ME/CFS is a common disease 
. It is often debilitating, but despite several decades of research its clinical manifestations still need to be more studied. All information which can contribute to the understanding of this disease is important.
The retroviruses which we study occur both in exogenous and endogenous form. The nonidentity between the human-derived XMRV and its closest relatives in the mouse genome is 5–7%, according to a BLAST search performed by JB (unpublished). Assuming that mice were the source of XMRV (a likely supposition) this small deviation, and the known high rate of variation in exogenous retroviruses, indicates a rather recent transmission to humans. Alternatively, all XMRV findings are due to contamination from a common source. The design of tests (PCR and serology) for infection with these viruses is highly dependent upon information regarding the possible spectrum of target viral sequences. A valuable resource is the not yet publicly available “RetroBank” 
. It is based on the computer program RetroTector© 
. The prototype of RetroBank currently contains c
a 40 000 retroviral sequences from 30 vertebrate genomes. The availability of this rich sequence source allowed us to evaluate current XMRV/HMRV nucleic acid based tests for expected range of retroviral detection. The result indicated that the gag
RTQPCRs described here, as well as the nested gag
should be able to detect most retroviruses related to MLVs in group G3 in the tree in and , i.e. they should have a similar detection range. The detection range of the INT RTQPCR 
seems to be much more narrow, which fits with the absence of amplification from mouse DNA with this RTQPCR. Nevertheless, the INT RTQPCR did not give a positive result, according to our criteria of repeatability with several different XMRV/HMRV specific PCRs. However, we got two initially weakly reactive results in two samples. The reason for the weakly reactive INT RTQPCR results is unknown. Despite that the nested gag
should have a similar detection range as both gag
RTQPCRs it did not become positive when samples weakly reactive with these RTQPCRs were retested with the nested gag
PCR. An explanation could be that the nested gag
PCR was not quite as sensitive as the RTQPCRs, which reached sensitivities of 1–10 target DNA copies, or that retroviral NA was somewhat degraded during subsequent freezing and thawing of the sample. The amplification range predictions indicate that the MLV related retroviral sequences, which are the ones reported in ME patients 
, should be detectable with the gag
RTQPCRs, whereas the INT RTQPCRs should be confined to the most XMRV-like targets, and should be less prone to false positivity due to mouse DNA contamination.
It should also be emphasized that a substantial portion of the 300 high scoring murine gammaretroviral proviruses belong to the here defined murine gammaretroviral groups G1 and G2. Both of these contain some proviruses which are completely intact and are strong candidates for being infectious. None of these should be detectable with current PCRs. Thus, we may not have seen the entire range of murine gammaretroviruses with zoonotic potential for humans.
The mouse retroviruses which have been reported to occur in humans are potentially pathogenic. This kind of viruses can give cancer (especially leukemia), encephalitis and immune deficiency in mice and other natural hosts. It is therefore logical to investigate their presence in patients with ME/CFS (encephalitis and immune deficiency) and in cancer (prostate cancer). However, the absence of a link to leukemia in humans is puzzling. Alternatively, XMRV/HMRVs could be passenger viruses without disease consequences. But the situation is interesting and should be followed up. Because of the great variation in results, methodological optimization is a high priority.
Contamination is an omnipresent hazard whenever supersensitive tests are employed. The three screening PCRs used here can detect 1–10 target molecules. There are three possible sources of contamination, i.e. false positivity in the PCRs, which should be considered, PCR amplimers, positive control DNA, and mouse DNA. In the first case, the controls for PCR amplimer contamination were non-template controls (PCR water; 1 to 4 samples per PCR round). None of them were positive. The three RTQPCRs are non-overlapping, and therefore cannot contaminate for each other. In the second case, the control for positive control DNA contamination, e.g. the XMRV VP62 clone, was sequencing. We sequenced amplimers from the few weakly reactive gag
and INT RTQPCRs. The gag
RTQPCR amplimers had exactly the sequence of the synthetic positive control DNA for the gag
RTQPCR, shown in . It had the sequence of XMRV VP62, but contained a characteristic 10 bp deletion. These weak reactions must have been due to contamination with this artificial DNA. Obviously, the number of non-template controls per PCR round was too small to detect this low frequency and low level of contamination. The sequences from the weakly and not repeatably positive INT RTQPCRs were identical to the target XMRV clone VP62 sequence for this PCR. The clone sequence is identical in many XMRVs. This makes it hard to distinguish true from false positivity due to contamination by sequencing. The third case, contamination with mouse DNA, was tested using the mouse mtDNA PCR and the IAP PCR. Likewise, MLV-like proviruses, high and low scoring, predicted to be detectable with most of the XMRV/HMRV PCRs, but not the INT RTQPCR which is rather strictly XMRV specific, occur in around 500 copies in the mouse genome (JB, data not shown). Therefore, even a slight mouse DNA contamination in samples subjected to XMRV/HMRV PCRs could cause a false positivity. Biologicals may contain much vertebrate DNA, which in its turn contains thousands of ERVs, possibly confounding sensitive and broadly targeted PCRs. Heparin, for example, contains DNA from the source animals (mostly pig and cow). The pig and cow genomes do however not contain proviruses expected to react in the PCRs used in this paper (JB, information from RetroBank 
). Moreover, patient samples (whole blood) analyzed in this paper were collected in EDTA tubes (PBMCs) or plain glass tubes (sera) and should not contain heparin. In a parallel titration, the sensitivities of the mouse DNA PCRs were lower than that of the gag
RTQPCRs. The absence of XMRV/HMRV positive results in our samples indicates that mouse DNA contamination was a small problem in our samples. However, the results of our bioinformatic search for MLV-like RV in vertebrate genomes, plus our experience of this issue, allows a few comments. There are similarities between the XMRV/HMRV and the “Human Retrovirus 5” (HRV5) stories 
. HRV5 is one of the so-called “rumor viruses” 
. It turned out to be a rabbit retrotransposon (RERV-H) whose DNA is abundant in rabbit sera 
. The rabbit genome contains around 700 copies of RERV-H 
. Any laboratory which handles rabbit sera is at risk of RERV-H contamination. In analogy with this, a low level of mouse DNA could be present in laboratory reagents or in the laboratory environment 
. However, non-template controls should also be positive then. They were not. A source of mouse DNA in our laboratory is uncertain and unlikely. As shown here, the likelihood of mouse DNA contamination causing reactions in the INT RTQPCR, where we had a few weak reactions, is lower than that of the other RTQPCRs.
The mouse genome contains three groups of high scoring gammaretroviral proviruses. Some of them may have zoonotic potential. The third group contains the highest proportion of structurally intact proviruses, and is the target of most PCRs used to detect XMRV/HMRV. Two new broadly targeted XMRV/HMRV PCRs were developed. They were employed on samples from ME/CFS patients and blood donors. False reactions due to contamination with synthetic target DNA were encountered. These could be classified as false by sequencing. The few remaining reactions did not fulfill our criteria for positivity, because they were not repeatable. The few weak and uncertain PCR reactivities encountered by us are very different from the high detection frequencies reported by others 
. It is possible that a higher amount of nucleic acid used per PCR could have given a higher frequency of XMRV/HMRV detection. However, many samples contained levels of nucleic acids similar to those used in other studies 
. Under the conditions used by us, we could not corroborate that XMRV/HMRV is frequent in Swedish ME/CFS/FM patients and blood donors.