ß-actin sequences were detected by PCR in the DNA extracts from all prostate tissues confirming their integrity for PCR testing. DNA from only two (5956 and 6203) of 162 patients (1.23%) tested positive for XMRV sequences in the initial screening (Fig. 1, Table 1). Of the 77 DNA specimens that were additionally tested in triplicate by the pol PCR assay, only one patient (5935) (1.3%) was found positive. This specimen was positive in only one of three replicates. Thus, the overall PCR prevalence was 3/162 or 1.85%. Patient 5956 was positive for gag and pol sequences in 3/5 and 4/6 repeat tests, respectively, and env sequences were also detected in 1/3 repeat tests (Table 2). Specimen 5956 was also positive in all three replicate pol PCR tests. Only XMRV pol and env sequences were detected in patient 6203 prostate tissue DNA in 1/4 and 1/1 repeat tests, respectively. Additional testing of DNA extracted from another prostate tissue section of 6203 was repeatedly negative, likely reflecting an uneven distribution of low copy XMRV in this tissue. DNA from all three patients repeatedly tested negative for murine mtDNA demonstrating the absence of contaminating mouse DNAs. To test for viremia we analyzed RNA extracts from plasma specimens from all three patients. All samples had undetectable sequences by the qRT-PCR and nRT-PCR tests.

Sequence analysis showed that patients 5935, 5956 and 6203 are infected with variant XMRV strains. The 168-bp pol sequences from all three patients showed 90.5–100% nucleotide identity to each other, 94–100% to XMRV, 91.7–98.8% to XMLV, 94–100% to PMLV, and 91–100% to ecotropic MLV (EMLV) in this short region. 164-bp env sequences from persons 5956 and 6203 were identical to each other and shared the highest nucleotide identity (94.9–100%) to XMRV and other xenotropic MLV strains, respectively, [25]. The env sequences from both persons were highly divergent from EMLVs sharing only about 46% nucleotide identity. 413-bp gag sequences were only amplified from the prostate tissue DNA of person 5956 and had about 98% nucleotide identity to XMRV but was identical to a xenotropic mERV found on mouse chromosome 8 (GenBank accession # AC127575).

Phylogenetic analysis of the env sequences inferred three distinct clusters with strong bootstrap support based on viral tropism as expected, since this region includes a portion of the viral surface membrane involved in cellular tropism (Fig. 2a). The env sequences from both patients clustered with XMLV sequences but were distinct from previously reported XMRV and prototypical PMLV (Fig. 2a). In contrast, the inferred phylogenetic relationships of pol sequences from all MLV groups showed that this region does not cluster by host range (Fig. 2b). The three pol sequences from the prostate cancer patients formed a well-supported lineage containing a neuropathogenic Moloney MLV recombinant (strain ts-1-92b, GenBank accession # AF462057) (Fig. 2b). XMRV pol sequences from previously reported prostate cancer patients (coded with VP), except two (VP29 and VP184, kindly provided by Joe DeRisi), clustered with those from persons with CFS (coded with WPI) with significant bootstrap support. VP29 and VP184 pol sequences were identical to each other but 1.2% divergent from other XMRV and formed a weakly supported cluster containing a mixture of EMLV and PMLV (Fig. 2b). These results demonstrate a broader diversity of XMRV in this region than previously seen, but are consistent with recombination with Moloney MLV as reported recently [7], [25]. VP29, VP79, VP86, VP88, VP90, and VP184 are prostate cancer patients reported in the Urismann et al. paper but for which the XMRV pol sequences are not yet available at GenBank [7]. Newly reported gag and pol sequences found in human tumor cell line DNA were located outside the regions used in our study and thus were not included in the analyses [25].

The gag sequence from patient 5956 clustered with XMRV sequences from a prostate cancer patient (VP88), a xenotropic mERV found on mouse chromosome 8, and a PMLV sequence identified in a blood donor (BD-28) by Lo et al. (Fig. 2c) [24]. This lineage was between clades containing XMRV from prostate cancer and CFS patients and most PMLVs and the remaining MLVs found in the Lo et al. study [24]. One prototypical PMLV (MCF1233) clustered with all EMLVs and two XMRVs, suggesting it is a recombinant in this region (Fig. 2c).

These phylogenetic results also suggest the short region of env targeted by our new PCR assay is useful for determining viral tropism for typing the MLV-related sequences. This information may be helpful in understanding the origin of the sequences identified in PCR-positive humans. In contrast, the gag and pol regions showed less clustering by tropism indicating viral recombination in these regions. For example, the gag sequences of prototypical XMLVs (MTCR), EMLVs (AKV, BM5eco), and PMLVs (MCF1233) clustered together with highly significant bootstrap support (Fig. 2a).

Nearly identical phylogenetic tree topologies for each gene region were obtained with both the NJ and ML methods. The XMRV sequences from both prostate cancer patients were also distinct from the PMLV sequences amplified from the murine cell line (RAW) used for preparing WB antigens demonstrating further that these are not laboratory contaminants (Fig. 2). These results confirm the presence of XMRV in both patients and demonstrate that XMRV diversity is greater than currently appreciated.

The XMRV sequences presented in the current paper are available at GenBank with the accession numbers HM003608–HM003612, and HQ116790. Sequences from other studies with XMRV-positive prostate cancer patients were not available at GenBank for inclusion in our analyses [9], [11], [12], [16].

All 162 prostate cancer patient plasma samples were found to be negative for antibodies to XMRV/MLV in the WB test, including plasma from persons 5935, 5956, and 6203 (Fig. 3). In most plasma specimens some level of nonspecific reactivity was observed to the uninfected antigen but without evidence of specific reactivity to Gag and/or Env proteins in the infected antigen blot (Fig. 3).

15 of the 162 persons (9.3%) were determined to be homozygous (QQ) for the R462Q RNase L mutation, 74 (45.6%) had the homozygous RR allele, and 73 (45.1%) were heterozygous (RQ) for the mutation (Table 1).

Patient 5935 is the heterozygous RQ RNase L genotype. He has a poorly differentiated adenocarcinoma with a Gleason's score of 6, T2C pathologic stage, PSA level of 4.6 ng/mL. Patient 5956 is the wild-type homozygous RR RNase L genotype. He had a poorly differentiated adenocarcinoma with a Gleason's score of 7, T3A pathologic stage, a 12.4 ng/mL PSA level. Patient 6203 is the heterozygous RQ RNase L genotype, and had a moderately differentiated adenocarcinoma with a Gleason's score of 6, T2C pathologic stage, and a 7.1 ng/mL PSA level. Thus, we were unable to confirm an association of XMRV infection with the homozygous QQ RNase L allele. Given the low XMRV prevalence, we did not have the statistical power to determine if there is an association of XMRV with tumor grade.

Anonymous, archived plasma and prostate tissue were available from 162 U.S. prostate cancer patients collected by the Fox Chase Cancer Center (FCCC) Biosample Repository Core Facility between 1997 and 2004. Whole blood specimens were collected at the time of pre-surgical testing or on the day of surgery prior to anesthesia from consenting cancer patients as part of a study approved by the FCCC Human Subjects Committee. Written informed consent was obtained from all participants. The FCCC Human Subjects Ethics Committee approved collection, storage, and future testing of the specimens collected from all consenting study participants. A non-research determination was approved for testing of the anonymized samples at CDC for XMRV and related viruses. The age at diagnosis of the 162 participants ranged from 36 to 71 years old with an average and median of 57 years. 86% were Caucasian, 12% African American, and 1% were Asian, not recorded, or other. The average and median serum prostate-specific antigen (PSA) levels were 7.22 ng/mL and 5.5 ng/mL, respectively. The tumor grade for 75 of 162 (46%) of the participants was moderately differentiated, while 54% had poorly differentiated tumors. Using the Gleason system, 42% of the study population had grade 5–6 cancer, 40% had grade 7, and 18% had high grade adenocarcinoma (Gleason's score 8–10). Pathologic staging classified the prostate tumors as T1C (0.6%), T2A (9.3%), T2B (3.1%), T2C (61.7%), T3A (17.9%), T3B (5.6%), and T4 (1.8%).

Plasma was aliquoted and stored at −80°C within 4 hours of blood collection. Plasma samples were aliquoted again at the CDC after thawing for serologic testing; the remaining aliquots were frozen at −80°C. DNA samples were prepared by phenol-chloroform extraction of prostate tissues at FCCC using standard procedures. Only human tissues are processed at FCCC. For selected samples the QIAamp DNA minikit (Qiagen, Carlsbad, CA) was used at the CDC. For plasma RT-PCR testing a volume of 0.5–1.0 ml was ultracentrifuged at 45,000 rpm to concentrate virus. 360 ul to 860 ul plasma supernatant was carefully removed and the pellet was resuspended in 140 ul of centrifuged plasma remaining in the tube and then RNA was extracted using the QIAamp Viral RNA minikit (Qiagen, Valencia, CA). Nucleic acid concentrations were determined by spectrophotometry using the Nanodrop instrument (Thermo Scientific, Wilmington, DE). Integrity of the DNA and RNA specimens was determined using ß-actin or GAPDH PCR, respectively, as described [35], [36]. All specimen preparation, tissue culture, and PCR testing was done in physically isolated rooms to prevent contamination.

To detect XMRV and other MLV variants we used separate PCR assays with primers specific for the detection of XMRV, or conserved primers for the generic detection of MLV and XMRV, as previously described [21]. DNA specimens were screened using two nested PCR assays. The first assay uses the PCR primers GAGOF, GAGOR, GAGIF, and GAGIR employed by Urismann et al. and by Lombardi et al. to detect 413-bp XMRV gag sequences in prostate cancer patients and in persons with CFS, respectively [7], [18]. The second PCR assay generically detects MLV and XMRV polymerase (pol) sequences about 216-bp in length using the primers XPOLOF, XPOLOR, XPOLIF, and XPOLOR [21]. Both the gag and pol PCR tests can detect 10 copies of XMRV plasmid DNA diluted in a background of 1 ug of human DNA [21]. Specimens testing positive for either gag or pol sequences were re-tested with both assays and were also tested with a third nested PCR test for XMRV envelope (env) sequences that is also generic for XMRV/MLV. The external XENVOF (5′ GGG GAT CTT GGT GAG GGC AGG AGC 3′) and XENVOR (5′ CAG AGA GAA CAG GGT CAC CGG GTC 3′) and internal primers XENVIF (5′ AGG GCT ACT GTG GCA AAT GGG GAT 3′) and XENVIR (5′ CCT TTT ACC CGC GTC AGT GAA TTC 3′) amplify a 215-bp env sequence. In addition, a subset of specimens (n = 77) for which sufficient DNA was available were tested in triplicate using the nested pol PCR assay to improve detection of low copy XMRV/MLV.

PCR was performed using 1 ug of prostate tissue DNA in a 100 ul reaction volume using standard conditions of 94°C for 30 sec, 50°C for 30 sec, and 72°C for 45 sec for 40 cycles on an ABI 9700 thermalcycler (Foster City, CA). PCR products were visualized by electrophoresis in an ethidium bromide-stained 1.8% agarose gel. To increase the sensitivity and specificity of the assays, amplified gag, pol, and env sequences were confirmed by Southern blot analysis using the biotinylated oligoprobes XGAGP2 (5′ ACC TTG CAG CAC TGG GGA GAT GTC 3′), XPOLP (5′ TTG ATG AGG CAC TGC ACA GAG ACC 3′), and XENVP (5′ TGG GCT CCG GTA GCA TCC AGG GTG 3′) and chemiluminescence detection. Like the XMRV gag and pol PCR assays [21], the new XMRV env PCR test was also capable of detecting 10 copies of XMRV plasmid in a background of 1 ug human DNA (30/32, 93.8%). We did not detect any XMRV/MLV sequences in PBMC DNA from 41 US blood donors using each of the three nested PCR tests [21].

Plasma samples from persons with positive XMRV PCR results in the prostate tissue DNA were tested further for viral RNA using two RT-PCR tests. The first assay, referred to as qpro uses MLV/XMRV generic protease (pro) Taqman primers Pro-UNV-F1 (5′ CCT GAA CCC AGG ATA ACC CT 3′) and Pro-UNV-R1 (5′ GTG GTC CAG CGA TAC CGC T 3′) and probes Pro-UNV-P1C (FAM5′ AGA TAC TGG GGC CCA ACA CTC CGT GCT GAC 3′BHQ1) and Pro-UNV-PR1 (FAM5′ CCT CCA GTA GCC CCT TGG ACC CAG GC 3′BHQ1). The second assay is the nested GAG RT-PCR test used by Urismann et al. and Lombardi et al. using the primers GAGOF, GAGOR, GAGIF, and GAGIR to detect XMRV RNA sequences in prostate cancer patients and persons with CFS [7], [18], respectively. For the Taqman RT-PCR test, 10 ul of RNA extracted from 50 ul plasma equivalents, 900 nM of primers and 250 nM of probes were used in a one step RT-PCR reaction mixture at 45°C for 20 min, 95°C for 10 min, followed by 55 cycles of 95°C for 30 sec, 52°C for 30 sec and 62°C for 30 sec. A plasmid containing the full-length genome of XMRV (VP62) and RNA extracted from tissue culture supernatants from the XMRV(VP62)-infected DU145 cell line (C7) (both kindly provided by Robert Silverman) were used as positive controls for the assay and for generating the standard curve for viral quantification [7]. The qRT-PCR pro assay has a sensitivity of 10 copies of XMRV protease RNA sequences per reaction in replicate dilutions (40/40, 100%) and a linear range from 10¹ to 10⁸ copies. Plasma from 32 US human blood donors and 32 HIV-1-infected persons tested negative by this assay.

For the nested gag RT-PCR tests, cDNA was synthesized with the XMRV-specific reverse primer GAG OR by using iScript Select cDNA Synthesis Kit (Bio-Rad, Hercules, CA) in a 20 ul reaction following the manufacturer's instructions. Briefly, 10 ul RNA representing extract from 126 ul plasma was mixed with 10 uM GAGOR, 2 ul GSP enhancer solution and Nuclease-free water in a 15 ul reaction volume. Following incubation at 65°C for 5 minutes and cooling to 4°C, 4 ul of 5× iScript Select reaction mix and 1 ul reverse transcriptase were added and incubated at 42°C for 1 hr. The reaction was stopped by heating to 85°C for 5 minutes. For first round PCR, 10 ul of the RT product was mixed with 0.2 uM of the external primers GAG OF and GAG OR, 2.5 mM of each dNTP, 2.5 mM MgCl2 using the Expand High Fidelity PCR System (Roche, Indianapolis, IN) and 5 ul 10× PCR buffer in a 50 ul reaction volume for 35 cycles at 94°C for 30 sec, 52°C for 30 sec, and 72°C for 45 sec. For nested PCR, 0.2 uM of the internal primers GAG IF and GAG IR in first round PCR reaction and run for 40 cycles with annealing temperature increased to 54°C. Both primary and nested PCR products were electrophoresed on 1.8% agarose gels and specific amplicons were detected by Southern blot as previously described. The nested gag RT-PCR assay had a sensitivity between 25 and 100 copies per reaction in replicate testing of XMRV RNA 92.5% (37/40) and 97.5% (39/40), respectively, but could also detect 5 copies in 50% (20/40) of the replicates. Plasma from 40 HIV-1 infected persons were all negative by this assay. 32 of these samples were negative by the pro qRT-PCR test.

Contamination of specimens with mouse DNA will produce a false positive PCR result since mouse DNA contains endogenous MLV sequences that amplify by the diagnostic XMRV/MLV PCR assays. Thus, to distinguish true infection from specimen contamination with murine DNA, all specimens testing positive for XMRV/MLV sequences were also tested for murine mtDNA COX2 sequences using a newly developed PCR assay. The primers MCOX2F2 (5′ TTC TAC CAG CTG TAA TCC TTA 3′) and MCOX2R1 (5′ GTT TTA GGT CGT TTG TTG GGA T 3′) and probes MCOX2PR1 (5′ FAM-CGT AGC TTC AGT ATC ATT GGT GCC CTA TGG T-BHQ 3′) and MCOX2P1 (5′ FAM-TTG CTC TCC CCT CTC TAC GCA TTC TA-BHQ 3′) were used in a two-step thermocycling of 95°C for 30 sec and 62° C 30 sec for 55 cycles on an iQ5 instrument (Bio-Rad, Hercules, CA). Dilutions of a plasmid containing murine COX2 sequences that was generated by PCR with the MCOX2F2 and MCOX2R1 of the murine macrophage cell line RAW264.7 (ATCC, Manassas, VI) was used as the assay standard. 0.5 ug–1 ug of prostate tissue DNA was tested using the MCOX2 primers. This assay was evaluated on PBMC DNA from 117 US blood donors and all samples tested negative, demonstrating 100% specificity of the assay. This test has a 90% and 100% sensitivity of detecting five and 10 copies of murine MCOX2 sequences in a background of 1 ug of human DNA, respectively. We were also able to detect 100–1,000 copies/cell of mtDNA sequences in the Raw264.7 mouse cell line which is equivalent to detecting a single mouse cell, assuming there are at least 1,000–10,000 mitochondria per cell.

PCR products were purified with QiaQuick PCR or gel purification kits (Qiagen, Valencia, CA) and were directly sequenced on both strands by using ABI Prism BigDye terminator kits and an ABI 3130xl sequencer (Foster City, CA). Sequences were aligned using Clustal W in the MEGA v 4.1 program [37]. Following manual editing and removal of indels, phylogenetic relationships were inferred using the neighbor joining (NJ) method implemented in MEGA v4.1 [37]. MLV phylogenies were also inferred using the program PhyML that implements the fast maximum likelihood (ML) method [38]. Support for the branching order was evaluated using 1,000 nonparametric bootstrap replicates.

A PMLV-infected HeLa cell line was used as antigen in the WB test as previously described in detail [21]. Briefly, PMLV-infected and -uninfected HeLa crude cell lysates were prepared as previously described [21], [39]. Plasma or serum samples were diluted 1∶50 and reacted separately to 150 ug of infected and uninfected cell lysates after protein separation through 4–12% polyacrylamide gels and then transfered to nytran membranes, as previously described [21], [39], [40]. Seroreactivity in human specimens was detected using peroxidase-conjugated protein A/G (Pierce, Rockford, IL) and chemiluminescence (Amersham, Uppsala, Sweden) [21], [39], [40]. Specimens were tested in parallel against control antigens from uninfected HeLa cells. Seroreactivity to bands in the WB blots from the infected antigens was compared to those in uninfected antigens to exclude nonspecific reactivity.

1∶500 dilutions of goat polyclonal antibodies to MLV whole virus and gp69/71Env (ATCC, Manassas, VI; VR-1537 and VR-1521, respectively) and a 1∶50 dilution of pre-immune goat sera were used as positive and negative controls, respectively, when testing of the human specimens. These two antisera have been shown to have high titers to MLV Gag (1∶32,000) and Env proteins (1∶8,000) in our WB assay, respectively [21]. In addition, using this antigen high titers to Gag and Env proteins (1∶64,000) were seen using a rabbit anti-XMRV polyclonal sera and the rat anti-spleen focus forming virus (7C10) monoclonal antibody (1∶32,000) that were previously used to detect XMRV protein expression and antibodies in prostate cancer and CFS patients, respectively [9], [18], [21]. Thus, samples were examined for seroreactivity to bands corresponding to Gag (p15, p30, pr68/80) and/or Env (gp69/71, p15E) proteins present in only the infected antigen and not the uninfected antigen. An absence of background reactivity in human samples was demonstrated previously using sera from 121 HIV and HTLV negative anonymous US blood donors collected in 1998, 13 HTLV-positive, 7 HIV-1-positive, six HIV-1/HIV-2 dual positive plasma, and pre-immune goat sera from ATCC [21].

DNA concentrations were first measured using the TaqMan RNase P detection kit (Applied Biosystems, Inc., Foster City, CA) following the manufacturer's instructions. 20 ng of DNA was used in the TaqMan RNase L Q462R SNP Detection Kit (C____935391_1_) using an ABI 7900HT Fast Real-time PCR instrument (Applied Biosystems, Inc., Foster City, CA) and conditions recommended by the manufacturer to determine the prevalence of wild type and mutant Q462R alleles in our study population. 20 ng of DNA in a 25 ul reaction volume was used to determine the SNP genotype. Genomic DNA from cell lines with known RNase L R462Q alleles were controls for the SNP assay (Jurkat, QQ; Raji, RR; MCF-7, RQ).