Next Gen sequencing provides evidence for MLVs and XMRV in prostate cancer! Eh? What's this all aboot?

In an epub that came up in my regular pubmed search today, we see that our Canadian friends have been doing next-gen sequencing of prostate cancer samples.  The publication is entitled "Next generation sequencing of prostate tumours provides independent evidence of XMRV contamination".  Groan. Seriously, how much can this one topic take.  But then I took a closer look at the data, and despite the title, they have some very interesting findings -- the manuscript is from Colin Collins' lab at the University of British Columbia and Vancouver Prostate Centre.  The paper, conveniently super-short, describes some results from genomic and transcriptomic sequencing of tissue samples from 9 prostate tumors or their metastatic deposits, and from 3 human tumor xenografts that were grown in mouse.  The authors conclude that because they detected XMRV in two of the human tumors that were not grown in mouse, but those tumors also have mouse mitochondrial DNA and were prepared on a different day to the other (XMRV negative) tumors, they detected their own contamination of the tissue.  ummm...OK.

Firstly, this study was kind of neat because as opposed to all the PCR-based methods to detect XMRV or MLVs in human tissues, next-gen sequencing is unbiased -- it will sequence whatever is there -- PCR primers are designed against a particular sequence, and if the conditions are not right or there has been some mutation of the original published sequence (as we'd expect from actively replicating or mutating viruses), the product may not amplify.  What these guys did was avoid that issue.  They used xenografts of human tissue implanted in mice as positive controls for what mouse viruses would look like.

 In  Table 1 of the manuscript they summarize the data- as expected, human tissues grown in and excised from mouse had mouse viruses (MLVs and apparently XMRV - I say apparently because it is not clear if the authors mean that the exact same sequence as VP62).  For MLV transcriptomes, there was between 327 and 443 reads per million sequence reads, and for XMRV there was between 169-220 per million reads.  So MLV and XMRV RNA was there.  DNA sequencing from two of the xenografts showed between 6 and 10 MLV DNA reads per million, and 2-4 XMRV DNA reads per million.  So detecting MLV and XMRV DNA copies using next-gen is obviously alot less sensitive than detecting it in the transcriptome - these tissues were contaminated (according to the paper) with significant mouse tissue, so we know there was genomically integrated viral DNA.  Therefore the sensitivity for MLV DNA was 30-70 fold less than for RNA, and for XMRV DNA detection was 40-80 fold less sensitive.  But that's for the controls.  Analysis of the human tumors resulted in ALL being positive for some level of MLV RNAseq reads.  Six samples were positive for XMRV.  The number of reads per virus type was much less than the xenografts grown in mouse -- the authors took the two tumors with the highest level of XMRV (each around 11 reads per million) and did next-gen DNA sequencing on them, as well as on one XMRV negative tumor.  They did not get any DNA reads positive for either MLVs or XMRVs.  But hang on, what was their sensitivity? By comparing their numbers of RNA positive reads, which ranged from as low as 0.1% of that seen in the tissues from mice -- so accordingly, we expect about 0.009 -0.16 MLV reads at the DNA level, if all things were equal.  To counter the argument that they might not have the sensitivity to exclude these viruses at the DNA level, the authors argue that published data suggests XMRV is expected to integrate on every chromosome within infected genomes. They reference Sam Chow's lab's paper from 2008, "Integration site preference of xenotropic murine leukemia virus-related virus, a new human retrovirus associated with prostate cancer".  But that same lab just published (click here) saying that at least from the cases they could tell, their samples were contaminated with DNA from infected DU145 cells.  And I'm pretty sure they never said that XMRV integrates on every chromosome in every infected cell.

So what to make of all this?  I don't know right now...it looks like there are MLV and even XMRV transcriptomes present in some prostate cancer tissues.  There might not have been the sensitivity to detect the viruses at the DNA level.  And assuming this data will be borne out by other next-gen sequencing studies, what does it all mean?  I guess we have to await some more next-gen sequencing to find out - meanwhile, if you want to read the paper - click here-- but you might need access to the journal of clinical microbiology....