Collagen networks within 3D PEG hydrogels support valvular interstitial cell matrix mineralization

Abstract

Enzymatically degradable hydrogels were designed for the 3D culture of valvular interstitial cells (VICs), and through the incorporation of various functionalities, we aimed to investigate the role of the tissue microenvironment in promoting the osteogenic properties of VICs and matrix mineralization. Specifically, porcine VICs were encapsulated in a poly(ethylene glycol) hydrogel crosslinked with a matrix metalloproteinase (MMP)-degradable crosslinker (KCGPQG↓IWGQCK) and formed via a thiol-ene photoclick reaction in the presence or absence of collagen type I to promote matrix mineralization. VIC-laden hydrogels were treated with osteogenic medium for up to 15 days, and the osteogenic response was characterized by the expression of RUNX2 as an early marker of an osteoblast-like phenotype, osteocalcin (OCN) as a marker of a mature osteoblast-like phenotype, and vimentin (VIM) as a marker of the fibroblast phenotype. In addition, matrix mineralization was characterized histologically with Von Kossa stain for calcium phosphate. Osteogenic response was further characterized biochemically with calcium assays, and physically via optical density measurements. When the osteogenic medium was supplemented with calcium chloride, OCN expression was upregulated and mineralization was discernable at 12 days of culture. Finally, this platform was used to screen various drug therapeutics that were assessed for their efficacy in preventing mineralization using optical density as a higher throughput read out. Collectively, these results suggest that matrix composition has a key role in supporting mineralization deposition within diseased valve tissue.

Graphical Abstract

1. Introduction

Calcification of valve leaflets is an end stage indicator of aortic valve stenosis (AVS) — a cell-mediated, progressive disease that results in reduced function of the valve and overall heart efficiency [1]. AVS affects ~2.8% of the population over the age of 75 with total valve replacement remaining the sole treatment option for patients with severe forms of the disease [2,3]. Non-surgical treatments, including drug-based therapies, that reverse disease onset or prevent calcification do not exist, in part because of an incomplete understanding of AVS progression. Thus, to complement clinical data and animal models of AVS, there is a need for in vitro culture systems that would enable dissection of the complex milieu of microenvironmental signals that may be playing a major role in disease progression.

The main population of cells that resides within heart valves are the valvular interstitial cells (VICs). VICs play an important role in valve tissue homeostasis but are also known to play a role in pathological valve calcification. Additionally, the cells lining the valve, valvular endothelial cells (VECs), have been known to play a key role in AVS progression. Previous studies have focused on the development of hydrogel co-cultures that mimic the valve cell organization in an attempt to understand VIC-VEC crosstalk [4-7]. While the results of VIC-VEC coculture reveal the beneficial effects of endothelial nitric oxide synthase (eNOS) and the role of cellular crosstalk in mediating cell phenotype [8-10], simplified experimental designs allow one to isolate specific cellular responses in a controlled environment. Therefore, this work is focused on the individual role of VICs and their specific response to microenvironmental changes.

Calcific nodules from diseased human valves are composed of different levels of calcium, phosphate, oxygen, and magnesium in various crystal structures, resulting in apatite formation [11,12]. In AVS, VICs express osteogenic markers, such as osteoblast-specific runt-related transcription factor 2 (RUNX2), which is an early marker of osteoblast commitment and differentiation in mesenchymal stem cells [13-16]. Under pro-mineralizing conditions, VICs express markers of mature osteoblasts, such as osteocalcin (OCN), a protein that is abundant in non-collagenous bone [17]. Transcription of OCN has been reported to be regulated in part by vimentin (VIM), an intermediate filament protein and well-known VIC fibroblast marker [18-20] where low levels of VIM are associated with a mature, differentiated osteoblast [21,22], Pro-mineralizing conditions for VICs that have been studied in vitro include osteogenic differentiation medium, which contains ascorbic acid (ASC), β-glycerophosphate (β-GP), and dexamethasone (dex), in addition to soft biomaterial culture systems [23,24].

Recent in vitro work aimed at exploring the progression of AVS demonstrated the importance of substrate dimensionality and stiffness as determinants of VIC phenotype [20,25-27]. VICs cultured on the surfaces of stiff materials readily activated to a highly contractile myofibroblast phenotype that formed Alizarin Red-positive nodules [28]. It has been reported that VIC-embedded soft collagen type I hydrogel matrices demonstrated an osteoblast-like VIC phenotype upon exposure to osteogenic medium, as measured by an increase in nodules and alkaline phosphatase (ALP). The Ras homolog family member A (RhoA) was shown to be a regulator of this process [29]. In addition, others have cultured VICs within gelatin-methacrylate-based hydrogels, and VICs were shown to respond to osteogenic medium via increases in RUNX2 gene expression and nodule formation [30] or by OCN gene expression [23]. However, detection of mineralization and late stage markers (OCN) within these VIC-laden matrices required several weeks [23,30].

In vivo, valve extracellular matrix (ECM) is thought to play an important role in mediating VIC phenotype and calcification [31,32]. About 50% of a healthy valve, as measured by dry weight, is comprised of collagen, most of which is collagen type I (~74%) [33]. Increased remodeling, disorganized deposition, and crosslinking of these collagen isoforms can lead to stiffening of valve leaflets [34]; however, it is unclear whether changes in collagen type and organization contribute to the mineralization that is observed in later stages of AVS [35-37]. While the VICs embedded in collagen hydrogels by Farrar et al. demonstrated increased osteogenic response, collagen mRNA expression decreased with time in osteogenic conditions [29]. Other reports show that diseased human valves have an overall decrease in collagen content, but an increase in collagen type I located around calcific nodules, thus indicating increased collagen turnover with disease [38]. While collagen is present ubiquitously in many tissues, the type, its crosslinking, and ratio of various isoforms changes with the progression of AVS, making it difficult to ascertain its specific role in valve ECM calcification.

With this in mind, we sought to engineer a 3D hydrogel system that is well suited for long-term culture of VICs [39]. Specifically, we employed a thiol-ene photoclick polymerization between a synthetic 8-armed 40 kDa poly(ethylene glycol)-norbornene hydrogel crosslinked with a matrix metalloproteinase (MMP)-degradable peptide crosslinker (KCGPQG↓IWGQCK) in the presence or absence of collagen type I to generate 3D hydrogels, which is well suited for VIC culture [26,27,40]. Porcine VICs were encapsulated within PEG hydrogels to permit longer-term culture, and the VIC osteoblast-like response was characterized via gene expression and protein fluorescence intensity (RUNX2, OCN, and VIM), contraction, and histology for mineral deposition. Toward better understanding of the role of matrix mediated calcification, an interpenetrating network of collagen type I was embedded within the PEG hydrogels, and VIC osteogenic response was measured similarly. Finally, this 3D model of VIC matrix calcification was used as a platform to screen for potential therapeutics that might inhibit calcification: osteoprotegerin (OPG), rosmarinic acid (RA), and zoledronic acid (ZA). Cumulatively, these data demonstrated the use of this degradable PEG + collagen hydrogel system to investigate the role of microenvironmental factors on VIC osteogenic potential, highlighting its use as a potential drug screening platform.

2. Materials and methods

2.1. Polymer Synthesis

8-armed 40 kDa poly(ethylene) glycol (PEG) norbornene was functionalized as previously described [41]. Briefly, amine functionalized 8-armed 40 kDa PEG (JenKem) was reacted with 48 molar equivalents of 5-norbornene-2-carboxylic acid (Sigma-Aldrich), 40 molar equivalents of O-(7-Azabenzotriazol-1-yl)-N,N,N,N-tetramethyluronium hexafluorophosphate (HATU, Chem- Impex Intl. Inc.) and 80 molar equivalents of N,N-diisopropylethylamine (DIEA, Sigma-Aldrich) in N,N-dimethylformamide (DMF, Fisher Scientific) overnight. Product was precipitated into cold ethyl ether (VWR), centrifuged, and washed a second time in ether. Prior to lyophilization, product was dried by vacuum, resuspended in deionized water, and dialyzed against water. ¹H nuclear magnetic resonance spectroscopy confirmed end-group functionalization; ¹H NMR (400MHz, CDCl 3, δ): 6.25-5.95 (m, 2H), 3.85-3.25 (m, PEG, 260H).

2.2. Rheological Measurements

Swollen hydrogel mechanical properties were assessed on a DHR3 rheometer (TA Instruments) using frequency sweep and strain sweeps. This ensured measurements fell within the linear viscoelastic regime. Studies were done at 37°C with an 8 mm standard geometry Peltier Plate tool coated with adhesive 600/P1200 sandpaper to prevented slippage.

2.3. VIC Isolation and Expansion

Male porcine hearts were acquired from Hormel Foods Corporation (Austin, MN, USA) within 24 hours of slaughter. Male VICs were used in these studies due to the purportedly increased propensity of calcification in males [42], and the cells in these studies were pooled from 30 male hearts. The aortic valve leaflets were extracted and placed into a solution of Earle’s Balanced Salt Solution (Sigma-Aldrich) containing 1.2% Penicillin/streptomycin (ThermoFisher Scientific) and Amphotericin B (ThermoFisher Scientific, 1 μg/mL Amphotericin B, 0.82 μg/mL sodium deoxycholate) at 37°C. The tissues were suspended in a solution of sterilized Type II Collagenase (Worthington Biochemical Corporation) at 250 u/mL in wash buffer and shaken at 37°C. An initial digest of 30 minutes followed by 30 sec of vortexing removed the endothelial cells. Then, a second digestion occurred in fresh collagenase solution for 70 minutes at 37°C, before vortexing for 2 minutes. The samples were strained through a 100 μm cell strainer to remove any remaining leaflet fragments. The isolated VICs were resuspended in M199 media (ThermoFisher Scientfic) supplemented with 15% fetal bovine serum (FBS, ThermoFisher Scientific), 1.2% Penicillin/streptomycin, and Amphotericin B (1 μg/mL Amphotericin B, 0.82 μg/mL sodium deoxycholate). VICs were plated on TCPS and expanded for up to a week, with media changes every other day. Cells were frozen in 45% expansion media (15% FBS M199), 50% FBS, and 5% DMSO (Sigma-Aldrich). All subsequent experiments were conducted using 10% FBS media supplemented with 1.2% Penicillin/streptomycin, and Amphotericin B (1 μg/mL Amphotericin B, 0.82 μg/mL sodium deoxycholate).

2.4. Cell Culture on TCPS

All cell experiments used male VICs between passage 2 and 3 (P2, P3s). For the Alizarin Red staining images presented in Figure 1a on TCPS, VICs were seeded directly onto a treated 6-well TCPS plate at ~30,000 cells/cm² in 10% FBS M199. Media was changed 24 hours post seeding (2 mL per well), and then every third day. GM consisted of 10% FBS M199, while OM (0.1 μM dex) contained 10 mM β-glycerophosphate (Sigma-Aldrich), 50 μg/mL L-ascorbic acid (Sigma-Aldrich), and 0.1 μM dexamethasone (Sigma-Aldrich). For the cell data presented in Figure 1b, VICs were seeded on a glass-bottom 24 well plate for imaging purposes at ~30,000 cells/cm². Glass bottom wells were treated with 0.1% gelatin in water (Sigma-Aldrich) for a minimum of 30 minutes before cells were seeded. Media was changed 24 hours post-seeding and then treated every three days. Cells were seeded in 10% FBS M199 overnight before subjection to experimental treatment conditions. At each timepoint, cells were fixed with 4% paraformaldehyde (PFA, Electron Microscopy Sciences) in PBS and stained with DAPI (Sigma-Aldrich). For the cell data presented in Supplemental Figure 4, VICs were seeded at 5,000 cells/cm² and cultured in either GM, or the zoledronic acid (ZA) at 0.5, 5, 50, or 500 μM diluted in GM for 3 days before fixing the cells in 4% PFA in PBS, and staining the nuclei with DAPI. Number of DAPI-stained nuclei was assessed from images acquired using an Operetta High-Content Imaging System (Perkin Elmer).

Figure 1. Long-term studies of the calcific phenotype of porcine valvular interstitial cells (VICs) are limited on tissue culture polystyrene (TCPS).

a) Representative images of calcium nodules (Alizarin red) present in valve tissue section from a 76-year-old human male with aortic valve disorder vs. VICs on TCPS at 7 and 15 days in osteogenic medium (OM) conditions. Scale bar = 0.5 mm. b) Average number of VICs per well over time in growth medium (GM) and OM (n ≥ 10 biological replicates, each replicate averaged 8 fields of view). c) RUNX2 and d) OCN mRNA gene expression levels of VICs in GM and OM conditions after 5 days on TCPS relative to RPL30 (n = 3 biological and technical replicates). Results analyzed using two-way ANOVA with Sidak’s multiple comparison test (b), or unpaired t-test (c,d). *p<0.05, **p<0.01, ***p< 0.001, ****p<0.0001.

2.5. VIC Encapsulation and Culture Treatments

VICs were encapsulated at 10 million cells/mL (P2s) within a matrix metalloproteinase (MMP)-degradable PEG hydrogel using photo-initiated thiol-ene polymerization into a 6 mm by 1 mm rubber mold placed on top of a teflon sheet. The polymer solution was prepared by mixing 8-armed 40 kDa PEG norbornene ([enes] = 6 mM), photoinitiator lithium phenyl-2,4,6-trimethylbenzoylphosphinate ([LAP] = 1.7 mM), CRGDS adhesive peptide ([thiols] = 1 mM, Bachem), and peptide cross linker KCGPQG↓IWGQCK ([thiols] = 3.8 mM, Bachem) to obtain a thiol:ene ratio of 0.8 and a storage modulus of 500 ± 76 Pa for all encapsulation studies. This polymer precursor solution was adjusted to a pH of ~7.0 with sterile 1M NaOH (Fisher Scientific). Hydrogels were polymerized for 3 minutes using 365 nm 2.5 mW/cm2 light and cultured in a solution of 10% FBS M199 media overnight prior to the introduction of any experimental conditions.

PEG + Collagen IPN hydrogels were prepared by first preparing a polymer solution of 8-arm 40 kDa PEG norbornene ([enes] = 6 mM), photoinitiator ([LAP] = 1.7 mM), CRGDS adhesive peptide ([thiols] = 1 mM, Bachem), and peptide cross linker KCGPQG↓IWGQCK ([thiols] = 3.5 mM, Bachem) to obtain a thiol:ene ratio of 0.75 in a solution with 1 mg/mL rat tail collagen type I (Fisher Scientific), and 10X Dulbecco’s phosphate buffered saline (DPBS, Thermo Fisher Scientific). The polymer mixture was prepared, the collagen solution (at 4° C) added, and slowly brought to neutral pH with the addition of 200 mM NaOH (Fisher Scientific). VICs were then added at 30 million cells/mL and mixed via gentle pipetting. Then, 30 μl of solution was added into the 6 x 1 mm rubber mold and polymerized under for 3 minutes under 365 nm 2.5 mW/cm2 light and cultured in a solution of 10% FBS M199 media overnight prior to any experimentation conditions.

Osteogenic medium was prepared using 10% FBS M199 supplemented with 10 mM β-glycerophosphate (Sigma-Aldrich), 50 μg/mL L-ascorbic acid (Sigma-Aldrich), and 1 mM dexamethasone (Sigma-Aldrich), unless otherwise noted. Calcifying medium (CM) was prepared by supplementing OM with 1 mg/mL of calcium chloride (Sigma-Aldrich). With the 0.2 mg/mL CaCl₂ present inherently in M199 medium, free Ca²⁺ levels in the CM condition were calculated to be ~0.4 mg/mL. Calcium inhibition treatments were prepared with the following, diluted into CM: zoledronic acid monohydrate (Sigma-Aldrich SML0223), rosmarinic acid (Sigma-Aldrich R4033), and recombinant human osteoprotegerin (Novus Biologicals, 185-OS-025). These therapeutics were all soluble in either H₂O or PBS.

2.6. mRNA Isolation and RT-qPCR Assessment of Gene Expression

After treatment of encapsulated VICs with different supplemented medias the gels were digested using Type II Collagenase (Worthington Biochemical Corporation) at 2 mg/mL for 30 minutes at 37°C. The samples were centrifuged at 1000 rpm for 5 minutes 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. The sample was centrifuged for 5 minutes at 1000 rpm, supernatant was removed, and the remaining pellet lysed for mRNA purification using an RNAeasy Micro kit (Qiagen) following manufacturer’s instructions. The concentration and quality of the mRNA was evaluated using a ND-1000 Nanodrop Spectrophotometer. RNAse-free water and mRNA were mixed with iScript Reverse Transcription Supermix (Bio-Rad) and cDNA was synthesized with an Eppendorf Mastercycler. Relative mRNA levels were measured using SYBR Green reagent (Bio-Rad) and an iCycler machine (BioRad). Results were normalized to the housekeeping gene RPL30 using the ΔΔCq method. The use of 60S ribosomal protein 30 (RPL30) as a housekeeping gene is based on the high stability expressed by this gene within samples of the same tissue [43,44]. Primers used in these studies are listed here:

Gene	Forward primer (5’-3’)	Reverse primer (5’-3’)
RPL30	AGATTTCCTCAAGGCTGGGC	GCTGGGGTACAAGCAGACTC
RUNX2	AACAACCACAGAACCACAAG	TGACCTGCGGAGATTAACC
OCN	TCTCTCCAGACCCCAGTG	GCCAGACAAGGACATTGAAG

2.7. Histological Preparation, Staining, and Quantification

Hydrogels for histological assays were fixed in 10% formalin (Sigma-Aldrich) for 30 minutes at RT, rinsed well with PBS, rinsed with Tissue-Tek OCT Compound (VWR), and then submerged in OCT for >48 hours at 4°C. Samples were then placed in a cryosection mold which was filled with OCT and then frozen by dipping in dry-ice chilled 2-methylbutane (Sigma-Aldrich). Samples were stored at −70°C until use. Samples were brought to −20°C and sectioned at 30 μM using a Cryostat (CM1850, Leica) onto glass slides (Colorfrost plus, Fisher Scientific) and stored at −70°C. Human tissue samples were purchased from OriGene (Cat. No. CS508070, sample ID FR0001995D, case ID CI0000005759). The valve tissue was from a 76-year old male patient diagnosed with coronary artery disease and hypertension undergoing a surgical valve replacement post myocardial infarction.

Sectioned samples were warmed to room temperature, outlined in PAP pen (Fisher Scientific), and hydrated with PBS (3 times 5 minutes), and then rinsed in DI water (3 x 2 minutes). For Von Kossa staining, samples were covered in 5% Silver Nitrate solution and put under a UV light for 60 minutes. Samples were rinsed in 2 minutes of running DI water before incubation in 5% Sodium Thiosulfate for 2 minutes. The slide was then rinsed under running DI water for 2 minutes. The slide was incubated in Nuclear Fast Red Solution for 5 minutes and again rinsed under running DI for two more minutes. The sample was dehydrated in 95% alcohol for 2x 1 minute, then 100% alcohol 3 times (2 x 1 minute, then 1 x 3 minutes), and then cleared in SafeClear (3 changes, 2 min each, Fisher Scientific). The coverslip was mounted using Permount Mounting Medium (Fisher Scientific), let dry for >48 hrs.

For Alizarin Red staining, samples were hydrated to DI as listed above and then Alizarin Red stain (Millipore Sigma) was applied at room temperature for 5 minutes. Solution was filtered with a 0.45 μm filter immediately before use. Excess dye was removed, and samples were dehydrated in 3 changes of acetone, 1 change of 1:1 acetone and SafeClear, and then 3 x3 min changes in SafeClear. Samples were mounted using Permount let dry for >48 hrs.

For immunofluorescence staining, slides were warmed to room temperature and permeabilized with 0.1% TritonX100 (Sigma-Aldrich) in PBS for 30 minutes. Samples were then blocked 5% bovine serum albumin (BSA, Sigma-Aldrich) in PBS for 1 hour at room temperature. Primary antibody (1:1000 for anti-RUNX2 ab23981, 1:1000 for anti-vimentin ab8069, or 1: 50 anti-OCN ab198228) was diluted in 5%BSA and incubated at 4° overnight in a humidity chamber. Samples were washed 3 x 5 minutes with PBST (PBS with 0.1% Tween20, Sigma Aldrich) before secondary antibody (1:300 AlexaFluor anti-rabbit 647, 1:300 AlexaFluor anti-mouse 488, 1:1000 DAPI) was incubated for 1 hour at room temperature in a humidity chamber. Samples were washed and kept in PBS for unmounted imaging. Sections stained for OCN were labeled with AlexaFluor 647 and false colored to yellow for clarity.

Histology samples were imaged at 20x with a Nikon Eclipse TE300 and a color camera. Von Kossa histology images were quantified using FIJI [45] and measuring integrated image intensity. Briefly, images were converted to grey 8-bit and a threshold was applied uniformly across conditions to yield binary images. The entire image intensity was quantified using the measure function and used for statistical analysis.

Immunofluorescent images were acquired using a 20x air objective on a Zeiss LSM 710 microscope. At minimum, 6 μm z-stacks were collected using 4 z-steps. The number of cells was quantified by using a custom MATLAB script. Mean immunofluorescent intensity of OCN/VIM and/or nuclear to perinuclear intensity of RUNX2 was assessed on maximum intensity projections of images using a custom MATLAB script. Briefly, nuclear RUNX2 signal was contained within the nuclear borders identified by DAPI signal, and perinuclear RUNX2 signal was contained within a narrow ring around each nucleus.

2.8. Contraction Assays

Cells were encapsulated as previously described and placed in 10% FBS M199. Samples were placed on a microscope slide on top of 10 mm graph paper, and photographs were taken from directly above the gels. Images were collected for each experiment on day 1, 3, 5, 7, 10, 13, and 15. Projected surface area was measured using FIJI. Two measurements per gel per day were collected and averaged to take the final projected surface area. Results were normalized to the initial gel diameter and represented as a percentage of this initial value.

2.9. Opacity Assessments

Samples were cultured in 10% M199 phenol free medium (Thermo Fisher) within an untreated glass bottom 24 well plate. For each timepoint, medium was removed, and samples were moved to the center of the well. Mean optical density was measured at 600 nm using area scan on the plate reader (BioTek Synergy H1), which is a common wavelength for determining light transmission of polymeric materials [46,47].

2.10. Calcium Assay

Calcium content was measured via Quantichrome Calcium Assay Kit (BioAssay Systems) as per manufacturer's instructions. To prepare samples, cell-laden gels are rinsed with PBS (no calcium added) for 3 x 5minutes, flash frozen, and lyophilized until dry. Then, 100 μL of 1 M HCl is added to the dried sample and stored at 4°C for 48 hours minimum. Samples were then diluted into 1 M HCl and 5 μL of each standard and sample were pipetted into a 96-well plate for comparison to a standard curve. Then, 200 μL of working reagent is added to each well, incubated at room temperature for 3 minutes, absorbance detected at 612 nm, and extrapolated to determine concentration of calcium in solution.

2.11. Caspase 3/7 Assay

Caspase-3/7 activity was assessed using CellEvent™ Caspase-3/7 Green Detection Reagent and was performed as per manufacturer's instructions (Invitrogen, Fisher Scientific). Briefly, caspase reagent was diluted to 8 μM into 5% FBS in PBS. Phenol free media was removed from the sample plate and 300 μl of the diluted reagent was added in each well. The samples were incubated at 37° C for 1 hour. 100 μl of the solution was pipetted into a 96-well plate twice (total of 200 μl), and fluorescence was read at excitation of 502 nm and emission of 530nm. After caspase reading, the gels were digested in collagenase and the pellet was resuspended for cell number counting and normalization.

2.12. Viability Assessment

VIC viability was assessed using Live/Dead staining (Thermo Fisher). Briefly, cells were encapsulated and placed in culture condition indicated, either OM, CM, or CM + 50 μM ZA drug treatment. Encapsulated cells were stained with calcein-AM at 0.5 μM and 4.0 μM of ethidium homodimer-1 for 30 minutes prior to confocal imaging within the live cell chamber using an Opera Phenix (Perkin Elmer). Cell viability was assessed via Harmony using the maximum intensity projection of the stack.

2.13. Statistical Analysis and Miscellaneous

Data are presented as mean +/− standard deviation with details regarding biological and technical replicates in the respective figure captions. Data were analyzed using Graph pad™ (Prism). Statistical tests and post hoc analyses are indicated in the appropriate figure captions. A table of all acronyms can be found in Supplemental Table 1.

3. Results

3.1. Tissue culture polystyrene (TCPS) has limitations for the long-term study of the VIC calcific phenotype

To identify cell culture conditions that recapitulate aspects of aortic valve calcification, cells were cultured using standard TCPS and osteogenic medium and evaluated for osteogenic markers. Aortic valve tissue from a 76-year-old human male patient diagnosed with an aortic valve disorder was stained for calcium deposition using Alizarin Red (Figure 1a, top panel). The representative valve section displayed calcium-positive nodules in discrete punctae, roughly 150 μm in diameter. Staining of this valve section was qualitatively compared to male porcine aortic VICs cultured on TCPS in osteogenic medium (OM; 10 mM β-glycerophosphate (β-GP), 50 μg/mL ascorbic acid (ASC), and 0.1 μM dexamethasone (dex) for up to 15 days. Alizarin Red staining of the TCPS-plated VICs revealed one large calcium-positive nodule after 7 days, and many calcium-positive cells that detached in a sheet to form a mass roughly 1.5 mm in diameter after 15 days. Neither timepoint in our studies on TCPS captured the calcium-positive morphology nor punctae pattern observed in the representative human tissue section. Upon observation of the sheets of detached cells, the average cell number for plated VICs was measured as a function of time in response to treatment with either growth medium (GM, 10% FBS M199) or OM (0.1 μM dex) (Figure 1b). GM conditions resulted in a steady increase of cell number over time, while the VICs cultured in OM (0.1 μM dex) began contracting and lifting off the plate after just 5 days of culture in sheets as demonstrated in the day 15 TCPS image. Error bars depict the standard deviation of the mean and illustrate the vast range of cell numbers observed between wells in a typical experiment under these conditions.

VIC mRNA was collected from TCPS plated VICs in both GM and OM (0.1 μM dex) culture conditions at day 5. Gene expression for RUNX2, an early marker of the osteogenic phenotype, was significantly increased in OM (0.1 μM dex) relative to GM (Figure 1c), indicating that VICs responded to the osteogenic cues by upregulating the expression of this osteogenic marker. In contrast, expression of OCN, a later marker of the osteogenic phenotype, was significantly decreased in the OM (0.1 μM dex) condition relative to GM (Figure 1d), indicating that VICs in these conditions do not express late stage markers of mature osteoblasts. Taken together, these results highlight some of the limitations in using TCPS to study osteogenic phenotypes of VICs, and motivated experiments with hydrogels to allow long-term culture of VIC populations and circumvent any issues with VIC contraction and detachment from the surface.

3.2. VICs encapsulated in soft, degradable PEG hydrogels respond to osteogenic medium but do not calcify

To permit longer-term culture and explore the osteogenic response of VICs, we encapsulated VICs within soft PEG hydrogels (G’ ≈ 500 Pa) containing a matrix metalloproteinase (MMP) degradable cross-linker (KCGPQG↓IWGQCK) and fibronectin-derived CRGD ligands to promote integrin binding (Figure 2a). VIC-laden hydrogels were cultured for up to 15 days in OM containing 0, 0.1 μM, and 1 mM dexamethasone (dex). Two concentrations of dexamethasone were tested, as this component of osteogenic medium is inconsistent throughout published results using VICs, whereas the β-GP and ASC component concentrations are less variable [28,48,49]. Gene expression was assessed at day 15 for RUNX2 (Figure 2b) and OCN (Figure 2c). RUNX2 gene expression increased at higher concentrations of dex, with the maximal response observed in OM containing 1 mM dex with a 9.4-fold increase relative to GM. While the VICs’ response to OM increased for this early marker of osteogenesis, OCN expression was not significantly different for any of the treatment conditions. Because maximal RUNX2 expression was detected in OM (1 mM dex), this concentration was selected for all OM conditions moving forward. Protein fluorescence intensity for RUNX2, OCN, and fibroblast marker vimentin (VIM) was assessed for VICs within PEG gels in OM via immunostaining of cryosectioned VIC hydrogel samples (Figure 2d). Nuclear localization of RUNX2 was significantly increased in OM compared to GM samples (Figure 2e), while there was no change in OCN or VIM intensity between VICs in GM and OM (Figure 2f,g).

Figure 2. Soft synthetic 3D PEG-based hydrogel cultures increase early osteogenic marker RUNX2, but not mineralization.

a) Schematic representation of 15-day osteogenic medium (OM) experiment of VICs encapsulated in 3D PEG hydrogels. b) RUNX2 and c) OCN mRNA gene expression levels relative to RPL30 of VICs in growth medium (GM) and OM with increasing dexamethasone (dex) dose (n = 3 biological and technical replicates). d) Representative images of immunostained VICs from sectioned PEG hydrogel samples in GM or OM (1mM dex) for osteoblast-like and fibroblast markers. Red, RUNX2 (runt-related transcription factor 2); yellow, OCN (osteocalcin); green, VIM (vimentin); blue, nuclei. Scale bar = 50 μm. e) RUNX2 nuclear to perinuclear fluorescence intensity analysis of VICs in GM and OM (1mM dex). f) OCN and g) VIM mean fluorescence intensity analysis of VICs in GM and OM (1mM dex) (n > 8 images compiled from 3 biological replicates). h) Representative images display lack of calcification via Von Kossa (VK) staining of sectioned PEG hydrogels, i) integrated density analysis of VK (n= 3 biological replicates. Minimum 3 images used for GM condition, and 12 images for OM), and j) opacity analysis (600nm) from encapsulated VICs in GM and OM (1mM dex) (n=3 biological replicates). Scale bar = 100 μm. (n=3 biological replicates). All data presented are normalized to corresponding GM condition. Results analyzed using one-way ANOVA with Dunnett’s multiple comparison test, or an unpaired t-test. *p<0.05, **p<0.01, ***p< 0.001, ****p<0.0001.

Cell-mediated hydrogel contraction was measured via projected surface area measurements over time (Supplemental Figure 1, as bulk contraction is indicative of phenotypic changes [23,25]. Results indicate that contraction occurred in the OM condition with the addition of dex at both high and low concentrations. OM condition without dex was not significantly different from the GM condition. Hydrogels cultured in GM and OM conditions were sectioned and stained for calcification via Von Kossa staining (Figure 2h). These data revealed no visible changes in calcification in the OM condition relative to GM; however, the integrated density analysis of Von Kossa sections demonstrated a 1.48-fold increase in positive calcification stain in hydrogels cultured in OM relative to GM (Figure 2i). Opacity, or optical density (OD, 600 nm), was measured as an additional metric for changes in matrix calcification. A similar technique of measuring changes in light transmittance has been previously implemented in detection of calcified areas of stem cells on TCPS [50]. No significant change in opacity was observed in GM vs. OM hydrogels at day 15 (Figure 2j). These results demonstrate that encapsulated VICs were responsive to exogenous osteogenic cues via increases in early markers of osteogenesis (RUNX2), but significant matrix calcification was not observed.

3.3. Incorporation of collagen interpenetrating networks (IPNs) within degradable PEG hydrogel microenvironment promotes VIC matrix calcification.

We next sought to investigate whether or not changes in matrix composition might accelerate or exacerbate differences in mineralization and expression of late stage osteogenic markers in encapsulated VICs. To investigate the role that matrix proteins might play in nucleating mineralization, we engineered our degradable PEG hydrogel platform to contain collagen type I as an interpenetrating network (IPN), as depicted in Figure 3a. The PEG + collagen IPN (referred to henceforth as PEG + Col) hydrogel composition was tailored to minimize differences in the shear storage modulus (G’) (Supplemental Figure 2a) and allow us to instead focus on the role of the collagen in influencing the VIC phenotype, as it is known that matrix modulus mediates VIC phenotypic changes [51]. First, to verify that VICs would respond to OM regardless of the hydrogel formulation, the projected surface area was measured over time up to 15 days. Results indicated that the PEG + Col hydrogel treated with OM contracted significantly more compared to the GM condition, although contraction was not dependent on the material system (Supplemental Figure 2b,c). Additionally, no significant changes in cell number were detected with the addition of collagen, or the treatment of OM relative to GM after 15 days in culture (Supplemental Figure 2d).

Figure 3. Collagen interpenetrating networks (IPNs) provide mineral nucleation sites within PEG hydrogels.

a) Schematic representation of 3D PEG + Col IPN hydrogels and 15-day VIC osteogenic medium (OM) experiments. b) Representative images of immunostained VICs in sectioned PEG + Col hydrogels in either growth medium (GM) or OM for osteoblast-like and fibroblast markers. Red, RUNX2 (runt-related transcription factor 2); yellow, OCN (osteocalcin); green, VIM (vimentin); blue, nuclei. Scale bar = 50 μm. c) RUNX2 nuclear to perinuclear fluorescence intensity analysis (n= minimum 14 images compiled from 3 biological replicates) and d) gene expression relative to RPL30 of VICs in GM and OM (n= 3 biological and technical replicates). e) OCN mean fluorescence intensity analysis (n= minimum 9 images compiled from 3 biological replicates) and f) gene expression relative to RPL30 of VICs in GM and OM (n= 3 biological and technical replicates). g) VIM mean fluorescence intensity analysis of VICs in GM and OM (n= minimum 14 images compiled from 3 biological replicates). h) Representative images of non-coalesced calcification via Von Kossa (VK) staining of sectioned PEG + Col hydrogels i) integrated VK density analysis (n=3 biological replicates, 9 images per condition), and j) opacity measurements (600nm) from VICs in GM and OM (1mM dex)(n= 3 biological replicates). All data presented are normalized to corresponding GM condition. Scale bar = 100 μm. Results analyzed using an unpaired t-test. **p<0.01, ****p<0.0001.

Expression of VIC fibroblast and osteoblast-like phenotypic markers were assessed in the PEG + Col system. Representative images of sectioned VIC-laden hydrogels after 15 days in GM or OM and stained for RUNX2, OCN, and VIM are shown in Figure 3b. RUNX2 was significantly localized to the nuclear space in OM treated samples (Figure 3c). Additionally, gene expression results showed a 12.6-fold increase in expression in the OM conditions relative to GM (Figure 3d). Neither OCN mean fluorescence intensity nor gene expression was significantly different between GM and OM conditions (Figure 3e,f). However, the mean fluorescent signal of fibroblast marker VIM was significantly lower in conditions cultured in OM compared to GM, indicating that VICs may become less fibroblast-like in PEG + Col, OM conditions (Figure 3g).

When comparing the PEG and PEG + Col hydrogel systems, RUNX2 gene expression was 1.4-fold increased in VICs cultured within the PEG + Col hydrogels at day 15 (Supplemental Figure 2e). This result indicated that VICs within the PEG + Col system may be having a stronger response to the osteogenic stimuli compared to the PEG system, so further characterization of the VIC osteogenic response and mineralization in the PEG + Col system was performed.

Histological sections of VICs within the PEG + Col system at day 15 in either GM or OM were stained with Von Kossa (Figure 3h). While more positive staining was observed in the samples cultured in OM compared to GM (Figure 3i), this staining did not coalesce into discrete black punctae, a pattern more indicative of calcium phosphate deposition. However, quantification of the Von Kossa staining revealed significantly higher positive staining in the OM conditions relative to GM conditions. Further comparison between the PEG and PEG + Col hydrogels higher levels of mineralization in the PEG + Col system (Supplemental Figure 2f). When measuring optical density, there were no changes in VIC-laden PEG + Col gels cultured in GM or OM (Figure 3j). Additionally, opacity was unchanged in PEG + Col hydrogels compared to unmodified PEG (Supplemental Figure 2g). Collectively, these results suggested that the incorporation of collagen as an IPN into degradable PEG increased markers of osteogenesis, but still did not result in robust matrix mineralization within 15 days.

3.4. Exogenous delivery of calcium promotes VIC mineralization in PEG + Col IPN hydrogel matrices

Previous studies have shown that the addition of CaCl₂ stimulates calcification of vascular smooth muscle cells and mesenchymal stem cells, therefore, we sought to further promote VIC mineralization and expression of mature osteogenic markers by exogenous addition of calcium chloride (CaCl₂) to the culture media [52,53]. VICs were encapsulated within the PEG + Col hydrogel system and exposed to OM + 1 mg/mL CaCl₂, henceforth referred to as calcifying medium (CM), for up to 12 days (Figure 4a). At that time, VIC-laden hydrogels became visibly opaque relative to the control without cells, suggesting high extents of calcification (Figure 4b).

Figure 4. Calcium chloride addition to osteogenic medium (OM) leads to significant matrix mineralization within PEG + Col hydrogels.

a) Schematic representation of 3D PEG + Col IPN hydrogels and 12-day VIC osteogenic medium (OM) + 1 mg/mL CaCl₂ (calcifying medium, CM) experiments. b) Representative images of no cell control and VICs in CM PEG + Col hydrogels at day 12. c) Representative images of black regions of calcification via Von Kossa (VK) staining of sectioned PEG + Col hydrogels and d) integrated VK density from VICs in OM and CM at day 12 (n=3 biological replicates, minimum 12 images averaged per condition). Scale bar = 100 μm. e) Opacity analysis (600nm) over time from encapsulated VICs in OM and CM (n= minimum 3 biological replicates), each sample normalized to its own day 1 value. f) Calcium levels deposited in PEG + Col hydrogels by VICs in OM and CM at day 12 (n= 3 biological replicates with 2 technical replicates). g) Representative images of immunostained VICs in sectioned PEG + Col hydrogels in OM or CM for osteoblast-like and fibroblast markers at day 12. Red, RUNX2 (runt-related transcription factor 2); yellow, OCN (osteocalcin); green, VIM (vimentin); blue, nuclei. Scale bar = 50 μm. h) RUNX2 nuclear to perinuclear fluorescence intensity analysis at day 12 (n= minimum 3 biological replicates, minimum 6 images per replicate) and i) gene expression relative to RPL30 over time of VICs in OM and CM (n= 3 biological and 3 technical replicates). j) OCN mean fluorescence intensity analysis at day 12 (n= minimum 3 biological replicates, minimum 6 images per replicate) and k) gene expression relative to RPL30 over time of VICs in OM and CM (n= 3 biological and 3 technical replicates). l) VIM mean fluorescence intensity analysis of VICs in OM and CM at day 12 (n= minimum 3 biological replicates, minimum 6 images per replicate). Data in (d,h-l) normalized to corresponding GM condition. Results analyzed using two-way ANOVA with Sidak’s multiple comparison (e), Dunnett’s multiple comparison test (i,k), or an unpaired t-test. *p<0.05, **p<0.01, ****p<0.0001.; * denotes significance between treatment conditions, while + denotes significance of CM with respect to day 3 expression levels.

To further quantify these differences, VIC-laden PEG + Col hydrogels cultured in CM were sectioned and stained with Von Kossa after 12 days (Figure 4c). While there was no positive black staining in the OM condition, the CM condition led to high amounts of positive black mineral deposition within the sample. Quantification of Von Kossa staining (Figure 4d) demonstrated a significant ~33-fold increase in overall density of the positive calcium phosphate staining in the CM condition relative to the OM condition. The opacity (600 nm) of VIC-laden hydrogels cultured in CM increased linearly with time and was significantly different from OM by day 7 (Figure 4e). Total calcium in hydrogels at day 12 was detected using a colorimetric calcium assay and corroborated the trends observed in optical density (Figure 4f), with a ~250-fold increase for hydrogels cultured in CM relative to OM. A no cell control proved that the opacification, increase in calcium deposition, and positive Von Kossa staining was cell-mediated and did not occur spontaneously in CM conditions (Supplemental Figure 3a,b,c, respectively). Optical density and calcium content were also assessed for PEG vs. PEG + Col hydrogels in CM at day 12, and confirmed that the addition of the IPN resulted in increased hydrogel mineralization (Supplemental Figure 3d,e). Taken together, the calcium assay and Von Kossa staining results confirm that higher optical density indicates increased VIC-mediated matrix calcification.

We next sought to determine whether the observed calcification may be partially apoptosis mediated. We explored this hypothesis using the PEG + Col network, given the increased mineralization observed relative to the original PEG hydrogel formulation. VIC Caspase 3/7 activity, an early marker of apoptosis, was assessed in OM or CM at day 12. CM conditions showed a significant 27.7-fold increase in caspase fluorescence compared to OM conditions (Supplemental Figure 3f). However, no significant changes in cell number were detected between PEG + Col gels treated with CM relative to OM (Supplemental Figure 3g). Together, these results suggest that the increased matrix calcification in these hydrogels is not necessarily apoptosis mediated.

VIC phenotype markers RUNX2, OCN, and VIM were assessed for VICs cultured within the PEG + Col system treated with CM and compared to OM condition (Figure 4g). Nuclear localization of RUNX2 (Figure 4h) was decreased for samples in CM at day 12; however, there was no significant differences in RUNX2 gene expression between OM and CM conditions when measured at day 3, 7, or 12 (Figure 4i). OCN mean fluorescence intensity was significantly increased for VICs cultured in CM relative to OM at day 12 (Figure 4j). Additionally, OCN gene expression was significantly increased for VICs in CM relative to OM at both 7 and 12 days (Figure 4k), and was also significantly increased with respect to levels at day 3 for both day 7 and 12 in CM. Moreover, VIM mean intensity decreased in the CM conditions compared to the OM conditions (Figure 4l), indicating a phenotypic shift away from the fibroblast phenotype. These results suggest that encapsulated VICs in CM have markers of mature osteoblasts, not only increased OCN gene and protein fluorescence intensity, but increased mineral deposition detected by histological staining, optical density measurements, and the colorimetric calcium assay.

3.5. Using the degradable PEG + Col model to screen for drug therapies to inhibit matrix calcification

Given the observation that our system readily enables quantifiable levels of calcification, we next sought to explore whether our platform could be used to screen potential pharmacological inhibitors of calcification. Osteoprotegerin (OPG), rosmarinic acid (RA) and zoledronic acid (ZA) were screened as calcification inhibitors and concentrations used in this study were selected based on efficacious values found in the literature. Specifically, OPG was used in concentrations from 1- 100 ng/mL, as the high end of this range has been proven to be non-cytotoxic and effective in in vitro studies to inhibit calcification, and the lower range is within the realm of concentrations detected circulating in vivo in humans (~3 ng/mL) [54-56]. RA reduced calcific nodules in vitro on vascular smooth muscle cells with treatment concentrations ranging from 40 - 80 μM, with concentrations as high as 270 μM for studying anti-fibrotic effects on liver slices [57,58]. Concentrations of ZA have been used in vitro ranging from 1- 50 μM, demonstrating no loss in viability, and inhibition of mineralization of calvarial osteoblasts at treatments of 50 μM [59]. Optical density (600 nm) was used to determine the efficacy of these therapeutics in reducing or preventing VICs from calcifying their microenvironments. VICs were encapsulated in the PEG + Col hydrogel matrix and cultured in CM while treated with the inhibitors, and the results were analyzed at day 12 (Figure 5a).

Figure 5. Cell-mediated 3D PEG + Col hydrogel mineralization is useful as a tool to screen anti-calcification therapeutics.

a) Schematic representation of 3D PEG + Col IPN hydrogels and 12-day anti-calcification drug screening experiments. b) Normalized opacity (600nm) from VICs in CM with indicated concentration of calcification inhibitor (zoledronic acid (ZA), rosmarinic acid (RA), or osteoprotegerin, (OPG)) (n = 3 biological and 2 technical replicates) and c) normalized calcium levels deposited by VICs in PEG + Col hydrogels in CM and indicated calcification inhibitor. Normalized to positive control calcifying condition of PEG + Col encapsulated VICs in CM (dashed line) (n= 3 biological replicates). d) Representative images of black regions of calcification via Von Kossa (VK) staining of sectioned PEG + Col hydrogels in CM and lack of calcification in CM + 50 mM ZA. Scale bar =100 μm. e) Representative images of immunostained VICs in sectioned PEG + Col hydrogels in CM or CM + 50 mM ZA for osteoblast-like and fibroblast markers. Red, RUNX2 (runt-related transcription factor 2); yellow, OCN (osteocalcin); green, VIM (vimentin); blue, nuclei. Scale bar = 50 μm. f) RUNX2 nuclear to perinuclear fluorescence intensity analysis, g) OCN and h) VIM mean fluorescence intensity analysis of VICs in CM and CM + ZA (n= minimum of 3 biological replicates, minimum 6 images per replicate). (f-h) Normalized to VICs in PEG + Col in OM (dashed line). Results analyzed using one-way ANOVA with Dunnett’s or Tukey’s multiple comparison test, or an unpaired t-test. *p<0.05, **p<0.01, ***p< 0.001, ****p<0.0001.; * denotes significance between treatment conditions (CM and CM + ZA), while + denotes significance of with respect to OM control (dashed line).

While both OPG and RA resulted in no significant change in opacity measurements relative to the control, ZA reduced opacity readings for all treatment concentrations (Figure 5b). Measurements at conditions with 0.5, 5, and 50 μM ZA resulted in roughly 27%, 39%, and 70% reduction in opacity readings, respectively, relative to the control. These results were confirmed using a colorimetric calcium assay, indicating that the lower optical density reading resulted in lower calcification within the hydrogel system, with a roughly 86% reduction in calcium content observed for the 50 μM ZA treatment condition (Figure 5c). Von Kossa staining further confirmed these results, as there was no positive black staining in the ZA treated samples, relative to the control without inhibitor treatment (Figure 5d).

Samples treated with ZA were cryosectioned and stained for markers of VIC phenotype, including RUNX2, OCN, and VIM (Figure 5e). Quantification of RUNX2 signal revealed neither CM nor CM + ZA had significantly different RUNX2 nuclear localization (Figure 5f). When compared to OM as a control (dashed line), the addition of ZA to CM resulted in significantly increased nuclear RUNX2 localization. Quantification of mean OCN fluorescence signal showed no differences between CM and CM + ZA conditions; however, CM treated samples were significantly higher than OM (dashed line) and this significance was not present in samples treated with ZA (Figure 5g). VIM mean fluorescence signal was significantly higher in samples cultured in CM + ZA relative to CM alone and was also significantly higher than the signal from OM treated samples (dashed line) (Figure 5h). These results indicated that ZA can prevent VIC calcification through modulating the VIC phenotype.

Cell viability of VICs within CM with 50 μM ZA was 82.7 ± 5.7% by day 5 and not significantly different from CM conditions, indicating that the lack of calcification was not due to cell death (Supplemental Figure 4a,b). Furthermore, we performed a concentration curve using ZA on TCPS at 0.5, 5, 50, and 500 μM in GM to confirm efficacy by determining changes in cell number (Supplemental Figure 4c). Taken together, these data suggest that our engineered 3D hydrogel cell culture system can be used as a drug-screening tool for identifying potential therapies for CAVS.

4. Discussion

The role that VICs and the local valve microenvironment play in the eventual calcification of valve leaflet tissue at late stages of AVS remains largely unknown. Thus, an in vitro culture system that mimics aspects of the native valve and promotes cell-mediated calcification is advantageous for elucidating factors to prevent or reverse valve calcification. Here, we explore the role that the microenvironment may play in valve calcification by designing and characterizing a 3D hydrogel system that promotes a VIC osteoblast-like phenotype and VIC-mediated calcification on an accelerated timescale. Our results confirm cell-mediated matrix mineralization and validate the use of this system as a higher throughput platform to screen relevant drugs to inhibit calcification.

In this work, we design and characterize a hydrogel system to promote matrix calcification by conserving a key aspect of the valve microenvironment via the addition of collagen type I, as collagens make up over 50% of the valve [33]. We hypothesized that the addition of collagen would act as a mineral nucleator to promote calcification and late stage osteogenic markers of VICs. To promote further calcification, we supplemented calcium chloride (CaCl₂) into the OM to simulate the high blood calcium observed in patients who are more at risk for developing cardiovascular disease [60,61]. The addition of CaCl₂ has been explored for its effect in stimulating calcification of vascular smooth muscle cells, mesenchymal stem cells, and whole valve tissues and thus, was applied to VICs in these experiments [52,53,62]. The 1 mg/mL CaCl₂ concentration of Ca²⁺ used in these studies is ~3 times higher than blood levels of a hypercalcemic patient [63], but allows for the accelerated calcification of the hydrogel matrix and reasonably timed experiments to better understand how changes in the microenvironment affect mineralization.

Cell-mediated calcification was dramatic in PEG + Col hydrogels cultured in CM, as seen in the bulk opacification and histological staining of the hydrogel with VICs relative to the acellular control within 12 days of culture. We show that opacity change of the hydrogel allows for an easy, higher throughput method to assess overall changes in calcification using a standard plate reader, as increased optical density (600 nm) was corroborated with increased Ca²⁺ content deposited within hydrogels.

While it is debated whether or not VICs become true osteoblasts, but they do express key genetic markers and functional outputs of an osteoblast-like phenotype in response to osteogenic-inducing stimuli [14,64]. We use a range of markers to characterize VIC phenotype, including RUNX2 as an early marker of osteogenesis, OCN as a later marker of osteogenesis, and VIM as a fibroblast marker. RUNX2 has been previously associate with an osteoblast-like phenotype of VICs [30,65,66], but as a transcription factor, nuclear localization is linked with transcriptional activity [67]. OCN expression is a component of bone formation and is linked to osteoblast-like phenotype of VICs [17,24], which as we mentioned previously, is regulated in part by vimentin [21,22]. Therefore, our results of VICs within PEG + Col hydrogels displaying increased gene expression of RUNX2 as well as nuclear localized RUNX2, increased OCN expression, decreased VIM, and increased matrix mineralization in response to exogenous osteogenic stimuli make a compelling case that these VICs are osteoblast-like.

In our studies, we see significant matrix calcification within 12 days, and significant changes in opacity within one week of the PEG + Col hydrogels in CM. These findings reduce culture times and increase markers of mature osteoblasts not only compared to our earlier studies in the absence of CaCl₂ and/or the secondary collagen IPN, but with respect to other findings in the field. Specifically, others using 3D hydrogel matrices to study VIC calcification have reported positive OCN gene expression within 21 days of osteogenic medium [23], positive Alizarin Red staining within 21 days [30], and positive Alizarin Red and gene expression for RUNX2 and OCN within 14 days [68]. Considering our 12-day timepoint to robust matrix calcification, we reduced time in culture by ~14% (or 50% when considering the day 7 opacity change) relative to these examples.

We hypothesized our 3D hydrogel culture platform would enable in vitro screening of potential therapeutics to assess their ability to inhibit calcification. To that end, we employed the PEG + Col hydrogel system to screen three potential inhibitors of valve calcification, using opacity as a readout followed by validation using a colorimetric calcium assay, histology, and immunofluorescent imaging. Specifically, osteoprotegerin (OPG), rosmarinic acid (RA), and zoledronic acid (ZA), previously investigated as anti-calcification therapeutics, were screened. OPG inhibits RANK:RANKL signaling, and therefore prevents bone demineralization and valvular calcification [69,70]. RA, an antioxidant compound found in medicinal plants has been shown reduce calcification of vascular smooth muscle cells in vitro through β-catenin and NF-κβ inhibition [57,71]. ZA belongs to a class of drugs called bisphosphonates that are used to treat osteoporosis and vascular calcification. Bisphosphonates are a synthetic analogue of naturally occurring pyrophosphate and that have a high affinity for calcium and inhibit osteoclast activity, preventing bone resorption [71,72].

Our results indicated that while OPG and RA had no significant effects on inhibiting mineralization of the 3D hydrogels in this study, ZA treatment was successful in reducing opacity and calcium content. Cell viability for treated samples remained >80%, indicating that ZA was having a biological effect rather than the decrease in opacity being attributed to treatment cytotoxicity. Interestingly, CM + ZA treatment reduced mean signal intensity of mature marker OCN yet increased VIM signal, suggesting that the ZA therapeutic promoted a fibroblast phenotype of the encapsulated VICs. ZA is reported to inhibit signaling pathways related to the prenylation of small signaling proteins such as Rho, Ras, and Rac [73], the inhibition of which has been shown in other studies to reduce or delay osteoblast differentiation and mineralization [23,74].

The design and characterization of 3D hydrogel culture systems as in vitro platforms to explore the role of the VIC microenvironment is valuable, but not without limitations. Here, we designed a culture platform with a singular cell type and ECM protein, while the native valve tissue is a complex composition of cells, including VECs, other ECM components, and turbulent blood flow. However, a simplified, experimental design allowed us to isolate specific cellular responses in a controlled environment by incorporating variables of interest, collagen in this case, into a synthetic network in the presence of a single cell type, VICs.

This materials-based system for 3D VIC culture designed for matrix mineralization and VIC osteoblast-like phenotype can be easily adapted for rapid screening of drug therapeutics using methods previously developed by our lab [75]. To the best of our knowledge, this is the first study of its kind using a combination of synthetic and natural polymers to create a 3D in vitro system to study robust VIC matrix calcification and osteogenic response without the use of mechanical stress or culture times that extend for >2 weeks.

5. Conclusion

In this work, we designed and characterized a 3D hydrogel system in order to explore the importance of the VIC microenvironment in promoting calcification. Soft, synthetic PEG hydrogels were modified with a collagen type I IPN as an in vitro platform to study how altering the matrix microenvironment affects functional mineralization outputs. The incorporation of the collagen IPN into the PEG hydrogel allows for the user to include whole proteins present in the valve microenvironment that nucleate mineralization deposition, but also maintain the tunable flexibility of the cellularly degradable PEG hydrogel. The addition of collagen into a 3D PEG hydrogel treated with osteogenic cues resulted in robust matrix calcification in under 2 weeks. The calcification can be characterized via high throughput optical density screens using a standard plate reader, rendering the technique a time and resource saving tool to screen for calcification in 3D cultures. To this end, we used this 3D hydrogel platform to screen for drug inhibition of VIC-mediated calcification and showed that a bisphosphonate, zoledronic acid, reduced calcification of the 3D hydrogel matrix.

Supplementary Material

1

Supplemental Figure 1. Projected surface area (SA) contraction over time for VICs in GM and OM with increasing dexamethasone dose, shown as percentage of day 1 measurement (n= 3 biological replicates through day 7 timepoint, n= minimum 2 biological replicates for day 10 and 13, n= minimum 1 biological replicate for day 15; 2 technical replicates at each measurement). Samples in OM conditions with high (1 mM) and low (0.1 μM) concentrations of dexamethasone (dex) contract significantly with respect to GM (p<0.001) and with respect to OM (0 dex) (p<0.01). Contraction between high and low dex was not significantly different. Results analyzed using one-way ANOVA with Sidak’s multiple comparison test at day 15 timepoint.

Supplemental Figure 2. a) Shear storage modulus (G’, Pa) for PEG and PEG + Col hydrogels (n = 3 technical replicates). b) Projected surface area (SA) contraction over time for VICs encapsulated in PEG + Col hydrogels in GM and OM (1mM dex), shown as percentage of day 1 measurement (n= 3 biological replicates through day 7 timepoint, n= minimum 2 biological replicates for day 10 and 13, n= minimum 1 biological replicate for day 15; 2 technical replicates at each measurement). c) Normalized projected surface area at day 15 for VICs within PEG and PEG + Col hydrogels in GM and OM conditions with respect to day 1 measurement (n= minimum 1 biological and 2 technical replicates). d) Number of cells per field of view for DAPI stained VICs within sectioned PEG and PEG + Col hydrogels in GM and OM conditions at day 15 (n= 3 biological replicates, 15 fields of view analyzed per condition). e) RUNX2 gene expression relative to RPL30 of VICs within PEG and PEG + Col in OM at day 15, normalized to GM condition for each hydrogel respectively (dashed line). f) Integrated Von Kossa density analysis for VICs within sectioned and VK stained PEG and PEG + Col hydrogels at day 15 (n= 3 biological replicates and a minimum of 9 images used per condition) and g) opacity analysis (600nm) for VICs within PEG and PEG + Col in GM and OM at day 15 (n= 3 biological replicates). (g,h) normalized to respective PEG GM condition. Results analyzed using a one or two-way ANOVA with Dunnett’s or Sidak’s multiple comparison test, or an unpaired t-test, when appropriate. *p<0.05, **p<0.01, ***p< 0.001, ****p<0.0001.

Supplemental Figure 3. a) Normalized opacity analysis (600nm) over time for VICs within PEG + Col in calcifying medium (CM) and no cell PEG + Col controls with respect to day 1 measurements (n= 3 biological technical replicates). b) Calcium levels deposited by VICs and no cell controls within PEG + Col in osteogenic medium (OM) and CM at day 15 (n=3 biological replicates). c) Representative image of lack of calcification of sectioned acellular PEG + Col hydrogel at day 12 in CM via Von Kossa (VK) staining. Scale bar = 100 mm. d) Normalized opacity analysis (600nm) and e) normalized calcium levels deposited within indicated culture condition at day 12 for VICs within PEG hydrogels ± Col IPN in OM and the presence (+) or absence (−) of CaCl₂ (n= 3 biological replicates). Values normalized to PEG + Col in OM. f) Caspase 3/7 mean relative fluorescence units (RFU) relative to cell number of VICs within PEG + Col in OM and CM at day 12 normalized to OM (n= 3 biological replicates). g) Number of cells per field of view for VICs within PEG + Col in OM and CM at day 12 (n= 3 biological replicates, 8 fields of view analyzed per condition). Results analyzed using one-way ANOVA with Dunnett’s multiple comparison test, or an unpaired t-test. *p<0.05, ****p<0.0001

Supplemental Figure 4. a) Percentage of live cells for VICs within PEG + Col hydrogels at day 5 in OM or CM ± 50 μM ZA (n= 3 biological replicates, average % alive cells calculated from 12 fields of view). b) Representative images of VICs within PEG + Col in OM, CM and CM + ZA for live/dead stain. Green, live cells (calcein); orange, dead cells (ethidium homodimer). Scale bar = 100 μm. c) Number of VICs on TCPS in growth medium (GM) supplemented with increasing zoledronic acid (ZA) dose (n = minimum of 7 biological replicates, each replicate is average number of cells/well as averaged over 6 fields of view). Results analyzed using one-way ANOVA with Dunnett’s multiple comparison test. * p<0.05, **p<0.01, ****p<0.0001.

Statement of significance.

The addition of a secondary collagen network to a synthetic PEG hydrogel network supported matrix mineralization by 3D encapsulated valvular interstitial cells (VICs) when cultured with osteogenic cues. Matrix mineralization of the PEG + Col hydrogel system, as identified with increased optical density, gene expression for late markers of osteogenesis, and histology, occurred in under 12 days. The use of this network drastically reduces culture time needed to study VIC calcification in a 3D hydrogel network. Using this culture platform that mimics aspects of the native valve and optical density as a higher throughput output for matrix mineralization, anti-calcification drug therapeutics were screened. Matrix mineralization reduced by 86% with the treatment of a bisphosphonate, zoledronic acid.

Data Availability

The raw data presented in this work are available from the corresponding author upon reasonable request.