There are many ways in which a PCR assay could be contaminated - that is well known. The thing is: Both the WPI and the FDA took appropriate measures so that if there would be contamination there, they would know. The WPI included negative controls in every experiment, they checked for mouse mtDNA, and perhaps their most strong evidence is that they have shown antibodies to a gammaretrovirus. Coffin can say all that he wants about these antibodies potential to be because of a friend MLV infection or something like that, but not only that it does not really matter (these are both MLVs, and as I understand it, the treatment would probably be the same) but also that it would be a very strange coincidence that while they were finding XMRV that was due to contamination, they, by chance, found F-MLV or whatever MLV that was really in the person.
The FDA also took appropriate measures to see that it is not contamination: They talk about contamination of the reagents, and that therefore their negative controls came up positive many time. Well, Dr. Lo used hundreds of negative controls, and they never, not even once, came up positive in the assay. They talk about presence of 22Rv1 DNA/RNA? Well, they might even say that this DNA/RNA got into Lo's patient samples while they were stored or processed, and much less to the healthy controls (which would be strange - it did get there, but in much less percentage?) - but how can they explain that when they took fresh samples 15 years later from 8 patients, 7 of them came up positive again? Does the contamination prefer the blood of the sick ones? Not that I'd expect that 22RV1 would contaminte Lo's samples with PMRVs, as it contains XMRV that is almost identical to VP62. Regarding the WPI's samples, again - they found antibodies, which are not produced, to my knowledge, in the lab - and moreover, as Dr. Mikovits said, they checked and as far as one can know, the 22Rv1 cell line (which does not exist in real human beings) was never present in the state of Nevada, ever (and their samples were taken and stored in the state of Nevada).
Next: Mouse DNA? Lo used an mtDNA assay that was extremely sensitive - more sensitive than Coffin's extremely sensitive IAP assay - and he also used Coffin's IAP and improved-IAP assays, and didn't find any mouse DNA contamination. As Dr. Lo showed, his mtDNA assay was 100 times more sensitive than his PCR assay with which he detected the PMRVs, so it makes no sense whatsoever that those were the results of mouse DNA contamination. Regarding the WPI's study, they also used an mtDNA assay on their samples and cell lines and found not contamintion, and again - their findings of antibodies a very, very strong piece of evidence.
So, regarding every suggesment for contamination there is an excellent answer for why it can't be a contamination.