finding is consistent with contamination of samples by an SV40 containing plasmid.25 Thirty-seven samples were further analyzed for differences in SV40 PCR amplicon length by use of the SV2741-SV4279 primer pair that flanks an engineered junction of SV40 present in expression plasmids.25 A 210-bp fragment, consistent with plasmid contamination of the sample, was present in one of the eight samples that were positive for SV40 large T-antigen intron (Fig. 1). The 1539-bp fragment consistent with intact SV40 genome was not amplified from any of the samples.
Negative controls included in all of the viral detection assays were negative for viral nucleic acid.
We obtained substantial evidence that oncogenic HPV and other pRb-inactivating DNA-tumor viruses do not play an appreciable role in the etiology of retinoblastoma regardless of RB1 genotype (Table III), in contrast to several recent provocative reports.13, 14, 15, 27 The DNA sequences that encode the viral oncogenes mediating transformation by HPV, HAdV, BKV and JCV were not detected in DNA purified from fresh-frozen retinoblastoma samples. For each of these viruses, expression of the pRb inactivating oncoproteins is necessary for viral-mediated transformation of the infected cell. A clonal association between virus and tumor cell is usually established via viral integration, and continued expression of viral oncogenes is necessary for maintenance of the malignant phenotype.28, 29, 30 The absence of viral sequences is therefore strong evidence against a role for these viruses in the etiology of retinoblastoma.
These findings therefore contradict previous reports in which HPV genomic DNA was detected in a subset of bilateral (presumed germline RB1 mutation) and unilateral nonfamilial (presumed nonhereditary) retinoblastoma tumors in Mexico and South America.13–15,, 27 Both Palazzi et al.13 and Orjuela et al.14 analyzed and detected only high-risk HPV genomic DNA (types 16, 18 and 35) in 12 (28%) of 43 and 14 (36%) of 39 tumors, respectively. Montoya-Fuentes et al.15 detected predominantly low-risk HPV6 and 11 in 42 (82%) and high-risk HPV types in 18 (35%) of 51 retinoblastomas. However, the E7 protein of low-risk HPV types is not capable of inactivating pRb function and, therefore, would not be expected to promote development of retinoblastomas.31 All authors described use of controls to exclude the possibility of specimen contamination in the laboratory. However, viral analysis was restricted to qualitative HPV detection by PCR. Data in support of high viral copy number, viral transcription or viral specificity to the tumor cell required to support an etiologic association was not reported.32 The source and etiologic significance of the detected HPV DNA in these prior studies therefore remains unclear. Possibilities include contamination of the archived paraffin block during processing in the clinical pathology laboratory. Alternatively, unrelated conjunctival HPV infection may explain these results. Peripartum vertical transmission of HPV infection to infants has been reported.33 However, although the incidence of cervical cancer is higher in South than in North America because of differences in screening practices, HPV infection prevalence is quite similar. We consider it less likely that the behavior of these viruses would differ by geography; however, we cannot entirely exclude the possibility of a critical cofactor (e.g. diet, ethnicity) that would substantially modify the behavior of these viruses. However, as an example, both endemic and sporadic cases of nasopharyngeal carcinoma are associated with Epstein-Barr virus despite the increased risk associated with ethnicity and cofactors such as diet. Therefore, geographic variation in etiology in North versus South American populations is therefore a less plausible explanation for these discrepant findings.
In addition to the absence of HPV genomic DNA, a role for HPV in retinoblastoma development would be contrary to what is currently known about the biology of HPV infection. HPV is trophic for the squamous epithelia of the skin, airway and anogenital tracts. There is no evidence that HPV can infect the retina or other neural tissue. HPV E7 did induce retinal tumors in transgenic mice in which the E7 gene is under the control of a developmentally regulated retinal promoter. Therefore, the HPV genome would need to be transcribed at critical time points during retinal development to promote retinoblastoma and would likely need to be acquired in utero. HPV is not a systemic infection and therefore vertical transmission via hematogenous dissemination is not likely. Although HPV genomic DNA has been detected in washed sperm,34 semen35 and amniotic fluid,36 this could be accounted for by contamination by anogenital epithelium.37 Recent reports of HPV replication within trophoblast cells consistent with dissemination via the cervical os remain unconfirmed.37, 38
In contrast to previous studies, we performed a detailed sequence analysis of the RB1 genotype in all tumors analyzed (Table II). Orjuela et al.14 reported no significant associations between the presence of HPV DNA and expression of pRb by immunohistochemistry. We found no evidence for HPV in retinoblastoma tumors, with or without biallelic inactivating RB1 mutations. This was despite the fact that the cases were specifically chosen to enrich the number of tumors in the study population in which no or only one RB1 mutation was detected. Mutations in other pRb family proteins (p107, p130) were not investigated in this study nor have they been implicated in human retinoblastoma in the literature. These proteins do not appear to compensate for loss of pRb function in the developing human retina as has been observed in murine models.39
We expanded our investigation to include an analysis of the possible role of pRb-inactivating human DNA tumor viruses other than HPV because retinal development could more plausibly be altered by these viruses based on their epidemiology and molecular biology. In contrast to HPV, the polyomaviruses BKV and JCV are hematogenously disseminated and can infect neural tissue and the retina. Furthermore, both JCV and BKV infections are highly prevalent in adults, establish latency in the central nervous system,40 may be reactivated in pregnancy41 and are possibly vertically transmitted.42 BKV and JCV DNA have been associated with central nervous system malignancies primarily in adults.43 However, the etiologic significance of these associations remains unclear.44 Although we are unaware of any prior association of BKV with retinoblastoma, BKV genomic sequences have been previously detected in an analogous solid tumor of neural origin in infants and young children, neuroblastoma.45 Recently, however, the presence of BKV and JCV-specific IgM antibodies indicative of newly acquired infection during pregnancy was not associated with risk of neuroblastoma in a case–control study of 115 index mothers with children diagnosed with neuroblastoma.46 Furthermore, BKV viremia was also not detected in the pregnant mothers by real-time PCR in this study, arguing against a role for BKV in this childhood tumor.46
Seroprevalence to HAdV, similar to that of human polyomavirus, is high. HAdV sequences detected in amniotic fluid and fetal tissues have been associated with fetal malformations.47 Although we used an assay capable of detecting 51 different HAdV types, false negative results could be attributed to loss of the hexon gene region during viral integration. Therefore, samples were also screened for the highly oncogenic HAdV-A 12, 18 and 31 by type-specific PCR targeted to the E1A oncogene, the amino terminal region of the genome that is almost invariably maintained. However, it remains possible that other adenoviruses of high (HAdV-D) and moderate (HAdV-B) oncogenic potential would not have been detected. Furthermore, in contrast to HPV, recent data suggest that HAdV may be able to act by a “hit and run” mechanism.48 The pRb inactivating oncoprotein of adenovirus E1A together with E1B or E4orf6/E4orf3 are necessary for the development of the transformed phenotype. Although most cell lines transformed by adenovirus maintain the presence of viral DNA through integration into the host cell genome, cells transformed by HAdV5 E1A and E4orf6 or E4orf3 can maintain a transformed phenotype after loss of adenoviral DNA in long-term culture, possibly because of cumulative viral-induced mutations and genetic instability.48 However, the absence of any viral sequences in samples argues against a role for adenovirus in retinoblastoma, given that adenovirus-negative cell lines arise from adenovirus-containing, transformed precursors.
Whether SV40 circulates in the human population as a result of contamination of the polio vaccine and contributes to the pathogenesis of human malignancies is a matter of considerable controversy.49, 50 Although we detected SV40 genomic DNA sequences in a small proportion of retinoblastomas using a single primer pair targeted to the SV40 large T-antigen exon, the low viral copy number, the uniform absence of the SV40 large T-antigen intron and the detection of plasmid-specific sequences in samples, argue that the sequences were derived from contaminating plasmid DNA. Our findings are consistent with those recently reported by Lopez-Rios et al.25 The region of the SV40 genome from base pairs 4100 to 4700 that we detected is present in a DNA fragment contained within numerous mammalian gene expression plasmids.51 The 8 retinoblastoma tumor specimens that were weakly positive for plasmid SV40 had been handled in molecular biology research labs where such plasmids are commonly used. A “hit and run” mechanism seems a less likely, alternative explanation for the very low viral copy numbers. This would require that all viral isolates in the tumors collected throughout North America would have undergone similar deletions of the SV40 T-antigen intron as well as the same deletion of a large fragment of the T-antigen exon, exactly as engineered in recombinant laboratory plasmids. The negative results cannot be ascribed to sequence variation at the primer site because the primers used to discriminate plasmid from native SV40 did not target the variable region of the SV40 genome. Furthermore, in animal models of SV40 induced neoplasia, SV40 T-antigen is expressed in all malignant cells.52In vitro, repression of SV40 T-antigen function results in cellular senescence and loss of the transformed phenotype,53 suggesting that a clonal-association with one or more copies of the SV40 T-antigen per tumor cell would be causally necessary. Previous studies that used quantitative PCR for detection of SV40 sequences in human tumors reported SV40 copy numbers substantially less than one viral copy per cell (1–100 copies per 0.5 μg of tumor DNA)54; 0.12 in ∼14,000 cells,55, 56 or have been negative.57, 58 Although it has been argued that low SV40 copy numbers in human tumors are still compatible with an etiologic role, such low copy numbers, in combination with the uniform absence of sequences found in native SV40, argue against a hit and run mechanism.
In summary, the etiology of the retinoblastoma tumors lacking both classical genetic or epigenetic alterations and pRb-inactivating tumor viruses remains unknown. Our data suggest that both somatic recombination and promoter methylation play an important role in the pathogenesis of sporadic retinoblastomas. There are several possible explanations for those retinoblastoma tumors that remain unexplained by our data, including genetic or epigenetic disruption of another component of the pRb pathway; an unidentified transacting factor such as a microRNA affecting RB1 expression; a disruption of the RB1 transcription that was not assayed in this study such as an intronic alteration causing premature termination or aberrant splicing; as yet unidentified pRb-inactivating tumor viruses or other unknown mechanism.
The authors thank Dr. A. Heim (Institut fur Virologie, Medizinische Hochschule Hannover, Hannover, Germany) for providing pGEM-T HAdV-2, Dr. E.O. Major (Laboratory of Molecular Medicine and Neuroscience, National Institute of Neurological Disorders and Stroke, National Institutes of Health, Bethesda, MD) for providing pBR322-BKV and pBR322-JCV, and Dr. R.P. Ricciardi (Department of Microbiology, School of Dental Medicine, University of Pennsylvania, Philadelphia) for kindly providing pCMV-Ad12E1A. The authors thank Dr. David E. Symer (National Cancer Institute, USA) for comments on the manuscript.
This work was supported in part by the Damon Runyon Cancer Research Foundation (M.L.G.); the National Cancer Institute of Canada with funds from the Terry Fox Run and the Canadian Cancer Society (B.G.); the Canadian Genetic Diseases Network and the Canadian Institutes for Health Research (B.G.); the Keene Retinoblastoma Perennial Plant Sale (B.G.); the Royal Arch Masons of Canada and the Canadian Retinoblastoma Society (B.G.).
- 26Adenoviridae: the viruses and their replication, 3rd edn. Philadelphia, PA: Lippincott-Raven, 1996..
|Target||Primer/probe sequence||GenBank Accession number||Base pairs||Fragment length||Reference|
|Human papillomavirus 16 E6|
|HPV 16-U||5′-AIGACTTTGCTTTTCGGGAT-3′||K02718||231–434||223||Gravitt et al.20|
|Human papillomavirus 18 E7|
|HPV 18-U||5′-ATGTCACGAGCAATTAAGC-3′||X05015||667–803||137||Gravitt et al.20|
|SV40 Large T antigen interon|
|SVINTfor||5′-AAGTAAGGTTCCTTCACAAAG-3′||NC001669||4690–4925||235||Lopez-Rios et al.25 Gillison et al.|
|SV40 large T antigen exon 2|
|SV.for3||5′-TGAGGCTACTGCTGACTCTCAACA-3′||NC001669||4372–4477||105||Carbone et al.24|
|BKV VP1 Gene|
|BKV VPf||5′-AGTGGATGGGCAGCCTATGTA-3′||V01109||670–764||94||Leung et al.22|
|JCV large T antigen|
|PEP-1||5′-AGTCTTTAGGGTCTTCTACC-3′||AB127351||4246–4407||151||Taoufik et al.23|
|Human Adenovirus hexon gene|
|HAdV AQ1||5′-GCCACGGTGGGGTTTCTAAACTT-3′||AY224392||18858–18989||133||Helm et al.21|
|Human adenovirus 12 E1A|
|HAdV12-for||5′-GAA CTG AAA TGA CTC CCT TGG-3′||X73487||507–659||152||Gillison et al.|
|HAdV12-rev||5′-CGG CAG ACT CCA CAT CAA GAT C-3′|
|HAdV 12 probe||5′-TAT TGG AGC ATT TGG TG-3′|
|Human adenovirus 18 E1A|
|HAdV18-for||5′-GGA CAG AAA TTA CTC CTT TGG-3′||AY490822||5–160||149||Gillison et al.|
|HAdV18-rev||5′-CAG TAA CGT CCA CAT CAA TAT C-3′|
|HAdV 18 probe||5′-ACA TAT TGG AGG ACT TGG TG-3′|
|Human adenovirus 31 E1A|
|HAdV31-for||5′-GAA CTG AAA TAA CTC CTT TGG-3′||AY490825||5–157||152||Gillison et al.|
|HAdV31-rev||5′-CGG CAG ACT CCA CAT CAA TAT C-3′|
|HAdV31 probe||5′-TAT TGG AGC ATT TGG TG-3′|
|Human endogenous retrovirus 3|
|PHP10-F||5′-CATGGGAAGCAAGGGAACTAATG-′3||M12140||1601–1736||135||Yuan et al.19|
|Virus||Standard control||PCR reaction (50 μl)||Thermal cycles|
|Restriction enzyme||Universal Master Mix||Each primer (μM)||Probe (μM)||Restiction digest||UNG digestion||Activation of AmpliTaq Gold||Cycles|
|HPV16||5-fold dilution series (from 105 copies) of pGEM containing full-length HPV 16 genome||1 unit of Bam H1||1X||0.2||0.1||37°C, 30 min||50°C, 2 min||95°C, 10 min||50|
|95°C, 15 s||55°C, 30 s|
|HPV18||5-fold dilution series (from 105 copies) of pGEM containing full-length HPV 18 genome||1 unit of EcoR1||1X||0.2||0.1||37°C, 30 min||50°C, 2 min||95°C, 10 min||50|
|95°C, 15 s||55°C, 30 s|
|HAdV hexon||5-fold dilution series (from 105 copies) of pGEM-T containing HAdV-2 hexon gene||1X||0.5||0.4||50°C, 2 min||95°C, 10 min||45|
|95°C, 3 s||55°C, 10 s|
|HAdV12||10-fold dilution series (from 107 copies) of pCRII containing HAdV12 E1A||1X||0.5||0.4||50°C, 2 min||95°C, 10 min||45|
|95°C, 15 s||53°C, 1 min|
|HAdV18||10-fold dilution series (from 107 copies) of pCRII containing HAdV18 E1A||1X||0.5||0.4||50°C, 2 min||95°C, 10 min||45|
|95°C, 15 s||52°C, 1 min|
|HAdV31||10-fold dilution series (from 107 copies) of pCRII containing HAdV31 E1A||1X||0.5||0.4||50°C, 2 min||95°C, 10 min||45|
|95°C, 15 s||53°C 1 min|
|BKV||5-fold dilution series (from 105 copies of pBR322 containing entire BKV genome||1X||0.2||0.05||50°C, 2 min||95°C, 10 min||50|
|95°C, 15 s||63°C, 1 min|
|JCV||5-fold dilution series (from 105 copies) of pBR322 plasmid DNA containing entire JCV genome||1X||0.4||0.05||50°C, 2 min||95°C, 10 min||50|
|95°C, 15 s||57°C, 1 min|
|SV40 large T-antigen exon||5-fold dilution series (from 105 copies) of pBR322 containing entire SV40 genome||1X||0.2||0.1||50°C, 2 min||95°C, 12 min||50|
|95°C, 15 s||55°C, 30 s|
|SV40 large T-antigen interon||5-fold dilution series (from 105 copies) of pBR322 containing entire SV40 genome||1X||0.5||0.4||50°C, 2 min||95°C, 10 min||45|
|95°C, 1 min||60°C, 45 s|