False negative results from using regular PCR reagents!

This is actually a new publication from our own lab, one of those "why did you run that control, and why did it turn out like that?" kind of moments, which led us to figure out that there are some serious problems with standard PCR techniques.  So what happened? We were testing human DNA samples to detect contaminating mouse DNA (recent reports have suggest mouse DNA gets into just about everything...including DNA isolation kits...) -- so we spiked some human DNA with varying amounts of mouse DNA to check our assay sensitivity-- then we wanted to test if carryover PCR product contamination (from previously generated reactions) could be eliminated by using a master-mix with uracil-DNA-glycosylase (UNG) -- it should do, because that's what is for -- so someone in the lab tested this, by adding mouse PCR product to the human DNA, but also added mouse DNA as well.  Huh! that reaction shouldn't tell us anything, because even if the PCR product was eliminated by the UNG, the mouse DNA would still amplify. Except that it didn't.  Even when the amount of PCR product added was minute, equivalent to femtoliters, and the mouse DNA was present in huge quantities (60ng) -- it still didnt' amplify!  We looked into this further to find that in fact, even primer-dimers from previously generated reactions will effectively inhibit new PCR reactions, even when the legitimate target is present in high copy number - who knew?  But how is this important?  Like all assays, its important to know what you are dealing with, the strenghths, and weaknesses of the technique -- in this case, if your target is limiting, you could have a huge problem, notably false-negative results.  We use the currently highly controversial example of murine leukemia virus-type sequences, that are either present (or not) in a variety of human tissue and serum samples, depending on who you talk to.  As we show in our paper - even two million copies of this virus can go undetected, if as much as 1 in 10000th of a microliter of a previously generated negative control PCR reaction contaminates the sample.  Maybe no one gets contaminated ever.  But as any lab that has ever tried to do nested PCRs  would likely attest to -- it is virtually impossible to never get contamination with previously generated reactions -- especially at the levels we are talking about here - anyway, read all about it, by clicking here (it's free!) http://www.biomedcentral.com/1756-0500/4/457/abstract