2.1. Extraction and Identification of Porcine Breast Tissue ECM Proteins
Proper handling of fresh tissues is critical for minimal degradation and maximal identification of native proteins within the tissues. Homogenization of the fresh porcine breast fatty tissues directly obtained from local slaughter house was conducted in a precooled homogenizer, with the container holding the tissues embedded in ice and sliced tissues mixed with ice-cold water as described in the materials and methods. The homogenized tissues were then decellularized with the non-ionic detergent triton X-100 or the zwitterionic detergent CHAPSO to remove the cellular contents while preserving the native and active states of the ECM proteins. Since porcine breast tissues contain rich fat that is difficult to remove compared to that within the mammary tissues of mice and human, we applied lipase in the detergent solution to maximize lipid removal and the decellularization efficiency (
Figure 1a). The detergents and lipase were removed with multiple rounds of washing with ddH
2O. The resulting ECM dry weight was about 2% of the total fresh tissues used for the extraction. Whole ECM protein extraction from the ECM was carried out using a gradient of concentrations of urea and thiourea solutions to solubilize proteins that could be dissolved under different urea solution conditions due to their sizes and native conformational states. After dialysis and concentration of the protein extract, the total protein amount within the extract was measured at about 90% of the ECM used for the extraction.
The proteomic composition of the porcine ECM was identified using liquid chromatography coupled with tandem mass spectrometry (LC-MS/MS). The majority of ECM proteins detected were different types of collagen, representing about 70% of the major proteins listed (
Figure 1b), with collagen I and III being the most abundant (~77% based on single chain ratios) among the total collagen content. While the sum of glycoproteins and proteoglycans accounted for about 5% of the ECM major proteins, the content of myosin and tropomyosin was about 19%, with other ECM proteins comprising the remaining 6%. Interestingly, when we compared these data with the proteomic profiles of mouse normal MFP ECM that we reported previously [
15], the ECM of rat mammary tissues [
12] and the ECM of human breast tissues adjacent (considered to be normal) to tumors [
16], it appeared that the ECM total collagen content of the human breast tissue ECM (about 80–85%) was in between of that of the porcine breast ECM and the rodent mammary ECM (about 85–90%). Another intriguing result was that higher myosin and tropomyosin content was detected in the porcine breast ECM compared to that of the rodent mammary ECM (less than 0.5%) and of the human breast ECM (about 1%) [
16]. The amounts of the overall glycoproteins and proteoglycans within the ECM of porcine, rodent, and human were similar at about 5–7%. These ECM compositional differences of the porcine, mouse, and human breast tissues could be due to the anatomical and physiological natures of the native tissues that are related to the functions of the tissues in each specific species. It is noteworthy that the reproductive cycle seems to play a role in mammary tissue ECM compositional changes [
12,
17]. Additionally, the differences in the buffers and methods used to extract native ECM proteins clearly have an impact on the types and amounts of proteins identified [
12,
15,
16,
18]. Future proteomic analyses of mammary ECM compositions of the aforementioned and additional species using the same extraction and analytical methods will potentially identify novel discrepancies upon cross comparing the data sets obtained from the same experiments.
2.2. Generation and Characterization of the Porcine Breast ECM Hydrogel Scaffold
To assess the impact of storage temperature of the ECM hydrogel on polymerization, the gels (8 mg/mL) directly stored at 4, −20, −80 °C, and flash froze in liquid nitrogen followed by storing at −80 °C were thawed on ice and dropped onto the bottom of a 100 mm tissue culture dish, which was then placed in a 37 °C incubator for polymerization. After 30 min of incubation, the gels from the different stocking conditions all gelatinized (
Figure 2a). When the polymerized gels were submersed in 1× PBS or DMEM, they retained their initial shapes and remained undissolved during the 10-day testing period (
Figure 2b). To maximize the preservation of the ECM proteins in their native forms and avoid protein degradation, we used liquid nitrogen flash freezing followed by storing at −80 °C as a standard approach for long-term stocking of the hydrogel solutions at different concentrations. Short-term storage on ice or at 4 °C for up to four weeks did not seem to affect the polymerization of the hydrogel and its performance in experiments as described below.
Generation of porous TMS using the ECM hydrogel was achieved using a combination of a gas foaming method [
19] and a freeze-gelation approach [
20] with modifications as described in the methods. This technique not only induced gelation of the hydrogel but also allowed production of interconnecting porous structures within the solidified gel. The sustainability of the fabricated porous scaffolds under regular tissue culture conditions (37 °C, 5% CO
2) was evaluated in 1× PBS or DMEM for 7 days. Our results showed that the porous scaffolds were stable and retained their shape during the period of testing (
Figure 2c,d). Hematoxylin and eosin (H&E) staining of the cross sections of the scaffolds showed inter-connective pores at the sizes of about 100–200 μm (
Figure 2e, left panel), highly resembling those of the decellularized porcine breast fatty tissue ECM (
Figure 2e, right panel).
In order to evaluate the mechanical resemblance of the reconstituted scaffolds to that of the matrix of native tissues, we used AFM to measure native porcine breast tissues, decellularized porcine breast ECM, and porous scaffolds generated using porcine breast ECM hydrogel. The AFM testing method on the stiffness of biological tissue structures, cells, and specific regions of ECM is not standardized. Thus, previous reports have shown different results depending on their individual testing conditions [
21,
22]. On a large scale, measurements of decellularized organ matrices with AFM have been reported [
23,
24]. On a fine scale, mechanical properties of breast cancer cells and their structures have been investigated using AFM [
25]. The probe tip size, indentation rate, and maximum force applied to samples should be optimized based on the mechanical strength of the samples. Thus, we optimized our testing parameters specifically for the breast tissues as described in the methods. The positioning of the probe tip for indention on the decellularized ECM and TMS has been illustrated in
Figure 2f-ii,f-iii.
Our AFM results showed that the average Young’s modulus of the native porcine breast tissues was 0.243 ± 0.027 kPa (
Figure 2f–i), which corresponded to a similar range of the compliance of human normal breast tissues [
25,
26] and is about 45–50% higher than that of mouse normal mammary tissues [
27,
28]. The Young’s modulus of the decellularized porcine breast ECM was 0.366 ± 0.061 kPa (
Figure 2f–i), which is about 50% higher than that of the porcine breast native tissues and stiffer than decellularized mouse normal mammary fatty tissues [
28]. A similar trend of higher Young’s modulus in decellularized tissues than in native tissues was observed in AFM measurement of human liver samples [
29]. Porous TMS generated using ECM hydrogel at the concentration of 40 mg/mL (data for the concentrations of 20 mg/mL and 60 mg/mL were not shown) demonstrated an average Young’s modulus of 0.411 ± 0.180 kPa (
Figure 2f–i), which is close to that of the decellularized porcine breast ECM. These data collectively indicate that the structural and mechanical properties of the reconstituted porous TMS highly resemble those of the decellularized native ECM and are suitable for spatial tissue cultures that closely mimic native breast tissue microenvironment.
2.3. ECM Support of Cell Surface Receptor Expression and Metabolomes in Spatial Culture
Our proteomics data showed that collagen I and III were the major protein components, whereas the main basement membrane protein laminin only accounted for a minimal amount of the porcine breast tissue ECM. Therefore, we carried out immunofluorescence (IF) staining of integrin β1 and β4 plasma membrane receptors for collagen I/III and laminin, respectively, in MM231 cells grown on the breast ECM hydrogel-coated glass coverslips for 24 h. The expression of focal adhesion kinase (FAK) on the surface of the cells was also immunostained to serve as an indicator of the adhesion sites of the cells. Our confocal fluorescence microscopy results showed that high levels of integrin β1 and low levels of integrin β4 were observed in the cells cultured on the hydrogel matrix (
Figure 3a,b), indicating that the cells attached to the matrix through integrin β1 and collagen I/III interactions.
To assess the capability of the porcine breast ECM hydrogel in capturing the metabolites secreted from cells grown on it, 4% porcine breast ECM hydrogel was added into the wells of 96-well plates at a thickness of 4 mm, followed by inserting porous TMS illustrated in
Figure 2c (2-mm thick, 6-mm diameter, 100-μm pore size) into the surface section of the gel before its polymerization in a 37 °C incubator. A total of 1.0 × 10
4 MM231 cells per well were seeded on the top of the porous scaffold and cultured in RPMI 1640 medium (Corning; 1×; 2 g/L of
d-glucose; 10% FBS) under optimal conditions (37 °C, 5% CO
2) for 3 days. The hydrogel samples underneath the porous scaffolds were collected, cross sectioned (
Figure 3c, left panel) and processed for mass spectrometry (MS) analysis of the metabolites collected within the gels as described in the methods.
Based on our MS spectral data, metabolite distribution patterns and relative abundance levels were determined using the SCiLS Lab MS imaging analysis software, which also grouped compounds (detected as ions with specific
m/
z values) into distinct ion distribution patterns across the scaffolds for comparison. A total of 41 distinct chemical features (i.e., the mass over charge ratios detected in MS) were selected based on their abundance distribution profiles and the positive scores (close to 1) resulted from the receiver operating characteristic (ROC) curve analyses in the MM231 cell culture-laden hydrogel samples compared to the medium-containing blank hydrogel samples. As illustrated by one of the forty-one results, the compound with
m/
z 663.024 was absent from the blank scaffold (
Figure 3c, right panel, top) but showed high accumulation with a distinct pattern of being more abundant near the edges of the ECM hydrogel scaffold than in the center. This compound had a ROC value of close to 1 (highly positive;
Figure 3d, top panel). The box plot for this compound (
Figure 3d, bottom panel) further demonstrated its preferential accumulation within the hydrogel samples of the MM231 cell cultures. Similar results were found in all 41 ions (data not shown), i.e., higher specific chemical abundances were detected in the hydrogel samples of the MM231 cell cultures with high ROC scores. Four box plots for the ions with top ROC scores from the 41 individual groups were shown in
Figure S1, which had
m/
z ratios at 663.024, 531.060, 839.043, and 768.371 with the ROC scores of 0.995, 0.996, 0.994 and 0.988, respectively. Together, these data demonstrate that the porcine breast ECM hydrogel scaffolds support matrix-associated cell membrane receptor expression and secretion of metabolites from the cells grown in the tissue-mimicking 3D space.
2.4. Applying Porcine Breast ECM Scaffold in Support of Spatial Cell Proliferation, Coculture of Cancer Cells and Stromal Cells, and Tumor Formation
To evaluate the proliferation of mammary epithelial cells on porcine breast ECM hydrogel, normal MCF10A or MM231 cells were seeded in 96-well plates coated with or without the porcine breast ECM hydrogel, collagen, and Matrigel, and cultured under optimal conditions (37 °C; 5% CO
2) for 7 days. Cell proliferation was measured using WST-1 reagent (Sigma-Aldrich, St. Louis, MO, USA) at different time points (Day 1, 3, 5, 7). Our results showed that both type of the cells proliferated faster on Matrigel compared to those on collagen and ECM hydrogel (
Figure 4a), possibly due to the compositional nature of Matrigel, which contains growth factors and underdefined cellular components derived from Engelbreth–Holm–Swarm (
EHS)
sarcoma sources [
15,
30]. In contrast, the growth of the cells cultured on the porcine breast ECM hydrogel is even slower than those on collagen but at comparable levels, suggesting certain degree of similarities of the two types of gels on supporting cell population expansion.
The spatial growth and proliferation of mammary epithelial cells within the breast ECM hydrogel were further assessed using an acini formation assay. Briefly, 1 × 10
4 MCF10A cells were mixed into 200 μL of 2% porcine breast ECM hydrogel and cultured on 8-well chamber slide under optimal conditions as described above for 7 days. Acini formation was assessed using light microscopy, and the structures of the acini were characterized with phalloidin and DAPI IF staining followed by confocal microscopy. Our results showed that MCF10A cells formed acini at different sizes, with larger ones having hollow centers (
Figure 4b). Compared to the traditional acini formation assay using Matrigel, the current method applies fewer matrix materials, simplified procedures, and shorter culture times. Importantly, the purified porcine total ECM hydrogel contains neither growth factors nor tumor cell products and has well-defined ECM proteins, lending promise for low background tissue cultures and adjustable options for adding desired culturing components within culture medium.
In order to observe the spatial expansion of cancer cell population and recruitment of stromal cells, we have devised a coculture system using the porcine breast ECM hydrogel and porous TMS (
Figure 2c,d) derived from the hydrogel. 1 × 10
4 GFP-MM231 cells were seeded on the porous scaffold (round, 2 mm thick, 4 mm diameter) placed in a well of 96-well plates. A layer of the ECM hydrogel (8 mg/mL) containing 1 × 10
4 RFP-HUVECs (human umbilical vein endothelial cells) was covered on top of the porous scaffold (
Figure 4c, top panels). After polymerization of the gel, 100 μL of 1× DMEM containing 10% FBS was added into the well (replicate samples were prepared). The two types of cells within the different yet mutually accessible compartments were cultured for 14 days. The distribution and migration of the cells within the co-culture system were imaged over time using confocal microscopy. Our data showed that both MM231 cells and HUVECs progressively increased their numbers and migrated out of their initial living compartment into adjacent areas (
Figure 4, middle and bottom panels). The interaction of the two types of cells was also increased, as exhibited by the yellow overlapping regions. Clearly, this compartmental culture approach using hydrogel and porous scaffold derived from the same native tissue ECM allows for the observation of certain cellular phenotypes, such as spatial cell migration and interaction, in an advanced tissue-mimicking environment that could be difficult to be captured in live tissues or other nontissue-specific culture models.
We next tested the efficiency of the porcine breast ECM scaffold in supporting tumor formation in vivo. 1 × 10
5 MM231 cells were seeded on a spherical porous TMS (4 mm diameter) and cultured for 24 h under optimal conditions as described before. Triplicate samples were prepared for the experiment. Then, the scaffolds were implanted separately into the mammary fat pad (MFP, 4th nipple region from the rostral side) of eight-week-old nulliparous NOD/SCID female mice. Tumor development was observed over a period of 4 weeks. The sizes of the tumors were dynamically measured using a caliper and reached an average of about 2-cm diameter at the end of week 4 post-implantation, at which point the tumors were collected, formalin fixed, cross-sectioned (10 μm thick), and H&E stained for histological examination as we reported previously [
15]. Our histological data showed that visible blood vessels had been grown into the tumors in most of the regions of the outer half of the tumor bodies (
Figure 4d,e, arrows), which were filled with mixtures of cancer cells and stromal cells clustered within or overlapped with ECM structures (
Figure 4e). This data indicates that the porcine breast ECM-based scaffold represents another tissue-specific platform, in addition to mice ECM-based scaffolds [
15], to support consistent formation of breast tumors in animals.