- Email: [email protected]
JC virus and human colon carcinoma: An intriguing and inconclusive association
GASTROENTEROLOGY 2003;124:268 –276
CORRESPONDENCE Readers are encouraged to write letters to the editor concerning articles that have been published in GASTROENTEROLOGY. Short, general comments are also considered, but use of the Correspondence section for publication of original data in preliminary form is not encouraged. Letters should be typewritten double-spaced and submitted in triplicate.
JC Virus and Human Colon Carcinoma: An Intriguing and Inconclusive Association Dear Sir: During the last few years there have been several reports emphasizing the association of polyomavirus and colon carcinomas.1–3 Others and ourselves have also been studying the presence of viral sequences and proteins in colon cancer, lymphomas, and other tumors with results that are usually contradictory and scarcely reproducible. Firstly, we wish to emphasize that we strongly believe that oncogenic viruses (HPV, HH8, Epstein–Barr, hepatitis B and C), and as well other ubiquitous viruses such as polyomaviruses (JC, BK, and SV40) and adenoviruses can, in significant percentages probably over 15%, be related to human tumors. In fact, SV40 has been associated with human mesotheliomas and other tumors, and JC and BK viruses with brain cancers. Nevertheless, it is difficult to detect viral sequences of polyomaviruses in human tumors; usually it is necessary to use nested PCR and, more importantly, it is difficult to reproduce results in different laboratories. With regard to the presence of JC virus in colon cancer and colon samples reported in the last few years, we tried to detect JC sequences in over 100 frozen human colon carcinomas and normal mucosa. We have attempted several methodological approaches, including nested PCRs. We used primers that amplify a conserved region of the large T antigen of polyomaviruses (PYV forward and PYV reverse).1–3 Moreover, a nested PCR with primers that amplify DNA sequences that coding the amino terminus of the JCV T antigen, was used.1 In all these PCRs, DNA samples were pretreated with topoisomerase I as previously described.1 In only one case were we able to detect JC sequences in a tumor sample. We were unable to detect specific staining of large T protein by immunohistochemistry (pAb416 Oncogene Science) in any case, including the PCR positive case. We ran parallel positive controls in all cases as a JCV Mad1 plasmid and DNA obtained from the COS cell line, which harbors SV40 virus. The SW480 cell line, reported to be positive to JCV, was repeatedly negative. To ensure that our results were accurate and that it was not a problem of sensitivity of the primers and the PCR strategy used in our laboratory, the samples and DNA of all the colon tumors were sent to the Diagnostic Service of the Microbiology National Centre of Spain. Indeed, they have recently reported 2 nested PCR methods that amplify a fragment of the large T antigen gene of polyomaviruses. The first method is a multiplex nested PCR for simultaneous detection and typing of polyomaviruses JCV, BKV, and SV40.4 The second one is a competitive nested PCR that use the internal control to exclude inhibitions for the PCR.5 Also, they used a PCR method that amplifies the transcription control region of JCV. The sensitivity of the 3 methods is even higher than the Taqman assay and is able to detect at least 10 copies of virus genomes per tube (Figure 1). Under these conditions, they were unable to detect any positive case in the colon samples. Moreover, in all colon tumors, we studied p53 protein and mutations, microsatellite instability, and clinicopathological features. About 40% of the tumors had mutated p53 and 12% had microsatellite instability. We can completely discard the possibility that the sensitivity of the nested PCRs used in this work is lower than others. Therefore,
there are 3 possible explanations for these contradictory results. First, those previously positive cases are due to PCR contamination or artifacts. The use of the PCR approach, especially after studying nested PCR and using plasmids as a control, it is relatively easy to suffer contamination and false positives. In fact, we have studied the presence of adenovirus regions in Ewing tumors and lymphomas also with contradictory results and, following a great amount of work and time to set up the PCRs, we only can talk about the inconclusive presence of adenovirus in Ewing tumors and lymphomas.6 Second, we can speculate that the epidemiology of polyomaviruses in the United States may be different from Europe. This presumption is difficult to uphold because these viruses are ubiquitous and therefore distributed worldwide.1 Moreover, other groups have found similar negative results in United States, including a report from the Fred Hutchinson Cancer Research Center in Seattle8 and other groups9 studying SV40 in lymphomas, with set of common primer for polyomaviruses, they also test a large series of colon cancer patients with negative results. Third, experimentally, and also in some human tumors, it has been demonstrated that some viruses can induce cellular transformation and, subsequently, the virus can be silenced and even lose the cellular genome during the progression of the tumor. This “hit-and-run” of virus in cellular transformation has to be considered in the discussion of virus and human cancer.7 All other criteria for providing links between a candidate virus and human cancer have been summarized by Blaho and Weinberg.10 Some of these interesting criteria are not demonstrated with polyomaviruses because there is no association between geographic distribution and viral infections and the incidence of these tumors. To summarize, we have been unable to demonstrate JC sequences in 100 colon carcinomas and normal mucosa. Even if the results that are contradictory to other reports could be based on the sensitivity of the PCR approach used, we can conclude that the number of JC virus sequences in human colon is extremely low and, then, an association between JC virus and colon cancer cannot yet be supported. J. HERNA´NDEZ LOSA V. FERNANDEZ-SORIA C. PARADA R. SANCHEZ-PRIETO ´ N Y CAJAL S. RAMO Department of Pathology Clinica Puerta de Hierro Madrid, Spain C.G. FEDELE A. TENORIO Diagnostic Microbiology Service Centro Nacional de Microbiologı´a Instituto de Salud Carlos III Madrid, Spain 1. Laghi L, Randolph AE, Chauhan DP, Marra G, Major EO, Neel JV, Boland CR. JC virus DNA is present in the mucosa of the human colon and in colorectal cancers. Proc Natl Acad Sci U S A 1999; 96:7484 –7489. 2. Ricciardiello L, Chang DK, Laghi L, Goel A, Chang CL, Boland CR.
January 2003
CORRESPONDENCE
269
4. Fedele CG, Avellon A, Ciardi M, Delia S, Tenorio A. Quantitation of polyomavirus DNA by a competitive nested polymerase chain reaction. J Virol Methods 2000;88:51– 61. 5. Fedele CG, Ciardi M, Delia S, Echevarria JM, Tenorio A. Multiplex polymerase chain reaction for the simultaneous detection and typing of polyomavirus JC, BK and SV40 DNA in clinical samples. J Virol Methods 1999;82:137–144. 6. de Alava E, Sanchez-Prieto R, Ramon Y, Cajal S. Adenovirus E1A and Ewing tumors. Nat Med 2000;6:4. 7. Nevels M, Tauber B, Spruss T, Wolf H, Dobner T. “Hit-and-run” transformation by adenovirus oncogenes. J Virol 2001;75:3089 – 3094. 8. Bigler J, Newcomb P, Morimoto L, Madelson M. Is JCV associated with colorectal cancer? Proc Am Assoc Cancer Res 2002;43: March. 9. Vilchez RA, Madden CR, Kozinetz CA, Halvorson SJ, White ZS, Jorgensen JL, Finch CJ, Butel JS. Association between simian virus 40 and non-Hodgkin lymphoma. Lancet 2002;359:817– 823. 10. Blaho JA, Aaronson SA. Convicting a human tumor virus: guilt by association? Proc Natl Acad Sci U S A 1999:96:7619 –7621. doi:10.1053/gast.2003.50033
Figure 1. Agarose gels showing the sensitivity of the nested PCR methods used in this study. (A) Multiplex-nested-PCR for polyomaviruses. Each of the cloned polyomavirus DNA was assayed between 0, 1, and 103 copies per tube. H2O is a negative control and MW is a DNA ladder. Each polyomavirus rendered an amplified fragment of different size (353 bp for BKV, 189 bp for JCV, and 135 bp for SV40). The primers used were previously described.5 (B) Competitive-nestedPCR for polyomaviruses. Serial dilutions (1–107 copies per tube, each processed in duplicate) of the cloned JCV DNAs were amplified in the presence of a fixed amount of internal standard competitor (100 copies per tube). MW is a DNA ladder and 0 is the nontarget DNA control. The upper band is the product of the target amplification (228 bp) and the lower band is the product obtained from the internal standard competitor (183 bp). The primers used were previously described.4 (C) Nested-PCR for amplification of the transcriptional control region of JCV. Decreasing dilutions of cloned JCV DNA (0.1– 103 copies per tube). H2O is the negative control and MW is a DNA ladder. The primers used were ORI 1: 5⬘- GATGAGCAACTTTTACACCTTG- 3⬘, ORI 2: 5⬘- CTTACCTATGTAGCTTTTGG- 3⬘, and the primers used in nested PCR were ORI 3: 5⬘ - GAAAAACAAGGGAATTTCCCTGGCC- 3⬘,ORI 4: 5⬘ - TGAGCTCTTTTTTTAGTTCCACTCC- 3⬘.
Mad-1 is the exclusive JC virus strain present in the human colon, and its transcriptional control region has a deleted 98-base-pair sequence in colon cancer tissues. J Virol 2001;75:1996 –2001. 3. Ricciardiello L, Laghi L, Ramamirtham P, Chang CL, Chang DK, Randolph AE, Boland CR. JC virus DNA sequences are frequently present in the human upper and lower gastrointestinal tract. Gastroenterology 2000;119:1228 –1235.
Reply. Losa et al. report in their letter that they have been unable to amplify JC virus DNA sequences using nested PCR in specimens of colon cancer, and question work that we have published.1–3 We acknowledge that differences in scientific results require open discussion to resolve discrepancies, and we will take this opportunity to rebut their negative findings. Our group has been testing the hypothesis that JC virus is involved in the generation of chromosomal instability in colorectal neoplasia for several years, and offers the admonition that is a technically demanding area. It is not always possible to explain why one group has difficulty reproducing results reported by another; however, differences in results are usually based on the technical details of the assays. We do not believe that PCR contamination or artifacts can account for the results reported by our laboratory. First, we repeatedly confirmed our findings using new sets of freshly prepared reagents, used new sets of tumor specimens, performed the PCRs in 3 different laboratory locations over time, and ran controls with every experiment. JC virus is a tiny (5.12 kb), closed, circular, super-coiled DNA virus, and these biophysical factors can confound attempts to amplify viral DNA by PCR. All of our data reported thus far have been obtained on DNA extracted from tissues that were either fresh or frozen in the absence of formalin or ethanol fixation. The differences are most likely due to the use of different primers by Losa et al. Sensitive detection of the viral DNA required the use of degenerate primers to overcome expected sequence variations of this virus. The degenerate primers were mentioned first in the legend to Table 3 of our first paper,1 and were described in detail in the second paper.2 Nested PCR was required to overcome the challenge of either low viral copy number or inefficient amplification because of the configuration of the viral DNA. However, the identity of the PCR products was identified by their migration in agar, Southern hybridization, and finally, cloning and sequencing of the PCR product.1 The targets amplified were unequivocally JC virus. Losa et al. report that they have used “PYV” primers, which map outside of the 520 bp region of the JC virus T antigen, which has been the focus of our work.1–3 These primers might be better for the detection of other polyomaviruses. It took time and persistence to achieve consistent results in our laboratory, and not every primer or technique will necessarily give the same results.
CORRESPONDENCE Readers are encouraged to write letters to the editor concerning articles that have been published in GASTROENTEROLOGY. Short, general comments are also considered, but use of the Correspondence section for publication of original data in preliminary form is not encouraged. Letters should be typewritten double-spaced and submitted in triplicate.
JC Virus and Human Colon Carcinoma: An Intriguing and Inconclusive Association Dear Sir: During the last few years there have been several reports emphasizing the association of polyomavirus and colon carcinomas.1–3 Others and ourselves have also been studying the presence of viral sequences and proteins in colon cancer, lymphomas, and other tumors with results that are usually contradictory and scarcely reproducible. Firstly, we wish to emphasize that we strongly believe that oncogenic viruses (HPV, HH8, Epstein–Barr, hepatitis B and C), and as well other ubiquitous viruses such as polyomaviruses (JC, BK, and SV40) and adenoviruses can, in significant percentages probably over 15%, be related to human tumors. In fact, SV40 has been associated with human mesotheliomas and other tumors, and JC and BK viruses with brain cancers. Nevertheless, it is difficult to detect viral sequences of polyomaviruses in human tumors; usually it is necessary to use nested PCR and, more importantly, it is difficult to reproduce results in different laboratories. With regard to the presence of JC virus in colon cancer and colon samples reported in the last few years, we tried to detect JC sequences in over 100 frozen human colon carcinomas and normal mucosa. We have attempted several methodological approaches, including nested PCRs. We used primers that amplify a conserved region of the large T antigen of polyomaviruses (PYV forward and PYV reverse).1–3 Moreover, a nested PCR with primers that amplify DNA sequences that coding the amino terminus of the JCV T antigen, was used.1 In all these PCRs, DNA samples were pretreated with topoisomerase I as previously described.1 In only one case were we able to detect JC sequences in a tumor sample. We were unable to detect specific staining of large T protein by immunohistochemistry (pAb416 Oncogene Science) in any case, including the PCR positive case. We ran parallel positive controls in all cases as a JCV Mad1 plasmid and DNA obtained from the COS cell line, which harbors SV40 virus. The SW480 cell line, reported to be positive to JCV, was repeatedly negative. To ensure that our results were accurate and that it was not a problem of sensitivity of the primers and the PCR strategy used in our laboratory, the samples and DNA of all the colon tumors were sent to the Diagnostic Service of the Microbiology National Centre of Spain. Indeed, they have recently reported 2 nested PCR methods that amplify a fragment of the large T antigen gene of polyomaviruses. The first method is a multiplex nested PCR for simultaneous detection and typing of polyomaviruses JCV, BKV, and SV40.4 The second one is a competitive nested PCR that use the internal control to exclude inhibitions for the PCR.5 Also, they used a PCR method that amplifies the transcription control region of JCV. The sensitivity of the 3 methods is even higher than the Taqman assay and is able to detect at least 10 copies of virus genomes per tube (Figure 1). Under these conditions, they were unable to detect any positive case in the colon samples. Moreover, in all colon tumors, we studied p53 protein and mutations, microsatellite instability, and clinicopathological features. About 40% of the tumors had mutated p53 and 12% had microsatellite instability. We can completely discard the possibility that the sensitivity of the nested PCRs used in this work is lower than others. Therefore,
there are 3 possible explanations for these contradictory results. First, those previously positive cases are due to PCR contamination or artifacts. The use of the PCR approach, especially after studying nested PCR and using plasmids as a control, it is relatively easy to suffer contamination and false positives. In fact, we have studied the presence of adenovirus regions in Ewing tumors and lymphomas also with contradictory results and, following a great amount of work and time to set up the PCRs, we only can talk about the inconclusive presence of adenovirus in Ewing tumors and lymphomas.6 Second, we can speculate that the epidemiology of polyomaviruses in the United States may be different from Europe. This presumption is difficult to uphold because these viruses are ubiquitous and therefore distributed worldwide.1 Moreover, other groups have found similar negative results in United States, including a report from the Fred Hutchinson Cancer Research Center in Seattle8 and other groups9 studying SV40 in lymphomas, with set of common primer for polyomaviruses, they also test a large series of colon cancer patients with negative results. Third, experimentally, and also in some human tumors, it has been demonstrated that some viruses can induce cellular transformation and, subsequently, the virus can be silenced and even lose the cellular genome during the progression of the tumor. This “hit-and-run” of virus in cellular transformation has to be considered in the discussion of virus and human cancer.7 All other criteria for providing links between a candidate virus and human cancer have been summarized by Blaho and Weinberg.10 Some of these interesting criteria are not demonstrated with polyomaviruses because there is no association between geographic distribution and viral infections and the incidence of these tumors. To summarize, we have been unable to demonstrate JC sequences in 100 colon carcinomas and normal mucosa. Even if the results that are contradictory to other reports could be based on the sensitivity of the PCR approach used, we can conclude that the number of JC virus sequences in human colon is extremely low and, then, an association between JC virus and colon cancer cannot yet be supported. J. HERNA´NDEZ LOSA V. FERNANDEZ-SORIA C. PARADA R. SANCHEZ-PRIETO ´ N Y CAJAL S. RAMO Department of Pathology Clinica Puerta de Hierro Madrid, Spain C.G. FEDELE A. TENORIO Diagnostic Microbiology Service Centro Nacional de Microbiologı´a Instituto de Salud Carlos III Madrid, Spain 1. Laghi L, Randolph AE, Chauhan DP, Marra G, Major EO, Neel JV, Boland CR. JC virus DNA is present in the mucosa of the human colon and in colorectal cancers. Proc Natl Acad Sci U S A 1999; 96:7484 –7489. 2. Ricciardiello L, Chang DK, Laghi L, Goel A, Chang CL, Boland CR.
January 2003
CORRESPONDENCE
269
4. Fedele CG, Avellon A, Ciardi M, Delia S, Tenorio A. Quantitation of polyomavirus DNA by a competitive nested polymerase chain reaction. J Virol Methods 2000;88:51– 61. 5. Fedele CG, Ciardi M, Delia S, Echevarria JM, Tenorio A. Multiplex polymerase chain reaction for the simultaneous detection and typing of polyomavirus JC, BK and SV40 DNA in clinical samples. J Virol Methods 1999;82:137–144. 6. de Alava E, Sanchez-Prieto R, Ramon Y, Cajal S. Adenovirus E1A and Ewing tumors. Nat Med 2000;6:4. 7. Nevels M, Tauber B, Spruss T, Wolf H, Dobner T. “Hit-and-run” transformation by adenovirus oncogenes. J Virol 2001;75:3089 – 3094. 8. Bigler J, Newcomb P, Morimoto L, Madelson M. Is JCV associated with colorectal cancer? Proc Am Assoc Cancer Res 2002;43: March. 9. Vilchez RA, Madden CR, Kozinetz CA, Halvorson SJ, White ZS, Jorgensen JL, Finch CJ, Butel JS. Association between simian virus 40 and non-Hodgkin lymphoma. Lancet 2002;359:817– 823. 10. Blaho JA, Aaronson SA. Convicting a human tumor virus: guilt by association? Proc Natl Acad Sci U S A 1999:96:7619 –7621. doi:10.1053/gast.2003.50033
Figure 1. Agarose gels showing the sensitivity of the nested PCR methods used in this study. (A) Multiplex-nested-PCR for polyomaviruses. Each of the cloned polyomavirus DNA was assayed between 0, 1, and 103 copies per tube. H2O is a negative control and MW is a DNA ladder. Each polyomavirus rendered an amplified fragment of different size (353 bp for BKV, 189 bp for JCV, and 135 bp for SV40). The primers used were previously described.5 (B) Competitive-nestedPCR for polyomaviruses. Serial dilutions (1–107 copies per tube, each processed in duplicate) of the cloned JCV DNAs were amplified in the presence of a fixed amount of internal standard competitor (100 copies per tube). MW is a DNA ladder and 0 is the nontarget DNA control. The upper band is the product of the target amplification (228 bp) and the lower band is the product obtained from the internal standard competitor (183 bp). The primers used were previously described.4 (C) Nested-PCR for amplification of the transcriptional control region of JCV. Decreasing dilutions of cloned JCV DNA (0.1– 103 copies per tube). H2O is the negative control and MW is a DNA ladder. The primers used were ORI 1: 5⬘- GATGAGCAACTTTTACACCTTG- 3⬘, ORI 2: 5⬘- CTTACCTATGTAGCTTTTGG- 3⬘, and the primers used in nested PCR were ORI 3: 5⬘ - GAAAAACAAGGGAATTTCCCTGGCC- 3⬘,ORI 4: 5⬘ - TGAGCTCTTTTTTTAGTTCCACTCC- 3⬘.
Mad-1 is the exclusive JC virus strain present in the human colon, and its transcriptional control region has a deleted 98-base-pair sequence in colon cancer tissues. J Virol 2001;75:1996 –2001. 3. Ricciardiello L, Laghi L, Ramamirtham P, Chang CL, Chang DK, Randolph AE, Boland CR. JC virus DNA sequences are frequently present in the human upper and lower gastrointestinal tract. Gastroenterology 2000;119:1228 –1235.
Reply. Losa et al. report in their letter that they have been unable to amplify JC virus DNA sequences using nested PCR in specimens of colon cancer, and question work that we have published.1–3 We acknowledge that differences in scientific results require open discussion to resolve discrepancies, and we will take this opportunity to rebut their negative findings. Our group has been testing the hypothesis that JC virus is involved in the generation of chromosomal instability in colorectal neoplasia for several years, and offers the admonition that is a technically demanding area. It is not always possible to explain why one group has difficulty reproducing results reported by another; however, differences in results are usually based on the technical details of the assays. We do not believe that PCR contamination or artifacts can account for the results reported by our laboratory. First, we repeatedly confirmed our findings using new sets of freshly prepared reagents, used new sets of tumor specimens, performed the PCRs in 3 different laboratory locations over time, and ran controls with every experiment. JC virus is a tiny (5.12 kb), closed, circular, super-coiled DNA virus, and these biophysical factors can confound attempts to amplify viral DNA by PCR. All of our data reported thus far have been obtained on DNA extracted from tissues that were either fresh or frozen in the absence of formalin or ethanol fixation. The differences are most likely due to the use of different primers by Losa et al. Sensitive detection of the viral DNA required the use of degenerate primers to overcome expected sequence variations of this virus. The degenerate primers were mentioned first in the legend to Table 3 of our first paper,1 and were described in detail in the second paper.2 Nested PCR was required to overcome the challenge of either low viral copy number or inefficient amplification because of the configuration of the viral DNA. However, the identity of the PCR products was identified by their migration in agar, Southern hybridization, and finally, cloning and sequencing of the PCR product.1 The targets amplified were unequivocally JC virus. Losa et al. report that they have used “PYV” primers, which map outside of the 520 bp region of the JC virus T antigen, which has been the focus of our work.1–3 These primers might be better for the detection of other polyomaviruses. It took time and persistence to achieve consistent results in our laboratory, and not every primer or technique will necessarily give the same results.