Background
Preclinical studies of breast cancer are limited by a lack of suitable models recapitulating aspects of human physiology and the biology of the human breast. Approximately 90 % of cancer treatments stemming from preclinical screens performed using xenografts in rodents fail during clinical trials [
1], highlighting intrinsic genetic, physiological [
2,
3] and morphological [
4] differences between humans and mice. The pig offers a promising alternative to traditional rodent models given they share pronounced genomic [
5] and biological [
6] similarities to humans. As such, pigs have increasingly become an integral species for translational research, particularly for preclinical toxicology studies and as a biomedical model for human cardiovascular, integumentary and gastrointestinal systems [
7].
While the mammary glands of female pigs have only been infrequently cited as a model for the human breast, they closely recapitulate several important aspects of human breast biology. Development of the mammary tissue in pigs from embryogenesis [
8] through puberty [
9] and gestation [
10] parallels that of the human breast [
11]. While pigs have an average of 10–14 mammary glands, each has multiple (2–4) galactophores that drain to the nipple and form the primary duct from which the parenchymal tissue develops [
12]. Humans also have multiple galactophores per nipple, while the mouse has only one [
13]. The histomorphology of the porcine mammary gland and human breast has been similarly described as having terminal ductal lobular units (TDLU) embedded within fibrous inter-and intralobular connective tissues [
9,
11], which contrasts to the simple ductal network and adipose-rich stroma of the mouse mammary gland [
4]. Importantly, intrinsic structural differences between the mammary glands of rodents and humans likely influence tumorigenic risk given that the stroma directs proliferative, morphogenic and hormonal responses by the epithelium [
14‐
16]. Furthermore, the relative abundance of different TDLU morphotypes in the human breast can influence breast cancer risk, where the least-differentiated TDLU type 1 (TDLU-1) is most prone to transformation [
17]. A porcine model of human breast cancer would stand to address many of these interactions that underlie breast development and tumorigenesis [
18]. Moreover, the size and positioning of the voluminous mammary glands will allow for the assessment of multiple treatments or endpoints within an animal and over time using serial biopsies [
19].
Further to the above, few reports detail methods to isolate and genetically manipulate the mammary epithelial cells (pMEC) in pigs. The objective of this study was to establish methods of lentivirus-mediated transformation of pMEC as a first step toward developing a novel model for human breast cancer. We hypothesized pMEC would undergo oncogene-induced transformation to yield tumors with a histopathology resembling human breast cancers. Herein, we report the successful lentiviral transduction of porcine mammary cells in vitro and tissue in vivo, formation of tumors by transformed pMEC in immunocompromised mice, and the precocious expansion of TDLU when transformed pMEC were isografted into the pig mammary gland.
Methods
Experimental design
We initially conducted experiments to determine the efficiency of using lentivirus for the transduction of pMEC in vitro and in vivo. In study one we sought to develop and optimize methods for the collection and dissociation of mammary tissue from nulliparous pigs for transduction in vitro. In study two we transduced pig mammary tissue in vivo by direct instillation of non-oncogenic lentivirus into the mammary gland duct or parenchyma. For study three, we sought to determine whether pMEC transduced with non-oncogenic lentivirus in vitro could develop typical mammary structures when transplanted back to the mammary fat pads of respective donor pigs. Finally, in studies four and five, pMEC were transformed in vitro by oncogenic lentivirus and either isografted to the mammary gland of donor pigs (study four) or xenografted to the mammary fat pad of immunocompromised mice (study five).
Animals
All experimental protocols for animal experimentation underwent prior ethical review and were approved by the UC Davis Animal Care and Use Committee following guidelines set forth by the Association for Assessment and Accreditation of Laboratory Animal Care and the Guide for the Care and Use of Agricultural Animals in Research and Teaching (protocol #17675). For study one and study five, mammary tissue was obtained at necropsy from healthy nulliparous Yorkshire × Hampshire pigs obtained from the specific pathogen-free swine facility at UC Davis when they were 3–5 months of age (n = 8 and n = 4, respectively). For study two (n = 9 pigs from two litters), study three (n = 8 pigs from two litters), and study four (n = 4 pigs from two litters) pigs were healthy, experimentally naïve 4 week-old Yorkshire × Hampshire females. For studies two, three and four, piglets were selected that possessed at least twelve mammary glands, which permitted an individual pig to carry experimental treatments and controls within separate mammary glands. Piglets were housed indoors in a temperature-controlled facility (25–27 °C), as littermate pairs, were fed twice daily and had ad libitum access to water. Pigs were monitored daily for any changes in behavior or health status. During surgical procedures, pigs were assessed for changes in body temperature, heart rate and respiration. All surgical procedures involving pigs were carried out in a disinfected surgical suite designed for accommodating large animals.
In study five, 20 experimentally naïve female
NOD scid gamma (NOD.Cg-
Prkdc
scid
Il2rg
tm1Wjl
/SzJ (NSG)) mice (The Jackson Laboratory, Sacramento, CA;
n = 4 per pMEC line) between 4 and 35 weeks of age (Table
1) were maintained in littermate groups with
ad libitum access to food and water. Mice were housed in a pathogen-free barrier facility under conditions of constant temperature (20–23 °C), humidity (45–65 %), and a 14 h light/10 h dark cycle. Tumor formation was assessed weekly by palpation, and tumor diameter recorded every 2d once they reached 1 mm diameter. During surgical procedures, mice were monitored for toe-pinch reflex and respiration rate. Surgical procedures were carried out within a disinfected biosafety cabinet to minimize pathogen exposure.
Table 1
pMEC were sorted to remove fibroblasts (CD140-), and some selected for expression of CD49f or tdTomato, and then transduced at various passages (PT) with one of the lentiviral constructs PGK-Tantigen (PGKT), CMV-tdTomato (CMVT) or PGKT-CMVT (PTCT)
ss071712 PGKT | CD140- | 1 | Cobblestone, no foci | 3 | 67d | 6 | 1×105 | Hydrogel | MG | No | N/a | 16.7 | N/a |
ss071712 CMVT | CD140- | 6 | 1×105 | Hydrogel | MG | No | N/a | 16.7 |
ss071712 PGKT | CD140- | 1 | 7 | 74d | 6 | 1×105 | Hydrogel | MG | No | N/a | 15.7 | N/a |
ss071712 CMVT | CD140- | 6 | 1×105 | Hydrogel | MG | No | N/a | 15.7 |
27-3 PTCT | CD140-tdTomato+ | 1 | Foci, radial outgrowth | 4 | 30d | 7 | 5×105 | Hydrogel | MG | No | <1 mm | 19.5 | Normal ductal epithelium |
27-3 PTCT | CD140-tdTomato+ | 7 | 5×105 | Hydrogel | SC (Rear) | No | N/a | 19.5 | N/a |
28-3 PTCT | CD140- | 1 | Elongated, some foci | 4 | 30d | 10 | 5×105 | Hydrogel | MG | No | 15.3 ± 1.5 | 18.5 | Fibrosis, vimentin positive |
28-3 PTCT | CD140- | 10 | 5×105 | Hydrogel | SC (Rear) | No | 12.3 ± 1.1 | 18.5 |
27-1 PTCT | CD140-tdTomato+ | 1 | Cobblestone, no foci | 4 | 30d | 11 | 5×105 | Hydrogel | MG | No | N/a | 36.4 | N/a |
27-1 PTCT | CD140-tdTomato+ | 11 | 5×105 | Hydrogel | SC (Rear) | No | N/a | 36.4 |
28-6 PTCT | CD140-tdTomato+ | 1 | Foci, rapid proliferation | 4 | 30d | 11 | 5×105 | Hydrogel | MG | No | <1 mm | 42 | Fibrosis, squamous epithelium |
28-6 PTCT | CD140-tdTomato+ | 11 | 5×105 | Hydrogel | SC (Rear) | No | N/a | 42 | N/a |
ss020513_1 PTCT | CD140- CD49f+ | 1 | Cobblestone, no foci | 4 | 102d | 3 | 4×105 | Hydrogel | SC (Shoulder) | No | 5.3 ± 0.6 | 32 | Glandular, squamous epithelium, CK8/18 positive |
ss020513_1 PTCT | CD140- CD49f+ | 242d | 3 | 1×106 | Matrigel | SC (Flank) | Yes | 6.0 ± 0.4 | 10 |
ss020513_2 PTCT | CD140- CD49f+ | 1 | Cobblestone, no foci | 4 | 3 | 8×105 | Matrigel | SC (Flank) | Yes | 6.2 ± 1.2 | 10 |
ss082112_PTCT | CD140- | 6 | Cobblestone, no foci | 4 | 59d | 10 | 1×106 | Matrigel | MG | Yes | <1 mm | 10.6 | Normal ductal epithelium |
ss082112_PTCT | CD140- | 6 | 11 | 1×106 | Matrigel | SC (Rear) | Yes | 7.21 ± 0.2 | 10.6 | Glandular, squamous epithelium |
Pigs in study two received daily 17β-estradiol injections (IM, 0.1 mg/kg; Sigma Aldrich, St. Louis, MO) for 7d after lentivirus instillation to stimulate MEC proliferation [
9]. Similarly, pigs in studies three and four received daily 17β-estradiol for 7d prior to excision of mammary tissue. Upon reinstillation of lentivirus-transduced cells, a 17β-estradiol/cholesterol pellet (0.05 mg/kg) was placed subcutaneously, reducing the need for daily hormone injections. All pigs serving as donors for mammary tissue received penicillin intramuscularly 24 h prior to tissue collection to reduce the potential bacterial contamination of cultures. NSG mice carrying pMEC lines ss020513_1 and ss020513_2 (
n = 4) and ss082112 (
n = 4) received a pellet containing 2 μg 17β-estradiol and 0.75 mg progesterone (Sigma-Aldrich) at the time of cell injection to promote the proliferation of engrafted cells.
Primary mammary cell isolation
In studies one and four, immediately following exsanguination of pigs, the skin was disinfected, the nipple retracted and ~1 g of mammary tissue excised (
n = 5 glands). In study three, pMEC were obtained by removing endogenous parenchyma from six mammary glands from five-week old pigs under isoflurane anesthesia using a cleared mammary gland procedure essentially as described [
20]. Analgesic (banamine, 2–5 mg/kg) was administered postoperatively.
Organoid preparation
Minced mammary tissue was digested (1.5 mg/ml collagenase A (Roche, San Francisco, CA; 75 μg/ml DNase I, Roche; 1 mg/ml hyaluronidase, MP Biomedicals, Santa Ana, CA) in growth media (10 % fetal bovine serum [FBS], DMEM/F-12, penicillin G/streptomycin sulfate/amphotericin B) at 37 °C for 3 h. Organoids (40–100 μm diameter) were plated in primary porcine mammary epithelial media (modified from MEGM [
21] as a 1:1 mix of MCDB170 (US Biological, Salem, MA) and DMEM/F-12 (CellGro, Manassas, VA) with penicillin G/streptomycin sulfate/amphotericin B, 0.5 % FBS, bovine insulin (7.5 μg/mL, Sigma-Aldrich), human EGF (5 ng/mL, Millipore, Billerica, MA), hydrocortisone (0.25 μg/mL, Sigma-Aldrich), human apo-transferrin (2.5 μg/mL, Sigma-Aldrich), ethanolamine (0.1 mM, Sigma-Aldrich), o-phosphoethanolamine (0.1 mM Sigma-Aldrich), bovine pituitary extract (35 μg/mL, Gemini Bio-Products, West Sacramento, CA), and lipid-rich bovine serum albumin (0.1 %, Gemini Bio-Products).
Cell culture
Primary pMEC were maintained in porcine mammary epithelial media or growth media, and were differentially trypsinized to reduce the number of contaminating fibroblasts [
21]. HEK293FT cells (Invitrogen, Grand Island, NY) were maintained in HEK293FT media (DMEM, 10 % FBS, penicillin G/streptomycin sulfate, non-essential amino acids and 1 mM sodium pyruvate). NIH/3 T3 cells (ATCC) were maintained in high glucose DMEM (Hyclone Laboratories, GE Healthcare Life Sciences, Logan UT) with 10 % FBS, 10 mM Hepes, 1 mM sodium pyruvate, penicillin G/streptomycin sulfate.
Vectors and virus
The human elongation factor 1α (EF1α)-tdTomato and phosphoglycerate kinase (PGK)-tdTomato plasmids were generated from pLVX-IRES-tdTomato (cytomegalovirus (CMV)-tdTomato; Clontech, Mountain View, CA) by cloning the EF1α promoter from pEF6/myc-His C (Invitrogen) or the human PGK promoter (GenBank:NG_008862.1) from the pMNDU3-PGK-Luc plasmid (UC Davis Vector Core, Sacramento, CA) by PCR (Additional file
1: Figure S1A). The PGK-T antigens (Tag)-CMV-tdTomato construct (Additional file
1: Figure S1B) was generated by first constructing pLVX-PGK-Tag. IRES-tdTomato was excised from pLVX-PGK-tdTomato to give pLVX-PGK. The Tag sequences encoding mouse polyomavirus small Tag (ST), middle Tag (MT) and large Tag (LT; GenBank:J02288) were amplified from p53.A6.6 (pPY-1; ATCC, Manassas, VA) and ligated into pLVX-PGK to generate pLVX-PGK-Tag. The CMV promoter was excised from pLVX-IRES-tdTomato and ligated into pLVX-PGK-Tag to generate pLVX-PGK-Tag-CMV. The tdTomato coding sequence was amplified from pLVX-IRES-tdTomato and ligated into pLVX-PGK-Tag-CMV to generate pLVX-PGK-Tag-CMV-tdTomato. All constructs were sequence verified.
Lentiviral supernatants for PGK-Tag-CMV-tdTomato were prepared by the University of California San Francisco viral core and titrated by flow cytometry (9.3 × 107 transduction units [TU]/ml). CMV-tdTomato viral supernatant was produced by triple transfection of HEK293FT cells in Opti-MEM I (Invitrogen) with packaging vector (16 μg/150 cm2 cells; pCMV-dR8.91, UC Davis Vector Core), envelope vector (3.2 μg/150 cm2 cells; pMDG-VSVG, UC Davis Vector Core) and transfer vector (16 μg/150 cm2 cells; pLVX-IRES-tdTomato) with Lipofectamine 2000 (Invitrogen). Lentiviral particles were concentrated by 100 kDa cut-off centrifugation (Millipore).
Viral stocks were titrated in HEK293FT cells using qPCR [
22]. The woodchuck hepatitis virus posttranscriptional regulatory element (WPRE) was used to measure integration and values normalized to the number of human albumin copies (WPRE Fwd GCGTCTGGAACAATCAACCT and Rev GGCATTAAAGCAGCGTATCC; hAlbumin [GenBank:152112963] Fwd GTGCTGCCTCGTAGAGTTTTCTG and Rev TCAATAGCCATGTGACCAGTGACT).
Fluorescence activated cell sorting
Primary pMEC cultures were expanded for 13-14d post-isolation with two differential trypsinizations using Accutase (Innovative Cell Technologies, Mira Mesa, CA) to remove fibroblasts, followed by Accumax (Innovative Cell Technologies) to dislodge pMEC. Single cells were incubated with phycoerythrin-conjugated anti- CD140a (BD Biosciences, Franklin Lakes, NJ) and/or biotinylated anti-human CD49f (AbD Serotec, Oxford, UK), followed by streptavidin-Alexa 488 (Jackson Immunoresearch) and propidium iodide (10 μg/ml) and sorted using a MoFlo cell sorter (Cytomation, West Lafayette, IN).
Viral transduction in vitro
Adherent pMEC were transduced overnight using a multiplicity of infection (MOI) of 100 for PGK-tdTomato, EF1α-tdTomato or CMV-tdTomato, or an MOI of 20 for PGK-Tag-CMV-tdTomato, along with polybrene (6 μg/mL; Millipore). For study four, cultures of CD140a- pMEC were transduced with PGK-Tag (ss071712), pLVX-PGK-Tag-CMV-tdTomato (ss082112, 27–1, 27–3, 28–3) or CMV-tdTomato (ss071712). Two pMEC lines (ss020513_1, ss020513_2) were sorted to be CD140a-/CD49f + then transduced with PGK-Tag-CMV-tdTomato. A subset of CD140a- pMEC transduced with PGK-Tag-CMV-tdTomato were sorted for tdTomato + either 7d (27–3, 28–3) or 17d later (27–1, 28–6) depending on the initial number of CD140a- pMEC.
Viral transduction in vivo
In study two, saline or lentivirus (CMV-tdTomato, 5 × 106 TU; EF1α-tdTomato, 1 × 107 TU and PGK-tdTomato, 5 × 106 TU with or without 6 μg/ml polybrene was instilled into the left (with polybrene; one gland/treatment) and right (without polybrene; one gland/treatment) thoracic and abdominal mammary glands of pigs (n = 9) via one of the two mammary ducts (intraductal) under isoflurane anesthesia. Additionally, saline or lentivirus suspension with 6 μg/ml polybrene was injected directly into the mammary parenchyma (20–25 mm subcutaneously; intramammary) in the four remaining mammary glands of each pig (one gland/treatment). Mammary glands were harvested at necropsy 5, 10 or 15 d later, and snap frozen or fixed in 4 % paraformaldehyde.
Ex vivo transduction and grafting
In study three, dissociated mammary organoids (n = 30) were transduced overnight (MOI = 100) with PGK-tdTomato, CMV-tdTomato, EF1α-tdTomato, CMV-Tag, EF1α-Tag, PGK-Tag or no vector (control) with polybrene (6 μg/ml). After 24 h (n = 4 pigs) or 8d (n = 6 pigs) cultures of organoids were trypsinized and resuspended in serum-free media. Donor pigs were anesthetized with isoflurane and cells reinstilled via two intramammary injections/gland (0.9–4.5×105 cells/injection; n = 2 glands/construct/pig) and each site closed with Vetbond (3 M, St. Paul, MN). The isografted mammary glands were harvested 3–5 weeks later, minced, and randomly divided for snap freezing in liquid N2 or fixation in 4 % paraformaldehyde.
In study four, lentivirus-transduced pMEC were injected into NSG mice under isoflurane anesthesia. A range of 0.4-1 × 10
6 pMEC resuspended in 20 μL of HyStem-C hydrogel (
n = 46 injected sites; Glycosan Biosystems, Alameda, CA) or Matrigel HC (
n = 24 injected sites; BD Biosciences) was injected either subcutaneously or directly into the mammary gland. For instillation of cells into the mammary gland, a small skin incision (~5 mm) was made to visualize accurate placement within the mammary fat pad. Mice were treated postoperatively with a single dose of analgesic (buprenorphine; 0.05 mg/kg) and were monitored once daily over 7d for changes in health and behavior. Table
1 summarizes the mice and cell injections used.
In vitro assays
Cell number was assayed using a methylene blue assay [
23]. Cells were plated in 96 well-plates at 2000 cells/well (
n = 6/cell line) on d0, and medium changed every 2d.
Cells (20,000/well) were resuspended in 0.35 % agar in growth medium and poured onto a base layer (0.7 % agar in growth media), with growth medium changed every 2d. After 21d, cells were stained (0.04 % crystal violet, 2.1 % citric acid), imaged and colonies >50 μm counted using ImageJ (NIH;
http://rsb.info.nih.gov/ij/).
Mammary gland and tumor whole mounts
Semi-thick tissue sections were dehydrated through graded ethanols to xylene (study three) or graded glycerol (study four) as described [
24]. Sections were imaged using a fluorescent dissecting microscope. The percent red area was calculated using Image J. Regions positive for red fluorescence were dissected and processed to paraffin for hematoxylin and eosin staining (H&E) and immunohistochemistry. Regions having dense ductal structures (study three, PGK-Tag mammary glands) were microdissected, paraffin-embedded and sectioned for histology (H&E) and genomic DNA extraction.
Western blotting
Cells (1–3 × 10
6) were lysed and sonicated in buffer with protease and phosphatase inhibitors, and western blots performed as described [
25]. Antibodies were from Santa Cruz Biotechnology (rat polyomavirus early, cyclin D1, and p53), Cell Signaling Technology (Rb, phospho Rb, MAPK1/3 and phospho MAPK1/3)) and Jackson ImmunoResearch (HRP-conjugated secondary antibodies).
Genomic DNA extraction and PCR detection of integration
Genomic DNA was extracted from paraffin-embedded tissues as described [
26] using combined heating and non-heating protocols. DNA quality was assessed using primers for the porcine prolactin receptor gene [
27]. The incidence of lentiviral integration in studies two and three was determined using primer sets specific to tdTomato (Clontech; tdTomatoFin Fwd CTCCGAGGACAACAACATGG and Rev CTTGGTCACCTTCAGCTTGG; CMVtdTomato Fwd AACACGATGATAATATGGTGAGCAAGGG and TdTomatointernal Rev GACAGCTTCTTGTAATCGGGGATGTC) or amplified a product specific for PGK-Tag (PGKTagspec5P TGAAGATGTAAAGGGTCAAATAGC and PGKspec3P-2 AAGGCATTAAAGCAGCGTATC).
RT-qPCR and qPCR (study two)
Total RNA was extracted and reverse transcribed as described [
28]. Primers spanned across exons of LEF1 (Fwd GACGAGCACTTTTCTCCAGGA, Rev TAATCTGTCCAACACCACCCG [GenBank:XM_005666939]), cyclin D1, (Fwd CCCTCCGTGTCCTACTTCAA, Rev CAGGCGGCTCTTTTTCAC [GenBank:AK400348)], cyclin A2, (Fwd TTGTGGGCACTGCTGCTATG, Rev GCAAGGACTTTCAAAACGAGGTG [GenBank:GQ265874]), MYC, (Fwd CGCTTTTTGGACGCTGGATT, Rev TTCTCCTCCTCGTCGCAGTA [GenBank:X97040]), RB1 (Rb), (Fwd ACGCCAACAAAAATGACTCC, Rev GTTGCCTCCTTCAGCACTTC [GenBank: JX099502]), TP53 (p53), (Fwd CCATCCTCACCATCATCACACT, Rev CTCTGTGCGGCGGTCTCT [GenBank:NM_213824]), P21, (Fwd GCAGACCAGCATGACAGATT, Rev TGTTTCCAGCAGGACAAGG [GenBank:XM_001929558]), P16, (Fwd GAGGGCTTCCTGGACACTTTG, Rev TGCAGTATCTCTGGGTTTCAATGA; [GenBank:AJ316067]) and 18S ribosomal RNA, (Fwd ACGGCTACCACATCCAAGGA, Rev CCAATTACAGGGCCTCGAAA [GenBank: AF179868]). All PCR products were sequenced. RT-qPCR was as described [
28], where relative transcript abundance was calculated using a 5-point standard curve obtained by 5-fold serial dilutions of a pMEC complementary DNA pool. The average relative expression for each sample was normalized to 18S ribosomal RNA levels [
29].
DNA was purified from tissues homogenized in Tri-Reagent (Molecular Research Center, Inc, Cincinnati, Ohio). Genomic DNA (40 ng) was amplified with tdTomatoFin primers using qPCR [
28]. Standard curves were generated from genomic DNA extracted from mouse mammary tumor cells (SSM-2) transduced with EF1α-tdTomato lentivirus and selected for tdTomato expression using a MoFlo cell sorter. The relative number of integrated virus particles was normalized to the corresponding level of 18S ribosomal DNA as a loading control for gene copy number (Fwd ACGGCTACCACATCCAAGGA, Rev CCAATTACAGGGCCTCGAAA [Genbank: NR_046261]).
Immunohistochemistry
Slides were prepared as described [
28], with modifications. Sections were incubated with anti-dsRED (1:50; Clontech), anti-human progesterone receptor (1:50; DakoCytomation, Carpinteria, CA), anti-human estrogen receptor (clone 6 F11, ThermoFisher Scientific, Waltham, MA), anti-vimentin (1:100; Millipore), anti-bovine cytokeratin 8/18 (1:2000; Fitzgerald Industries, Acton, MA), or anti-Ki67 (clone Ab-4; 1:1000; ThermoFisher Scientific) in 5 % horse serum in PBS at 4 °C overnight and detected with NovaRED (Vector Laboratories) or DAB (Invitrogen).
Statistics
Differences were assessed by two-way ANOVA, and
P-values calculated using Students
t-tests. Data is presented as means ± SEM, with significance at
P < 0.05. For animal studies, individual mammary glands were treated as the experimental unit. Animal group sizes were selected to provide >80 % power to detect differences, taking into consideration our experience with assessing porcine mammary gland morphology [
29] and the typical engraftment characteristics and rates of primary bovine and human MEC in immunodeficient mice [
30,
31].
Discussion
Here we report the first successful lentivirus-mediated transgenesis and transformation of primary pMEC
in vitro and
in vivo. Given significant structural and functional similarities between the mammary glands of pigs and humans [
9], our approach is a first step toward a promising animal model in which to investigate the tissue-level cellular and environmental interactions behind human breast development and oncogenesis.
A variety of genetically-engineered mouse models has enabled the identification of various genes involved in mammary cancer initiation and progression, yet the resulting tumors often differ in their pathology compared to human breast cancers [
33]. This discordance may reflect the greater frequency at which mouse tumors develop from alveolar structures [
33] compared to those of the human breast that often arise from less-differentiated TDLU-1 and -2 [
17]. As such, investigations into the initiation and progression of tumors within TDLU have been hampered by the absence of similar structures in the mouse mammary gland alongside the challenges associated with obtaining human tissues. Development of a pig breast cancer preclinical model stands to complement recent advances to optimize the limitations of mouse models such as the addition of human stromal elements to the mouse mammary fat pad [
34] and reconstitution of the mouse mammary epithelium with human preneoplastic cells [
35].
Our techniques enable transformation of MEC derived from the TDLU of pre- and peripubescent pigs that are synonymous with TDLU-1 and -2 in the human breast [
9]. Moreover, vacuum-assisted mammary biopsies [
19], which are impractical in rodents [
36], will allow serial sampling of a pig model of breast cancer to study the response to cancer treatment over time, or the influence of environmental and lifestyle factors on breast cancer risk. While a porcine pre-clinical model has limitations including increased cost and housing requirements [
36], these drawbacks are countered by the potential ability to authentically model normal human breast development and breast cancers within a context that is biologically and physiologically more similar to the human.
The human breast and pig mammary gland have a similarly complex stromal microenvironment, which contrasts to that of the rodent mammary gland [
18]. Our results highlight the influence of stromal factors on the occurrence and rate of pMEC tumor formation
in vivo. Similar to transformed human breast epithelial cells [
37], transformed pMEC failed to form tumors in the microenvironment of the mouse mammary fat pad despite developing tumors ectopically. Our data corroborates previous evidence showing that elements of the mouse mammary fat pad are inadequate to support proliferation of non-native epithelium [
18]. These data, alongside work by others [
34,
31,
30], adds to evidence highlighting differences between species in the stroma-directed behavior of the mammary epithelium. Indeed, attempts to reconstruct human [
31] or bovine [
30] mammary tissues in the mouse mammary fat pad by xenografting tissue fragments or dissociated MEC failed to generate species-specific TDLU.
Given that the pig mammary gland is rich in connective tissues, co-inoculating irradiated pig stromal fibroblasts with transformed pMEC may have increased their viability when xenografted into the mouse mammary fat pad, as for human and bovine MEC [
34,
38]. Along these same lines, Matrigel promoted transformed pMEC to form tumors, consistent with the stromal requirements of xenografts of human breast cancer cells [
39]. The matrix proteins found in Matrigel can support the growth of cells in a foreign environment, recapitulating the connective-tissue rich stroma of the human breast and pig mammary gland [
18,
40]. However, unlike in humans, mammary carcinomas are extremely rare in pigs [
41]. This low incidence may reflect innate repressive effects of the stroma, which can restrain the tumorigenicity of undifferentiated mouse embryonic carcinoma cells [
42] and prevents marginally abnormal human mammary organoids developing hyperplasias in humanized mouse mammary fat pads [
34]. The structure and morphology of tumor stroma associated with xenografts differs from normal stroma [
43], suggesting that changes within tumor stroma may be more permissive. Accordingly, transformed porcine dermal fibroblasts readily formed undifferentiated sarcomas in the mammary glands of immunosuppressed pigs [
44], while we found that isogenic transplanted pMEC transformed by Tag failed to form palpable tumors in immunocompetent pigs (data not shown). While immunosuppression may have increased the take of transformed pMEC [
44], it is also possible that transformation of stromal cells is also required to induce mammary carcinomas in pigs. As such, future work to overcome these repressive mechanisms and develop the pig as an authentic breast cancer model may open up hitherto unexplored mechanisms to defeat breast cancer.
We transformed pMEC by lentivirus-mediated expression of murine Tag proteins. While Tag or similar viral oncoproteins have not been directly implicated in human breast carcinogenesis, Tag activates various oncogenes and bypasses p53 and Rb [
45,
46] as in many human breast cancers [
47,
48]. Mammary tumors that develop from germline transgenic MT expression in the mouse mimic invasive human breast carcinomas with progressive loss of ER coincident with increased invasiveness [
49]. Herein, Tag-transformed pMEC developed estrogen receptor- negative, proliferative tumors when xenografted to mice. Interestingly, pMEC co-injected with Matrigel into estrogen-treated mice displayed a shorter latency to tumor formation, raising the possibility that transformed pMEC may have initially been hormone-sensitive. Given that mice carrying pMEC co-injected with hydrogel were not treated with estrogen, we cannot ascribe any differences in tumor formation rates to hormone supplementation. Nevertheless, our observations are consistent with similar findings in a mouse explant model of transgenic MT [
50].
Previous reports indicate that primary human MEC require disruption of multiple pathways for complete transformation [
37,
51,
35] whereas rodent cells can be transformed by two oncogenes [
52]. Along these lines, primary pig fibroblasts [
44] and human breast epithelial cells [
51] were transformed by sequential transduction with dominant negative p53, activated CYCLIN-dependent kinase complex (Cyclin D1/CDK4), c-Myc, H-Ras and telomerase. Our approach provides for the efficient induction of tumorigenesis in primary pMEC by lentivirus-mediated expression of Tag, which upregulates multiple oncogenic pathways. This approach led to the formation of heterogeneous tumors, perhaps as a consequence of lentiviral transduction. Along these lines, transgenic expression of MT in the mouse mammary gland induced homogenous luminal, solid adenocarcinomas [
53] whereas low levels of lentivirus-directed MT transformed both luminal and basal MEC, leading to concomitant heterogeneity among the tumors [
54]. One further explanation may be variable Tag expression within pMEC subpopulations. Consistent with this postulate, the degree of Ras expression in transformed primary human MEC correlated with tumorigenicity
in vivo [
37]. Interestingly human MEC transformed with Ras-SV40 ST/LT-hTERT formed poorly differentiated squamous tumors when xenografted to the mammary fat pad of immunocompromised mice [
37], similar to our findings with xenografted pMEC transformed by Tag. Although squamous differentiation is rare among human breast carcinomas, it is a phenotype that has been linked to Wnt activation [
55], which has been associated with basal-like and receptor negative breast cancers [
56]. Transformation of pMEC by alternative oncogenes may yield tumors with distinct histopathologies and hormone receptor profiles. In addition, our data indicate that selecting pMEC for CD49f + enhanced the subsequent formation of luminal breast carcinomas, yielding tumors with luminal differentiated epithelium, whereas unenriched pMEC developed as normal ductal structures or fibrous tumors with and without squamous epithelium. It is possible that selection for CD49+/EpCAM+ cells in combination with culture conditions to increase ErbB3 expression over EGFR expression may increase the likelihood of generating luminal adenocarcinomas [
35].
Acknowledgements
We thank K. Parker, C. Sehnert and D. Sehnert for assistance with animals and facilities, E. Martínez-Alvarez, N. Breuner, A. Hanson, S. Santos, T. Lazzari and K. Scheier for technical assistance. O. Aina, N.E Hubbard, J.E. Walls at the UC Davis Center for Comparative Medicine performed the H&E stains, immunohistochemistry and provided imaging assistance. This work was supported by a grant from the Department of Defense (BC098018, to R.C. Hovey and R.D. Cardiff), and the University of California, Davis Henry A. Jastro Research Scholarship Award (A.R. Rowson-Hodel). We gratefully acknowledge the infrastructure support of the Department of Animal Science, College of Agricultural and Environmental Sciences, and the California Agricultural Experiment Station of the University of California-Davis.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made.
The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
AR-H and JT prepared the manuscript and RH critically reviewed the manuscript. AR-H, JT, RM and RH designed the experiments, AR-H, RM and JT conducted the experiments and AB and RC provided technical expertise and reviewed tumor histopathology. All authors read and approved the final manuscript.