Osteopontin activity modulates sex‐specific calcification in engineered valve tissue mimics

Abstract Patients with aortic valve stenosis (AVS) have sexually dimorphic phenotypes in their valve tissue, where male valvular tissue adopts a calcified phenotype and female tissue becomes more fibrotic. The molecular mechanisms that regulate sex‐specific calcification in valvular tissue remain poorly understood. Here, we explored the role of osteopontin (OPN), a pro‐fibrotic but anti‐calcific bone sialoprotein, in regulating the calcification of female aortic valve tissue. Recognizing that OPN mediates calcification processes, we hypothesized that aortic valvular interstitial cells (VICs) in female tissue have reduced expression of osteogenic markers in the presence of elevated OPN relative to male VICs. Human female valve leaflets displayed reduced and smaller microcalcifications, but increased OPN expression relative to male leaflets. To understand how OPN expression contributes to observed sex dimorphisms in valve tissue, we employed enzymatically degradable hydrogels as a 3D cell culture platform to recapitulate male or female VIC interactions with the extracellular matrix. Using this system, we recapitulated sex differences observed in human tissue, specifically demonstrating that female VICs exposed to calcifying medium have smaller mineral deposits within the hydrogel relative to male VICs. We identified a change in OPN dynamics in female VICs in the presence of calcification stimuli, where OPN deposition localized from the extracellular matrix to perinuclear regions. Additionally, exogenously delivered endothelin‐1 to encapsulated VICs increased OPN gene expression in male cells, which resulted in reduced calcification. Collectively, our results suggest that increased OPN in female valve tissue may play a sex‐specific role in mitigating mineralization during AVS progression.

osteopontin (OPN), a pro-fibrotic but anti-calcific bone sialoprotein, in regulating the calcification of female aortic valve tissue. Recognizing that OPN mediates calcification processes, we hypothesized that aortic valvular interstitial cells (VICs) in female tissue have reduced expression of osteogenic markers in the presence of elevated OPN relative to male VICs. Human female valve leaflets displayed reduced and smaller microcalcifications, but increased OPN expression relative to male leaflets. To understand how OPN expression contributes to observed sex dimorphisms in valve tissue, we employed enzymatically degradable hydrogels as a 3D cell culture platform to recapitulate male or female VIC interactions with the extracellular matrix. Using this system, we recapitulated sex differences observed in human tissue, specifically demonstrating that female VICs exposed to calcifying medium have smaller mineral deposits within the hydrogel relative to male VICs. We identified a change in OPN dynamics in female VICs in the presence of calcification stimuli, where OPN deposition localized from the extracellular matrix to perinuclear regions. Additionally, exogenously delivered endothelin-1 to encapsulated VICs increased OPN gene expression in male cells, which resulted in reduced calcification. Collectively, our results suggest that increased OPN in female valve tissue may play a sex-specific role in mitigating mineralization during AVS progression.

K E Y W O R D S
calcification, osteopontin, sex differences, synthetic hydrogels, valvular interstitial cells

| INTRODUCTION
Aortic valve stenosis (AVS) is a cell-mediated progressive disease that affects 2.8% of the US population, but males have a twofold excess risk of developing this degenerative disease. [1][2][3] Macroscopic and microscopic analyses of stenotic human aortic valves demonstrate that females exhibit more fibrosis whereas males show a higher propensity for calcification during the progression of aortic valve disease. [4][5][6] Likewise, recent clinical data further support that females have increased hemodynamic dysfunction that results in more severe AVS, yet have lower calcification relative to males. 7 The underlying biological factors that determine sex differences in cardiac disease and more specifically, aortic valve stenosis remain poorly understood and are multivariate. 8,9 Identifying the molecular and cellular mechanisms involved in valve fibrosis and calcification is key to understanding AVS progression and developing future precision treatments that consider patient sex.
Valvular interstitial cells (VICs) are known to be the primary mediators of calcification in the aortic valve, and the sex of VICs may impact how calcification manifests and evolves in male and female valve tissue. 10 In vitro results suggest VICs activate to a pro-calcific osteoblast-like phenotype in response to osteogenic cues, such as matrix stiffening, serum phosphate levels, and pro-calcific cytokines. 11,12 However, sex is a variable that influences VIC phenotype. For example, sex-specific in vitro models of valve disease have been used to explore sex differences in early osteogenic markers, gene expression, inflammatory cues in patient sera, and the role of X-chromosome inactivation in VIC phenotype response. [13][14][15][16] Additionally, significant decreases in gene expression have been observed in vitro for pathways involved in calcification and ossification in female VICs relative to males. 14 While it is debated if VICs become true osteoblasts, VICs express markers of osteogenesis including the transcription factor RUNX2, an early marker, as well as osteocalcin (OCN), and osteopontin (OPN). 17 Although there is evidence that female VICs in their local microenvironment are inherently less predisposed to calcification, the processes behind these observed dimorphisms remain unclear.
OPN is a bone sialoprotein that is widely used as a marker of the osteogenic phenotype, but it is also considered to have anti-calcific properties. 18 Prior work has shown OPN likely mediates calcification of the aortic valve microenvironment, as OPN is present to varying degrees in calcified valves, but absent or very low in healthy valve tissue. 19 Recent clinical data suggest that OPN plasma levels may be a biomarker for severity of chronic heart failure, as high levels correlate with increased risk of mortality 20 and severity of aortic valve calcification, 21 yet may be cardioprotective post-MI. 22 Studies demonstrate OPN as a potential inhibitor of calcification, as OPN is an acidic phosphoprotein that binds calcium to regulate apatite crystal growth. [23][24][25] OPN has several post-translational modifications, including cleavage and phosphorylation sites, lending OPN to bind calcium with high affinity and co-localize to regions of biomineralization extracellularly. 26 OPN also plays a role in other calcific diseases and can modulate the response of multiple cell types 27,28 to prevent, or reverse, ectopic calcification, depending on its phosphorylation state. 29,30 As such, the pleiotropic role of OPN in regulating calcification has yet to be fully characterized in different contexts, with limited understanding as to how OPN regulates sexually dimorphic disease progression in valve tissue. Previous studies have suggested that OPN expression is enhanced by endothelin-1 (ET-1) in cell types such as cardiomyocytes and osteoblasts, 31,32 yet other work suggests ET-1 inhibition decreases OPN mRNA expression but also reduces calcification. 33 Thus, while the role of ET-1 and OPN in valve calcification remains unknown, ET-1 offers an upstream target to control OPN and investigate its effects on the distinct pro-fibrotic versus pro-calcific AVS phenotypes.
For mechanistic studies involving VICs, hydrogel culture systems have become valuable tools to selectively activate VICs and investigate their transition to myofibroblasts and osteoblast-like phenotypes. 15,34,35 For example, the VIC osteoblast-like phenotype, as indicated by expression of RUNX2, can be promoted when cultured within 3D gelatin methacrylate and hyaluronic acid hydrogel when exposed to osteogenic environmental cues. 36 Alternatively, VICs cultured on compliant substrates more readily activate to an osteoblast-like phenotype, as measured via OCN-positive nodule formation. 37 Furthermore, soft 1-kPa 3D hydrogels promoted the osteoblast-like phenotype of VICs relative to a stiffer 3-kPa hydrogel formulation. 38 Taken together, VIC osteogenesis may be investigated in vitro using soft 3D materials and may be used to clarify the effects of OPN on VIC phenotype.
Here, we aimed to explore the sex-specific role of OPN in mediating calcification of aortic valve tissue. We hypothesize that OPN reduces calcification in female valve leaflets relative to male valves.
To this end, we use a degradable PEG-based hydrogel with an interpenetrating network (IPN) of collagen type I that mimics aspects of the native valve microenvironment. We have shown previously that incorporating collagen type I, a component of valve tissue, into these PEG hydrogels allows for cell-mediated mineralization processes to occur in this synthetic system. 39 We employ this 3D hydrogel culture system as a precision biomaterial 40  Slides were rinsed for 2 min under running DI water and then incubated for 2 min in 5% sodium thiosulfate. Slides were again rinsed for 2 min under running DI water and incubated for 5 min in nuclear fast red solution. A final rinse under running DI water was performed before the sample was dehydrated in 95% alcohol, 2 Â 1 min, and then 100% alcohol (2 Â 1 min, then 3 min) and cleared in SafeClear (three changes, 2 min). A coverslip was mounted using Permount (Sigma-Aldrich) and the slide dried for >48 h before imaging.
For immunofluorescence staining, slides were fixed as listed above in 4% PFA and rinsed in PBS 3 Â 5 min, before permeabilization with 0.1% TritonX100 (Sigma-Aldrich) in PBS for 30 min at room temperature.
Secondary antibody was diluted in PBS (1:300 for proteins of interest, 1:1000 DAPI) and incubated at room temperature for 1 h in a humidity chamber. Samples were kept in PBS for unmounted imaging. Porcine valve leaflet tissues were collected within 24 h of slaughter, isolated from the valve cusp, and flash frozen using Tissue-Tek OCT Compound (VWR) in a cryomold on top of dry ice. Samples were brought to À20 C and sectioned at 30-μm onto glass slides (Colorfrost plus, Fisher Scientific) and stored at À20 C until use.
Samples were prepared for IF staining using the same protocol used for human tissues.

| mRNA isolation and gene expression via RT-qPCR
To extract mRNA, hydrogels were digested in type II collagenase (Worthington Biochemical Corporation) at 2 mg/ml. Samples were centrifuged for 5 min at 1000 rpm and supernatant was removed. The sample was resuspended in 10% FBS phenol-free M199 media (Thermo Fisher Scientific) and strained through a 100-μm cell strainer.
Samples were centrifuged again at 1000 rpm for 5 min. Supernatant was removed and the remaining pellet was lysed for mRNA isolation. the housekeeping gene L30 was used for normalization (Table 1). ab1791) incubating overnight at 4 C. Membranes were rinsed 3 Â 10 min in 5% milk in TBST, incubated with secondary conjugated antibody (Jackson ImmunoResearch; 1:5000 anti-Rb for OPN and 1:10,000 for HIS3) diluted in 5% Milk in TBST and rocked at room temperature for 1 h. Chemiluminescence was detected using Pierce ECL Plus solution (ThermoFisher Scientific) and ImageQuant LAS 4000 detector was used to assess relative protein expression to HIS3.

| Embedding and sectioning of hydrogel samples
Hydrogels were fixed in 10% formalin (Sigma-Aldrich) for 30 min, rinsed with multiple changes of PBS, and further rinsed in Tissue-Tek OCT Compound (VWR). The samples were then submerged in OCT for 48 h at 4 C. Samples were centered in a cryosectioning mold filled with OCT and were frozen in 2-methylbutane (Sigma-Aldrich) cooled by dry ice. Samples were stored at À70 C and brought to À20 C at time of use. Samples were sectioned (30 μm for hydrogels) using a Cryostat (CM1850, Leica) and placed on glass slides (Colorfrost plus, Fisher Scientific) and stored at À70 C.

| Image acquisition and analysis
Immunofluorescence (IF) images were acquired using a 20x air objective on a Zeiss LSM 710 confocal microscope. At minimum, 6-μm zstacks were collected using four z-steps. IF images were analyzed with FIJI 42 by creating an ROI and measuring the mean fluorescence intensity of each image.
For cell:ECM intensity ratios, a custom MATLAB (2017a) script was applied to analyze signal localization. Briefly, a median intensity z- projection image was created to improve nuclear segmentation and to accurately represent the OPN signal throughout the entire image volume. Nuclei were identified using DAPI signal, and those regions were dilated to include a small area of pixels surrounding each nucleus ("cellular" signal). Mean OPN signal intensities were then calculated for two distinct pixel subsets in each image: "cellular" pixels (identified by DAPI), and "ECM" pixels (all non-cellular pixels in the image). The data are presented as a ratio of mean OPN intensity values, which were calculated by dividing the cellular mean intensity by the ECM mean intensity (i.e., cell:ECM).
Histological samples stained with VK were imaged at Â20 with a Nikon Eclipse TE300 microscope and color camera. VK histology images were quantified using FIJI to measure integrated image intensity. 30 Briefly, images were converted from RGB to 8-bit grayscale, then a threshold was uniformly applied across conditions to yield binary images. Image intensity was quantified using the measure function and used for statistical analysis.
Additionally, these binary images were processed using a radially averaged autocorrelation function in FIJI to calculate the correlation length of VK puncta in each image ( Figure S1). Correlation length is determined by integrating the radial autocorrelation function, and describes the maximum radial distance at which features (i.e., VK puncta vs. surrounding tissue) are well segregated. 43,44 In other words, the correlation length is a measure of feature size; images with larger features will correlate at greater distances. Pixel values were converted using a measured conversion of 2.73 pixels/μm. Additional discussion regarding the utility of spatial autocorrelation measurements is included in the Supporting Information Material S1.  To better assess the distribution and spatial organization of VKpositive punctae, we utilized radial autocorrelation analysis to determine the correlation length of each sample. Radial autocorrelation analysis calculates the average "similarity" between pixels compared across increasing radial distances; two pixels of the same binary value have an autocorrelation of 1, and pixels of opposite binary values have an autocorrelation of À1. Generally, autocorrelation curves begin near 1 (as even images with small feature size will autocorrelate highly at short comparison distances) and decrease quickly with increasing spatial lag. Samples with large features (i.e., increased frequency and density of VK-positive signal) autocorrelate highly over longer distances, which can be summarized by integrating the area under these curves to determine the correlation length (see Figure S1 for more details). Autocorrelation curves were generated for female and male tissue sections (Figure 1d), and clearly illustrate differences in spatial organization of calcification between heavily calcified samples compared to sparser VK staining. Integrating these curves displays these same trends (Figure 1e)  While OPN is an osteogenic marker, it may also be involved with inhibiting calcification growth. 26

| Female VICs encapsulated within PEG + Col hydrogels have increased OPN gene expression
We next encapsulated porcine VICs within 3D degradable hydrogels that recapitulate key aspects of the aortic valve microenvironment, including relevant osteogenic and calcific phenotypes depending on media conditions. 39 Our goal was to test if sex differences were

| OPN localization shifts in female VICs cultured in PEG + Col hydrogels and treated with calcifying medium
Female or male VICs were encapsulated within PEG + Col IPN hydrogels and exposed to CM (OM + 1 mg/ml CaCl 2 ) for 12 days. Sec- ( Figure 5b). We then quantified the intensity of OPN signal with respect to distance from the nuclei using a custom MatLab script (Figures 5c and S4) for the OM or CM conditions. Mapping the cell: ECM mean intensity ratio of signal showed that while there was no significant differences between males in OM and CM, the change in OPN localization between females cultured in OM and CM was significantly different. This indicated that local changes in the OPN distribution may play a significant role in matrix mineralization in valve tissue; the intensity of the OPN signal near the nuclei for female VICs was significantly increased in CM relative to OM.

| Exogenous treatment of ET-1 results in sexspecific increase in OPN expression while reducing matrix calcification
To test our hypothesis that OPN was mediating the calcification within the 3D hydrogel system, we sought to induce OPN gene expression in VICs via exogenous stimulation with ET-1. ET-1 has been shown to induce OPN expression in cell types such as cardiomyocytes 46 and osteoblasts. 47 Female or male VICs were encapsulated within the PEG + Col hydrogel system and cultured for  Collectively, we show that females have increased levels of OPN expression, as shown in both human valve tissue ( Figure 2) and within our in vitro hydrogel cultures using female porcine cells ( Figure 3).

| DISCUSSION
Although only OPN mRNA, and not protein, was significantly increased in our hydrogel system, this may be related to OPN's diverse post-translational modifications and functions. Even after protease digestion, OPN fragments retain biological activity, 51 meaning that cell-secreted thrombin and MMPs can dramatically alter the availability of the complete OPN protein while lower molecular weight cleavage products preserve its cell-binding function even in the absence of its RGD and polyaspartate sequences. 30 These fragments have been observed to contain newly exposed active domains 52 Figure 6). We suspect ET-1 is a critical regulator of sex-specific fibro-calcification, as our current work demonstrates a link between ET-1 and OPN expression, and previous work suggests that ET-1 increases myofibroblast activation in female VICs via genes that escape XCI. The results presented in this manuscript are some of the first studies examining the sex-specific regulation of OPN in the context of ET-1, a biochemical factor critical to driving AVS progression.
We also showed that OPN increased, but calcification decreased, in both human tissues and in a PEG + Col hydrogel as a simplified model of AVS (Figures 3 and 4). One explanation for our observations might be related to the increased presence of a rich and disorganized collagen matrix in female valve tissue that was observed in Figure 2.
Collagen type I is known to bind OPN, 45

CONFLICT OF INTEREST
The authors declare no conflict of interest.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.