22Rv1 cells were obtained from the American Type Culture Collection (ATCC) and Detectors of Exogenous Retroviral Sequence Elements (DERSE) XMRV indicator cell line was kindly provided by Dr. Vineet N. KewalRamani (National Cancer Institute, Frederick, Maryland, USA). Both cell lines were maintained in RPMI 1640 medium with 10% FBS (Invitrogen, Spain). Cell-free XMRV was harvested from 22Rv1 supernatant, clarified by centrifugation; aliquots were stored at −80°C until used. Zidovudine (AZT) was purchased from Sigma-Aldrich (Spain) and Raltegravir (RAL) was obtained through the AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, NIH; from Merck & Company, Inc.

Human tonsils were obtained from patients undergoing therapeutic tonsillectomies. All procedures followed the Helsinki Declaration in 1975, as revised in 1983, and were approved by the Ethics committee of the Hospital Germans Trias i Pujol. All individuals provided their written informed consent. The specimens were dissected into ∼2–3 mm³ blocks, and cultured in RPMI 1640 medium supplemented with 15% fetal bovine serum, 1 mM sodium pyruvate (Invitrogen Life Technologies, Spain), 0.1 mM minimal essential medium with nonessential amino acids (Invitrogen) and a mixture of antibiotics on top of collagen sponge gels at the medium-air interface as previously described [88].. Nine individual blocks were placed on the gelfoam in a well of a six-well plate. Lymphoid tissue blocks were left uninfected or were infected with 5 µL of XMRV stock (4×10⁸ RNA copies/mL) for each tissue block. Cultures were conducted in the absence or the presence of reverse transcriptase or integrase inhibitors, AZT (10 µg/mL) or RAL (5 µg/mL), respectively. Fresh drug was added at every medium change. Each experimental condition was composed of three wells for a total of 27 blocks whose culture media were pooled. Culture medium was harvested every 3–4 days, centrifuged and both, the pellets with the cells migrating out of the tissue, and the clarified supernatant were frozen at −80C until use. On days 14 to 22 after infection, cells were mechanically isolated from uninfected and infected tissue blocks and immunophenotyped. Cell Pellets were also stored at −80°C for further analyses.

Viral infection was evaluated at different time points in the cells migrating out of the tissue and in tissue cells at the end of the culture by analyzing the presence of viral DNA by conventional PCR. DNA was isolated using QIAamp DNA Blood kit (Qiagen, Spain) according to the manufacturer's instructions, and gag gene was amplified as previously described [20]. Briefly, PCR was conducted with 5 µL of extracted DNA and the primers 419F and 1154R using Accuprime Taq DNA Polymerase (Invitrogen, Spain). The cycles were 10 min at 94°C (30 sec at 94°C, 30 sec at 57°C, 1 min 68°C)×40 cycles and 10 min at 68°C. Then, samples were electrophoresed in 2% agarose gels containing SYBR safe DNA in TAE buffer.

To quantify the absolute XMRV DNA copy numbers, standard plasmids for XMRV-gag and for the single-copy CCR5 gene were used. Both plasmids were constructed by conventional PCR using AmpliTaq High Fidelity (Invitrogen, Spain) and the following primers: Q445F and Q528R previously described for the detection of XMRV gag [6] and CCR5-576F (5′-TCATTACACCTGCAGCTCTCATTT-3′) and CCR5-726R (5′- ACACCGAAGCAGAGTTTTTAGGAT-3′) for CCR5 detection. The two PCR products purified with QIAquick PCR Purification Kit (Qiagen, Spain) were cloned into a PCR-Script Amp Cloning Kit (Agilent, Spain) according to the manufacturer's instructions. The plasmids were purified by standard methods, their concentrations were measured using a fluorometer and the corresponding copy numbers were calculated. Quantitative real-time PCR was performed with DNA isolated from the cells migrating out the tissue at different time-point during the culture and in tissue cells at the end of the culture in duplicate, using a TaqMan Universal Master Mix (Applied Biosystems, Spain), the primers described above and the probes: F480PRO; (5′-FAM-ACAGAGACACTTCCCGCCCCCG-MGBNFQ-3′) and CCR5-661T (5′-VIC-CTGGTCCTGCCGCTGCTTGTCA-TAMRA-3′) to detect XMRV gag and CCR5, respectively. A ten-fold serial dilution of the two standard plasmids, ranging from 1×10¹⁰ to 1×10⁻¹ copies/reaction was used in duplicate to construct standard curves. All real-time PCR reactions were performed on a ABI Prism 7000 (Applied Biosystems, Spain).

Quantitative measurements of XMRV viral RNA in culture supernatants were done using a quantitative real-time RT-PCR assay. Total RNA was extracted from 280 µL of culture supernatant at different time points during the culture (QIAamp Viral RNA kit, Qiagen, Spain) and one-step quantitative reverse transcription PCR was performed using the Ag-path-ID One-Step RT-PCR Kit from Ambion (Applied Biosystems, Spain) using the primers, probes and standard curves described above.

PCR products obtained by conventional PCR from tissue cells (see above) were cloned using the PCR-Script Amp Cloning Kit. Over 10–30 colonies were selected for each product and sequenced using the Big-Dye Terminator Cycle Sequencing and the ABI 3100 sequence analyzer (Applied Biosystems, Spain). All sequences were assembled, aligned and edited using the Sequencher v.4.2 and GeneDoc v.2.6.001 software. The alignments were uploaded to the Hypermut tool on the Los Alamos National Laboratories website (http://www.hiv.lanl.gov/content/sequence/HYPERMUT/hypermut.html) to identify hypermutated sequences [91].

To analyze infectivity of supernatants from XMRV-infected tonsils we used two different approaches. First, infection of the DERSE XMRV indicator cell line was performed. The development of DERSE cells involved the transfection of LNCaPs cells with a MLV reporter vector encoding puromycin resistance and a CMV enhancer/promoter driven GFP reporter gene whose transcription was antisense to the vector mRNA (KewalRamani VN, unpublished data). DERSE XMRV indicator cells were plated at a density of 40,000 cells/well in 48-well plates and allowed to grow overnight. Next day, the cells were infected with culture supernatant (50 µL) of previous tonsils infected and harvested at days 16 and 22 post-infection (donor 402). As a positive control we infected one well with the 22Rv1 supernatant stock. The culture was splitted at days 3, and 7 and maintained for 10 days. The percentage of GFP-expressing cells was evaluated by fluorescence microscopy and flow cytometry. In addition, new tonsils histocultures were re-infected with supernatant harvested at day 16 and 22 (donor 401 and 402) from previous histocultures infections. The infection was done at day 0 as described above although here the tissue was re-infected with the same amount of virus at day 1 of culture. The culture was maintained 28 days and culture medium, cells migrating out of the tissue and tissue cells were harvested as previously described.

Single cell suspensions were prepared as described above. Tissue lymphocytes at the end of the culture were immunophenotyped using monoclonal antibodies against the following surface antigens: CD3, CD4, CD8, CD45RO, CD38 and HLA-DR for the analysis of T cell, and CD19, IgD, CD38, CD27, from Becton Dickinson, Barcelona, Spain) and CD10 for B cells (from eBioscience, Barcelona, Spain). Multicolor analysis was performed with a LSRII flow cytometer (Becton Dickinson). Data were acquired with the BD FACSDiva software and analyzed with Flow Jo software.

Culture supernatants clarified by centrifugation and collected at different time points during culture were assayed using a Cytometric Bead Array kit (CBA, BD Biosciences, Spain). Chemokines (CXCL8/IL-8, RANTES, CXCL9/MIG, CCL2/MCP-1, and CXCL10/IP-10) were measured using the Human Chemokine kit according to the manufacturer's instructions using a LSRII flow cytometer (Becton Dickinson).

To evaluate the ability of XMRV to infect human lymphoid tissue, infectious XMRV harvested from prostate carcinoma cell line 22Rv1 was used to infect tonsillar explants ex vivo. After 14 days post-infection, we analyzed the presence of XMRV gag proviral DNA in tissue blocks. Cells were mechanically isolated from uninfected and infected tissue and the presence of XMRV gag proviral DNA was clearly observed in infected tissue cells by conventional PCR at the end of each experiment (Figure 1A). The specificity of the infection was evaluated using AZT and RAL, two antiretroviral drugs known to inhibit XMRV reverse transcriptase (RT) and integrase, respectively [92], [93]. Both drugs completely inhibited XMRV DNA detection in tissue cells, indicating that infection is the result of specific reverse transcription and integration processes (Figure 1A). To assess the kinetics of the infection we quantified the XMRV proviral DNA in cells migrating out the tissue, which were collected every 3–4 days. After 3 days of culture, migrating cells were already XMRV positive and remained positive until the last day tested (Figure 1B). The RT inhibitor AZT markedly inhibited the presence of XMRV proviral DNA in all time points. Conversely, modest effects were observed after 7 or 14 days in the presence of the integrase inhibitor RAL (Figure 1B). Cells from the mock-infected tissue were negative in all cases.

Once we confirmed the entry and the reverse transcription of XMRV in the human lymphoid tissue, we accurately evaluated the level of the infection by setting up a quantitative real-time PCR to measure the absolute XMRV DNA content. Total XMRV gag copies per 1×10⁶ cells were obtained for several tissues by CCR5 normalization after 22 days of infection (Figure 2). A clear increase in the amount of proviral DNA present in tissue cells was observed when comparing with the histocultures infected in the presence of antiretroviral drugs. The treatment with AZT and RAL resulted in almost complete inhibition (>97%) at most time points and donors tested (Figure 2A). In migrating cells, the increase in the infection was detected after 9–10 days of culture, although the proviral DNA content underwent a significant increase at later points (after 16 days post-infection, Figure 2B). The presence of AZT in the culture reduced proviral DNA level to negative values from day 3 to the end of cultures. However, in the presence of RAL, proviral DNA values were comparable to the infected tissue without drug and remained stable until day 7–13, declining at the last days of culture (Figure 2B). This different behaviour in the inhibition between both drugs could be explained by the fact that our PCR system detected not only the integrated DNA, but also unintegrated DNA. AZT would totally block viral DNA synthesis, resulting in decrease of both total and integrated DNA. On the contrary, when an integration inhibitor like RAL is used, unintegrated viral DNA accumulates in the nuclei in the form of 1-LTR and 2-LTR circles [94]. These episomal forms may explain the detection of XMRV DNA at an almost constant value over time in our infected tissue in the presence of RAL.

The observed increase in viral DNA content in both, tissue and migrating cells, indicated that XMRV might establish a productive infection in our histoculture model. To evaluate this possibility, XMRV RNA released into the medium, starting on day 5 post-infection, was quantified in culture supernatants by using real-time qRT-PCR. The cumulative production in the infected cultures from day 9 to day 22 post-infection is shown in Figure 3A. XMRV RNA was mainly detected in infected tissues that were cultured without antiviral drugs. The total production of viral RNA in AZT and RAL containing cultures constituted only the 8 and 4%, respectively of the infected tissues without drug (Figure 3A). The evaluation of replication kinetics revealed a decrease in RNA viral levels up to day 9 in cultures with or without drug, representing the washing out of the starting viral inoculums (Figure 3B). Thereafter, RNA virus release experienced a continuous increase over time in infected tissue without drug from day 9 to the last day of culture, reaching levels up to 2×10⁶ copies/mL after 22 days of culture. This increase in RNA virus release, ranging from 16- to 21-fold in the different tested donors, was completely blocked by AZT or RAL (Figure 3B). These findings could indicate that XMRV production, which is sensitive to RT and integrase inhibitors, occurs in tonsillar tissue.

A severe restriction in the spreading of the XMRV infection has been shown in PHA-activated human PBMCs [54]. Since human lymphocytes express restriction factors, such as APOBEC 3G and 3F, [84], [95] we sought to determine whether these factors impacted on XMRV replication in tonsils. DNA from infected tissue cells at the last day of the experiment was isolated and a 731-base pair fragment of the XMRV gag region was amplified. The PCR products were cloned and their DNA sequences were compared to the XMRV VP42 consensus. A total of 80 clones derived from 5 different donors were analyzed (Figure 4A). An initial analysis revealed that 46 sequences (57.5%) had mutations, although only 32 (40%) of them harbored more than one mutation. To identify the hypermutated sequences we used the LANL Hypermut 2.0 program [91], using a p-value<0.05. Compared to XMRV VP42, a total of 15 (18.8%) sequences were identified as hypermutated, with a range of 0 to 31.2% among different donors. In total, 180 changes were detected with a mutation frequency of 3×10⁻³/nt (180 mutations for 58480 nucleotides sequenced) of which 142 were G-to-A changes (Figure 4B). Substitutions in the GG dinucleotide context (GG-to-AG), which is the specific target site for APOBEC 3G [81], [82], were more prevalent (76%) than in the APOBEC 3F-preferred context (GA-to-AA) (20%), suggesting that XMRV in our ex vivo infection is mainly restricted by APOBEC 3G (Figure 4). At the amino acid level, mutations caused both, changes in the amino acid sequence and introduction of stop codons.

To further characterize the XMRV infection in tonsillar tissue, we investigated possible pathogenic effects. It has been shown that HIV-1 severely depletes CD4⁺ T-cells in ex vivo infected human lymphoid tissue [88], [89], with a depletion of 77 and 97% of CD4⁺ T-cells relative to uninfected samples after 8 and 13 days post-infection, respectively [96]. Therefore, we addressed whether XMRV replication may alter tissue lymphocytes. Despite the apparent infection of the tissue, we did not observed changes in the percentage of the main lymphocyte populations after 9 days of infection (Figure 5). XMRV infection did not modify the percentage of CD3⁺ T-cells (63±11% and 64±8% in XMRV⁻ and XMRV⁺ tissue, respectively), nor CD4⁺ T-cells (47±7% vs 48±5%), nor CD8⁺ T-cells (20±12% vs 19±11%). The percentages of B cells were also similar between both tissues, CD19⁺ cells (29±16% vs 27±14%) (Figure 5). A deeper analysis of T-cell subsets showed that XMRV infection did not modify the naïve/memory cell ratio, in both T and B cells, nor immune activation markers, evaluated by the expression of CD38 and HLA-DR in both CD4⁺ and CD8⁺ T-cells (data no shown). None of these parameters was modified after 28 days in culture (data no shown).

In human lymphoid tissue infected with other viruses such as HIV and HHV-6 a modulation in the chemokine secretion by viral replication has been documented [90], [97], [98]. To address this possibility, five chemokines from a designed Human Chemokine kit, some of them previously described to be altered in patients with CFS [99], [100] were measured by using a quantitative cytometric bead array on the supernatants collected during the course of 22 days of viral infection The total production of those chemokines did not significantly change by XMRV infection in the absence or presence of antiviral drugs (Figure 6).

The increase in DNA and viral RNA together with the presence of hypermutated sequences suggests an active replication. To evaluate if the supernatant of infected tonsils histocultures contained replication-competent virus, we performed a viral infectivity assay using DERSE XMRV indicator cells, which were infected with those supernatants. The indicator cell line DERSE XMRV will only produce GFP following infection with a replication-competent virus since the gfp gene is functionally reconstituted upon reverse transcription. DERSE cells were infected with 50 µL of supernatants harvested at day 16 and 22 (donor 402, Figure 3B) which contained 23,000 and 99,000 copies of XMRV RNA, respectively and with XMRV from 22Rv1 supernatant stock (40,000 copies of RNA) and the GFP expression was monitored by fluorescence microscopy every 2–3 days of infection. GFP⁺ cells were detected in the control culture (infected with 22Rv1 supernatant stock) as early as 3 days (data not shown) and a high number of cells were positive at day 7 post-infection (Figure 7A). At this time, GFP expression was also observed when cells were infected with the supernatants of infected tonsil, (Figure 7A middle and lower panels). No GFP expression was detected in mock-infected cells. To quantify the percentage of cells infected with each supernatant, DERSE cells were harvested and analyzed by flow cytometry at different days after infection. At 7 days post-infection a high number of GFP⁺ cells (28%) were detected when a 22Rv1 supernatant stock was used (Figure 7B). In the infections with the supernatants from day 16 and 22, and in agreement with the data observed by microscopy, a lower number of GFP⁺ cells were detected (4% and 7%, respectively), although the levels were significantly higher than those observed in uninfected wells (Figure 7B) at day 7. Following culture passage, an increase in the number of GFP+ cells was observed in all the infections, reaching similar values in the infections with both supernatant than in the infection with the 22Rv1 stock (21% and 27% of GFP+ cells with supernatants from day 16 and 22, respectively, and 29% of GFP+ cells with 22Rv1 supernatant. Figure 7B). The results from this viral infectivity assay suggested that XMRV viral particles, which were able to replicate and spread were presents in the tonsillar supernatants.

Furthermore, we analyze the ability of tonsillar supernatants to re-infect a new histoculture. In order to maximize the infection, tissue from two different donors was infected at day 0 and re-infected with the same amount of virus at day 1. We use each time 45 µL (5 µL for each tissue block) of XMRV 22Rv1 stock (4×10⁸ copies/mL), donor 402 supernatant day 16 (4.7×10⁵ copies/mL) and donor 402 supernatant day 22 (2×10⁶ copies/mL) (see Figure 3A). After 28 days of infection we observed a very large number of copies of proviral DNA in tissue cells infected with 22Rv1 supernatant and almost a total inhibition with AZT. However, none of the tonsil supernatants used to re-infect the tissue was able to infect this new tissue and only a residual amount of proviral DNA (even lower than that detected in the infection of 22Rv1+AZT) was observed, mainly in the supernatant from day 22 (Figure 8A). Consistent with the lack of detection of proviral DNA in tissue cells, we found only a residual level of both, proviral DNA in cells migrating out of the tissue (Figure 8B) and XMRV RNA released into the medium (Figure 8C).