Hosts: Vincent Racaniello, Alan Dove, and Rich Condit
Vincent, Alan, and Rich talk about XMRV integration sites in prostate tumor DNA, the decline effect and scientific method, and the first virus of Caenorhabditis nematodes.
Click the arrow above to play, or rightight click to download TWiV #123 (67 MB .mp3, 93 minutes).
Subscribe (free): iTunes, RSS, email
Links for this episode:
- Analysis of XMRV integration sites from human prostate cancer (Retrovirology)
- Integration site preference of XMRV (J Virology)
- The Decline Effect and the Scientific Method by Jonathan Lehrer (New Yorker)
- Why most published research findings are false (PLoS Medicine)
- Cochrane Reviews
- Richard Feynman’s 1974 Caltech Commencement Address (pdf) (thanks, Bill!)
- First virus of Caenorhabditis (PLoS Biology)
- Nodaviruses at ViralZone
- Spandrels (thanks, Welkin!)
- TWiV on Facebook
- Letters read on TWiV 123
Weekly Science Picks
Rich – Einstein: His Life and Universe by Walter Isaacson (Aspen Institute)
Alan – Hi-definition microscopy movies in 3D
Vincent – Stan Maloy interview with Beatrice Hahn on the origins of HIV and malaria
Send your virology questions and comments to firstname.lastname@example.org.
How does contamination explain the immunohistochemistry of XMRV in PC?
You should listen to TWiV #113 with Alan Rein. He gives a good explanation of antibody cross-reactivity and the importance of using HPLC-purified reagents. The short version: antibody assays can be noisy unless they’re done under very stringent conditions.
I did a lot of in situ and IHC during the early HPV days. I agree that both assay types can be tricky and noisy. But the in situ hybridization figures presented in Silverman’s 2006 PLoS paper and the IHC in Singh’s 2009 PNAS paper look like they are detecting specific signal more so than noise.
But is that specific-looking signal coming from a virus, or from a human antigen that cross-reacts with the antibody?
From Singh’s paper with the 100% and 1% infected, it looks like the signal is virus specific. Silverman’s in situ are more difficult to see, but I would think that if the DNA probe was hybridizing to human/endogenous targets, everything would be lit up.
I agree. Wouldn’t there be much more signal in every cell if there was cross binding to human or endogenous targets?
No human antigen can cross react with a monoclonal antibody to SFFV env. Only SU specific sequences will react. If a reaction occurs it is a MULV class virus.
10 Copies per 660 diploid cells. (Schalberg et al, 2009)
You really should listen to the Alan Rein episode again. He goes into a lot of this, and specifically discusses Ila’s work. Human antigen expression levels vary all the time, so cross-reactivity can look an awful lot like a real signal. It’s a potential problem in all IHC.
I posted this on the blog, but it needs mentioning again.
There is substantial scientific evidence that HGRVs are human pathogens and none whatsoever that they are not.
Firstly, XMRV integration sites were first discovered in prostrate DNA taken directly from patients. No cell lines of any kind involved. Cell line contamination not possible there.
Also, the study in question is unable to say where the cell line has been “Analysis of 15 nude mouse strains indicated that none contained XMRV, but some strains potentially used to passage the xenograft contained both PreXMRV-1 and PreXMRV-2.”
The abstracts posted in CROI present conclusions as fact when they are no such thing. The DU145 cell lines were screened for the presence of XMRV by Dong et al in 2007, and no XMRV was present as provirus or cDNA. Therefore, if integrated sequences were found within the cells, then the direction of transfer was from the human prostate DNA, which contained those sequence, to the cells in question. Similarly, the apparent appearance of XMRV, after the production of xenografts can be explained by the fact that the PCR system used was not sensitive enough to detect XMRV in the original tissue. We certainly have enough examples of that. The testosterone pellets used in the process however would ensure rapid replication of XMRV to the levels which the PCR system used could detect it.
This is the reference showing how an increase in the level of NF-kappa B increases the replication rate of XMRV, published in the peer reviewed journal of virology.
How does an increase in the level of NF-kappa B increase the replication rate of a contaminant?
This is the reference showing the elevated rate of NF-kappa B in DU145 cells.
If DU145 cells have XMRV integrated into their DNA, can anyone explain why the XMRV does not transcribe and the fact that DU145 cells do not express XMRV, given the high levels of nf-kappa b and oxidative stress known to exist within these cells?
Finally, Kim et al. has found that XMRV is capable of making three insertions within a 100 kilobase region, and that the integration points for all of the 472 regions mapped were different. If the nucleotide positions were identical to the two sites identified in Garson et al. contamination must have come from the patient sample, not the DU145 cells.
Now Kearney et al. has apparently been able, and I will use Alan Dove’s word’s here, “to detect XMRV in infected macaque blood (i.e. she can see it when it’s there) but unable to find it in any clinical specimens, including two that Lombardi et al. had reported as positive.” Now the titres in the infected monkeys would be massively higher than in human PMBC’s, with a natural infection pattern. That is artificial inoculation versus natural infection, and can be used to determine tissue tropism.
The question is, did Kearney use the Lombardi assay to see if she could locate XMRV, which would be adhering to the scientific method?