Introduction

Worldwide, approximately 800,000 incident cases of breast cancer are diagnosed each year [1]. For several decades, the retrovirus Mouse Mammary Tumour Virus (MMTV) or a close relative (collectively termed ‘MMTV’) has been suggested as a possible aetiological agent for human breast cancer. Aspects of the biology of MMTV and some recent reports suggest a strong case. MMTV was the first virus shown to cause cancer in mammals [2]. In the mouse, MMTV can be transmitted through the germline or via milk to suckling pups. Tumorigenesis occurs due to disruptive non-site-specific proviral integration in mammary epithelia during puberty and pregnancy [3]. MMTV env sequences have been reported at relatively high frequencies in unselected breast cancer specimens but at dramatically lower frequencies in healthy breast tissues from affected or unaffected women or in other cancer tissues [4–12]. Frequencies in breast cancer tissues of 20–40% have been reported for women in the USA, Italy, Mexico, Brazil, Argentina and Australia, and frequencies of 12, 10–17 and 0.8% have been reported for women in Japan, China and Vietnam, respectively. The geographical variation in the incidence of female breast cancer has been reported to correlate with the regional dominance of the mouse species Mus domesticus, suggesting a possible zoonotic connection [13]. MMTV env sequences have been reported to be especially common in higher grade breast cancers [9], breast cancers diagnosed during gestation (62%) [14] or lactation [6], and cases from multiple-case families [6, 15], as well as in male breast cancers (62%) [9]. MMTV transcripts and the MMTV sag gene have also been reported as common in breast cancer specimens [16, 17], and a complete 9.9 kb provirus has been reported [18]. It was recently reported that MMTV is capable of active infection of the human aneuploid mammary epithelial cell line Hs578T [19].

However, MMTV’s candidature as an aetiological agent for human breast cancer is controversial [20–23]. Null findings have been reported by studies of UK, Austrian, Swedish, Japanese and Italian breast cancer specimens [24–28]. In apparent discord are reports of 13.9 or 74% env-positivity in Tunisian breast cancer specimens [29, 30] and 4.2 or 24% in Mexican cases [31, 32].

The implications for public health, should MMTV eventuate as a causative agent for human breast cancer, are great. We sought to validate previous reports by attempting to detect MMTV env sequences in a population-based sample of breast cancer specimens from the Australian Breast Cancer Family Study (ABCFS) using a specifically engineered nested PCR. Studied specimens were enriched for higher grade breast cancers and cases for which diagnosis occurred relatively soon after childbirth.

Materials and methods

ABCFS specimens

Thirty-eight 5 μm and twelve 3 μm Formalin Fixed Paraffin Embedded (FFPE) breast cancer tissue sections representing breast cancers from ABCFS participants were selected to enrich for breast cancers of relatively high histological grade and cancers from cases in which there was a relatively short time between the woman’s last childbirth and her date of breast cancer diagnosis. H&E stained neighbouring sections were used to confirm the presence of breast cancer tissue [33, 34].

p203 and human genomic DNA controls

The plasmid p203 in E. coli was purchased from the American Type Culture Collection (ATCC). p203 harbours the relevant portion of MMTV targeted by the nested PCR described here. A small-scale plasmid preparation was performed according to ATCC instructions and using the QIAprep Spin Miniprep Kit (Qiagen, Doncaster, VIC, Australia). p203 concentration was determined using the Quant-It system (Invitrogen, Mulgrave, VIC, Australia) and Qubit instrument (Invitrogen, Mulgrave, VIC, Australia) according to the manufacturer’s instructions. QIAamp Kit (Qiagen, Doncaster, VIC, Australia)-purified human genomic DNA from a healthy control individual whole blood sample was sourced and the concentration determined using the Quant-IT system with the Qubit instrument.

DNA extractions

Formalin Fixed Paraffin Embedded tissues (full face sections) were scraped from slides onto aluminium foil using scalpel blades and transferred to microfuge tubes using syringe needles. Separate foil, blades and needles were used for each specimen. DNA was extracted using the QIAamp DNA FFPE Tissue Kit (Qiagen, Doncaster, VIC, Australia) according to the manufacturer’s instructions but with the following exceptions: histolene was used instead of xylene, two ethanol spin/wash steps were employed instead of one to facilitate thorough washing without losing tissue during aspiration of supernatant, no RNaseA was used, and elution was performed using 50 μl ATE buffer.

β-Globin PCR specimen adequacy check

GH20 and PC04 β-globin PCR primers [35] were ordered from Integrated DNA Technologies (IDT) at standard desalting grade. PCRs were performed in 30 μl reaction volumes containing 1.5 U AmpliTaq Gold (Applied Biosystems, Mulgrave, VIC, Australia), 200 μM dNTPs (Bioline, Alexandria, NSW, Australia), 100 nM PC04 and GH20 primers and 5 μl template DNA and overlaid with 25 μl molecular biology grade mineral oil (Sigma-Aldrich, Sydney, NSW, Australia) added from a fresh aliquot by Gilson pipette. PCR was performed using the following thermal parameters: 95°C for 10 min, followed by 40 cycles of 94°C for 45 s, 56°C for 45 s and 72°C for 45 s, followed by 72°C for 5 min. PCR products (268 bp) were analysed using agarose gel electrophoresis.

Nested PCR primers and primer design

ClustalW2 multiple sequence alignments of MMTV env were performed using GenBank-deposited sequences to include a range of MMTV strains and sequences reported to be derived from human tissue specimens (see Fig. 1). Primer sites were selected with a view to yielding small amplicons (less than 200 bp), limiting the degree of mismatch across the aligned sequences, especially near the 3-prime terminus, targeting a GC content approaching 50% without excessively long GC runs, and avoiding unintended ‘off target’ priming from HERV-K10 (the reported most similar human endogenous retrovirus) or the human genome, excessive primer secondary structure formation and primer-dimer formation. Primer-BLAST, OligoAnalyzer 3.1 and ClustalW2 were employed to facilitate these design requirements. Remaining points of mismatch with respect to the alignment group were addressed by introducing degeneracy at corresponding primer positions. 5-prime clamp sequence was included to facilitate primer participation by all members of the degenerate primer ‘pool’. Resulting first round PCR primers were (5′ to 3′, lower case indicates 5′ clamp sequence): DPenvF2: ctatagggGGYTACTTTGGGATTTCTCCCTTC and DPenvR2: ctatagggACCAWGTAGGTTCTACCCATC. Resulting second round (nested) PCR primers were (5′ to 3′): DPenvF3: ctatagggTCGCCTARTRTAGATCAGTCAGA and DPenvR3: ctatagggCATCYTGCYTCATACCATCGATGAAC. Nested PCR primers were ordered from IDT at standard desalting grade. Anticipated product sizes from PCRs employing DPenvF2 and DPenvR2 or DPenvF3 and DPenvR3 are 166 and 123 bp, respectively. MMTV env sequences spanned by these PCRs are 150 and 107 bp, respectively.

Fig. 1
figure 1

ClustalW2 multiple sequence alignment of MMTV env within the human breast cancer-associated region [4]. The range of MMTV strains includes strains reported to be derived from human breast cancer tissue (prefixed ‘H’). Database accession codes for the respective sequences are NC_001503, D16249, AF228551, AF228550, AF071010, X01811, AF239172, AF346816 and AF243039. Boxed arrows indicate nested PCR primer binding sites used in this study. These are annotated with the degenerate bases included in the primers (Y, R or W). The vertical arrow indicates degeneracy designed into primer DPenvR3 based on the sequences AY600606, AY600599, AY600611, AY600614, AY600604, AY600605, AY600597 and AY600608 which, together with AY600615, AY600613, AY600610 and AY600598 [9], were also aligned to aid primer design (not shown)

Full size image

Nested PCR

For assay sensitivity assessment, nested PCRs (PCR1 containing DPenvF2 and DPenvR2, followed by PCR2 containing DPenvF3 and DPenvR3) were performed using different cycle numbers (30 or 35 cycles for each PCR) or annealing temperatures (50–56°C) using decreasing copies of p203 (tested down to five copies) against a backdrop of 10 ng of MMTV-negative human genomic DNA, followed by gel electrophoretic analysis. All conditions clearly detected five copies of p203 against a backdrop of 10 ng MMTV-negative human genomic DNA. In all cases 10 ng MMTV-negative human genomic templates and water controls yielded negative results. Conditions selected to screen the breast cancer specimens were: PCR1: 30 μl reaction volumes containing 1.5 U AmpliTaq Gold (Applied Biosystems, Mulgrave, VIC, Australia), 200 μM dNTPs (Bioline, Alexandria, NSW, Australia), 100 nM DPenvF2 and DPenvR2 primers and 5 μl template DNA overlaid with 25 μl molecular biology grade mineral oil (Sigma-Aldrich, Sydney, NSW, Australia) added from a fresh aliquot by Gilson pipette. PCR1 was performed using the following thermal parameters: 95°C for 10 min, followed by 35 cycles of 94°C for 45 s, 50°C for 45 s and 72°C for 45 s, followed by 72°C for 5 min. PCR2: 30 μl reaction volumes containing 1.5 U AmpliTaq Gold (Applied Biosystems, Mulgrave, VIC, Australia), 200 μM dNTPs (Bioline, Alexandria, NSW, Australia), 100 nM DPenvF3 and DPenvR3 primers and 2 μl PCR1 product template overlaid with 25 μl molecular biology grade mineral oil (Sigma-Aldrich, Sydney, NSW, Australia) added from a fresh aliquot by Gilson pipette. PCR2 was performed using the following thermal parameters: 95°C for 10 min, followed by 35 cycles of 94°C for 45 s, 55°C for 45 s and 72°C for 45 s, followed by 72°C for 5 min. These conditions were selected as sufficiently relaxed to enable efficient specific amplification in cases of unforeseen primer-template mismatch whilst yielding ‘clean’ product profiles. PCR2 product was sequenced using DPenvF3 as the sequencing primer and BigDye3.1chemistry (Applied Biosystems, Mulgrave, VIC, Australia).

Agarose gel electrophoresis

2.4% Agarose TBE gels containing ethidium bromide were loaded with 5 μl PCR products alongside 250 ng of 100 bp ladder (New England Biolabs distributed by Genesearch, Arundel, QLD, Australia). Following electrophoresis, gels were photographed under UV light transillumination.

Results and discussion

Fifty ABCFS specimens were tested for ‘sample adequacy’ using PCR to amplify β-globin sequence. Strong bands of 268 bp following PCR and agarose gel electrophoresis were interpreted as being derived from adequate specimens, and the corresponding specimens were subsequently tested for MMTV env sequences by nested PCR. A relatively conservative approach was applied, faint products of 268 bp or no apparent products of 268 bp were interpreted as being derived from inadequate specimens and the corresponding specimens were not subsequently tested for MMTV env sequences. Of the 50 original specimens, 42 were deemed adequate (Fig. 2a), derived from thirty-four 5 μm and eight 3 μm sections.

Fig. 2
figure 2

2.4% Agarose TBE gels depicting the assessment of ABCFS specimen adequacy following β-globin PCR (a) and MMTV env sequences following nested PCR (b). ‘M’ indicates 250 ng 100 bp ladder marker; ‘−’ indicates a no template control; numbers 1–50 indicate the 50 ABCFS specimens screened for specimen adequacy; ‘F’ indicates a specimen interpreted as inadequate; A1–A42 indicate the 42 adequate specimens tested for MMTV env sequence by nested PCR; ‘+’ indicates a ‘five copies of p203 with 10 ng MMTV-negative human genomic DNA’ control

Full size image

All 42 adequate specimens represented in this study were collected from women living in the Melbourne (22/42) or Sydney (20/42) metropolitan areas that were diagnosed with invasive breast cancer before the age of 40 years. The mean age at time of diagnosis was 35.1 years (28–39 years), and 83.3% (35/42) of specimens exhibited infiltrating ductal profiles. 76.2% of the specimens (32/42) presented as histological grade 3 cancers, the remainder being grade 2 cancers. Overall, 78.6% of women (33/42) had given birth and 90.9% of specimens (30/33) were from women who had breastfed. Of the women who had given birth, the median time from the last birth to date of diagnosis was 54 months (0–244 months). Based on previously reported prevalences of MMTV env detection for Australian breast cancers (30–40%) [8, 9], reported increased prevalence for higher grade breast cancers [9] and reported increased prevalence for cases diagnosed during gestation (62%) [14] or lactation [6], approximately 30–50% of the 42 breast cancer specimens included in this study might have been expected to harbour MMTV env sequences, i.e. approximately 12–21 specimens. From the Poisson distribution, the probability of observing zero prevalence is 0.000006 when the true prevalence is 30%, and 0.05 when the true prevalence is 7%. That is, the true prevalence for Australian women is highly unlikely to be as high as previous reports, though we cannot exclude a low prevalence.

The nested PCR regimen used in this study to look for MMTV env sequences proved to be robust and extremely sensitive. In all experiments conducted, each using fresh dilutions of p203 to yield five plasmid copies against 10 ng MMTV-negative human genomic DNA as a parallel sensitivity control for separate PCR runs, MMTV env was detected as a strong 123 bp band following agarose gel electrophoresis. ‘No template’ controls always yielded profiles free of visible PCR product. Sequencing and BLAST analysis of the assay sensitivity control products confirmed the detection of MMTV env-derived sequences. None of the 42 adequate breast cancer specimens tested yielded visible MMTV env-derived products (Fig. 2b). Given the nature of the sample set tested and the demonstrated high sensitivity of the nested PCR regimen employed, including design features and PCR conditions to account for unforseen discrepancies between potential sample set MMTV env sequences and sequences included in the primer design alignment group, it seems unlikely that any of the 42 adequate ABCFS specimens tested contained MMTV env sequences. These findings appear inconsistent with previous reports of associations between MMTV and human breast cancer and, in particular, reports of 30–40% prevalence for unselected Australian breast cancer specimens [8, 9]. Given the risk of PCR contamination and the potential to contribute to false findings [22], we recommend caution when interpreting findings derived from highly sensitive PCR regimens which are not consistently supported by other studies.

Mouse Mammary Tumour Virus has not been found in human breast milk and is not detected in healthy tissue adjacent to cancerous tissue in humans, unlike in the mouse, and the histopathology of murine MMTV tumours is dissimilar to human breast cancers (hyperplasias vs. typically invasive ductal carcinomas) [36]. In a large study, post-menopausal women who were breastfed as infants did not have a higher risk of breast cancer compared with those who were not [37], and a separate study has reported that human breast cancer rates have increased at the same time that breastfeeding rates have decreased [38]. Pregnancy in humans appears to have a protective effect with respect to breast cancer risk, unlike MMTV-associated breast cancer in the mouse [39, 40], and serological conversion to yield antibodies cross-reactive with MMTV in human breast cancer patients remains to be demonstrated at a population level [41] despite the fact that mice produce strong antibody responses to MMTV. Further, Transferrin Receptor 1 acts as an MMTV entry receptor in the mouse but not in humans [42, 43].

The above observations together with the findings presented in this study and a growing number of null findings from independent groups internationally suggest that MMTV is unlikely to be a causative agent for female breast cancer. Alternative sources of evidence, for example, derived from universally accepted specific serological analysis will probably be required to definitively demonstrate an association between MMTV and female breast cancer. Consistent demonstration by independent groups of such an association and the detection of MMTV infection prior to the development of disease will be required to demonstrate causality [44, 45].