During the last year an enormous amount of data about the origin and the prevalence of XMRV have indicated that this virus has a recombinant origin and it is not circulating in the human population 
, although many questions about the biology and physiopathology of this virus remain still unclear.
Despite all these data and considering, i) the susceptibility of humans cells to the XMRV infection 
, ii) the contradictory data on experimental infection of macaques by XMRV 
, iii) the high-titer production of XMRV by the 22Rv1 cell line, widely used in laboratory, and iv) the existence of XMRV/human contacts in laboratory personnel involved in cell culture facilities, it could be relevant to develop new experimental models for the study of XMRV pathogenesis in humans alternative to the use of non-human primates.
The main objective of the present study was to investigate whether the human lymphoid tissue might be a target for XMRV derived from 22Rv1 supernatant. We used histocultures of human tonsillar tissues that have been reported as a useful model for studying various aspects of the pathogenesis of other viruses 
. The ex vivo
infection with XMRV allowed us to assess the replication of the virus, in addition to the pathogenic effects of this infection in human lymphoid tissue.
Tonsillar tissue after 14–22 days of infection become efficiently and specifically infected by XMRV. In contrast to previously described findings using in vitro
infected PBMCs, the proviral copy numbers in our ex vivo
infected histoculture increased overtime from day 2 to day 22 which could indicate that new target cells become infected throughout the culture. This increase in the proviral DNA content was concomitant to the increase in XMRV RNA released into the culture medium, suggesting that this XMRV RNA is associated with infectious particles that may lead to the spread of the infection. When the presence of infectious viral particles in the supernatants of infected tonsils was evaluated, we observed that we could recover replication-competent XMRV from these supernatants by infecting the indicator DERSE XMRV cell line. In contrast, those supernatants were unable to establish an infection in fresh lymphoid tissue, suggesting that with the passage of the virus through several rounds of infection, the innate antiviral restriction factors may abrogate its infectivity. Nevertheless, it should be noted that these supernatants have a much lower amount (2 to 3 log-copies of RNA/mL less) that the 22Rv1 supernatant used as a positive control of infection. Indeed, a high viral titer is required to establish an infection in human PBMCs 
, so it is possible that viruses released by tonsils, although infectious in the DERSE assay, did not reach a sufficient titer to establish a new histoculture infection. In addition, the increase in proviral copy numbers, both in the tissue and the migrating cells, could be fully explained by cell divisions of infected cells, which in turn would lead to an increase in the level of released XMRV RNA.
Although the absolute levels of XMRV RNA released varied from donor to donor, the replication was similar to those observed in tonsillar tissue infected with some strains of HIV (R5 strains) (ranging between 2 and 5×106
RNA copies/mL) 
, but lower than viral production of more pathogenic strains of HIV (×4 strains), which range from 15 to 60×106
RNA copies/mL 
. However, for HIV has never been reported whether this amount of virus produced is sufficient to establish a new productive infection in a new histoculture.
Our observation that we could not establish a new infection in tonsillar tissue may indicate that, in ex vivo
infected human lymphoid tissue, the virus is restricted by innate host restriction factors such the APOBEC family, since hypermutation, the introduction of excessive G-to-A substitutions by those host proteins that impair the viral replication, has also been observed in our experimental model. The sequence analysis of our clones showed that 19% of the sequences were hypermutated. This percentage, although lower than the 77 and 65% of hypermutated sequences found with the infection of two APOBEC 3G/3F positive cell lines (CEM and H9, respectively) 
, is similar with the level of hypermutagenesis described in the infection of PHA-stimulated PBMCs 
. In addition, the mutation frequencies obtained from our sequences (3×10−3
/nt) were higher than those found in 22Rv1 proviral DNA (4.9×10−4
/nt) or with the infection of an APOBEC 3G/3F negative cell (CEM-SS) (8.4×10−4
/nt), although they were similar to those found in the infection of H9 cells (6.6×10−3
. Moreover, although G-to-A mutations can occur in one of four different dinucleotide contexts (GG, GA, GC or GT), APOBEC 3G induces twice as many GG-to-AG as GA-to-AA changes 
, while APOBEC 3F primarily induce GA-to-AA 
. In tissue, most of the G-to-A changes occurred in a GG dinucleotide context. Overall, our data indicate that XMRV infection would be restricted, mainly by APOBEC 3G, as described in in vitro
infected PBMCs or human cell lines. Nonetheless, in lymphoid tissue, the virus could somehow be able to evade or overcome these restriction factors, as a low amount of infectious virus is released. There are various non-excluding explanations for this observation. The most obvious conjecture could be that the level of APOBEC 3G is lower in lymphoid tissue than in peripheral PBMCs and not enough to block virus replication. However, all APOBEC family members are expressed widely in hematopoietic cell populations and in tissues, particularly in tonsils 
. Secondly, despite the fact that our data show that mechanically isolated lymphocytes from tissue and migrating cells are XMRV DNA+
, the cell population responsible for the viral production remains unknown. In our histocultures, XMRV could infect other cell types that do not express these restriction factors, being a non-lymphoid cell the major viral target as it was shown in prostate cancer tissues 
or in tissues from experimentally infected monkeys 
. Further work is necessary to define the cellular population that would sustain this low viral production in ex vivo
In ex vivo
HIV infected tonsils, virus replication causes a profound depletion of the viral target cells (CD4+
T cells). In our histocultures, we did not observe changes in the percentages in any of the major lymphocyte populations or modifications in the release of inflammatory chemokines. Yet, in XMRV infected macaques the virus did not induce any lymphocyte depletion, rather there was an increase in the frequency of circulating B and NK cells in blood 
In conclusion, in the absence of any confirmed human XMRV infection and with the conflicting results of infection in animal models, which may not accurately mimic human XMRV infection, in our ex vivo cultured human lymphoid tissue, XMRV is able to infect tissue cells and produce infectious viruses, even though they were unable to establish a new infection in fresh tonsillar tissue. The XMRV replication is largely controlled by innate antiviral restriction factors and probably further contained by the adaptative immune response, although the long term presence of proviral DNA and viral RNA in the lymphoid tissue remains absolutely unknown. Hereby, laboratories working with XMRV producing cell lines should be aware of the potential biohazard risk of working with this replication-competent retrovirus.