Ethics statement

All experimental procedures were performed in accordance with relevant guidelines and regulations and were approved by the Institutional Animal Care and Use Committee (IACUC) at Cornell University (#2013–0022). Tissues obtained from equine (Equus caballus) and canine (beagles, Canis lupus familiaris) donors were recovered after euthanasia. All donors were euthanized for reasons unrelated to this study and all tissues were extracted post-mortem, therefore, IACUC approval was not required for the mammary gland tissue collections.

Generation and cryopreservation of mammary tissue fragments for xenotransplantation

Mammary gland tissues were recovered from clinically healthy, intact, virgin animals, that were euthanized for reasons unrelated to this study. Tissues were obtained from equine (Equus caballus) donors (various breeds) aged between 2 to 16 years, consisting of 5 cm² equine mammary gland tissue pieces bordering the mammary gland compartments, and canine (Canis lupus familiaris) donors (beagles) aged ~ 15 months, consisting of 2 cm² mammary gland tissue pieces from the mammary duct near the nipple. The large age range in equine tissue donors was due to the need to opportunistically obtain scare equine tissues for xenotransplantations, thus, tissues were extracted from animals of variable age. Notably, studies assessing multiple xenotransplanted tissue types have shown that donor age may impact xenograft function and behavior [34], however, it is well-acknowledged that mammary gland tissues from aged donors can still regenerate fully functional mammary glands [35]. All tissues were extracted at Cornell College of Veterinary Medicine (CVM) or at the Cornell Animal Health Diagnostic Center (AHDC). Reproductive history was validated by caretaker/veterinarian documentation records.

Mammary tissue fragments were generated as previously described [4, 36, 37]. Briefly, tissues were mechanically digested using surgical scissors and then enzymatically digested for 3 h at 37°C in a solution of Dulbecco’s modified Eagle’s medium (DMEM) (cat #: 10-014-CV) /Ham’s F12 (cat #: 10-080-CV) (Corning Inc., Sommerville, MA) (DMEM/F12) (50/50), containing collagenase (300 U/ml) (cat #: LS004176) (Worthington, Lakewood, NY), Hyaluronidase (100 U/ml) (cat #: H3506) (Sigma Aldrich, St. Louis, MO), 5% fetal bovine serum (FBS) (cat #: S11550) (R&D Systems, Minneapolis, MN), insulin (5 μg/ml) (cat #: 12585–014) (Sigma Aldrich), and hydrocortisone (1 μg/ml) (cat #: H2270) (Sigma Aldrich). Digested tissues were then triturated with a 25 ml serological pipette (cat #: 75816–090) (VWR, Radnor, PA) to complete digestion. Mammary tissue fragments were then washed in a solution of Hank’s Buffered Saline Solution (HBSS) (cat #: 21-022-CM) (Corning Inc.) containing 2% FBS (R&D Systems) (2% HFBS), then suspended in a solution of FBS (R&D systems) plus 10% dimethyl sulfoxide (DMSO) (cat #: D2650) (Sigma Aldrich), frozen in cryovials at -80°C in isopropanol cryochambers for 24 h, and then stored in liquid nitrogen for long term storage. Prior to use, cryovials were thawed at 37°C for 1–2 min, and mammary tissue fragments were washed in sterile PBS (cat #: 21-040-cv) (Corning Inc.) and kept on ice until xenotransplanted. Tissue fragments of ~200–400 μm in diameter were selected for xenotransplantation and inserted into the cleared mammary fat pad (MFP), as described below, within 0.5 to 3 h after thawing.

Primary mammary fibroblast culture and injections

Primary early passage (≤ passage, P5) mammary fibroblasts, isolated from cryopreserved mammary tissue fragments and maintained in (DMEM) (Corning Inc.) containing 10% fetal bovine serum (FBS) (R&D Systems) and 1% Pen-Strep (cat #: 30-002-CI) (Corning Inc.) (fibroblast culture media), were cultured for ≤ 14 days prior to being injected into cleared mammary fat pads [38]. For fibroblast isolation, mammary tissue fragments were thawed and resuspended in HBSS (Corning Inc.) containing 2% FBS (R&D Systems) (2% HFBS), then washed in HBSS (Corning Inc.). Fragments were then triturated for 5 min in pre-warmed 0.25% Trypsin-EDTA (cat #: 25-053-CI) (Corning Inc.) and washed in 2% HFBS. To generate single cell suspensions, Dispase II (final conc. 5 U/ml) (cat #: D4693) (Sigma Aldrich) and DNase I (final conc. 0.25 mg/ml) (cat #: LS002138) (Worthington) were added to the solution and fragments were triturated for 5 min. After adding 2% HFBS, single cell suspensions were passed through a 100 μm cell strainer (cat #: 352360) (Corning Inc.), and then a 40 μm cell strainer (cat #: 352340) (Corning Inc.), to remove debris. Cells were counted using a hemacytometer and plated at 3 x 10⁶ cells/well on an adherent 6-well tissue culture plate (cat #: 3516) (Corning Inc.) for 1 h to enrich for adherent fibroblasts [4]. Cells that remained suspended were separated and cultured as mammosphere-derived epithelial cells (MDEC, see below), and adherent fibroblast cultures were trypsinized and re-plated to semi-confluency for injection. After trypsinization, fibroblasts were resuspended at 5 x 10⁵ cells in 50 μl of 1:1 HFH (HBSS (Corning Inc.), 2% FBS (R&D Systems), 0.01M HEPES (cat #: 25-060-CI) (Corning Inc.)):Matrigel Matrix Basement Membrane (8.4 mg/ml protein concentration) (cat #: 354234) (Corning Inc.) [24, 29], and kept on ice until injection into the cleared fat pad. Injections were performed using a 28G needle (cat #: 329424) (BD Pharmingen, Franklin Lakes, NJ). Vehicle injections were performed using a cell-free solution of 50 μl of 1:1 HFH (HBSS (Corning Inc.), 2% FBS (R&D Systems), 0.01M HEPES (Corning Inc.)): Matrigel Matrix Basement Membrane (8.4 mg/ml protein concentration) (Corning Inc.). Following implantations, equine and canine primary mammary fibroblasts were allowed to engraft for 2 weeks prior to analysis and/or mammary tissue fragment xenotransplantation.

Cleared fat pad surgeries and mating of immunodeficient mice

NOD-scid gamma (NSG) (NOD.Cg-Prkdc^scid Il2rg^tm1Wjl/SzJ) mice were purchased from the Cornell Progressive Assessment of Therapeutics (PATh) patient-derived xenograft (PDX) Facility or the Jackson Laboratory (Bar Harbor, ME) and maintained on-site under sterile conditions. All mice were supplied with food (e.g., standard chow) and water ad libitum. Fat pad clearing surgeries were performed on 3–4-week-old female mice that were anesthetized with isoflurane (cat #: 200–129) (Dechra Pharmaceuticals, Northwich, UK) mixed with O₂ (3% vol/vol in induction chamber, then maintained at 2–2.5% vol/vol on nose cone) and received a subcutaneous injection of ketoprofen (3 mg/kg) (cat #: 07-803-7389) (Patterson Veterinary Supply, Loveland, CO), followed by an additional abdominal subcutaneous injection of bupivacaine (5 mg/kg) (National Drug Code #: 0409-1163-18) to reduce post-procedural pain. Timelines for surgical procedures are detailed in S1 Fig. Timeline for baseline surgeries (without additional interventions, i.e., without induced pregnancy or fibroblast injections) is presented in S1a Fig. All surgical procedures were performed on a warming pad within a sterile biosafety cabinet. While under anesthesia, the developing mammary epithelia in the 4^th inguinal MFPs were removed to de-epithelialize (clear) the MFPs [21, 24, 29]. Excised portions of the 4^th inguinal MFPs were mounted on glass slides and stained by acetocarmine (1.25 mg/ml carmine powder (cat #: C1022) (Sigma Aldrich) in 45% Glacial acetic acid (cat #: A38-212) (Thermo Fisher, Waltham, MA) (see below)) to visualize extracted murine mammary glands and confirm successful clearing. Immediately after clearing, a small incision was made in the remaining fat pad and a donor equine or canine mammary tissue fragment (~200–400 μm in diameter) containing mammary parenchyma and stroma was inserted. Incisions were subsequently sealed with surgical staples (cat #: INS750344) (Cellpoint Scientific, Inc., Gaithersburg, MD) and mice recovered on a warming pad until mobile. Following surgery, mice were given an oral antibiotic suspension of sulfamethoxazole (0.4mg/ml) and trimethoprim (0.08mg/ml) (cat #: 0121-0854-16) (PAI Pharma, Greenville, SC), dissolved in drinking water, for a minimum of 10 days. All mice received additional subcutaneous injections of ketoprofen (3 mg/kg) (Patterson Veterinary Supply) at 24 and 48 h after surgery to alleviate pain. Mice that underwent surgery were housed in cages of ≤ 4 mice and were monitored daily for signs of distress and to ensure proper healing of the abdominal incisions. No signs of distress were observed during the xenograft engraftment period and no challenges were observed (e.g., infection, tumor formation).

For mating procedures, female mice bearing xenografts (6 weeks post-surgery) were paired and placed in mating cages with an adult male NSG mouse. Females were checked for vaginal copulation plugs daily to determine days post-coitus (dpc) and monitored for changes in external appearance to determine gestation progression [39]. For experiments that assessed the mammotropic effects of circulating pregnancy hormones, pregnant mice were euthanized at 18 dpc (S1b Fig). For mice that received primary mammary fibroblast injections, fibroblasts were injected immediately post-clearing, and mammary tissue fragment xenotransplantation surgeries occurred 2 weeks after MFP clearing and fibroblast injections, when mice were aged 5-6-weeks old (S1c Fig). All mice were euthanized by CO₂ asphyxiation, with CO₂ flow maintained for 60 s after respiratory arrest, followed by cervical dislocation. Cleared MFPs (4^th inguinal MFPs) containing mammary xenografts were then removed and fixed for analysis. All xenografts were assessed ~ 9 weeks after xenotransplantation surgeries, unless otherwise noted.

Histological staining and immunohistochemistry (IHC)

MFPs containing xenografts were fixed in 10% neutral-buffered formalin (10% NBF) (cat#: HT501128-4L) (Sigma Aldrich) for 24 h at room temperature (RT), stored in 70% EtOH (cat#: 2716GEA) (Decon Labs, Inc., King of Prussia, PA), embedded in paraffin, and sectioned for histological analysis. Picro-Sirius (Sirius) red (cat #: VB-3017) (VitroView, Rockville, MD) and Masson’s Trichrome (MTC) (cat #: HT15-1KT) (Sigma Aldrich) stainings were performed, as per manufacturer’s instructions. Hematoxylin and eosin stainings (H&E) were performed by the Animal Health Diagnostic Center (AHDC), Cornell University College of Veterinary Medicine.

For immunohistochemistry (IHC) analysis of β-lactoglobulin (β-LG), Cytokeratin-14 (CK14), Cytokeratin-18 (CK18), α-Smooth Muscle Actin (α-SMA), and Estrogen receptor-α (ERα) presence, formalin-fixed paraffin embedded (FFPE) tissue sections were deparaffinized in xylene (cat #: 89370–088) (VWR)and rehydrated in EtOH (Decon Labs, Inc.) and H₂O. Rehydrated tissues were permeabilized in Tris-buffered saline (TBS), containing 0.5% Tween-20 (cat #: J20605-AP) (Sigma Aldrich) when assessing cytoplasmic antigens and in 0.3% Triton X-100 (cat #: J66624) (Thermo Fisher) when assessing nuclear antigens. Antigen retrieval, using a citric acid buffer (10mM citric acid (cat #: C-8532) (Sigma Aldrich), 0.5% Tween-20 (Sigma Aldrich), pH = 6.0), was performed by immersing tissues in 95–100°C buffer for 20 min. Tissues were blocked with a TBS solution containing 1% bovine serum albumin (BSA) (cat #: A3608) (Thermo fisher) and 10% normal goat serum (cat #: 102643–594) (VWR) for 30 min at 37°C. Primary antibodies were diluted in a TBS solution containing 1% BSA (Thermo Fisher) and incubated overnight at 4°C. Endogenous peroxidase activity was blocked by incubating tissues in 0.3% H₂O₂ (cat #: H325-500) (Thermo Fisher) for 15 min at RT. Secondary antibodies were diluted as described above (TBS containing 1% BSA (Thermo Fisher)), and incubated for 1 h at RT. Antibody information and dilution concentrations are detailed in S1 Table. To improve immunolabelling signal for CK14 and CK18, tissues were incubated with anti-mouse IgG (H+L), biotin-SP (Jackson, West Grove, PA) for 1 h at RT, washed, and then incubated with Streptavidin, Horseradish peroxidase (HRP)-conjugated (Sigma Aldrich) for 1 h at RT. Colorimetric signals were developed with 3-Amino-9-Ethylcarbazole (AEC) substrate kit (cat #: 551015) (BD Pharmingen) and tissues were counterstained with Gill’s hematoxylin (cat #: 6765007) (Thermo fisher). Isotype control antibodies (S1 Table) were incubated with tissue serial sections at identical concentrations as corresponding primary antibodies to ensure that immunofluorescent (IF) and IHC signals were not due to non-specific binding. IHC labeling of vimentin and Marker of proliferation Ki-67 (Ki-67) was performed by Cornell AHDC. Briefly, FFPE tissue sections were processed using a Bond Max automated IHC stainer (Leica Biosystems, Nussloch, Germany). Ki-67 signal was developed using the BOND Polymer Refine Red Detection system (cat #: D59390) (Leica Biosystems), and vimentin signal was developed using the BOND Polymer Refine Detection system (cat#: D59800) (Leica Biosystems). Diluent-only (Bond Primary Antibody Diluent) (cat #: AR9352) (Leica Biosystems) controls (no 1° ctrl) were performed alongside Ki-67 and vimentin IHC analyses to ensure signals were not derived from non-specific binding. IHC labelled tissue sections were mounted using Glycergel Aqueous Mounting Medium (cat #: C0563) (Dako, Glostrup, Denmark).

For IF labeling, the protocol was followed as described above. Fluorescent signal was developed using an Alexa Fluor-488 conjugated antibody (S1 Table), counterstained with DAPI dilactate (4′,6-diamidino-2-phenylindole) (1:20000) (cat #: 76482–848) (AAT Bioquest, Inc, Pleasanton, CA), and tissues were imaged using an Olympus Fluoview FV3000 confocal microscope (Olympus, Tokyo, Japan) and processed using ImageJ software (1.54f) (Java 1.8.0_172 (64-bit)) (https://imagej.nih.gov/ij/). To quantify fluorescence intensity, corrected total cell fluorescence (CTCF) was performed. IF images were converted to grayscale and individual xenograft mammary ductal structures were outlined using ImageJ software to determine structure area and integrated density (IntDen). Background mean fluorescence was determined by measuring mean fluorescence within non-fluorescent regions of each tissue section. CTCF = IntDen–(Structure area*Mean fluorescence of background) [40, 41]. CTCF for each whole xenograft was determined by averaging the CTCF of each xenograft ductal structure within each tissue section.

Clear unobstructed brain/body imaging cocktails (CUBIC) tissue clearing and light sheet imaging

CUBIC-based tissue clearing was performed as described [42]. Briefly, MFPs containing xenografts were fixed in 4% paraformaldehyde (PFA) (cat #: J19943-K2) (Thermo Fisher) for 24 h, washed in PBS (Corning Inc) (3 x 1 h) and placed in CUBIC Reagent 1A (N,N,N’,N’-tetra-kis(2-hydroxypropyl)-ethylenediamine (5% (w/w)) (cat#: T0781) (TCI, Tokyo, Japan), triton-X100 (10% (w/w)) (Thermo Fisher), urea (10% (w/w)) (cat #: BP169-500) (Thermo fisher), and NaCl (25 mM) (cat#: S271) (Thermo fisher) in distilled water) on a rotator for 5 days at 37°C, with solution changed every 24–48 h. For IF labeling, tissues were washed in PBS (Corning Inc) then blocked in blocking solution (triton-X100 (0.5% (w/v) (Thermo Fisher), 10% normal goat serum (v/v) (VWR)) overnight at 4°C. Tissues were then immersed in Alexa Fluor^™ 488 Phalloidin (Invitrogen, Waltham, MA) diluted in blocking solution (1:500) for 5 days at 4°C. Tissues were then washed in PBS (Corning Inc) and immersed in CUBIC Reagent 2 (sucrose (44% (w/w)) (cat#: S5-500) (Thermo fisher), urea (22% (w/w)) (Thermo Fisher), triethanolamine (9% (w/w)) (cat #:90279-100ML) (Sigma Aldrich), triton-X100 (0.1% (w/w) (Thermo Fisher)) in distilled water) for 2 days at 37°C. To finish clearing and in preparation for light sheet imaging, MFPs were embedded in Ultrapure Agarose (1.5% (w/v)) (cat #:16500–500) (Invitrogen) in CUBIC Reagent 2 within polymethylmethacrylate square spectrophotometer cuvettes (cat #: 111157) (Globe Scientific, Mawah, NJ). Fluorescent images were taken using a Light Sheet microscope (LaVision Biotech, Bielefeld, Germany) designed around an Olympus MVX-10 zoom macroscope (Olympus). Xenograft tissues were imaged using wavelengths at 488 nm (Alexa Fluor^™ 488 Phalloidin) and 561 nm (autofluorescence) and 3D renderings and z-stack (10 μm optical thickness) videos were generated using Arivis Vision 4D software (version 3.4) (Zeiss, Dublin, CA). Structural boundaries and morphology of mammary xenografts were defined in Arivis Vision 4D software (version 3.4) (Zeiss) by using anatomical mammary tissue landmarks (e.g., mammary lumina, bilayered mammary epithelia). Xenograft vascularization was assessed by differences in autofluorescence intensity and structural morphology, and manually outlined.

Nucleic acid extraction and quantitative RT-qPCR analysis

To validate the presence of equine or canine genomic DNA (gDNA), DNA was extracted from FFPE tissue scrolls containing cleared MFPs bearing mammary xenografts. Paraffin was removed using xylene (cat#: 9460–11) (VWR) and EtOH (cat #: E7023-500ML) (Sigma Aldrich) and DNA was extracted using the DNeasy Blood and Tissue kit (cat #: 69506) (Qiagen, Hilden, Germany), as per manufacturer’s instructions. PCR amplicons were generated using species-specific primers for secretory carrier membrane protein 3 (Scamp-3) (murine-specific) and beta-2-microglobulin (B2M) (either equine- or canine-specific) and analyzed on 2% agarose gels. For real-time quantitative (RT-q) PCR, RNA was extracted from paraffin scrolls using the FFPE RNA extraction kit (cat #: 73504) (Qiagen), as per manufacturer’s instructions. cDNA was synthesized using M-MLV Reverse Transciptase (cat #: PR-M170A) (Thermo fisher) and RT-qPCR reactions were performed using SYBR^® Green Supermix (cat #: 172–5124) (Bio-Rad, Hercules, CA). All RT-qPCR reactions were performed using a QuantStudio 3 Real Time PCR system (Thermo fisher). Target gene expression was normalized to GAPDH expression and fold change (FC) was determined using the 2^-ΔΔCt method then transformed using the Log₂(FC) function. Undetected samples were analyzed as Cycle threshold (Ct) = 40. Primers used for gDNA PCR and RT-qPCR are listed in S2 Table.

Acetocarmine staining of wholemount tissues

MFPs containing extracted mammary epithelia and de-epithelialized (cleared) MFPs bearing xenografts were excised, spread evenly on a glass slide (cat #: 48311–703) (VWR), and air-dried for 5 min. Tissues were fixed in 10% NBF (Sigma Aldrich) for 24–48 h and stored in 70% EtOH (Decon Labs, Inc.). For staining, tissues were first rinsed in H₂O for 15 min, dehydrated in 1 h changes of 95% and 100% EtOH (Decon Labs, Inc.), and cleared in xylene (cat #: 89370–088) (VWR) for 2 h, then submerged in acetocarmine solution (1.25 mg/ml carmine powder (Sigma Aldrich) in 45% Glacial acetic acid (Thermo Fisher)) for 48 h. After staining, tissues were rinsed, dehydrated, and cleared as described above, and mounted using Permount mounting media (cat #: 022–208) (Thermo fisher). When assessing the presence of blebbing structures on xenografts, all acetocarmine stained tissues were assessed by an observer blinded to each condition. Wholemounted xenograft surface area (mm²) was measured using ImageJ software (1.54f) (Java 1.8.0_172 (64-bit)) (https://imagej.nih.gov/ij/). Briefly, the ImageJ scale feature was calibrated to a defined distance using a scale micrometer, then the xenograft structural boundaries were traced by an observer blinded to each condition at equal magnifications to measure surface area (mm²).

Mammosphere-derived epithelial cell (MDEC) culture, fibroblast conditioned media generation (CM), and BrdU proliferation assay

Mammosphere-derived epithelial cells (MDEC) populations were generated as previously described [4, 37, 43] from mammary tissue fragments used in this study. Briefly, mammary tissues and single cell suspensions were isolated as described above and 3 x 10⁶ cells were plated in adherent 6-well culture plates (Corning Inc.). Primary cells incubated for 1 h over two cycles, and non-adherent cells, which were depleted of mammary fibroblasts, were collected and plated in suspension to form mammospheres [43]. Mammospheres were then seeded on adherent T75 flasks (cat #: 07-000-228) (Thermo Fisher), incubated at 37°C and 5% CO2, in media containing DMEM/F12 (50/50) (Corning Inc.) with 10% FBS (R&D Systems), 2% B27 (cat #: 12587–010) (Thermo Fisher), 1% antibiotic/antimycotic (cat #: 30-004-CI) (Invitrogen), 10 ng/mL basic fibroblast growth factor (cat #: SRP4037) (BioVision, Milpitas, CA), and 10 ng/mL epidermal growth factor (cat #: E9644) (Sigma Aldrich) (EpSC medium) to generate MDEC cultures.

To generate primary mammary fibroblast conditioned media (CM), early passage (≤ P5) equine or canine fibroblasts were seeded at 1.33 x 10⁴ cells/cm² in T25 culture flasks (cat #: 07-000-226) (Thermo Fisher) and incubated overnight at 37°C in 5% CO2 in fibroblast culture media. Fibroblast culture media was removed, then cell monolayers were washed twice in sterile 1x PBS (Corning Inc). Four ml of serum-free (sf) DMEM (Corning Inc.) was added to fibroblast cell monolayers and incubated for 24 h to generate fibroblast CM. CM was removed and centrifuged twice at 300 x g for 10 min, with supernatants moved to a sterile tube after each spin to remove cell debris before proceeding to BrdU proliferation analysis.

BrdU proliferation analysis was performed using the BrdU Cell Proliferation ELISA kit (colorimetric) (cat #: ab126556) (Abcam, Cambridge, UK), as per manufacturer’s instructions. Briefly, biological replicates (n = 3) of equine and canine MDEC cultures were plated in triplicate at 5 x 10³ cells/well in a 96-well cell culture plate (cat #: 3596) (Corning Inc.) and allowed to recover for 24 h. EpSC media was removed and fibroblast CM containing 1X BrdU reagent was added, after which cells were incubated for 24 h. Cells were then washed, fixed, and BrdU was labeled, as per manufacturer’s instructions. BrdU incorporation was assessed via spectrophotometric analysis using a Tecan Infinite M200 Pro Microplate Reader (Tecan, Männedorf, Switzerland). Colorimetric reads were taken at 450 nm and background reads (550 nm reads, 450nm reads from “no BrdU” control wells) were subtracted prior to statistical analysis.

Statistical analysis

Student’s two-tailed t-tests were used to determine statistical significance unless otherwise noted. Paired two-tailed t-tests were performed when assessing matched samples for BrdU proliferation analysis. An alpha level of 0.05 was used for all tests. (* P < .05, ** P < .01, *** P < .001, and **** P < .0001). All above statistical analyses were performed using GraphPad Prism (version 10.1.2 (324)) (GraphPad Software, La Jolla, CA).

Xenotransplanted mammary tissue fragments from equine and canine donors engraft in murine mammary fat pads and establish mammary xenografts

Developing mammary glands were extracted from pubescent (3–4 weeks old) immunodeficient NSG mice to de-epithelialize (clear) the mammary fat pads (MFPs) [21, 24, 29], and mammary tissue fragments from equine or canine donors were bilaterally xenotransplanted into the residual stroma (Fig 1a. Successful clearing of the MFP was confirmed by acetocarmine staining of the removed murine endogenous mammary ductal tree (Fig 1b). Mammary tissue fragments from both species ranged from ~ 200–400 μm in diameter and contained both parenchymal and stromal components, as shown by mammary ducts with characteristic mammary lumina surrounded by a fibrous supportive stroma (Fig 1c). Nine weeks after the xenotransplantation surgery (S1a Fig), MFPs containing xenografts were extracted for downstream analyses. Based on the presence or absence of mammary xenografts within formalin-fixed paraffin-embedded (FFPE) serial sections and acetocarmine stained wholemounts, we report a ~ 50% engraftment success rate (Fig 1d).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Overview of mammary fat pad clearing and xenotransplantation procedure.

(a). Diagram depicting the mammary fat pad (MFP) clearing and xenotransplantation approach. (b). Acetocarmine-stained pubescent murine mammary gland extracted from the 4^th MFP of a 3-week-old NOD scid gamma (NSG) mouse. Dotted line indicates location of incision that separated the developing mammary gland from the remaining fat pad. Scale bar = 500 μm. (c). Representative mammary tissue fragments from equine and canine donors stained with H&E. Arrowheads indicate regions of fibrous supportive stroma and asterisks (*) indicate representative examples of mammary ductal structures. Scale bar = 100 μm. (d). Pie chart of percent (%) engraftment success rate of xenotransplantations, as determined by histological analysis. Counts determining engraftment success rate per experiment are presented in S3 Table.

https://doi.org/10.1371/journal.pone.0298390.g001

Mammary xenografts from both equine and canine mammary donors were observed in cleared MFPs (Fig 2). Wholemount staining showed that equine and canine xenografts resemble mammary lobules containing multiple clustered acini that are structurally distinct from the branching mammary ductal tree characteristic of the mouse mammary gland (Fig 2a). These regionally-confined structures lack detectable ductal branching and resemble mammary xenografts derived from other mammalian donors such as human [29] and cow [15, 24, 27]. Hematoxylin and eosin (H&E) staining showed that both equine and canine xenografts are comprised of multiple acinar-like structures containing hollow lumina that are surrounded by a layer of supportive stroma (Fig 2b). To confirm the correct species of origin and to assess the presence of secretory luminal cells in the mammary xenografts, immunohistochemistry (IHC) was performed to detect β-lactoglobulin (β-LG), a whey protein secreted in the milk of ruminants, horses, and dogs, but absent in humans and rodents [44, 45]. The presence of β-LG in luminal cells and within lumina verified that these structures were derived from equine or canine donor tissues (Fig 2c). Importantly, the mammary xenografts showed structural (S2a Fig) and β-LG presence (S2b(i) Fig) similarities when compared to donor species glands. To further validate tissue engraftment and species identity, genomic DNA (gDNA) was extracted from paraffin scrolls of MFPs that were confirmed by H&E analysis to contain or not contain mammary xenografts. PCR analysis using species-specific gDNA primers confirmed the presence of equine or canine gDNA in sections containing xenografts, but not in those without xenografts (Fig 2d). As expected, PCR reactions using murine-specific Scamp3 primers resulted in amplicons (246 bp) in all samples containing mouse gDNA, and equine- and canine-specific B2M primers resulted in amplicons (equine B2M: 442 bp; canine B2M: 303 bp) in MFPs containing xenografted tissues and the equine or canine positive controls (Fig 2d).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Histological analysis and species identity confirmation of equine and canine mammary xenografts.

(a). Acetocarmine-stained murine mammary gland and equine and canine mammary xenografts within the cleared MFPs. Arrowheads indicate xenografts and dotted black lines designate xenograft borders. Scale bar = 500 μm (b). H&E staining of tissue sections containing murine mammary gland and mammary xenografts. Arrowheads indicate xenograft stroma and asterisks (*) indicate representative examples of xenograft lumina. Scale bar = 500 μm. (c). IHC analysis of β-lactoglobulin (β-LG) labeling in murine mammary gland and mammary xenografts. Scale bar = 500 μm. (d). PCR verification of xenograft species identity using species-specific primers (Murine (Mu): Scamp3, Equine (Eq): B2M, Canine (Ca): B2M). Genomic DNA extracted from FFPE sections containing (+) or lacking (-) mammary xenografts and from respective donor mammary primary cells as positive control. NTC = “no template” control. Raw gel images are presented in S3 Fig.

https://doi.org/10.1371/journal.pone.0298390.g002

Mammary xenografts proliferate and display parenchymal and stromal characteristics similar to donor equine and canine mammary glands

Equine and canine mammary xenografts were characterized by IHC using functional and structural markers (Fig 3). The proliferation marker Ki-67 (Ki-67) was used to assess proliferation of the engrafted mammary tissue fragments, indicating survival within the host MFP. Ki-67⁺ cells were observed in both equine and canine mammary xenografts with presence localized primarily to the inner luminal cell layer (Fig 3a). To determine if the mammary xenografts recapitulate the structural architecture and protein labeling patterns observed in donor glands, markers for myoepithelial cells (i.e., α-smooth muscle actin (α-SMA) and cytokeratin-14 (CK14)) and luminal cells (i.e., cytokeratin-18 (CK18)) were evaluated (Fig 3b). Similar to what is observed in equine and canine mammary glands [31, 32] (S2b(ii) Fig), α-SMA and CK14 labeling was restricted to the outer myoepithelial cell layer and CK18 labeling was restricted to the inner luminal cells of the xenografts, indicating a bilayered architecture (Fig 3b). Presence of estrogen receptor-alpha (ERα) was also assessed, as mammary gland responsiveness to mammotropic hormones (e.g., estrogens) is a defining feature of mammary gland growth and functionality [46]. As expected, we observed ERα presence in the nuclei of luminal cells of both equine and canine xenografts (Fig 3b), mirroring characteristics of the donor glands [31, 32] (S2b(ii) Fig) and indicating that these mammary xenografts may be responsive to hormonal stimulation.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Mammary cell proliferation and parenchymal characteristics.

(a). IHC analysis of Ki-67 presence in equine and canine xenografts. No primary antibody control tissues (No 1° ctrl) are presented. (b). IHC analysis of α-smooth muscle actin (α-SMA), cytokeratin-14 (CK14), cytokeratin-18 (CK18) and estrogen receptor-α (ERα) labeling in equine and canine mammary xenografts. Representative tissues treated with isotype controls are presented. Arrowheads indicate positive IHC labeling (red colorimetric indicator) Scale bar = 100 μm.

https://doi.org/10.1371/journal.pone.0298390.g003

To visualize mammary xenografts in a three-dimensional space and assess vascularization, MFPs containing mammary xenografts were rendered optically transparent using a Clear Unobstructed Brain Imaging Cocktails (CUBIC)-based approach and imaged using light sheet microscopy [42]. In concordance with observations of acetocarmine- and H&E-stained xenograft tissues using 2D imaging techniques (Fig 2a and 2b), 3D imaging also showed that equine and canine xenografts consist of multiple acini confined to the xenotransplantation site (Fig 4a, S1–S3 Videos). Importantly, autofluorescent imaging, which is sufficient to define anatomical boundaries within tissues [47, 48], revealed blood vessels surrounding and penetrating equine and canine xenografts, indicating endothelial cell recruitment (Fig 4a, S1–S3 Videos). The extensive vasculature within the xenograft stroma of both species was corroborated by H&E staining (Fig 4b). Performing IHC analysis to evaluate vimentin presence, a filamentary protein present in both myoepithelial cells [49] and stromal fibroblasts [31, 32], showed that the mammary parenchyma of each xenograft was surrounded by an extensive fibroblast-rich stroma (Fig 4c), further demonstrating structural similarity to what is observed in donor mammary glands. Important to note is that: (i) the antibody used for vimentin labeling is non-reactive in mouse [50], as indicated by a lack of vimentin labeling in vimentin-rich murine tissues (i.e., skin and mammary gland [51–54]) following IHC (S4 Fig), and (ii) positive labeling was localized solely to the xenotransplantation site and absent in other extracellular matrix (ECM)-rich, fibrous regions of the MFP, suggesting that the observed stroma of equine and canine xenografts is largely derived from the stromal cells of the xenotransplanted tissue fragments.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Xenograft structure, vascularization, and stromal characteristics.

(a). Volume renderings of equine and canine mammary xenografts generated following wholemount light sheet imaging. Images captured with autofluorescent signal (561 nm). White box denotes mammary xenograft and dotted red lines highlight surrounding vasculature. Scale bars = 500 μm. (b). H&E analysis of equine and canine mammary xenografts. Dotted lines and arrowheads indicate blood vessels. (c). IHC analysis of vimentin presence in equine and canine mammary xenografts and donor mammary tissues. Scale bars = 100 μm.

https://doi.org/10.1371/journal.pone.0298390.g004

Host pregnancy upregulates expression of milk-associated genes and alters mammary xenograft proliferation

Pregnancy promotes the production of circulating ovarian mammotropic hormones and growth factors that drive mammary development [55, 56]. It is well-established that many of these bioactive factors are species cross-reactive [29, 57–59]. To validate hormone responsiveness, and thus, functionality of the mammary xenografts, host mice were mated and xenografts were assessed 18 days post-coitus (dpc) (S1b Fig), a period prior to parturition during which circulating levels of mammotropic ovarian hormones are high and the murine mammary gland has developed to facilitate lactation [29, 60, 61] (S5 Fig). Gross anatomical changes in mated mice at 18 dpc are shown in S5a Fig, and representative developmental changes of the murine mammary gland within the 3^rd inguinal mammary fat pad are presented (S5b Fig) to indicate typical murine mammary development at this timepoint. Changes in expression of the milk protein-associated genes β-casein (CSN2) [9, 46] and β-lactoglobulin (LGB1) [9] were assessed by RT-qPCR analyses of MFPs containing xenografts using equine or canine-specific primers. Equine and canine xenografts recovered from pregnant hosts showed an upregulation in both CSN2 and LGB1, albeit this did not reach significance for the equine xenografts due to a high variation in gene expression in the nonpregnant group (Fig 5a). H&E imaging showed the presence of thickened eosinophilic deposits, suggestive of lipid synthesis and proteinaceous fluid production [29, 56] within the lumina of equine, but not canine, xenografts (Fig 5b), with slightly greater abundance of deposits observed in the equine xenografts from pregnant hosts. Despite a lack of dense thickened deposits, eosinophilic secretions were still present within the lumina of the canine xenografts (Fig 5b). IF imaging of β-LG in nonpregnant and pregnant equine (Fig 5c(i)) and canine (Fig 5d(i)) xenografts revealed a baseline level of β-LG presence within the lumina regardless of host reproductive state. Quantification of fluorescence within the mammary xenograft lumina using a corrected total cell fluorescence (CTCF) approach revealed increased presence of β-LG labelling within both equine (Fig 5c(ii)) and canine (Fig 5d(ii)) xenografts from pregnant hosts, mirroring the expression trends of the RT-qPCR analyses (Fig 5a). Notably, significance in fluorescence intensity between nonpregnant and pregnant equine and canine xenografts did not reach statistical significance due to intra-group variations in CTCF (Fig 5c(ii) & 5d(ii). Regardless of the minor responses of the equine xenografts to host pregnancy, these data indicate functional mammary differentiation and lactation of xenografts when exposed to increased concentrations of host-derived circulating hormones via pregnancy.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Mammary xenografts respond to endogenous mammotropic hormone stimulation.

(a). RT-qPCR analyses of lactation-associated genes, β-casein (CSN2) and β-lactoglobulin (LGB1), MFPs containing equine or canine xenografts from nonpregnant or pregnant host mice. Fold change (FC) determined via the 2^-ΔΔCt analysis method and transformed via the Log₂(FC) function. Error bars represent standard deviation (SD). Statistical significance was assessed using unpaired (Student’s) t-test (P ≤ 0.05). (b). H&E analysis of xenografts in nonpregnant or pregnant host mice. Arrowheadss indicate thickened eosinophilic deposits within equine xenografts. Scale bar = 100 μm. (c). Immunofluorescent (IF) labelling of β-lactoglobulin (β-LG) in equine xenografts from nonpregnant and pregnant mice (i) and corrected total cell fluorescence (CTCF) intensity quantitation (ii). (d). IF labelling of canine xenografts from nonpregnant and pregnant mice (i) and CTCF intensity quantitation (ii). Immunoglobulin G (IgG) antibody was used as isotype control on xenografts from pregnant mice. Tissues counterstained with DAPI (4′,6-diamidino-2-phenylindole). Scale bar = 100 μm. Error bars represent standard deviation (SD). Statistical significance was assessed using unpaired (Student’s) t-test (P ≤ 0.05).

https://doi.org/10.1371/journal.pone.0298390.g005

It is well-established that in addition to lactogenesis, pregnancy also drives increased cell number and glandular surface area [62]. A characteristic of this process is increased proliferative index of epithelial cells as the gland develops to facilitate lactation [63–65]. To that end, IHC labelling of Ki-67 in nonpregnant and pregnant equine and canine xenografts (Fig 6a(i)) showed a significant increase in the percentage of Ki-67⁺ mammary luminal cells was observed in canine, but not equine, xenografts recovered from pregnant mice (Fig 6a(ii)). To assess xenograft structural alterations, wholemounts containing equine and canine mammary xenografts from nonpregnant and pregnant mice were stained with acetocarmine, which showed a lobule-like, spherical morphology confined to the xenotransplantation site irrespective of donor species or host pregnancy status (Fig 6b). To determine if host pregnancy drives growth, xenograft surface area was assessed [28]. A trend towards increased average xenograft surface area following pregnancy was observed, especially in the canine xenografts; however, sizes were highly variable within groups and thus, statistical significance was not reached (Fig 6c). Although the reasons for this remain elusive, the varied xenograft responses could be due to horses, dogs, and mice displaying in vivo differences in circulating hormone concentrations during pregnancy [66–68], or due to differences in mammary developmental stages between equine and canine xenografts in response to ovarian hormones, which would not be captured in these experiments as all xenografts were recovered and analyzed at the same time point; 18 dpc. Collectively, and despite some discrepancies between both mammals, these data demonstrate that mammary xenografts responded to hormonal stimulation, indicating xenograft functionality.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Xenograft whole mounts and size analysis of xenografts in response to endogenous mammotropic hormone stimulation.

(a). IHC analysis of Ki-67 positivity in mammary xenografts derived from nonpregnant or pregnant host mice. Arrowheads indicate representative Ki-67⁺ cells. Scale bar = 100 μm. (i). Quantification (%) of Ki-67⁺ mammary luminal cells. (ii). Error bars represent standard deviation (SD). Statistical significance was assessed using unpaired (Student’s) t-test (P ≤ 0.05). (b). Acetocarmine-stained mammary xenografts from nonpregnant and pregnant host mice. Dotted black lines designate xenograft borders. Scale bar = 500 μm. (c). Surface area (mm²) quantifications of equine (i) and canine (ii) mammary xenografts. Error bars represent standard deviation (SD). Statistical significance was assessed using unpaired (Student’s) t-test (P ≤ 0.05).

https://doi.org/10.1371/journal.pone.0298390.g006

Preimplantation of mammary fibroblasts does not promote mammary ductal branching in equine and canine xenografts

While mammary lobule-like structures are commonly observed following equine and canine mammary cell xenotransplantation, indicated by xenografts consisting of clustered acini (Fig 2a and 2b, S1–S3 Videos), ductal branching into the MFP is typically absent [24, 29]. Previous reports using human mammary epithelial cells showed that pre-implantation of species-matched fibroblasts within the MFP were capable of driving xenograft ductal morphogenesis [29, 38]. This is based on the knowledge that soluble ligands produced by stromal fibroblasts, such as fibroblast growth factor (FGF), fibroblast-derived ECM components, such as collagen-I, and fibroblast-derived matrix remodeling proteins are essential for mammary developmental and ductal outgrowth [69–71]. As seen in S1c Fig, we performed experiments by implanting cultured mammary gland-derived primary species-matched fibroblasts into cleared MFPs to generate a stromal environment that would be more conducive for ductal formation. First, we confirmed increased collagen deposition and successful engraftment of equine and canine fibroblasts in MFPs that were recovered 2 weeks following implantations by (i) staining with Picro-Sirius (Sirius) red and Masson’s trichrome (MTC) and (ii) performing IHC for vimentin presence using the antibody clone that is non-reactive in mouse [50] (S4 Fig). MFPs that received fibroblast implantations from both species were characterized by dense, collagen-rich populations of vimentin⁺ stromal fibroblasts (Fig 7a), indicating that implanted fibroblasts successfully engrafted and deposited collagen. As expected, these dense regions were absent in MFPs that received the Matrigel vehicle injections (Fig 7a). Second, we xenotransplanted equine and canine mammary gland tissue fragments 2 weeks after the species-matched fibroblast implantations and analyzed the xenografts 9 weeks later (S1c Fig). In contrast to previous reports with immortalized human fibroblast pre-implantations [29, 38], we did not observe ductal growth in equine or canine xenografts in the presence of species-matched fibroblast engraftment prior to mammary xenotransplantations, as assessed by acetocarmine staining of wholemounts. Specifically, equine mammary xenografts within fibroblast pre-implanted MFPs showed a lobule-like morphology confined to the xenotransplantation site, which was similar to wholemounts that received vehicle injections (Fig 7b) or no pre-implantation procedure at all (Fig 2a). Despite a similar overall lobule-like morphology, it was noticed that some equine xenografts showed more blebbing formations after fibroblast pre-implantation (< 20% of xenografts) when compared to vehicle control (0% of xenografts in the vehicle group) (Fig 7b). The morphology of the canine xenografts was also lobule-like in both fibroblast pre-implanted and vehicle groups, with interconnected outgrowth structures observed (Fig 7b). Both vehicle and fibroblast pre-implanted groups displayed similar proportions of blebbing formations (~35–50% for both groups), indicating that this morphology in canine xenografts may be a species-specific characteristic. When quantifying 2D surface areas, there were no statistically significant differences between treatment groups for both species, despite a trending increase in surface area in the fibroblast pre-implanted groups (Fig 7c).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Pre-implantation of mammary fibroblasts and assessment of ductal morphogenesis.

(a). Sirius red and Masson’s Trichrome (MTC) histological stains, and IHC analysis of vimentin presence in vehicle-injected and mammary fibroblast pre-implanted MFPs (2 weeks post-injection). Scale bar = 200 μm. (b). Acetocarmine-stained mammary xenografts from vehicle-injected and mammary fibroblast pre-implanted MFPs. Black dotted lines designate xenograft borders. Arrowheads indicate regions of xenograft blebbing. Scale bar = 500 μm. (c). Surface area (mm²) quantifications of acetocarmine stained mammary xenografts within vehicle-injected or fibroblast pre-implanted MFPs. Error bars represent standard deviation (SD). Statistical significance was assessed using unpaired (Student’s) t-test (P ≤ 0.05). (d). BrdU assay assessing equine and canine MDEC proliferation over 24 h treatment with species-matched mammary fibroblast CM compared to control base media (DMEM). Symbols (circle, square, and triangle) represent individual MDEC cultures of both species. Error bars represent standard deviation (SD). Statistical significance was assessed using paired two-tailed t-tests (P ≤ 0.05).

https://doi.org/10.1371/journal.pone.0298390.g007

Due to the trending increase in xenograft surface area, an in vitro experiment using a BrdU-ELISA assay was performed as a proxy to assess the capacity of paracrine signaling by primary mammary fibroblasts to drive mammary epithelial cell proliferation. Mammosphere-derived epithelial cells (MDEC), which are heterogenous cultures of primary mammary cells derived through the same approach used to generate mammary tissue fragments for xenotransplantation surgeries [43], were treated with species-matched primary fibroblast conditioned media (CM) and subjected to the BrdU-ELISA to determine changes in mammary cell proliferation (Fig 7d). Interestingly, MDEC cultures treated with primary mammary fibroblast CM showed decreased proliferation when compared to MDEC cultures treated with control base media (DMEM), suggesting that paracrine signaling from pre-implanted mammary fibroblasts is most likely not a major driver of mammary xenograft growth in vivo.