Collagenase (from Clostridium histolyticum), transforming growth factor β (TGF-β), sodium dihydrogenphosphate (NaH₂PO₄), L-ascorbic acid, hexadecylpyridinumchloride, propidiumiodide, FITC-dextran (40 kDa) and JC-10 mitochondrial staining kit from Merck (Darmstadt, Germany), Alizarin Red from Cell Systems (Troisdorf, Germany), OsteoSense 680 EX from Perkin Elmer (Waltham, USA) and Calcein-AM from Thermo Fisher (Waltham, USA) were used for culture media, sample preparation and staining.

Human VICs were obtained from 10 patients undergoing surgical AV replacement. The study was approved by the local Ethics Committee (Medical Faculty, University Leipzig, registration number 128/19-ek) and all patients gave written informed consent in accordance with the Declaration of Helsinki. AVs were minced and enzymatically digested in a serum-free collagenase (125 U/ml) solution under steady rotation at 15 rpm at 37°C for a total of four hours. Cells were passed through a 40 µm cell strainer and pelleted by centrifugation. VICs were cultured in high glucose DMEM (Thermo Fisher, #61965059) supplemented with 10% FBS (PAN Biotech, Aidenbach, Germany) and 1% penicillin/streptomycin (Thermo Fisher, Waltham, USA), further termed as growth medium (GM). Cells were cultured in a humidified atmosphere at 37°C and 5% CO₂ and passaged at 90% confluency. Experiments were performed in passages 3 to 4 at 100% confluence in GM. Control cultures were grown in high glucose DMEM (+5% FBS + 1% penicillin/streptomycin) referred to as control medium (CM). Calcification was induced using PM consisting of high glucose DMEM (+5% FBS +1% penicillin/streptomycin) supplemented with 2 mM NaH₂PO₄ (pH 7.4) and 50 µg/ml L-ascorbic acid. Fibrosis was induced using FM containing high glucose DMEM (+5% FBS + 1% penicillin/streptomycin) supplemented with 0.5 ng/ml TGF-β.

Impedance spectroscopy measurements were performed using an ACEA xCELLigence® Real-time Cell Analyzer (RTCA) DP instrument (ACEA Bioscience, San Diego, USA). The RTCA system holds 3 “E-Plate 16” electrode arrays and a computer-based control unit. The E-Plates exhibit gold interdigital electrodes covering 80% surface of each well. 7000 cells were seeded per well and immediately continuously recorded for impedance changes. Values of the cell-free electrodes were set as blank. Rising impedance values indicated increasing electrode coverage, first via cellular adhesion and secondly due to cell proliferation. Constant impedance values represented confluent cellular monolayers. At this stage, CM, FM or PM treatment for a total of 20 days was started. Experimental cellular impedance (CI) values were recorded every 60 min for 20 days. Data were normalized to the last value that was measured before the experiment (t = 0). Growth and experimental media were replaced every 2 to 3 days.

Calcium-containing cellular mineral depositions were quantified using Alizarin Red staining at day 0 and then every other day until day 20 and finally on day 21 in CM, FM or PM. VICs were washed with PBS, fixed with 10% formaldehyde and stained with 2% Alizarin Red for 15 min. The cells were then washed with deionized water and air-dried. Afterwards, hexadecylpyridinumchloride (100 mM) was added and incubated in the dark at room temperature for 3 h. The resulting solution was diluted 10-fold and optical density was measured at 540 nm using a microplate reader (Tecan, Männedorf, Switzerland). Data was normalized to t = 0.

For visualization of VIC-driven hydroxyapatite (HA) deposition, a near-infrared, bisphosphonate-based molecular probe (OsteoSense 680 EX) was used. Therefore, 2 nmol OsteoSense 680 EX solution was added to the cell cultures for 24 h and subsequently visualized at 680 nm using a fluorescence microscope (Keyence, BZ-X800). The OsteoSense-stained area was quantified using ImageJ (31).

Pro-collagen type 1 alpha 1 (pro-COL1A1) secretion was measured in cell culture supernatants using ELISA according to the manufacturer's recommendations (R&D Systems, Minneapolis, USA, #DY6220-05).

Calcein-AM (2 µM) and propidiumiodide (PI) (3 µM) were added to the VIC cultures at the indicated time points to quantify viable and dead cells. Cells were incubated at 37°C and 5% CO₂ for 15 min. Viable and dead cells in CM or PM were visualized at 525 nm and 620 nm, respectively, using a microplate reader (Tecan, Männedorf, Switzerland).

As a hallmark of cell death, mitochondrial transmembrane potential Δψm was determined in PM or CM treated VICs at 0, 3, 7 and 12 days using the “Mitochondrial Membrane Potential Kit” (Merck, Darmstadt, Germany) following the manufactureŕs instructions. A ratio of red-to-green fluorescence was determined expressing the amount of dead cells in relation to viable cells.

Permeability of the VIC monolayer due to CM or PM treatment was determined according to Bischoff et al. (32) at indicated time points (t = 0, 3, 7, 12, 15, 18, 21 days). In brief, VICs were grown on the upper compartment of Transwell inserts (12 mm diameter, 0.4 µm pore size; Greiner BioOne, Kremsmünster, Austria) to confluency. To measure macromolecular transmission through the cell monolayer, 1 mg/ml FITC-Dextran in CM was added to the upper compartment and incubated for 30 min. Then, 50 µl samples from the lower compartment were subjected to fluorescence measurement at 525 nm. FITC-Dextran standard curve ranging from 1000 µg/ml to 0.001 µg/ml in CM was used for quantification.

Gene expression of ACTA2 in CM- or PM-treated VIC cultures at day 0, 7 and 21 was analyzed using qRT-PCR. Takyon NoRox Sybr Mastermix Blue (Eurogentec, Lüttich, Belgium) on a BioRad CFX system (BioRad, Hercules, USA) was used. Exon-spanning primers were designed with an annealing temperature of 60°C (5´-3´sequences for forward: CGTTACTACTGCTGAGCGTG and reverse: CGATGAAGGATGGCTGGAAC). Standard curves were used to calculate copy numbers and reaction efficiency. Measurements were done in triplicates and specimen with standard deviation > 0.3 were excluded. Hypoxanthine phosphoribosyltransferase 1 (HPRT1, (5´-3´sequences for forward: CTCATGGACTGATTATGGACAGGAC and reverse: GCAGGTCAGCAAAGAACTTATAGCC) as a stable housekeeping gene was used for expression normalization of the target gene.

Actin cytoskeleton alterations were analyzed using Phalloidin-iFluor 594 reagent (Abcam, Cambridge, UK) at day 0, 7, 14 and 21. Cells were fixed and F-actin fibers were stained with Phalloidin for 60 min at room temperature. Nuclei were stained with 10 µg/ml Hoechst 33342 (Invitrogen, Waltham, USA) for 10 min at room temperature. Visualization was done using a Keyence BZ-X800 microscope.

Statistical analyses were performed using GraphPadPrism 6 (San Diego, USA). Data were checked for normal distribution using Shapiro-Wilks test and are presented as mean ± standard error of the mean (SEM). “n” accounts for the number of biological replicates (derived from different donor valves) in the experiments and is declared for each dataset. ELISA and qRT-PCR raw data were median-normalized, when significant plate-to-plate median differences indicated batch effects.

Group differences were analyzed using one-way analysis of variance (ANOVA) or Student´s t-test. Time traces were assessed by two-way repeated measures ANOVA. Bonferroni correction was used as post-hoc test. All p-values <0.05 were considered statistically significant. Pearson's correlation coefficient (r) was calculated to test for associations between data of 2 different techniques.

Non-invasive EIS was used to screen PM- and FM-induced cellular responses in VICs (Figure 1A). The EIS signal of VICs in CM was stable for 14 days and then declined to 67.9 ± 5.8% at day 20. In VICs treated with FM, the EIS signal declined slowly from day 5 down to 53.0 ± 3.6% at day 20. Interestingly, in PM, VICs showed a biphasic EIS time profile, which differed significantly from EIS time traces of CM- or FM-treated VICs. PM instantly reduced the cellular impedance signal to a minimum at 17.1 ± 3.3% at day 8 (p = 0.039, “phase 1”). During the following 12 days, termed “phase 2”, the signal increased significantly up to a maximum of 159.3 ± 40.1% on day 20 (p < 0.001 vs. minimal EIS signal, p = 0.14 vs. day 0).

PM in a cell-free environment also had an impact on EIS signals showing gradually increasing signals (see Supplementary Figure S1).

CM- and FM-treated VIC cultures showed comparable collagen type 1 secretion and comparable, negligible Alizarin Red absorption (Figures 1B,C). PM treatment significantly reduced the collagen type 1 secretion of VICs 3.3-fold after 14 days (p = 0.035) compared to CM treatment. PM VICs had significantly reduced collagen type 1 secretion of 3.2- to 8.6-fold compared to FM treated VICs (p < 0.05) during the complete observation period and a 3- to 4-fold reduction compared to CM treated VICs (p < 0.05) at day 14 and 21 (Figure 1B). Compared to FM- and CM-treated VICs, PM stimulated massive extracellular deposition of calcium-containing mineral material, expressed by 4.5- to 8-fold increased Alizarin Red absorption during the complete observation period (p < 0.001, Figure 1C). Pearson correlation coefficients (r) were calculated to detect potential associations between the PM-specific EIS signals and the VIC-driven collagen secretion and mineral deposition (Figures 1D,E). As EIS signal was biphasic, r was calculated for phase 1 (day 0–8) and phase 2 (day 12–21) separately. A very high negative association was found for the increasing Alizarin Red absorption and the decreasing EIS signal with r = −0.95 (p = 0.015) in phase 1. Additionally, the EIS signal in phase 1 was moderately correlated to the decreasing collagen type 1 secretion with r = −0.67 (p = 0.22). A very high, positive correlation of the EIS signal in phase 2 with the increasing Alizarin Red absorption with r = 0.97 (p = 0.008) was found.

Additional staining of PM-stimulated hydroxyapatite (HA) deposition by VICs using a near-infrared molecular probe confirmed the Alizarin Red results. However a less intense staining was evident due to specificity to HA (Figures 1F,G). Quantification of the OsteoSense-positive area revealed higher HA-deposition of VICs in PM compared to CM.

The impact of experimental media on cellular alterations, viability and cytotoxicity, and therewith also on the EIS signal, was investigated. Visualization of viable (Calcein-AM⁺, green) and dead VICs (PI⁺, red), respectively, illustrated specific reactions to CM or PM (Figure 2A). VICs in CM maintained round, mostly Calcein-AM⁺ cell clusters during the observation period. Starting from day 4, dead VICs next to and inside the cell clusters with increasing frequency were seen up to 21 days. In contrast, PM-treated VICs formed viable, Calcein-AM⁺ cell clusters only until day 4, when the amount of dead cells (PI⁺) started to increase. From day 6 on, cluster formation was disrupted resulting in a homogeneous distribution of viable and dead VICs. Importantly, the PM Calcein-AM fluorescence signal, indicative for viability, was significantly reduced by 38.3 ± 7.3% from day 13 on up to 52.1 ± 3.9% (p < 0.001 vs. CM) at day 21 (Figure 2B). Additionally, the mitochondrial transmembrane potential Δψm of PM-treated VICs revealed significant mitochondrial hyperpolarization up to 48.1 ± 20.8% (p = 0.038) at day 3 and 92.3 ± 25.4% (p = 0.001) at day 12 compared to CM-treated VICs (Figure 2C). Macromolecular permeability, as a measure of monolayer disruption, of PM-treated VIC monolayers showed a biphasic time trace (Figure 2D). In PM-treated VIC monolayers macromolecular permeability was significantly enhanced up to 1.4-fold at day 7 (p = 0.001 vs. CM) and decreased constantly reaching its minimum of a 1.9-fold reduction at day 18 (p < 0.001 vs. CM). From day 6 on, VICs in PM had a spindle-like morphology with long and thin cell bodies, whereas VICs in CM exhibited a flattened, rounded morphology (Figure 3A). Based on this observation of morphological alterations, the relative amount of activated VICs in CM and PM was evaluated by analyzing ACTA2 (alpha smooth muscle actin) gene expression as a marker of myofibroblastic activation (Figure 3B). ACTA2 expression in CM was 6-fold higher at day 7 and 2-fold higher at day 21 compared to PM (both p < 0.001). Phalloidin staining of F-actin fibers confirmed an abundance of stress fibers of VICs in FM and a lack of stress fiber assembly in PM, which is in line with ACTA2 gene expression (Figure 3C).

EIS was used to analyze donor- and sex-specific impedance patterns (Figure 4 and Supplementary Figure S2). First, EIS was used to determine the doubling time as a measure of proliferation potential of VICs in GM. On average, female VICs were found to exhibit a 13 ± 4 h longer doubling time than male VICs (p = 0.017). In general, the PM-specific, biphasic EIS time profile was observed for all patientś VICs (Figure 4A and Supplementary Figure S3). However, male VICs entered phase 2 significantly earlier at day 8.7 ± 0.4, while female VICs ended phase 1 not until day 11.2 ± 1.0 (p = 0.03). In addition, in male VICs the decrease of cellular impedance in phase 1 was significantly more pronounced (minimum: 7.4 ± 4.2%) than in female VICs (minimum: 26.5 ± 4.4%, p = 0.003) (Figure 4B). A high donor-specific heterogeneity of EIS phase 1 and 2 characteristics was observed. Seven out of 10 donors showed moderate phase 2 enhancement (+74% to +107% total growth) (Figures 4B,C). One female donor showed lowest phase 2 enhancement (+2.1 ± 0.9%), another female donor showed strongest phase 2 enhancement with a total increase of +914.4 ± 66.7%. In contrast, the strongest male phase 2 enhancement was +349.4 ± 32.7%. Averaged phase 2 EIS time traces of male and female VICs PM were comparable with an increase of +156.1 ± 20.0% and +226.2 ± 61.2% (p = 0.80), respectively.

Collagen type 1 secretion (Figure 4D) was comparable from male and female VICs in CM, FM and PM (p > 0.05). Alizarin Red staining showed comparable calcium-containing mineral deposition of male and female VICs (p > 0.05).

The observed PM-specific biphasic EIS signal with an initial drop followed by a secondary increase indicates differential electrode coverage during the cultivation period (32). A moderate positive Pearson correlation coefficient between the phase 1 EIS decrease and the declining collagen 1 secretion suggests that suppression of collagen production contributed to the initial EIS decrease. Whether the decline in collagen secretion is due to a direct effect of PM on collagen secretion or secondary to a loss of collagen-producing cells remains to be elucidated. In contrast, in FM EIS as well as collagen secretion were constant over time. Thus, EIS is of limited utility to monitor profibrotic ECM modulation in vitro. This is supported by a study of Fuentes-Velez et al. who demonstrated that fibrotic remodeling in hepatic tissue cultures is not associated with EIS alterations (33). But, synergistic effects including the modulation of collagen secretion and beginning ECM mineralization accompanied by major cell loss may contribute to the Phase 1 EIS signal.

Phase 2 EIS increase in PM was correlated with VIC-driven ECM mineralization as confirmed by HA-specific near-infrared molecular probe staining and Alizarin Red staining of calcium-containing mineral deposits (34). It was previously demonstrated that EIS signals increase when HA is formed and decrease when HA degrades (35). HA, carbonated HA and other calcium-, phosphate- and carbonate- containing HA are common organic compounds in AVs from patients with severe AS (36).

In summary, we conclude that the PM-specific EIS phase 2 can be assigned to ECM mineralization. In addition, PM alone showed a slow gradually increasing EIS time trace due to low, passive calcific matter formation on the electrode surface. This is a first hint towards an active acceleration of a passive calcification process by VICs. Since PM in a cell-free environment showed slight passive calcific precipitation, the biphasic EIS time trace only occurred in the presence of VICs and induced ECM calcification in a significantly more pronounced fashion.

Collagen deposition in vivo is well-documented to prepare the ground for ECM mineralization of the AV by providing a supportive structure for mineral material aggregation (37, 38). Importantly, PM simultaneously decreased the collagen secretion in favor of ECM mineralization comprehending FCAVD pathobiology in vivo (5, 39).

As mentioned above, the moderate correlation between EIS and suppression of collagen secretion might not solely explain the initial phase 1 EIS drop. Hence, intracellular responses of VICs to PM treatment were analyzed as well. Phase 1 EIS was characterized by a decrease of viable VICs due to mitochondrial hyperpolarization and apoptosis-mediated cell loss, which was accompanied by a high macromolecular permeability illustrating the disruption of the VIC monolayer integrity. Hyperpolarization of the mitochondrial membrane potential is a hallmark of apoptotic cell death and can be induced by high extracellular inorganic phosphate (40) as present in PM. High extracellular inorganic phosphate can alkalinize the cytosol (40), which in turn forces mitochondrial uptake of inorganic phosphate, resulting in enhanced mitochondrial superoxide generation (40–42). Mitochondrial superoxide represents the largest intracellular reservoir of reactive oxygen species (40). Thus, PM may induce oxidative stress-induced VIC damage. In vivo, high serum phosphate levels are known to induce AV calcification by stimulating complex intracellular signaling towards apoptosis and osteo-chondrogenic gene expression (43, 44). Oxidative stress and apoptosis thereby serve as initiators of calcific nodule formation in FCAVD (5, 45, 46). Conclusively, a phosphate-induced alkalization of the VICs cytosol with subsequent oxidative stress mediated apoptosis might be an underlying pathomechanism in patients with hyperphosphatemia, contributing to the initiation of ECM mineralization and calcific nodule formation. In vitro, VIC apoptosis has been described for osteogenic media containing β-glycerophosphate (12, 47).

Besides cell loss, media-dependent differences in cellular morphology were observed in our study. Whereas VICs in CM displayed heterogeneous cellular phenotypes, VICs in PM exhibited predominantly uniform spindle-like, thin, long cell bodies. In conjunction with the significant decrease of ACTA2 gene expression and abolished stress fiber assembly, one can hypothesize that PM-stimulated blocking of myofibroblastogenesis has occurred.

In summary, decline of myofibroblastogenesis, suppression of collagen synthesis, apoptosis and ECM mineralization create the PM-specific, biphasic EIS signal observed in vitro (Figure 5). In line with these results, the phenotype-specific cells that survived the apoptosis induce ECM mineralization. These processes have been described in and around calcific nodules in the AV leaflets in vivo (39). Therefore, VICs cultured in PM represent a suitable in vitro model of FCAVD, mimicking in vivo-like features in a reasonable time period.

EIS is preferable to standard methods, since it records changes in cellular electrical properties in a non-invasive environment and thereby allows real-time data acquisition of intra- and extracellular alterations due to external stimuli (49). The versatility of EIS was further demonstrated in our study when assessing sex- and donor- specific responses to PM. EIS demonstrated higher proliferative capacity of male VICs (see Supplementary Figure S2) with higher initial sensitivity to PM, indicating donor sex dependency. Additionally, EIS revealed a strong inter-individual heterogeneity of ECM mineralization in vitro, wherein strongest variability occurred in female VICs (see Supplementary Table S1). Further studies are required to determine whether this is a general characteristic and which hormonal status or comorbidities may underlie this observation. Finally, EIS analyses unveiled that individual VIC calcification potential and velocity in vitro are strongly donor specific.