Cells and DNA used to test cross-reactivity of BLV primers and anti-BLV p24 monoclonal antibody (mAb) with other viruses are listed with their sources in Tables 1 and 2. Except for fetal lamb kidney (FLK) cells, cell lines were stored in liquid nitrogen until use. Before experiments were performed, the species of origin of all cell lines was authenticated by using the 2-pronged method (cytochrome oxidase housekeeping gene primers) (11).

Coded human samples were acquired from the Cooperative Human Tissue Network, a National Cancer Institute–supported tissue bank. Specimens were selected, without regard to patient age, race, or diagnosis, from archived breast tissues acquired from female patients who underwent breast surgery during 2000–2005 at participating hospitals in 4 catchments areas: Birmingham, Alabama; Pennsylvania; Ohio; and Oakland, California. The Institutional Review Board of the University of California, Berkeley (Berkeley, CA, USA) approved human subject use. Bovine control tissue came from the University of Wisconsin, Madison (Madison, WI, USA), as described (10).

PCR primers (Table 3) were examined for BLV specificity by using BLAST (http://blast.ncbi.nlm.nih.gov/Blast.cgi) to search for similar sequences in the nucleotide collection database, which includes exogenous viruses and human endogenous retroviruses such as HERV-K. The search was optimized for highly similar sequences.

For DNA extraction and quality control, cell lines were rinsed with Dulbecco phosphate-buffered saline (DPBS) and pelleted (500 × g for 3–5 min); DNA was then extracted by using the QIAamp DNA Mini Kit (QIAGEN, Hilden, Germany) according to the manufacturer’s cell protocol. DNA from human tissue specimens was extracted from frozen or deparaffinized formalin-fixed paraffin-embedded (FFPE) sections (5 µm thick) by using the QIAamp DNA Mini Kit according to the manufacturer’s tissue protocol. Overnight proteinase K digestion was extended 3–6 h to result in complete digestion, free of visible tissue particles. Extracted DNA quality was confirmed by amplification of a housekeeping gene sequence: human glyceraldehyde-3-phosphate dehydrogenase (GAPDH) for human, rhesus monkey, baboon, and bat material; murine GAPDH for mouse and rat cell lines; and bovine GAPDH for bovine, ovine, and feline cell lines (Table 4). Molecular contamination of extracted human DNA by BLV control DNA was monitored by using sheep-specific primers for the FLK cell line and plasmid vector primers for the C72/taxNeo line (bovine mammary epithelial cells stably transfected with BLV tax gene) (12), the only 2 BLV-containing cell lines previously grown in the laboratory in which testing was conducted.

For nested L-PCR, extracted DNA (0.85 µg) was added to 50 μL of PCR mix (2.0 mmol/L MgCl₂, 0.2 mmol/L dNTPs, 0.025 U/µL Taq polymerase [all from Promega, Madison, WI, USA], and 0.2 μmol/L outer primers for each BLV gene [Table 3]) in Hot Start Micro 50 tubes (MolecularBio Products, San Diego, CA, USA). For the second (nested) round of amplification, 2 μL of the first-round PCR product was added to a new tube of the same reaction mix with inner primers for the corresponding genome region (Table 3). Cycling conditions are given in Table 3. Reaction mix and conditions, including MgCl₂ concentrations, were optimized by using the BLV-positive FLK fibroblasts and BLV-negative bat lung fibroblasts (Tb₁Lu). Sensitivity of the nested L-PCR was determined by using the housekeeping gene GAPDH, which occurs in humans as 1 copy/cell (13). DNA was extracted from 1 million cells of the human cell line MCF7 (Table 1) and serially diluted to an end point equivalent of 1 cell. Nested L-PCR performed on each dilution established the sensitivity at ≤10 copies.

L-PCR, which requires initial tissue digestion to extract DNA, is prone to molecular contamination. We reduced this probability by separating PCR procedure steps into 3 locations: 1) preparation of PCR mixes in a dedicated, locked, DNA-free room with entrance and ventilation system separate from the main laboratory and with dedicated laminar flow hood, small equipment, plasticware, and reagents; 2) addition of template DNA in a different room in a small hood dedicated to specimen handling, with DNA easily decontaminated by UV light and DNA decontamination solution (RNase AWAY; Molecular BioProducts, San Diego, CA, USA) applied to equipment and surfaces before and after handling each specimen; and 3) processing and addition of positive control samples last, and only in a fume hood (velocity = 127 fpm) vented to atmospheric air, thus preventing contaminating aerosols from entering the general work area.

For sequencing, BLV amplicons obtained by nested L-PCR were separated by electrophoreses in a 1% agarose gel at 100 V, excised from the gel, and cleaned by using the Zymoclean Gel DNA Recovery Kit (Zymo Research, Irvine, CA, USA). The requested sample of 100 ng of DNA in 12 µL water and 1 μL of the 5 μmol/L sequencing primer stock solution of the inner primer pair for each genome region (Table 3) was submitted to the University of California, Berkeley, DNA Sequencing Facility for direct sequencing. For tax, inner primer 2 was used to obtain a longer product. DNA for sequencing was obtained from 2–5 amplifications and was sequenced at least once in forward (5′) and reverse (3′) directions. Sequences were checked against corresponding electropherograms. Variations from the reference sequence (GenBank accession no. EF600696) were considered valid only if they matched in both forward and reverse directions. Targeted sequences were relatively short because formalin causes DNA breaks, making it difficult to obtain sequences of >130 nt (14).

Nested in situ PCR (IS-PCR), adapted from Nuovo (15), was used to identify which cell types within tissues were BLV positive. Thick suspensions of detached, rinsed control cells were smeared on enhanced adherence glass microscope slides, air dried, and fixed for 16–18 h with 10% buffered neutral formalin. To enhance entry of PCR mix into cells, samples were made permeable by digestion with 2 mg/mL pepsin in 0.1 N HCl (40–80 min for tissue sections; 20 min for control cell smears), pepsin inactivation solution (100 mmol/L Tris-HCl, 100 mmol/L NaCl, pH 7.4) applied for 1 min, followed by a DPBS rinse and 5 min in absolute ethanol. Samples, run in duplicate, were surrounded with a 15 × 15 mm frame seal chamber (Bio-Rad, Hercules, CA, USA); 60 μL of PCR mix were then placed into the chamber, and the plastic cover was sealed over the frame. The PCR mixture was 4.0 mmol/L MgCl₂, 0.4 mmol/L dNTPs, 1 μmol/L primers (Operon Biotechnologies, Huntsville, AL, USA), 0.06% bovine serum albumin, 8 μmol/L digoxigenin-11-dUTP (dig) (Hoffman-La Roche, Basel, Switzerland), and 0.053 U/μL Amplitaq Gold DNA Polymerase (Applied Biosystems, Foster City, CA, USA), a Taq polymerase activated only at ≥92°C, designed to reduce false positives from nonspecific DNA repair by Taq at cooler temperatures (15). Primers were the same as those used for L-PCR (Table 3). Slides were placed into an IS-PCR machine (Hybaid Thermo OmniSlide; Cambridge Biosystems, Cambridge, UK) for amplification. After each round, covers and chambers were removed, and slides were rinsed in DPBS. After the second round, endogenous peroxidase was quenched 30 min in 3% H₂O₂ in methanol. Label incorporated into PCR products was detected by anti-dig antibodies in an avidin-biotin-immunoperoxidase reaction (Hoffman-La Roche). The chromagen was diaminobenzidine. Outcome measurement was a semiquantitative judgment of color density of cells: 1+, light tan; 2+, medium tan; 3+, dark brown; 4+, almost black. Ratings of ≥2+ were considered positive. No donor information or results of L-PCR were available at the time of slide evaluation.

Initially, the BLV-positive control cell line (FLK) and the BLV-negative cell line (Tb₁Lu) were used to optimize the reaction and ensure no false-positive reaction in the negative cell line. Control testing was run simultaneously with each batch of human tissue assays: 1) positive control, a smear of BLV-positive cells (FLK cell line) reacted with complete PCR mix; 2) negative controls, a smear of FLK cells and an adjacent serial section of each specimen reacted with PCR mix minus primers, to rule out false-positive reactions unique to each tissue resulting from unquenched endogenous peroxidase, nonspecific reaction of the sheep antibodies used in the final immunoperoxidase detection, or nonspecific DNA repair by Taq polymerase; 3) permeabilization control for entry of PCR mix into cells, with an adjacent serial section of each tissue reacted with PCR mix different from that for IS-PCR by omission of primers, 4.5 mmol/L concentration for MgCl₂, and use of a different Taq polymerase (Promega) that reacts nonspecifically at cooler temperatures (4°C–50°C) to repair DNA. Specimens were scored positive only if mammary epithelial cells were positive and the background control slide (adjacent section) without primers was negative for the corresponding area of mammary epithelial cells. Specimens were scored negative only if the sample and its background control section were negative and the permeabilized control exhibited dig incorporation into cell nuclei resulting from Taq polymerase DNA repair, confirming entry of PCR mix into cells (15).

IHC testing for BLV p24 was performed by using formalin-fixed cell smears and deparaffinized FFPE tissue sections (5 μm) on superadherent microscope slides. Samples were quenched of endogenous peroxidase for 30 min in 3% H₂O₂ in methanol and rinsed in DPBS; antigens were then unmasked by incubation in citrate buffer (0.1 mmol/L sodium citrate, 0.04 mmol/L citric acid, pH 6.0) for 25 min at 95°C in individual plastic containers to prevent cross-transfer of tissue material. Unmasking was followed by a DPBS rinse, and an avidin-biotin-immunoperoxidase procedure (Vectastain Elite ABC Kit; Vector Laboratories, Burlingame, CA, USA) was performed according to the manufacturer’s instructions. Blocking serum was 1.5% fetal horse serum in DPBS. Primary antibody in blocking serum (1:10) was a hybridoma-produced mouse mAb against BLV p24 (AIDS Research and Reference Reagent Program, National Institute of Allergy and Infectious Diseases; www.aidsreagent.org) (16); specificity was validated by immunoblot reactivity with purified recombinant p24 (7). Secondary antibody was 1.5% horse anti-mouse IgG (Vector) in blocking serum. The chromagen was diaminobenzidine. Negative controls were the BLV-negative control cell line Tb1Lu and adjacent human tissue sections reacted with fresh hybridoma medium (diluted 1:10 in blocking buffer) substituted for primary antibody. The positive control was FLK, which is known to replicate BLV. Outcome measurement was a semiquantitative judgment of color density, as described for IS-PCR.

Figure 1

Figure 1. Amplification of bovine leukemia virus (BLV) genome regions in human breast tissue specimensNested liquid-phase PCR, using primers from 5 BLV genome regions, was used to amplify products from DNA extracted from...

Figure 2

Figure 2. Partial sequences of 5 genome regions of bovine leukemia virus (BLV) DNA isolated from human breast tissue samples: A) long terminal repeat; B) group-specific antigen; C) polymerase; D) envelope; E) trans-activating...

Human samples were selected from 219 FFPE breast tissue samples; 97 (44%) samples had positive results for BLV by nested IS-PCR that used primers from the most conserved BLV genome region, tax. Five tax-positive samples and 1 tax-negative sample (as a control) were chosen for in-depth molecular analysis on the basis of BLV status and sample size large enough and with enough mammary epithelial cells to yield sufficient material for multiple assays and extensive quality control tests. BLV was detected in DNA extracted from human tissues by using nested L-PCR. All 5 tax-positive samples chosen were positive for the LTR region but showed varying results for other BLV genome regions (Table 5; Figure 1). Sequences of all samples positive for BLV had high identity (E value <1.2) only to BLV nucleotide sequences deposited in GenBank, which suggests that these isolates did not represent some other entity. Variations from the BLV reference sequence were infrequent, and all involved base substitutions (Figure 2).

Figure 3

Figure 3. Localization of bovine leukemia virus (BLV) in human breast tissue and bovine mammary epithelium samples detected by in situ PCR for the BLV tax region and immunohistochemical testing for p24 capsid...

Nested IS-PCR was used to identify the cell type in which the PCR product was localized. Figure 3 shows IS-PCR results obtained by using tax primers for the BLV-negative human sample (no. 143), 1 of the BLV-positive samples (no. 010), the positive and negative cell line controls, and a BLV tax-positive bovine mammary gland sample. The site of the amplified BLV DNA was the secretory mammary epithelium, identified by an anatomic pathologist (H.M.J.). Table 5 summarizes the results for L-PCR and IS-PCR (tax) testing of the 6 tissue samples that were studied in depth.

Validation of the IS-PCR results for a subset of 7 samples (3 tax negative and 4 tax positive) was performed by an independent laboratory by using control cell smears and coded FFPE sections sent from our laboratory with no information about the human patients, tissue pathology, or our results. The detection method was PCR in situ hybridization, in which the PCR occurs in situ but with no label incorporated during amplification (15). Labeled probes specific for the BLV tax region were applied after amplification. The independent laboratory confirmed with 100% concordance the results we obtained by using nested IS-PCR and L-PCR for the tax genome region (Table 6).

Of the 215 human breast tissue samples tested by IHC, 12 (6%) had positive results for BLV p24. The reaction was confined to secretory mammary epithelium, and distribution of both BLV DNA and p24 in mammary epithelial cells was duct and lobule specific; that is, virtually all cells in the duct or lobule would be BLV positive but neighboring ducts or lobules could be completely negative. A BLV p24–positive breast tissue sample is shown in Figure 3, panel F.

Figure 4

Figure 4. Test results showing lack of cross-reactivity of bovine leukemia virus (BLV)–specific primers with representatives of all mammalian and avian retrovirus subfamilies and human exogenous and endogenous viruses previously identified in human...

GenBank searches indicated complete identity (E value ≤1.1) of PCR primers (Table 3) only with BLV sequences. L-PCR further substantiated primer specificity (Figure 4; Table 1) by demonstrating no cross-reactivity (amplicon generation) by using template DNA from cell lines harboring representatives of all mammalian and avian oncogenic retroviral subfamilies and human lentiviruses; viruses previously reported in human breast tissues (Epstein-Barr virus, human papillomavirus, mouse mammary tumor virus–like sequences) (17); and human endogenous retrovirus HERV-K (Table 1). Extracted DNA met quality control standards of exhibiting a housekeeping gene (GAPDH; Figure 4). mAb specificity was supported by lack of cross-reactivity with cell lines or by tissues replicating viruses previously identified in human breast tissues (Epstein-Barr virus, human papillomavirus, and mouse mammary tumor virus–like sequences) (17) or that cause chronic human infections (hepatitis B virus, human T-cell leukemia viruses 1 and 2, HIV 1 and 2, human herpesviruses 1 and 2, and cytomegalovirus) (Table 2).

Several methods were used to test for cell and molecular contamination. The no-template-DNA control run with each assay ruled out contamination in commercial DNA isolation columns and PCR reagents; in addition, contamination of human DNA samples by FLK DNA was ruled out by testing human DNA with sheep-specific primers (11), and contamination of FLK with other cell lines, including human, was ruled out by using species-specific primers (11). Sequences from the human isolates also provided strong evidence against contamination; none were exact matches with each other or with any Genbank BLV sequence, which includes the FLK-positive control cell line and bovine specimens previously sequenced in our laboratory. Furthermore, the FLK stock cell line has a signature base substitution (bp 5194; GenBank accession no. EF600696) in the env region; this substitution is unique among sequences deposited in GenBank and was not detected in the human BLV isolates we investigated (Figure 2). The positive results obtained by using in situ methods (IHC and IS-PCR) demonstrating the signal within individual formalin-fixed mammary epithelial cells are further evidence that the positive reactions do not represent contamination of PCR mix. Previous reports have documented that contaminating DNA cannot cause a false-positive reaction with PCR in situ hybridization in FFPE tissues (15). Negative controls (adjacent tissue sections without BLV-specific primers or antibody) run with each in situ assay were all negative, which suggests that positive signals were not false positives resulting from nonspecific reactions.

During the 4 decades since BLV was identified in cattle, there has been considerable interest in determining whether humans could become infected with BLV, especially because cattle are a major food source. As in early serologic studies, previous cellular and molecular studies reported no evidence of human infection with BLV. An explanation for some of the negative findings (18–22) could be that sufficiently sensitive reagents and techniques such as PCR and sequencing were not available when the studies were conducted. Most previous studies also focused on leukocytes, the cell type involved in bovine leukemia/lymphoma (18–27). We focused instead on mammary epithelium, in which we had detected BLV DNA and protein in cattle (10). In situ techniques (i.e., IS-PCR, PCR in situ hybridization, and IHC) enabled confirmation that BLV was localized within mammary epithelial cells.

Evidence for BLV DNA and protein in humans is not surprising. Many viruses, including those that are oncogenic, are known to cross species naturally, and most microbial species pathogenic in humans are speculated to have had a animal origin at some point in human evolution. Once in the human population, most zoonotic viruses can be transmitted among humans, a process that poses the most serious threat to human health (28). BLV is known to cross species readily; the virus infects capybara, zebus, and water buffaloes naturally and sheep, goats, pigs, rabbits, rats, and chickens experimentally (1). Human cells (fibroblasts) are susceptible to infection with BLV in vitro (29).

The lack of gag, pol, and env sequences in some of the BLV-positive panel samples and the presence of LTR and tax sequences in all of them is consistent with results reported for the closely related HTLV-1 (30). Deletions in gag, pol, and env were observed in HTLV-1 isolates from 25.7%–56% of adult T-cell leukemia patients, and frequency increased with clinical progression of leukemia (30). Such deletions are postulated to be advantageous to the virus by enabling escape from immune surveillance. In contrast, the LTR and tax regions of HTLV-1 are highly conserved. Analogous results were observed in specimens from cattle infected with BLV; deletions involving parts of gag and env and all of pol were frequent (2).

Overall, the human BLV isolates differed from the reference sequence by only a few base substitutions, a finding that fits well with the biology of deltaretroviruses, which have infrequent interhost transmission and remain largely latent within the host, probably as a strategy to escape the host’s immune response (31). Infectivity of deltaretroviruses occurs primarily by cell–cell contact, not by extracellular virions (32), and BLV virions have not been found in the peripheral blood of infected cattle (2). In our study, evidence of the capsid protein p24 was infrequent (12/215 [6%]) but consistent with the concept that, in some humans, BLV could be replicating.

Mutation rate tends to be lower for latent viruses than for viruses that escape immune attack by rapidly mutating to outpace the host’s immune response (e.g., HIV [2,31]). Four of the 7 LTR base substitutions we observed (in 3 human BLV isolates) occurred in the PU box, a short (12 bp) protein-binding region that, when mutated, decreases basal gene expression (2,33). This expression would facilitate evasion of the host’s immune response and would thus be likely to be selected for during virus evolution. Transcription in BLV-infected bovine lymphocytes is rare, identified in only ≈1/50,000 peripheral blood cells (2).

In summary, multiple lines of evidence support the conclusion that the BLV DNA and protein we found are more likely to represent the in vivo presence of BLV in humans than to represent some other virus, molecular laboratory contamination, or an artifactual nonspecific reaction. In view of the potential public health implications of BLV in humans, future research should address how humans acquire BLV infection, how frequently BLV infection occurs in different populations, and whether the virus is associated with human disease.