Background:  Xenotropic murine leukemia virus–related virus (XMRV) has recently been associated with human prostate cancer (PC) and chronic fatigue syndrome (CFS). However, other studies have failed to detect XMRV in PC and CFS patients. A human PC cell line, 22Rv1, produces XMRV that is virtually identical to virus isolated from PC and CFS patients. The 22Rv1 and CWR-R1 cell lines were derived from a human PC xenograft, CWR22, which was serially passaged in nude mice. To evaluate the genetic variation and evolutionary potential of XMRV, nucleic acid extracts of early and later passages of the CWR22 xenografts and CWR-R1 were analyzed.

Methods:  DNA isolated from early (3rd and 7th) and later passage CWR22 xenografts consisted of a mixture of tumor DNA and nude mouse DNA. Short tandem repeat analysis was used to confirm that the tumor DNA were derived from the same person as the 22Rv1 and CWR-R1 cell lines. XMRV-specific PCR primers and qPCR assays were developed and used for the analysis of xenograft and nude mouse nucleic acids. We explored the origin of XMRV by analyzing xenograft-associated nude mouse DNA and DNA from other nude mouse strains; the data revealed the presence of two previously undescribed endogenous proviruses, PreXMRV-1 and PreXMRV-2, which contained >3.2-kb stretches of their genomes with ~99.92% identity to XMRV.

Results:  PCR assays showed that both cell lines and later passage xenografts contained XMRV but the early passage xenografts did not, indicating that XMRV was not present in either the original CWR22 tumor or associated nude mouse tissue, but became prevalent in later passage xenografts. Retroviral recombination between PreXMRV-1 and PreXMRV-2 involving a few template switching events can generate a replication-competent virus that differs from XMRV by only 5 nucleotides (99.94% identity). 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.

Conclusions:  We conclude that XMRV was not present in the original CWR22 prostate tumor but was generated by recombination between PreXMRV-1 and PreXMRV-2 during in vivo passages of the CWR22 xenograft. The probability of an identical recombinant arising multiple times is vanishingly small, raising the possibility that contamination of human samples with XMRV originating from the 22Rv1 cell line is responsible for its reported association with PC and CFS.