Vitamin D3 regulates estrogen’s action and affects mammary epithelial organization in 3D cultures

Vitamin D3 (vitD3) and its active metabolite, calcitriol (1,25-(OH)2D3), affect multiple tissue types by interacting with the vitamin D receptor (VDR). Although vitD3 deficiency has been correlated with increased incidence of breast cancer and less favorable outcomes across ethnic groups and latitudes, randomized human clinical trials have yet to provide conclusive evidence on the efficacy of vitD3 in treating and/or preventing breast cancer. When considering that carcinogenesis is “development gone awry”, it becomes imperative to understand the role of vitD3 during breast development. Mammary gland development in VDR KO mice is altered by increased ductal elongation and lateral branching during puberty, precocious and increased alveologenesis at pregnancy and delayed post-lactational involution. These developmental processes are largely influenced by mammotropic hormones, i.e., ductal elongation by estrogen, branching by progesterone and alveologenesis by prolactin. However, research on vitD3’s effects on mammary gland morphogenesis focused on cell proliferation and apoptosis in 2D culture models and utilized supra-physiological doses of vitD3, conditions that spare the microenvironment in which morphogenesis takes place. Here, using two 3D culture models, we investigated the role of vitD3 in mammary epithelial morphogenesis. We found that vitD3 interferes with estrogen’s actions on T47D human breast cancer cells in 3D differently at different doses, and recapitulates what is observed in vivo. Also, vitD3 can act autonomously and affect the organization of MCF10A cells in 3D collagen matrix by influencing collagen fiber organization. Thus, we uncovered how vitD3 modulates mammary tissue organization independent of its already known effects on cell proliferation.


Introduction 31
Breast cancer remains a major cause of mortality among women worldwide. Epidemiological studies 32 have shown that key stages during breast development are particularly susceptible to the effects of 33 carcinogens. For instance, women aged 10-19 years who were exposed to atomic bomb radiation in 34 Hiroshima in World War II showed an excess of breast cancer cases at the age of prevalence compared 35 to similarly exposed women aged 35 years and older (1). Likewise, women exposed to diethylstilbestrol 36 during fetal life have a higher risk of breast cancer compared to unexposed women (2), and women 37 exposed to DDT in the womb showed a four-fold higher risk of breast cancer in adulthood (3). This 38 phenomenon has also been observed in rodents; namely, rats exposed to NMU around puberty have a 39 100% incidence of tumors, but the incidence rate falls to just 10% when exposed after 90 days of age (4). 40 Rodents exposed in utero to low doses of BPA have also shown a higher incidence of mammary gland 41 tumors in adult life (5,6). These "windows of susceptibility" coincide with key milestones of 42 organogenesis and/or tissue remodeling, buttressing the notion that carcinogenesis is "development 43 gone awry."(7) 44 Vitamin D3 (VitD3), and its active metabolite calcitriol (1,25-(OH) 2 D 3 ), has been primarily studied in the 45 context of normal and diseased bone development (8). However, research over the past few decades 46 have shown that vitD3 can affect multiple tissue types, including the mammary gland (8). For instance, 47 DMEM/F12 with phenol red, 5% equine serum, 20 ng/ml epidermal growth factor (EGF), 0.5 µg/ml 101 hydrocortisone, 0.1 µg/ml cholera toxin, and 10 µg/ml insulin. Experiments performed using MCF10A 102 cells used the same medium. In all cases, cells were incubated at 37⁰C in 6% CO 2 and 100% humidity. 103 Dose response curves to calcitriol 104 Dose-response curves to calcitriol in 2D culture were performed in 24-well plates. Cells were seeded at a 105 density of 35,000/well in media and allowed to attach. For MCF10A cells, seeding media was changed to 106 media containing different doses of calcitriol after 24 hours of seeding. For T47D cells, propagation 107 media was removed 48 hours after seeding, and substituted with CD-FBS medium containing different 108 doses of calcitriol with or without E2. After 6 days, cell numbers were determined using the SRB assay 109 (20). 110

3D cultures 111
Collagen type I gels were formulated at a final concentration of 1 mg/mL as described previously (21). 112 Cells were suspended in the gel solution at a density of 75,000 cells per gel and 1.5 mL of mixture (per 113 well) was poured into 12-well plates. The mixture was allowed to congeal for 30 min at 37⁰C, and 1.5 mL 114 of appropriate media was added to each well (CD-FBS media containing E2 +/-calcitriol for T47D, 115 MCF10A media +/-calcitriol). Gels were detached as previously described (22). Cultures were 116 maintained either for 1 week to measure total cell yield or 2 weeks for morphological assessments. At 117 each endpoint, gels were harvested and processed as described by Speroni et al (21). Briefly, to count 118 cell numbers, cells were extracted by digesting 3D gels with collagenase and then lysed to obtain nuclei 119 that were then counted using a Coulter Z1 particle counter (Beckman Coulter, CA). For morphological 120 assessments, gels were harvested at 2 weeks, fixed with 10% phosphate-buffered formalin, and either 121 embedded in paraffin for histological analyses or whole mounted and stained with carmine alum to 122 visualize epithelial structures.

Whole mount Analysis 124
Whole-mounted gels stained with carmine alum were imaged using a Zeiss LSM 800 confocal 125 microscope for automated morphometric analysis as described by Paulose et al (23). Briefly, ~1 mm 2 126 area of the gel periphery was imaged to a depth of ~100 µm. Resulting images were stitched together 127 and analyzed using Software for Automated Analysis (SAMA) (23) and statistical analyses of 128 morphometric parameters was performed using GraphPad Prism software. 129

Picrosirius staining 130
Formalin-fixed, paraffin-embedded gels were sectioned using a microtome at 5µm thickness. Gel 131 sections were then rehydrated and stained with picrosirius red solution to visualize collagen fibers and 132 counter-stained with Weigert's hematoxylin as described by Junqueira et al (24). Stained sections were 133 observed under polarized light using a Zeiss Axioskop 2 Plus microscope. 134

Statistics
Tukey's post hoc test were performed to determine differences in the dose-response curves, number of 146 elongated structures in MCF10A gels and MCF10A gel diameters; one-way ANOVA with post hoc 147 Dunnett's 2-sided t-test was performed for cell yields in T47D 3D gels. Kruskal-Wallis test was performed 148 to determine differences in the morphometric parameters of structures. Chi-square analysis was 149 performed to compare distributions of T47D epithelial structures in different volume categories. 150 Unpaired t-test with Welch's correction was used to analyze RT-PCR data. For all statistical tests, results 151 were considered significant at p < 0.05. 152

Results 153
Effects of calcitriol on estrogen-induced cell proliferation 154 T47D cells exposed to different doses of vitD3 showed a decrease in cell numbers at the end of a 6-day 155 period only when in the presence of estrogen in both 2D and 3D cultures (Fig. 1). However, the reduced 156 cell yield was observed at the 50 nM and 100 nM doses of vitD3 alone in 2D (Fig. 1A), whereas a reduced 157 cell yield was observed only at the 100 nM dose in 3D (Fig. 1B). Based on our inverted microscope 158 visualization of dead "floaters" in these cultures, the decrease in cell yield, especially in 2D culture, can 159 be attributed to cell death. In contrast, we observed an increase in cell numbers at lower calcitriol doses, 160 such as 10 and 25 nM in 3D culture, a phenomenon that has not been previously reported. Separately, 161 calcitriol activity in these cells was confirmed by performing RT-PCR for CYP24A1 gene transcripts; there 162 was a dose-dependent increase following calcitriol treatment (Supp. Fig. 1). 163

Effects of calcitriol on Estrogen-induced epithelial organization 164
Previous work from our lab had shown that E2 induces T47D cells to form mostly elongated shaped 165 structures when embedded in a 3D rat tail collagen type I matrix (21). Using this same model, we 166 investigated how vitD3 affects the organization of T47D cells in the presence of 0.1 nM E2. As shown in Fig. 2, calcitriol affects the organization of these cells, mainly by affecting the volume of the epithelial 168 structures, in a dose-dependent manner (Supp. Fig. 2). More specifically, the change in the organization 169 is seen through a re-distribution of different sized structures in the population -the 50 nM calcitriol 170 dose results in a higher number of smaller structures whereas the 10 nM dose results in a slight, non-171 statistically significant increased number of larger structures (Fig. 2B). In contrast, calcitriol alone does 172 not seem to affect these cells when added to the hormone-depleted CD-FBS medium, thus suggesting 173 that either the interactions with E2 are responsible for the effects observed, or the effect of calcitriol is 174 independent of E2 but is unobservable as the cells are not proliferating in the absence of E2.  (27). Using this model, we investigated vitD3's effects on the 204 organization of these cells; however, we utilized a rat tail type I collagen matrix (Fig. 6). Previous work 205 from our lab (28) has shown that epithelial organization in 3D culture is influenced by the species of the 206 collagen type I used in the extracellular matrix formulation. Consistent with those and other findings 207 (29), we observed mostly acinar structures with few ductal structures in a rat tail type I collagen matrix 208 (Fig. 6A). Interestingly, we also observed that calcitriol increases the number of elongated structures in 209 3D culture, with the highest number observed at the 10 nM dose, followed by a reduction at higher 210 doses (25 & 50 nM, Fig. 6B). 211

Effects of calcitriol on MCF10A epithelial organization
Confocal images of MCF10A 3D gels were analyzed for changes in the morphological parameters of the 213 epithelial structures upon calcitriol treatment. Exposure to calcitriol resulted in flatter and less spherical 214 structures; these effects were significant at the 50 nM dose (Fig. 7). Calcitriol also resulted in a decrease 215 in the volume of these epithelial structures, most significantly at 50 nM dose (Supp. Fig. 4), which can be 216 attributed to the increase in flatness with increasing calcitriol doses. 217

Effects of calcitriol on collagen organization 218
MCF10A cells organize collagen fibers in the 3D gels, as a prerequisite for organizing into ducts or acini 219 (30). Because of the changes in the morphological parameters of the epithelial structures, we 220 investigated collagen fiber organization in 3D gels using picrosirius staining. Polarized light microscopy of 221 FFPE sections stained with picrosirius revealed that calcitriol has a non-monotonic effect on collagen 222 fiber organization in this 3D model (Fig. 8). Calcitriol at 10 nM dose reduced the number of organized 223 collagen fibers. In contrast, 25 and especially 50 nM doses showed an increase in the amount of 224 organized collagen fibers. Additionally, at these doses, organized fibers were more uniformly distributed 225 throughout the 3D gels, especially in areas distal from epithelial structures (data not shown). These 226 observations may explain why calcitriol treatment also resulted in increased contraction of the 3D gels 227 in a dose-dependent manner after 2 weeks in culture (Supp. Fig. 5). 228

Discussion 229
Despite evidence linking vitD3 deficiency to increased risk of breast cancer and worse clinical outcomes 230 in patients, randomized clinical trials have yet to confirm the efficacy of vitD3 as a preventive or 231 therapeutic option in this disease (8). Experimentally, while VDR KO mice do not develop tumors 232 spontaneously, mammary glands from these mice exhibit a striking phenotype of excessive and 233 precocious development at key stages (15,16). This suggests that vitD3 plays an important role in the 234 development of the normal mammary gland. When considering that carcinogenesis is "development gone awry" (7), an understanding of the role of vitD3 in this process may provide worthy therapeutic 236 options for breast cancer patients. 237 The VDR is expressed in the mammary gland at the different stages of postnatal development that are 238 largely influenced by the mammotropic hormones E2, Prg and Prl. These hormones have well-239 characterized effects on the morphogenesis of the gland; for example, E2 stimulates ductal elongation, 240 Prg increases lateral branching and Prl induces alveologenesis (17). Although the VDR KO phenotype of 241 the gland has been described (15,16), no reference has been made so far to the interactions between 242 E2, Prg and Prl, and vitD3 in a 3D environment in which morphogenesis takes place. To fulfill this need, 243 we have utilized two different 3D culture models to tease out vitD3's effects that are either dependent 244 or independent of its interactions with E2. We noticed that calcitriol exhibits a non-monotonic dose 245 response only in 3D cultures, a phenomenon not previously described. This is in line with current 246 knowledge regarding steroid hormone activity and also favors the notion that vitD3 functions as a 247 steroid hormone. 248 In the presence of estrogen in 3D culture, calcitriol increases total cell yield at 10 nM dose whereas it 249 decreases total cell yield at 100 nM dose (Fig. 1B). This reduction in cell yield can be attributed to cell 250 death given our observation of floater cells in both 2D and 3D cultures. Comparable evidence was found 251 when the role of vitD3 was explored in apoptosis (8,25). We also observed that calcitriol constrains the 252 effects of E2 on mammary epithelial morphogenesis without affecting total cell yield, more specifically 253 on the organization of epithelial ductal structures in 3D conditions. Consistent with our finding, the 254 mammary glands of CYP24A1 KO mice, which cannot metabolize calcitriol, exhibit stunted development 255 (31); Zinser et al (16) report that VDR KO mammary glands exhibit increased ductal elongation at 256 puberty. Of note, in both of these models, the proliferative capacity of the epithelial cells was not 257 affected. Given that E2 is responsible for ductal elongation during puberty, our results recapitulate In order to investigate whether vitD3 autonomously affects epithelial cells beyond its interactions with 268 E2, we utilized an estrogen-independent 3D culture model. MCF10A cells, considered to portray a 269 "normal-like" behavior in vitro, organize into mostly acinar and form some ductal structures in the 3D 270 collagen matrix (27). We observed that vitD3 retains its non-monotonic effects on morphogenesis even 271 in an estrogen-independent 3D culture model. We showed that this organization of MCF10A cells is 272 affected in a dose-dependent manner when treated with calcitriol in 3D cultures. These cells showed 273 greater sensitivity to calcitriol in 3D when compared to 2D cultures, with cell death increasing in a dose-274 dependent manner upward from a 10 nM dose (Fig. 5). 275 Mechanical forces are the main mediators of shape during morphogenesis. Previously, we have shown 276 that mammary epithelial cells embedded in a type I collagen matrix manipulate the collagen fibers 277 around them in the process of organizing into complex shapes such as ducts and acini (21,30); these 278 epithelial cells exert mechanical forces that act on collagen fibers and on other cells. As fibers organize, 279 they constrain the cells on their ability to move and to proliferate (33). We have also shown that 280 hormones distinctively influence the way epithelial cells organize collagen fibers, and consequently 281 determine the shape of the structures formed (21,34). Based on these results, we hypothesized that 282 vitD3 would also affect fiber organization. MCF10A cells treated with increasing concentrations of 283 calcitriol in 3D cultures increase the contraction of gels. Treatment with calcitriol also decreased the 284 number of cells in the gels in a dose-dependent manner. Gel contraction is dependent on the number of 285 cells present in the gel and on the manipulation of collagen fibers by the cells. While the lower number 286 of cells can account for the smaller sizes of structures observed in the 3D gels treated with calcitriol, this 287 does not explain the increased contraction of these gels. Picrosirius staining revealed that even though 288 there are fewer cells in the gels at 50 nM dose, there is a more uniform distribution of organized fibers 289 throughout the gel (Fig. 8). As our lab and others (35) have described, organized fibers are responsible 290 for transmission of forces and more organization of fibers leads to increased anisotropy in the 3D 291 environment. The increased contraction of the gels can therefore be explained by the transmission of 292 forces across long distances by the cells, a phenomenon previously reported (36). 293 We also observed that at 10 nM dose of calcitriol there was the least amount of organized fibers and the 294 greatest number of elongated structures (Fig. 8). On closer observation, the calcitriol treated gels 295 contained a lower number of branched, elongated structures (ductal, tubular) and a higher number of 296 unbranched, elongated structures (cord-like). Additionally, when compared to untreated gels, increase 297 in calcitriol dose resulted in shorter and thinner elongated structures. We have previously reported that 298 MCF10A cells embedded in a collagen type I matrix form ductal/tubular structures in the periphery of 299 the gel and cord-like structures in inner areas (22). In the calcitriol treated gels, in addition to the inner 300 areas, cord-like structures were also observed in the periphery. The arrangement of collagen fibers by 301 the cells is affected by a multitude of factors that include physical constraints. Therefore, in order to 302 fully elucidate the differential organization of structures depending on the calcitriol dose used, 303 additional measurements of local biomechanical parameters in the calcitriol-treated 3D gels is required.
All models, by definition, are simplified versions of the object being modeled. They are used precisely 305 because they reduce the number of variables considered relevant to explain a phenotype. Thus, like all 306 experimental models, 3D culture models have their limitations. For example, the use of established cell 307 lines, which are considered rather stable is dictated by the limitations of using freshly isolated primary 308 cells. Isolated human primary cells are not efficient in forming biologically relevant structures in collagen 309 or ECM matrices in vitro; only a small percentage of them express mammotropic hormone receptors and 310 they lose their potential to form structures shortly after being placed in culture (37)(38)(39) Figures   Fig. 1. Calcitriol affected total cell yield of T47D cells differently in 2D and 3D culture conditions, and only in the presence of E2 (0.1 nM). (A) Calcitriol reduced total cell yield starting at 50 nM and higher doses in 2D culture (*p<0.05, one-way ANOVA). (B) Calcitriol resulted in increased cell numbers at 10 and 25 nM doses, but decreased total cell number at 100 nM dose (*p<0.05 compared to E2, one-way ANOVA).   . MCF10A cells showed differential sensitivity to calcitriol depending on culture conditions. (A) 100 nM calcitriol significantly decreased total cell numbers in 2D culture, but in 3D culture (B), the effects were observed starting at 10 nM (*p<0.05, one-way ANOVA).