This article presents a methodological critique of the study carried out by Robinson et al (1). In essence this article submits that the work in question departs so profoundly from the work of other research groups that the results must be treated with caution. In particular these results are at best open to interpretation and at worst may be the product of experimentally induced artifacts.
Initially the results of the study by Robinson et al appear unremarkable. The detection rate by PCR is in line with that reported by Danielson et al but some 400% lower than that achieved by IHC using XMRV specific probes (2, 3). This of course means that any conclusion made by PCR alone regarding the association of XMRV and mouse DNA is profoundly unsafe. One must also point out that Robinson et al did not find higher rates of XMRV infection in cancerous versus non cancerous sequences which is a serious departure from the results of other workers. In fact it is such a notable departure that an explanation is needed.
The association between XMRV and mouse DNA in this study is imperfect whereas in a contaminant scenario it would not be. The authors theorise that this may be due to the inability of their PCR assay to detect XMRV in low copy numbers.
This hypothesis was unfortunately not put to the test by using IHC which has been shown to be far more sensitive than PCR. As PCR is by some considerable margin the least sensitive method for detecting XMRV in prostate cancer (2, 3) and chronic fatigue syndrome (4) no association between the presence of mouse DNA and XMRV can be safely made.
The authors expressed surprise that their PCR approach produced gag sequences which did not contain the 24-nt deletion characteristically present in the XMRV sequences cloned to date, as they expected their primers to be specific to VP62 gag. This raises a number of issues. Firstly, despite the authors' expectations they provide no evidence that their primers were specific, merely that they were capable of detecting XMRV gag (5, 6). The only study which has detected XMRV in prostate cancer using primers set for unique sequences in the XMRV genome is Schlaberg et al 2009 (2).
It is therefore something of a mystery why researchers concerned about mouse contamination did not engage in a literature search to establish whether the primers used in their experiments were in fact specific to XMRV sequences which are unique to the virus. This is particularly vital as if the Schlaberg primers did not produce the same results this study would fall. Secondly the presence of a recombinant with a different gag sequence is entirely consistent with the behaviour of XMRV-related viruses in vivo and is to expected in tumour tissue (7). This occurs via the interaction of the nucleic acid of the exogenous viruses and the endogenous viruses which permanently occupy mammalian genomes.
Indeed, the clone reported by the authors which contained a 24 base pair deletion in glycogag would be a classic example of a viral variant caused by the recombination of XMRV and an endogenous MLV virus. The authors were in a position to clarify the issue themselves which sadly they did not. XMRV has unique sequences in the U3 integrase and SU region.
From the authors report they had access to sequences or whole clones containing the information needed to confirm whether these were recombinants or not. It seems astonishing that they did not provide this information in the paper.
Finally,the authors by their own admission expected a primer sequence constructed against a region between U5 and the 5' end of GAG to be only capable of amplifying XMRV These primers were part of a specific assay developed by Urisman et al (2006) (5) but do not compliment any region that is unique to XMRV. It is possible that these authors misunderstood this essential point.
It is difficult to fathom therefore why the authors would conclude that these expected PCR findings were evidence of mouse contamination when a far more parsimonious explanation was available.
Indeed, one would argue that the authors (if aware of recent research conducted by Danielson et al 2010) (8) should have been surprised that primer sequences that they believed were specific to XMRV gag would have produced any product at all. The important point to note is that at this juncture in the study there were no results suggestive of any mouse contamination at all.
Next we turn to the section of the study where the authors test the ability of their PCR assays (IAP and mtDNA) to detect mouse DNA in mouse cell lines. The cell lines named are McCoy and RAW264.7 (RAW). It is not clear whether both cell lines were of mouse origin. The McCoy cell line used could have been human in origin. If so, it would have been spiked, as the authors stated that they amplified mouse DNA from both cell lines.
If they were both mouse cell lines then other considerations enter the equation. McCoy mouse cell lines express a polytropic MLV (9) and the RAW cell line was initially transformed by Abelson MLV. These cell lines have the potential of introducing other viruses into the experimental environment.
In any event, the presence of mouse DNA in these cells must be considered as the source of contamination reported in this experiment. This seems to be a parsimonious explanation as there was no evidence of contamination prior to the use of these cell lines.
Finally, we examine the claim made by the authors that IAP was a much more sensitive assay than mtDNA PCR in detecting mouse DNA in transformed cell lines. The normal criticism of this statement would centre on unfounded extrapolation from in vitro to in vivo conditions. There is however more to consider here.
The environment of transformed and immortalised cells is one of extremely high levels of oxidative stress (10). Mitochondrial DNA is much more prone to oxidative damage (due to its nudity) than genomic DNA (11). Indeed studies have demonstrated that such cell lines are either depleted of mtDNA (12) or that the mitochondrial genome is highly mutated (13).
Therefore the amount of mtDNA actually recoverable by PCR in this experiment may well have been minimal. Thus before any claims regarding the relative sensitivity of the two assays were made, the integrity of the mitochondrial DNA should have been investigated.
Moreover while the authors demonstrated that their IAP assay could not amplify human DNA they did not demonstrate that it could not amplify IAP sequences in prostate tumour tissue. This is particularly remiss of them as IAP sequences have only been detected in PMBCS in cases of Sjiornes syndrome but occur at a high frequency in the DNA of people suffering from cancer (14).The greatest concern however relates to the known ability of retroviruses to package IAP sequences into their genomes.Hence a positive response to a IAP assay may well just indicate XMRV or PMLV infection.
The following is a quote from A Dusty Miller expressing his view about the IAP test in a communication to people with ME/CFS. “However, unpublished results suggest that IAP sequences might be transferred by retroviruses, so finding IAP sequences in human samples might simply reflect IAP transfer by the XMRV and XMRV-like viruses we are trying to detect”
In conclusion, the findings of this study are inconsistent with the reports of other workers. Given the potential for the accidental introduction of contaminant DNA into this experiment, the conclusions cannot be safely generalised. Similarly the claim that IAP is superior to ,or even as reliable , as mtDNA PCR in vivo is not supported by the data cited as alternative explanations for the findings are readily available.
References: 1) MJ Robinson, OW Erlwein, S Kaye, J Weber, O Cingoz, A Patel, MM Walker , W-J Kim, M Uiprasertkul, JM Coffin and MO McClure; Mouse DNA contamination in human tissue tested for XMRV; Retrovirology 2010, 7:108doi:10.1186/1742-4690-7-108 2) Schlaberg R, Choe DJ, Brown KR, Thaker HM, Singh IR (2009) XMRV is present in malignant prostatic epithelium and is associated with prostate cancer, especially high-grade tumors. Proc Natl Acad Sci USA 106:16351–16356 3) RS Arnold, NV Makarova, AO Osunkoya, S Suppiah, TA Scott, NA Johnson, SM Bhosle, D Liotta, E Hunter, FF Marshall, H Ly, RJ Molinaro, JL Blackwell, JA Petros; XMRV Infection in Patients With Prostate Cancer: Novel Serologic Assay and Correlation With PCR and FISH; Urology - April 2010 (Vol. 75, Issue 4, Pages 755-761, DOI:10.1016/j.urology.2010.01.038) 4) JA Mikovits, Y Huang, MA Pfost, VC Lombardi, DC Bertolette, KS Hagen and FW Ruscetti; Distribution of Xenotropic Murine Leukemia Virus-Related Virus (XMRV) Infection in Chronic Fatigue Syndrome and Prostate Cancer; AIDS Rev 2010; 12: 149-52 5) Urisman A, Molinaro RJ, Fischer N, Plummer SJ, Casey G, et al. 2006 Identification of a Novel Gammaretrovirus in Prostate Tumors of Patients Homozygous for R462Q RNASEL Variant. PLoS Pathog 2(3): e25. doi:10.1371/journal.ppat.0020025 6) B Dong, S Kim, S Hong, J Das Gupta, K Malathi, EA Klein, D Ganem, JL DeRisi, SA Chow, and RH Silverman; An infectious retrovirus susceptible to an IFN antiviral pathway from human prostate tumors; PNAS January 30, 2007 vol. 104 no. 5 1655-1660 10.1073/pnas.0610291104 7) H van der Puttena, W Quinta, J van Raaija, ER Maandaga, IM Verma and A Bernsa; M-MuLV-induced leukemogenesis: Integration and structure of recombinant proviruses in tumors; Cell, Volume 24, Issue 3, 729-739, 1 June 1981 doi:10.1016/0092 8674(81)90099-4 8) BP Danielson, GE Ayala and JT Kimata; Detection of Xenotropic Murine Leukemia Virus-Related Virus in Normal and Tumor Tissue of Patients from the Southern United States with Prostate Cancer Is Dependent on Specific Polymerase Chain Reaction Conditions; J Infect Dis. (2010) 202 (10): 1470-1477. doi: 10.1086/656146 9) Fong CK, Yang-Feng TL, Lerner-Tung MB; Re-examination of the McCoy cell line for confirmation of its mouse origin: karyotyping, electron microscopy and reverse transcriptase assay for endogenous retrovirus; Clin Diagn Virol. 1994 Apr;2(2):95-103. 10) RH Burdon, V Gill and C Rice-Evans; Oxidative Stress and Tumour Cell Proliferation; Free Radical Research 1990, Vol. 11, No. 1-3 , Pages 65-76 11) C Richter, J W Park, and B N Ames; Normal oxidative damage to mitochondrial and nuclear DNA is extensive; PNAS September 1, 1988 vol. 85 no. 17 6465-6467 12) J-I Hayashi, M Takemitsu and I Nonaka; Recovery of the missing tumorigenicity in mitochondrial DNA-less HeLa cells by introduction of mitochondrial DNA from normal human cells; Somatic Cell and Molecular Genetics Volume 18, Number 2, 123-129, DOI: 10.1007/BF0123315 13) M Bakhanashvili, S Grinberg, E Bonda, AJSimon, S Moshitch-Moshkovitz and G Rahav; p53 in mitochondria enhances the accuracy of DNA synthesis; Cell Death and Differentiation (2008) 15, 1865–1874; doi:10.1038/cdd.2008.122 14) W Seifarth, H Skladny, F Krieg-Schneider, A Reichert, R Hehlmann, and C Leib-Mösch; Retrovirus-like particles released from the human breast cancer cell line T47-D display type B- and C-related endogenous retroviral sequences; J Virol. 1995 October; 69(10): 6408–6416
Robinson et al flaws in study design
21 January 2011
This article presents a methodological critique of the study carried out by Robinson et al (1). In essence this article submits that the work in question departs so profoundly from the work of other research groups that the results must be treated with caution. In particular these results are at best open to interpretation and at worst may be the product of experimentally induced artifacts.
Initially the results of the study by Robinson et al appear unremarkable. The detection rate by PCR is in line with that reported by Danielson et al but some 400% lower than that achieved by IHC using XMRV specific probes (2, 3). This of course means that any conclusion made by PCR alone regarding the association of XMRV and mouse DNA is profoundly unsafe. One must also point out that Robinson et al did not find higher rates of XMRV infection in cancerous versus non cancerous sequences which is a serious departure from the results of other workers. In fact it is such a notable departure that an explanation is needed.
The association between XMRV and mouse DNA in this study is imperfect whereas in a contaminant scenario it would not be. The authors theorise that this may be due to the inability of their PCR assay to detect XMRV in low copy numbers.
This hypothesis was unfortunately not put to the test by using IHC which has been shown to be far more sensitive than PCR. As PCR is by some considerable margin the least sensitive method for detecting XMRV in prostate cancer (2, 3) and chronic fatigue syndrome (4) no association between the presence of mouse DNA and XMRV can be safely made.
The authors expressed surprise that their PCR approach produced gag sequences which did not contain the 24-nt deletion characteristically present in the XMRV sequences cloned to date, as they expected their primers to be specific to VP62 gag.
This raises a number of issues. Firstly, despite the authors' expectations they provide no evidence that their primers were specific, merely that they were capable of detecting XMRV gag (5, 6). The only study which has detected XMRV in prostate cancer using primers set for unique sequences in the XMRV genome is Schlaberg et al 2009 (2).
It is therefore something of a mystery why researchers concerned about mouse contamination did not engage in a literature search to establish whether the primers used in their experiments were in fact specific to XMRV sequences which are unique to the virus. This is particularly vital as if the Schlaberg primers did not produce the same results this study would fall. Secondly the presence of a recombinant with a different gag sequence is entirely consistent with the behaviour of XMRV-related viruses in vivo and is to expected in tumour tissue (7). This occurs via the interaction of the nucleic acid of the exogenous viruses and the endogenous viruses which permanently occupy mammalian genomes.
Indeed, the clone reported by the authors which contained a 24 base pair deletion in glycogag would be a classic example of a viral variant caused by the recombination of XMRV and an endogenous MLV virus. The authors were in a position to clarify the issue themselves which sadly they did not. XMRV has unique sequences in the U3 integrase and SU region.
From the authors report they had access to sequences or whole clones containing the information needed to confirm whether these were recombinants or not. It seems astonishing that they did not provide this information in the paper.
Finally,the authors by their own admission expected a primer sequence constructed against a region between U5 and the 5' end of GAG to be only capable of amplifying XMRV These primers were part of a specific assay developed by Urisman et al (2006) (5) but do not compliment any region that is unique to XMRV. It is possible that these authors misunderstood this essential point.
It is difficult to fathom therefore why the authors would conclude that these expected PCR findings were evidence of mouse contamination when a far more parsimonious explanation was available.
Indeed, one would argue that the authors (if aware of recent research conducted by Danielson et al 2010) (8) should have been surprised that primer sequences that they believed were specific to XMRV gag would have produced any product at all. The important point to note is that at this juncture in the study there were no results suggestive of any mouse contamination at all.
Next we turn to the section of the study where the authors test the ability of their PCR assays (IAP and mtDNA) to detect mouse DNA in mouse cell lines. The cell lines named are McCoy and RAW264.7 (RAW). It is not clear whether both cell lines were of mouse origin. The McCoy cell line used could have been human in origin. If so, it would have been spiked, as the authors stated that they amplified mouse DNA from both cell lines.
If they were both mouse cell lines then other considerations enter the equation. McCoy mouse cell lines express a polytropic MLV (9) and the RAW cell line was initially transformed by Abelson MLV. These cell lines have the potential of introducing other viruses into the experimental environment.
In any event, the presence of mouse DNA in these cells must be considered as the source of contamination reported in this experiment. This seems to be a parsimonious explanation as there was no evidence of contamination prior to the use of these cell lines.
Finally, we examine the claim made by the authors that IAP was a much more sensitive assay than mtDNA PCR in detecting mouse DNA in transformed cell lines. The normal criticism of this statement would centre on unfounded extrapolation from in vitro to in vivo conditions. There is however more to consider here.
The environment of transformed and immortalised cells is one of extremely high levels of oxidative stress (10). Mitochondrial DNA is much more prone to oxidative damage (due to its nudity) than genomic DNA (11). Indeed studies have demonstrated that such cell lines are either depleted of mtDNA (12) or that the mitochondrial genome is highly mutated (13).
Therefore the amount of mtDNA actually recoverable by PCR in this experiment may well have been minimal. Thus before any claims regarding the relative sensitivity of the two assays were made, the integrity of the mitochondrial DNA should have been investigated.
Moreover while the authors demonstrated that their IAP assay could not amplify human DNA they did not demonstrate that it could not amplify IAP sequences in prostate tumour tissue. This is particularly remiss of them as IAP sequences have only been detected in PMBCS in cases of Sjiornes syndrome but occur at a high frequency in the DNA of people suffering from cancer (14).The greatest concern however relates to the known ability of retroviruses to package IAP sequences into their genomes.Hence a positive response to a IAP assay may well just indicate XMRV or PMLV infection.
The following is a quote from A Dusty Miller expressing his view about the IAP test in a communication to people with ME/CFS.
“However, unpublished results suggest that IAP sequences might be transferred by retroviruses, so finding IAP sequences in human samples might simply reflect IAP transfer by the XMRV and XMRV-like viruses we are trying to detect”
In conclusion, the findings of this study are inconsistent with the reports of other workers. Given the potential for the accidental introduction of contaminant DNA into this experiment, the conclusions cannot be safely generalised. Similarly the claim that IAP is superior to ,or even as reliable , as mtDNA PCR in vivo is not supported by the data cited as alternative explanations for the findings are readily available.
References:
1) MJ Robinson, OW Erlwein, S Kaye, J Weber, O Cingoz, A Patel, MM Walker , W-J Kim, M Uiprasertkul, JM Coffin and MO McClure; Mouse DNA contamination in human tissue tested for XMRV; Retrovirology 2010, 7:108doi:10.1186/1742-4690-7-108
2) Schlaberg R, Choe DJ, Brown KR, Thaker HM, Singh IR (2009) XMRV is present in malignant prostatic epithelium and is associated with prostate cancer, especially high-grade tumors. Proc Natl Acad Sci USA 106:16351–16356
3) RS Arnold, NV Makarova, AO Osunkoya, S Suppiah, TA Scott, NA Johnson, SM Bhosle, D Liotta, E Hunter, FF Marshall, H Ly, RJ Molinaro, JL Blackwell, JA Petros; XMRV Infection in Patients With Prostate Cancer: Novel Serologic Assay and Correlation With PCR and FISH; Urology - April 2010 (Vol. 75, Issue 4, Pages 755-761, DOI:10.1016/j.urology.2010.01.038)
4) JA Mikovits, Y Huang, MA Pfost, VC Lombardi, DC Bertolette, KS Hagen and FW Ruscetti; Distribution of Xenotropic Murine Leukemia Virus-Related Virus (XMRV) Infection in Chronic Fatigue Syndrome and Prostate Cancer; AIDS Rev 2010; 12: 149-52
5) Urisman A, Molinaro RJ, Fischer N, Plummer SJ, Casey G, et al. 2006 Identification of a Novel Gammaretrovirus in Prostate Tumors of Patients Homozygous for R462Q RNASEL Variant. PLoS Pathog 2(3): e25. doi:10.1371/journal.ppat.0020025
6) B Dong, S Kim, S Hong, J Das Gupta, K Malathi, EA Klein, D Ganem, JL DeRisi, SA Chow, and RH Silverman; An infectious retrovirus susceptible to an IFN antiviral pathway from human prostate tumors; PNAS January 30, 2007 vol. 104 no. 5 1655-1660 10.1073/pnas.0610291104
7) H van der Puttena, W Quinta, J van Raaija, ER Maandaga, IM Verma and A Bernsa; M-MuLV-induced leukemogenesis: Integration and structure of recombinant proviruses in tumors; Cell, Volume 24, Issue 3, 729-739, 1 June 1981 doi:10.1016/0092 8674(81)90099-4
8) BP Danielson, GE Ayala and JT Kimata; Detection of Xenotropic Murine Leukemia Virus-Related Virus in Normal and Tumor Tissue of Patients from the Southern United States with Prostate Cancer Is Dependent on Specific Polymerase Chain Reaction Conditions; J Infect Dis. (2010) 202 (10): 1470-1477. doi: 10.1086/656146
9) Fong CK, Yang-Feng TL, Lerner-Tung MB; Re-examination of the McCoy cell line for confirmation of its mouse origin: karyotyping, electron microscopy and reverse transcriptase assay for endogenous retrovirus; Clin Diagn Virol. 1994 Apr;2(2):95-103.
10) RH Burdon, V Gill and C Rice-Evans; Oxidative Stress and Tumour Cell Proliferation; Free Radical Research 1990, Vol. 11, No. 1-3 , Pages 65-76
11) C Richter, J W Park, and B N Ames; Normal oxidative damage to mitochondrial and nuclear DNA is extensive; PNAS September 1, 1988 vol. 85 no. 17 6465-6467
12) J-I Hayashi, M Takemitsu and I Nonaka; Recovery of the missing tumorigenicity in mitochondrial DNA-less HeLa cells by introduction of mitochondrial DNA from normal human cells; Somatic Cell and Molecular Genetics Volume 18, Number 2, 123-129, DOI: 10.1007/BF0123315
13) M Bakhanashvili, S Grinberg, E Bonda, AJSimon, S Moshitch-Moshkovitz and G Rahav; p53 in mitochondria enhances the accuracy of DNA synthesis; Cell Death and Differentiation (2008) 15, 1865–1874; doi:10.1038/cdd.2008.122
14) W Seifarth, H Skladny, F Krieg-Schneider, A Reichert, R Hehlmann, and C Leib-Mösch; Retrovirus-like particles released from the human breast cancer cell line T47-D display type B- and C-related endogenous retroviral sequences; J Virol. 1995 October; 69(10): 6408–6416
Competing interests
None declared