Total protein lysates were prepared from all 60 cell lines. Viral envelope protein (Env, also known as gp70) was detected by immunoblotting with monoclonal antibody 7C10 [61], which was raised against the envelope protein of Friend spleen focus-forming virus and which cross-reacts with XMRV and other X-MLV envelope proteins. A positive signal was detected in the lysate of only 1 cell line, EKVX (Table 1 and Figure 1A), a lung adenocarcinoma– derived cell line [26]. As a second means of screening for virus, PCR for XMRV and X-MLV– related gag and env sequences was performed. In concordance with the immunoblotting results, only EKVX yielded prominent bands of the expected size with both the env and gag primers, 533 bp and 731 bp, respectively (Table 1 and Figures 1B and 1C). This result is consistent with Hue et al. [60], who detected MLV-like sequences in EKVX by PCR. To rule out the possibility that EKVX cultures or derived materials were contaminated with mouse cells or DNA containing endogenous X-MLVs that could give rise to the env and gag products, we used nested PCR as described by Ono et al. [62] to test for the presence of mouse mitochondrial DNA. Although we could reliably detect 1/10⁶ equivalents of mouse genomic DNA spiked into human genomic DNA, no mouse mitochondrial PCR products were observed with up to 250 ng of EKVX genomic DNA (data not shown). We were also unable to detect mouse DNA contamination using a single-round (45 cycles) PCR assay for intracisternal A particles (IAP) [63] on up to 600 ng of EKVX genomic DNA (Figure 1D). The mouse IAP products are ~200-300 bp; the ~100 bp band in the HCT-116 lane may be a primer dimer product, as it was observed frequently in no template controls in additional experiments.

We observed a faint ladder of background bands in all cell lines with the env primers. Therefore, we tested a subset of first round env products with nested primers, and a band of the appropriate size was only detected with EKVX (data not shown). The first round background bands were not observed in two representative cell line DNA samples (HCT-116 and DU145) spiked with a small amount (1/10⁶) of genomic DNA containing the target DNA sequence (data not shown). Because no mouse DNA was detected in HCT-116 or DU145 genomic DNA using the IAP assay (Figure 1D), the background bands may have been the result of mispriming on human DNA sequences. A few cell lines gave rise to products of incorrect size with the gag primers (e.g., HCT-116 in Figure 1C), but no corresponding envelope PCR product was observed, and no product was detected using nested primers for gag (data not shown).

Lung can serve as a point of entry for virus infection. A search of the GEO database [64] reveals numerous examples of expression of the X-MLV/XMRV receptor XPR1 [65] in human lung cells. Although potential contamination with mouse sequences was not ruled out, XMRV sequences were reported in human respiratory secretions [66], suggesting that lung could be a target for MLV-like viruses. However, all 8 of the other lung cancer cell lines in the NCI-60 panel and 4 additional lung cancer cell lines [NCI-H1666, NCI-H650, NCI-H2122, NCI-H358; obtained from M. Lerman, NCI (data not shown)] that we tested by immunoblotting were negative for MLV-like virus.

Cell-free supernatants were prepared from cultures of EKVX cells. Reverse transcriptase activity was detected in the EKVX supernatant using manganese, but not magnesium (data not shown), indicative of a mammalian type C retrovirus reverse transcriptase [67]. The ability of the EKVX viral supernatant to infect and spread in human cells was assessed using the human prostate cancer cell line LNCaP. LNCaP cells pretreated with polybrene were incubated with cell-free EKVX supernatant or fresh media (negative control). Cells were subcultured and total protein lysates were prepared at 10 and 16 days post-infection. Analysis by immunoblotting with monoclonal antibody 7C10 and with goat anti-Rauscher MLV p30 indicated the presence of capsid and envelope viral proteins in the lysate (Figure 2), indicating the establishment and spread of infection from the EKVX supernatant.

The nucleotide sequences of the gag and env PCR fragments amplified from EKVX genomic DNA were determined. Sequences of the EKVX gag and env fragments were more closely related to X-MLVs than to XMRV (98% vs. 87% identity for gag; 99% vs. 92% identity for env). Based on these results, primers were designed to amplify additional regions in the LTR, gag, and env based on the sequence of DG-75 X-MLV. The sequences of these PCR products were determined and are aligned with representative XMRV and X-MLV sequences in Figure 3. The LTR-gag region of the EKVX virus lacks several of the defining characteristics of XMRV while aligning with 99% identity with the X-MLV sequence; most notably, it lacks the 14 bp deletion found in U3 and the short additions and 24 bp deletion found in the gag-leader of XMRV (Figure 3A). The primer binding sites for the EKVX sequence and the X-MLV sequence shown in Figure 3A are complementary to Gln-tRNAs, whereas that for XMRV is complementary to Pro-tRNA [68]. Gln-tRNA binding sites have been reported for numerous endogenous MLVs [69,70,71]. Also of note within the gag-leader region, MLV sequences contain an alternative (CTG) start codon, upstream of the ATG start codon, that yields a glycosylated precursor (glyco-Gag) [72]. In contrast to XMRV, which has an in-frame stop codon that precludes glyco-Gag synthesis [55], there is no stop codon between the CTG and ATG gag start codons and they are in frame with each other, suggesting that the EKVX virus should produce glyco-Gag. The env region (Figure 3B) lacks the trinucleotide G insertion found in XMRV, and the env-LTR region (Figure 3C) is missing the same 14 bp deletion in U3 as noted in Figure 3A. Together, these sequence data confirm that the EKVX virus is more closely related to X-MLVs than to the reported sequences of XMRV.

Xenotropic MLVs are so named because they can infect non-mouse species but cannot infect most laboratory strains of mice. X-MLV sequences are present as endogenous viruses in all laboratory and some house mouse strains [73,74,75,76,77], and endogenous viruses can be activated to produce infectious virus under certain conditions such as chemical or radiation exposure or immune induction [78,79,80,81,82]. Because EKVX was derived by passage through a nude mouse [26], the virus in the cell line may have been present in the original tumor or may have been acquired from nude mouse cells producing X-MLV. Several instances of human tumors and cell lines passed through nude mice having acquired viruses from the host have been reported [83,84,85,86,87,88,89]. However, not all cells passaged in mice acquire a virus. Indeed, four other cell lines of the NCI-60 panel are known to have been established after passage through mice (Table 1), and these tested negative for X-MLVs. The propensity for acquisition of a virus may depend on the strain of mouse, the means of immune suppression, the characteristics of the xenografted tumor, and whether the experimental protocol includes any additional factors (chemical exposure, radiation) that could promote the activation of an endogenous virus. Suzuki et al. [86] found that cells from 6/9 tumors transplanted into nude mice produced infectious murine type-C virus. Lusso et al. [85] found that cells recovered from all 6 human hematopoietic tumors studied that were transplanted into splenectomized, irradiated, and anti-asialo-GM1–treated nude mice acquired X-MLV infections. In other studies in which a variety of solid tumors were implanted as xenografts in nude mice or mice treated with mouse thymocyte antiserum, only 1/9 [87], 1/12 [84], and 2/11 [88] xenografts were associated with type C retroviruses. Intriguingly, Tralka et al. [88] reported IAP production in the human osteosarcoma cells that had acquired X-MLV infection. In a recent study, Zhang et al. [89] found evidence that 6/23 (mouse DNA-free) human cell lines established following passage through mice had been infected with X-MLV. Because of the possibility that the EKVX human lung cancer cell line acquired its X-MLV upon passage through nude mice, our study sheds no light on the controversy of whether X-MLV infections occur naturally in the human population.

The identification of a human infectious retrovirus in a commonly handled cell line is of concern on multiple counts: 1) unknown risk of infection and health risks to persons exposed to the virus, 2) unanticipated influence of infection and expression of viral proteins on experimental results [85,86], and 3) risk of spread of infectious virus to other cultured cell lines [89]. Infection of cells with X-MLV significantly altered the interactions of HIV-1 with those cells compared to their uninfected counterparts [85]. With respect to the potential for unintended spread of the virus among cultures, two reported examples of X-MLV in human cell lines that had not been transplanted into mice appear to have resulted from vertical transmission of virus from another infected cell line [90,91]. Zhang et al. [89] found that non-xenograft cell lines maintained in a xenograft-free facility showed no evidence of MLV infection, whereas 17% of non-xenograft cell lines cultured in laboratories that also maintained xenograft cultures became infected with MLV. Our study indicates that the X-MLV present in the EKVX cell line has not spread to other cell lines of the NCI-60 panel maintained by the DTP.

Our results confirm the findings of Sfanos et al. [92], who published the results of a similar study while this manuscript was in preparation for publication. They tested 58 of the 60 cell lines of the NCI-60 panel, and like us, found EKVX to be the only one of these cell lines expressing an X-MLV. An important difference between their study and ours is that the cell lines tested by Sfanos et al. were obtained from the NCI and then cultured in their own laboratory, whereas our initial screen was conducted on materials (DNA and cell pellets) directly supplied by the DTP, providing an additional level of confidence that the virus is present in the source cell line maintained by the DTP. Additionally, we utilized a different envelope antibody and different PCR primers, further strengthening the validity of the findings of both papers.

Frozen cell pellets and genomic DNA were obtained from the DTP. EKVX cells were obtained from the DTP, and LNCaP cells were obtained from the laboratory of F. Ruscetti, NCI. Both cell lines were cultured in RPMI supplemented with penicillin/streptomycin and 10% fetal bovine serum.

Total protein lysates were prepared as previously described [93], and 30 μg were subjected to electrophoresis on 4-12% Bis-Tris gels (Invitrogen, Carlsbad, CA) with MOPS buffer and transferred to nitrocellulose. Monoclonal antibody 7C10 [61], polyclonal goat antisera against Rauscher MLV p30 (NCI, Bethesda, MD), and β-actin antibody (Novus Biologicals, Littleton, CO) were used as primary antibodies. Detection was accomplished with horseradish peroxidase-labeled secondary antibodies and enhanced chemiluminescent substrate (Thermo Scientific, Rockford, IL). XMRV-infected LNCaP cell lysate, provided by F. Ruscetti, was used as a positive control.

PCR for XMRV gag sequences was carried out on 75 ng of genomic DNA in a 25 μl reaction using primers 419F and 1154R and conditions as previously described [94] except the final concentration of each primer was 0.2 μM and the extension step was shortened to 45 s. PCR for XMRV env sequences was carried out similarly with primers 5922F (5′-GCTAATGCTACCTCCCTCCTGG) and 6454R (5′-GGCCCTACATTGAGGACCTGG) with an annealing temperature of 58°C. Hot-StartIt FideliTaq (USB, Cleveland, OH) was used for PCR. No template (water) controls and an XMRV-infected positive control (provided by F. Ruscetti) were included. In some cases, 150 ng genomic DNA was spiked with 1.5 pg or 0.15 pg (0.15 pg/150 ng approximates 1 provirus copy per cell) plasmid DNA containing the XMRV env sequence (provided by F. Ruscetti). Nested PCR was carried out on 1.5 µl of the first round product in a 25 μl reaction as previously described for gag [94] and at an annealing temperature of 58°C using primers 5942F (5′-GGGGACGATGACAGACACTTTCC) and 6271R (5′-GAGCCCACTGAGGAATCAAAACAG) for env. PCR products were analyzed by electrophoresis on 1.5-2% agarose gels and staining with ethidium bromide.

Nested PCR was carried out as described [62] to test for the presence of mouse mitochondrial DNA in 250 ng EKVX genomic DNA. Single-round PCR for intracisternal A particles (IAP) for 45 cycles on up to 600 ng of EKVX genomic DNA was carried out as published [63].

Additional PCR products were generated for sequencing using Finnzyme Phusion polymerase (New England Biolabs, Ipswich, MA) with up to 200 ng genomic DNA in 100 µl reactions containing the HF buffer supplied by the manufacturer, 200 μM dNTPs, and 0.5 μM each primer. Primers were designed to amplify regions in the LTR, gag, and env based on the sequence of DG-75 X-MLV (GenBank accession no. AF221065). Primers and conditions used were: DG-75 7762F (5′-CAAGCTAGCTGCAGTAACGCCATT) and DG-75 287R (5′-CAGACGCAGGCGCAAACATTAGAT) with initial denaturation for 30 s at 98°C followed by 35 cycles of denaturation at 98°C for 8 s and annealing/extension at 72°C for 15 s, followed by a final extension at 72°C for 7 min; DG-75 7607F (5′-AAAGGCAGAATTTCGGTGGTGCAG) and DG-75 7968R (5′-TGCTGGTTCCGCTTTATCTGGGTA) with the same conditions; and DG-75 7945R (5′-TACCCAGATAAAGCGGAACCAGCA) and DG-75 664R (5′-AGGGTCAGACTCAAAGGAGTGGTT) with initial denaturation for 30 s at 98°C followed by 35 cycles of denaturation at 98°C for 8 s, annealing at 70°C for 20 s, and extension at 72°C for 26 s, followed by a final extension at 72°C for 7 min. Products were separated on 2% agarose gels and extracted using a QiaexII Gel Extraction Kit (Qiagen, Valencia, CA).

Sanger sequencing was performed by the Laboratory of Molecular Technology, SAIC-Frederick. Sequences were assembled using Geneious v5.1.6 [95]. Sequences were aligned with representative XMRV and X-MLV sequences using Clustal W2 (www.ebi.ac.uk/Tools/msa/clustalw2) [96,97]. Sequences have been deposited with GenBank with accession numbers JN861040, JN861041, and JN861042.

Supernatants were collected from cultures of EKVX cells 24 hours after a media change and were passed through 0.45 μm filters. For virus infection, subconfluent LNCaP cells in 6-well dishes were pretreated with polybrene (4 μg/ml), then the media was removed and replaced with cell-free EKVX supernatant or fresh media (negative control) containing polybrene (2 μg/ml). After 5 hours incubation, the media was changed in each well. Cells were subcultured and total protein lysates were prepared at 10 and 16 days post-infection. Analysis for viral proteins by immunoblotting was carried out.