The following is an analysis of the paper, ‘Phylogenetic analysis of MLV sequences from longitudinally sampled Chronic Fatigue Syndrome patients suggests PCR contamination rather than viral evolution’. This paper attempted to find evidence for laboratory contamination in the second positive human gammaretrovirus (HGRV) ME/cfs study, Lo et al. (2010).
It should be noted at the outset that the samples in Lo et al. repeatedly tested negative for mouse genomic and mitochondrial DNA. Greg Towers latest paper (Katzourakis, 2011) focuses on one rare instance where an IAP assay arguably produced a false negative result. This publication is the latest in a sequence of reports attempting to prove that HGRVs are harmless contaminants. This is the antithesis of the scientific method. Once again we have a paper seeking to overrule experimental scientific evidence by the use of unvalidated computer simulations involving assumptions that are highly unlikely to apply to class switching (the creation of a new class of retrovirus by recombination) in the evolution of gammaretroviruses.
The reader of this analysis is invited to believe that for some reason the IAP and mitochondrial assays used to examine the Lo samples failed to detect mouse contamination, which in the mind of the author must have been present. As, in the mind of the author (Towers), the sequences must be PCR introduced contamination. This evidence of contempt prior to investigation ill becomes a scientist. Research by a person lacking an open mind and seeking to confirm a prejudicial viewpoint, cannot, in any sense of the word, be described as scientific. This person is however willing to accept studies using PCR assays that merely rely on evidence regarding theoretical limits of detection to indicate effectiveness. This is despite the fact that numerous studies exist in virology that prove assays demonstrating excellent theoretical limits of detection fail to detect a target in a known clinical positive sample.
“The shape of a phylogenetic tree reflects the evolutionary processes under which it has grown”
The above statement has no specific meaning. There are different methods of statistical analysis, which produce different phylogenetic trees. Therefore, it depends at which level of a tree (superficial or fine detail) the data is considered. The shape of a tree may be similar, but the detailed relationship between sequences can be very different. Greg Tower’s argument is based on quite fine detail. The positioning of sequences within clades on the basis of a bootsrap analysis, as opposed to the maximum likelihood analysis used in this study, can be very different.
Lo et al. produced an analysis that did not put CFS 1-3 into the modified polytropic group.
“Phylogenetic analysis of the human-derived MLV sequences as well as a variety of known MLV sequences indicates that the three patient derived sequence types, CFS 1-3, fall within the modified polytropic MLV clade. The overall shape of the phylogentic tree, including the three main groupings (polytropic, modified polytopic and xenotropic) and the relationships between them, are consistent with previous studies based on full-length proviruses (12, 13)”
These are the references 12 and 13
12. Hue, S., E. R. Gray, A. Gall, A. Katzourakis, C. P. Tan, C. J. Houldcroft, S. McLaren, D. Pillay, A. Futreal, J. A. Garson, O. G. Pybus, P. Kellam, and G. J. Towers. 2010. Disease-associated XMRV sequences are consistent with laboratory contamination. Retrovirology 7:111.
13. Jern, P., J. P. Stoye, and J. M. Coffin. 2007. Role of APOBEC3 in genetic diversity among endogenous murine leukemia viruses. PLoS Genet 3.2014-22.
The above requires a little translation. Firstly, the whole proviruses mentioned did not include any exogenous ectropic polytropic or modified polytropic viruses.In short they were almost exclusively endogenous proviruses Reference 12 is also a paper led by Towers. Thus, using the same qualitative inputs and the same computer program, he is getting the same results. This in no way demonstrates objective accuracy, The analysis is clearly driven by a preconception that HGRVs are endogenous mouse viral sequences and the sequences are classified to conform to this prejudicial view.
This would appear to be an appropriate juncture to make a comment about the classification of polytropic and modified polytropic MuLVs. This classification was actually devised by Coffin and Stoye and refers to endogenous MuLVs. According to those authors, a modified polytropic endogenous MuLV is a modified polytropic if it contains a 24 nucleotide deletion in the env gene. Other authors refer to exogenous MuLVs as modified polytropic if they have smaller deletions and a number of base substitutions. The definition of a modified polytropic is hence somewhat nebulous and is in reality polytropic viruses with deletion sequences. These terms are therefore subjective labels, and thus it is difficult to see how classifying sequences on the basis that they are similar to subjectively labeled endogenous retroviral sequences has any scientific meaning.
“Under a model of within-patient viral evolution, we would expect all of the 2010 daughter sequences to branch from the parental CFS type 1 sequence with longer but approximately equidistant branches. This is true for all other longitudinally sequenced viruses, reviewed in (24).”
When reference 24 is consulted, one notes that, with the exception of HIV, the work focuses on the within-patient evolution of single stranded RNA viruses that produce high titres and very high rates of replication. HIV is the only retrovirus thus far which has been longitudinally sequenced. Thus, for some reason Towers expects a gammaretrovirus to replicate at the same rate and reach very high titres in an infected host. This expectation is clearly misplaced. A vast literature exists demonstrating that gammaretroviruses are slowly replicating low titre viruses. Towers is treating this model as though it was an accurate representation of gammaretroviral evolution, but it does not even apply to other retroviruses, such as deltaretroviruses or beta retroviruses. In fact, the whole paper is built on the assumption that human gammaretroviruses behave like HIV, without any scientific evidence provided in support of that belief.
“We assessed the fit of the data to this model by inspection of the phylogenetic tree and by maximum likelihood based model testing. The phylogenetic placement of the longitudinal sequences does not fit this expected model. When the sequences from the new time point are examined, we find that 5 of 6 are phylogenetically distinct from the parental sequence and from each other. The more recent longitudinally sampled sequences are shown in red (Fig. 1).”
The phylogenetic model mentioned above is based on the assumption that the differences between the sequences are created by DNA base substitution (Markov models). It is an entirely theoretical construct. The scale in Figure 1 is expanded, which maximizes the impression of a very large dissimilarity between the sequences. We have already discussed the fact that the phylogenetic boundaries between polytropic and modified polytropic are based on endogenous MuLVs and are artificial because they are not a product of evolutionary change. The scientific perspective would focus on sequence variation and how many nucleotide base changes have occurred, rather than focus on linguistic terminology.
“In fact, the two most distantly related sequences from these longitudinal patient samples are about as different from each other as the biggest distance possible within the polytropic clade”
If they are within the same narrowly constructed polytropic clade, then they can hardly be described as distantly related. Such colorful use of language in a scientific journal would seem to be inappropriate.
“Thus these new patient-derived MLV sequences show tremendous variation from the parental CFS type 1 sequence and as such are extremely unlikely to have evolved from the CFS”
This statement is simply incorrect. In absolute terms the sequence changes are relatively minor. The reader is referred to Figure 1.
“The probability that the data are consistent with a model of within-patient evolution can be explicitly tested by comparing the maximum likelihood phylogenetic tree directly derived from the data with a tree in which sequences from the second time point are constrained to cluster with sequences from the first time point. 123 The difference in likelihood of these two topologies was compared using the Shimodaira Hasegawa test”
Using this test (Shimodaira Hasegawa) when one of two typologies to be compared is one of maximum likelihood and the typologies have not been determined a prior completely invalidates the mathematical assumptions that underpin this test.
“To calculate the chance of a modified polytropic virus evolving into a polytropic virus we reconstructed the common ancestors of the polytropic and modified polytropic…”
This demonstrates a complete lack of understanding concerning the mechanisms underlying the generation of genetic variation in MLVs. Polytropic viruses do not evolve from ancestors by nucleotide substitution as assumed in the statistical model used in this publication. Polytropic retroviruses are formed by the recombination of ecotropic MLVs with polytropic endogenous env sequences found within the host genome. Recombination could easily account for the changes of the sequences found in patients with ME/cfs over time. One would hope that a retrovirologist would realize that the changes in the sequences could not be formed using the mechanism assumed by a computer simulation before spending vast amounts of money engaging in an exercise to prove that these changes could not be a result of nucleotide substitutions. One would also hope that a scientist would at least consider that the assumptions built into this model were incorrect where the generation of sequence variation in gammaretroviruses are concerned. The use of a statistical test in an inappropriate manner, so that the null hypothesis is invalidated, enabling its rejection, is tantamount to incompetence. Desperately ill patients are entitled to better than this.
