Tumor necrosis factor alpha and interleukin 1 beta suppress myofibroblast activation via nuclear factor kappa B signaling in 3D-cultured mitral valve interstitial cells

Abstract

Mitral valve disease is a major cause of cardiovascular morbidity throughout the world. Many different mitral valve pathologies feature fibrotic remodeling, often accompanied by an inflammatory state. Mitral valve fibrosis is mediated by valvular interstitial cells (VICs), which reside in the valve leaflets and often differentiate into myofibroblast-like cells during disease conditions. In this study, we investigated the effects of tumor necrosis factor alpha (TNF-α) and interleukin 1 beta (IL-1β) on mitral VICs, since these pro-inflammatory cytokines have been shown to exert pleiotropic effects on various cell types in other fibrotic disorders. Using biomimetic three-dimensional culture systems, we demonstrated that TNF-α and IL-1β suppress myofibroblast differentiation in mitral VICs, as evidenced by gene and protein expression of alpha smooth muscle actin and smooth muscle 22 alpha. Addition of TNF-α and IL-1β also inhibited mitral VIC-mediated contraction of collagen gels. Furthermore, inhibition of NF-κB, which is downstream of TNF-α and IL-1β, reversed these effects. These results reveal targetable pathways for potential development of pharmaceutical treatments for alleviating fibrosis during mitral valve disease.

1. Introduction

Mitral valve disease affects over 1% of the United States population and significantly contributes to cardiovascular morbidity and mortality, particularly in elderly populations, by disrupting the unidirectional flow of blood from the left atrium to the left ventricle of the heart [1]. Mitral valve disease can be caused by connective tissue disorders, infections, congestive heart failure, or other underlying pathologies, and can impair the valve’s ability to open during diastole (mitral stenosis) or close during systole (mitral regurgitation) [2]. The only corrective treatments for severe valve disease are surgical repair or replacement [3], but these procedures are often poorly tolerated by severely ill patients [4]. Due to incomplete mechanistic understanding, there are no pharmaceutical treatments for mitral valve disease that address the underlying valve dysfunction.

Structurally, the mitral valve consists of two leaflets that are encircled by a fibrous annulus and are attached to the ventricular wall by chordae tendineae. Within the leaflets, fibroblast-like cells known as valvular interstitial cells (VICs) actively maintain the extracellular matrix (ECM) of the valve. Whereas VICs are quiescent in healthy valves, they become “activated” during valve disease, differentiating into myofibroblast-like cells that remodel the valvular ECM. Markers of activated VICs include alpha smooth muscle actin (αSMA) and smooth muscle 22 alpha (SM22α), which are overexpressed in myofibroblasts and smooth muscle cells but not quiescent fibroblasts [5]. Furthermore, activated VICs are more contractile than quiescent VICs, exerting increased traction forces on the ECM via cytoskeletal αSMA stress fibers [6].

Prolonged activation of VICs can result in fibrosis, a maladaptive remodeling response characterized by excessive collagen deposition and increased tissue stiffness. Fibrosis contributes to a number of heart valve diseases, including rheumatic heart disease and functional mitral regurgitation. Rheumatic heart disease (RHD), a post-inflammatory response triggered by Streptococcus pyogenes infections, induces fibrotic remodeling in mitral valve leaflets via the transforming growth factor beta (TGF-β) signaling pathway, leading to mitral stenosis [7]. Recently, it has been shown that fibrosis also contributes to functional mitral regurgitation (FMR), a condition that affects over 50% of patients with heart failure [8]. The root cause of FMR is myocardial remodeling, which distorts the geometry of the mitral valve annulus and chordae tendineae such that the valve cannot close properly. Through mechanisms that are not yet fully characterized, these alterations cause fibrotic remodeling in the valve leaflets [9–12], exacerbating FMR by preventing the leaflets from adequately enlarging to compensate for the distorted valve geometry and mechanical environment [13–16]. Understanding the signaling pathways that regulate VIC activation and fibrosis during mitral valve disease could enable the discovery of new therapeutic strategies for RHD and FMR.

Two proposed modulators of fibrotic remodeling during heart valve disease are tumor necrosis factor alpha (TNF-α) and interleukin 1 beta (IL-1β). These cytokines are secreted by infiltrating macrophages after tissue damage, recruiting additional leukocytes and initiating an inflammatory cascade that leads to fibrosis if left unresolved [17]. Serum levels of TNF-α and IL-1β are also elevated in patients with heart failure [18,19], suggesting a role for chronic inflammation in the progression of this disorder. The effects of TNF-α on intracellular signaling are mediated by two distinct receptors: TNFR1, which is expressed by all cell types and promotes inflammation and tissue degeneration, and TNFR2, which is selectively expressed and promotes regeneration and homeostasis [20]. On the other hand, IL-1β signaling occurs via its ubiquitously-expressed receptor IL-1R1 [21]. Through a network of intracellular signaling effectors, the TNF-α and IL-1β signaling pathways converge on nuclear factor kappa-B (NF-κB), a transcription factor family that coordinates the expression of genes related to cellular survival, proliferation, and differentiation during the inflammatory response [22].

Although TNF-α and IL-1β indirectly drive fibrosis by promoting inflammation, the direct effects of TNF-α and IL-1β on myofibroblasts are tissue- and context-specific. For example, TNF-α stimulates collagen synthesis in intestinal fibroblasts [23], but TNF-α inhibits the expression of collagen, αSMA, and the pro-fibrotic cytokine TGF-β1 in dermal fibroblasts [24]. Similarly, overexpression of IL-1β leads to pulmonary fibrosis [25], but IL-1β suppresses myofibroblast activation and TGF-β signaling in cardiac fibroblasts [26] and downregulates collagen I and III expression in intestinal smooth muscle cells [27]. As for heart valves, some in vitro studies have proposed that TNF-α and IL-1β accelerate calcific aortic valve disease, another heart valve disease marked by adverse ECM remodeling, by promoting osteoblastic differentiation in aortic VICs [28–30]. However, other studies have suggested that TNF-α and IL-1β promote aortic VIC quiescence [31]. It is important to note that these results are not necessarily generalizable to mitral valve disease, as aortic and mitral VICs have different gene expression signatures due to their distinct embryonic origins [32] and respond differently to inflammatory stimuli [33,34]. The effects of TNF-α and IL-1β on mitral VIC activation have not yet been characterized.

In vitro models of VIC biology are necessary for generating mechanistic insights about heart valve disease. Unfortunately, VICs cultured on polystyrene tissue culture plates become artifactually activated [35], which may mask the effects of experimental interventions. To overcome this limitation, VICs can be cultured in 3D hydrogel scaffolds that mimic the heart valve microenvironment. Scaffold materials may include biomacromolecules that are present in valves, such as collagens and glycosaminoglycans, as well as synthetic polymers such as poly(ethylene glycol) (PEG), which can be biofunctionalized with ECM-derived peptides to direct cellular behavior [36]. Two peptides that are commonly incorporated into PEG scaffolds are the matrix metalloprotease-degradable sequence GGGPQGYIWGQGK and the integrin-binding adhesion motif RGDS [37–39]. Although several research groups have used these platforms to model aortic valve disease [40–46], there have been comparatively few 3D culture models of mitral valve disease. In this study, we use collagen- and PEG-based 3D culture platforms to investigate the influence of TNF-α and IL-1β on myofibroblast activation of mitral VICs.

2. Materials and methods

2.1. Isolation of primary mitral valve interstitial cells (VICs)

Mitral valve leaflets were dissected from porcine hearts (from 6-month-old, young adult pigs) obtained from a commercial abattoir (Animal Technologies, Tyler, TX). To mitigate the effects of biological variability, leaflets from 4 male and 4 female pig hearts were pooled together for each batch of cells. First, leaflets were trimmed to remove annular tissue and chordae tendineae. The clear zone of the anterior leaflet, which is less prone to fibrotic remodeling than other leaflet regions and contains a phenotypically distinct population of VICs [47], was also removed. Next, the leaflets were incubated in a collagenase II solution (600 U/ml, Worthington Biochemical, Lakewood, NJ) for 10 min at 37 °C and scraped with a sterile cotton swab to denude the endothelium. To isolate VICs, denuded mitral valve leaflets were minced, enzymatically digested in a solution containing collagenase III (300 U/ml, Worthington Biochemical), hyaluronidase (50 U/ml, Worthington Biochemical), and dispase (0.5 U/ml, Stemcell Technologies, Vancouver, Canada) for 4 h at 37 °C, and filtered through a 70 μm cell strainer. After isolation, mitral VICs were expanded for 3 passages in VIC growth media consisting of 50:50 DMEM/F12 (ThermoFisher, Waltham, MA), 10% bovine growth serum (BGS; HyClone, ThermoFisher), 1% antibiotic-antimycotic (ThermoFisher), and 0.01 M HEPES (SigmaAldrich, St. Louis, MO).

2.2. PEG-peptide conjugation

Poly(ethylene glycol) (PEG) was biofunctionalized with RGDS (Cayman Chemical, Ann Arbor, MI) and GGGPQG↓YIWGQGK (“PQ”, GenScript, Piscataway, NJ) peptides as previously described [37]. Briefly, 3.4 kDa monoacrylated PEG succinimidyl valerate (PEG-SVA, Laysan Bio, Arab, AL) was reacted with RGDS (1.2 mol RGDS per 1.0 mol PEG) or PQ (1.0 mol PQ per 2.1 mol PEG) overnight in HEPBS buffer (Santa Cruz Biotechnology, Dallas, TX) at pH 8.0 to generate monoacrylated PEG-RGDS and diacrylated PEG-PQ-PEG. Next, these solutions were dialyzed against 4 changes of ultrapure water in a 3500 MWCO dialysis membrane (Repligen, Waltham, MA) to remove unreacted precursors. After dialysis, the PEG-RGDS and PEG-PQ-PEG solutions were sterile-filtered, frozen, and lyophilized.

2.3. PEG encapsulation and 3D culture

To construct the 3D cultures, mitral VICs were suspended at 5 × 10⁶ cells/ml in a prepolymer solution comprised of 4% w/v PEG-PQ-PEG and 2 mM PEG-RGDS in a white light photoinitiator system as previously described [48]. This photoinitiator system consisted of 1.5% v/v triethanolamine, 10 mM Eosin Y, and 0.35% v/v 1-vinyl-2-pyrillidinone in HEPES-buffered saline. Cell suspensions were pipetted into circular elastomeric molds (500 μm thickness, 8 mm diameter), sandwiched between glass slides, and exposed to white light (160 kLux; UltraTow LED Floodlight, Northern Tool and Equipment, Burnsville, MN) for 90 s to crosslink the PEG scaffolds. After crosslinking, the cell-seeded scaffolds were maintained for 3 days in VIC growth media (50:50 DMEM/F12 + 10% BGS) to wash out unreacted precursors, enable hydrogel swelling, and promote cellular growth and attachment. Subsequently, constructs were treated with 10 ng/ml TNF-α or 10 ng/ml IL-1β (R&D Systems, Minneapolis, MN) for 2 days in low-serum media (50:50 DMEM/F12 + 1% BGS). These doses were selected based on their effects on aortic VICs in previously-published hydrogel culture systems [31]. For NF-κB inhibition studies, constructs were additionally treated with the NF-κB inhibitor PS-1145 (40 μM; Tocris Bioscience, Bristol, United Kingdom). A vehicle control of cells treated with 0.4% dimethyl sulfoxide (DMSO) was also performed.

2.4. RNA isolation and quantitative reverse-transcriptase PCR

To isolate RNA, cell-seeded constructs were flash-frozen in liquid nitrogen, thawed, and homogenized in Trizol (ThermoFisher) using a TissueLyser II (Qiagen, Hilden, Germany). Total RNA was purified from Trizol lysates using DirectZol spin columns (Zymo Research, Irvine, CA) according to the manufacturer’s protocol. RNA concentrations were quantified using a NanoDrop 2000 spectrophotometer. Starting from 200 ng RNA per reaction, cDNA was synthesized using the High-Capacity cDNA Reverse Transcription Kit (ThermoFisher). Quantitative reverse transcriptase PCR was performed using the iTaq Universal Probes Supermix (Bio-Rad, Hercules, CA) with TaqMan primers specific to porcine ACTA2 (probe ID: Ss04245588_m1), TAGLN (Ss03373216_g1), TGFB1 (Ss04955543_m1), COL1A1 (Ss03373340_m1), COL3A1 (Ss04323794_m1), and GAPDH (Ss03374854_g1). Two technical replicates were run for each biological sample.

2.5. Protein isolation and western blotting

To isolate protein, cell-seeded constructs were flash-frozen in liquid nitrogen, thawed, and homogenized in RIPA lysis buffer (ThermoFisher) using a TissueLyser II. Protein concentrations were quantified using the Pierce BCA kit, and equal protein quantities were loaded into NuPage 4%–12% Bis-Tris gels (ThermoFisher). Subsequently, proteins were separated via electrophoresis (250 V for 25 min in NuPage MES-SDS Running Buffer) and transferred onto a PVDF membrane (30 V for 1 h in NuPage Transfer Buffer). Afterwards, the membranes were blocked for 1 h in Intercept Blocking Buffer (Li-Cor Biosciences, Lincoln, NE) and probed overnight at 4 °C with primary antibodies against αSMA (Abcam, Cambridge, United Kingdom; catalog number ab7817; 1:1000 dilution) and GAPDH (Cell Signaling Technology, Danvers, MA; catalog number 2118S; 1:1000). The next day, the membranes were washed 3 times with Tris-buffered saline (TBS) + 0.1% Tween, probed with secondary antibodies conjugated to Cy3 or Alexa Fluor 647 (Jackson ImmunoResearch, West Grove, PA) for 1 h at room temperature, and imaged using an Azure Biosystems c600 system. Densitometry analysis of protein band intensity was performed in ImageJ (NIH, Bethesda, MD). All antibodies were validated to ensure that signal levels were within the linear detection range.

2.6. Immunostaining, viability staining, and image quantification

To quantify cellular viability, cell-seeded constructs were incubated for 30 min in a solution containing 2 μM calcein AM and 4 μM ethidium homodimer-1 in PBS (Live/Dead Viability/Cytotoxicity Kit, ThermoFisher) and imaged immediately via confocal microscopy. To quantify the percentage of activated VICs, constructs (350 μm thickness, 5 mm diameter) were fixed in 10% formalin for 45 min, permeabilized in 0.25% Triton-X for 15 min, and blocked in 3.5% bovine serum albumin (BSA) overnight at 4 °C. Next, scaffolds were incubated with primary antibodies against αSMA (Abcam, ab7817; 1:200 dilution) overnight at 4 °C under gentle agitation. After primary antibody incubation, scaffolds were washed 4 times with phosphate-buffered saline (PBS) + 0.1% Tween and incubated with Alexa Fluor 488-conjugated secondary antibodies (ThermoFisher; 1:200 dilution) overnight at 4 °C. Finally, samples were counterstained with DAPI and Alexa Fluor 555-conjugated phalloidin (ThermoFisher) overnight before imaging via confocal microscopy (Nikon A1-RSI). Three randomly selected fields were imaged per PEG scaffold, and the percentage of αSMA positive cells in each field (normalized to total cells) was quantified using an automated CellProfiler analysis pipeline [49]. Cells were classified as positive or negative for αSMA based on their αSMA intensity normalized to phalloidin counterstain intensity [31].

2.7. Collagen contraction assay

To assess changes in VIC contractility, a collagen gel contraction assay was performed. Mitral VICs were suspended in a mixture of rat tail collagen I (VWR), sodium hydroxide, and serum-free M199 media (ThermoFisher) according to published protocols [50], such that the final density was 2 × 10⁶ VICs/ml in 2.5 mg/ml collagen I. This mixture was pipetted into low-adhesion 96-well plates (80 μl/well) and crosslinked for 1 h in a humidified incubator at 37 °C. Next, collagen gels were detached from the edges of the wells using a needle, and VIC media containing 10% BGS was added. Gels were treated with 10 ng/ml TNF-α, 10 ng/ml IL-1β, or vehicle alone, in the presence or absence of 40 μM PS-1145. Subsequently, the collagen gels were allowed to contract freely for 2 days, and the contraction percentage was quantified in ImageJ by normalizing the final gel area to the initial area.

2.8. Mechanical testing

To characterize the stiffness of the scaffolds, circular PEG hydrogels (500 μm thickness, 8 mm diameter) with (n = 3) or without (n = 3) encapsulated cells were subjected to unconfined compression to a strain of 30% at a rate of 0.015 mm/s using a Bose Electroforce ELF 3200 mechanical tester, while reaction force was recorded with a 1000 g load cell (Bose Electroforce, Eden Prairie, MN). For comparison, mitral valve tissue samples from the posterior leaflet (n = 5) and the rough zone of the anterior leaflet (n = 6) were taken using a circular 8 mm biopsy punch and subjected to the same testing conditions. Load and displacement were converted to engineering stress and strain by normalizing to the sample cross-sectional area and height, respectively. Compressive moduli were calculated as the least-squares linear fit slope of the stress/strain curve between 5% and 15% strain.

2.9. Statistical analysis

RNA and protein fold changes were calculated by first normalizing expression levels to GAPDH and then normalizing these values to the control group. Statistical analysis was conducted in RStudio using the Kruskall–Wallis non-parametric test, with post-hoc analysis conducted using the Mann–Whitney U test. Results with p < 0.05 were considered statistically significant. To ensure reproducibility, all experiments were independently repeated at least twice on different days using different batches of primary mitral VICs.

3. Results

3.1. Mitral VICs remain viable and form adhesions within biofunctionalized PEG hydrogels

First, we characterized the viability and morphology of mitral VICs within biofunctionalized PEG hydrogels. Between cellular encapsulation and cytokine treatment, all hydrogels were maintained for three days (two media changes) to wash out unreacted precursors and promote cellular attachment. After this three-day washout period, live/dead staining revealed that 81 ± 13% of VICs in the scaffolds were viable (Fig. 1A). Cytoskeletal visualization via phalloidin staining (Fig. 1B) showed that VICs were attached to the scaffolds via dendritic extensions, as reported in other 3D fibroblast cultures [51]. Furthermore, the compressive moduli (stiffnesses) of the scaffolds were comparable to those of mitral valve leaflets (scaffolds alone: 6 ± 1 kPa; scaffolds with VICs: 4 ± 1 kPa; anterior leaflets: 6 ± 2 kPa; posterior leaflets: 6 ± 2 kPa; p = 0.45 across groups) (Fig. 1C).

Fig. 1.

Mitral VICs remain viable and form adhesions within biofunctionalized PEG hydrogels. (A) Left: mitral VIC viability in 4% w/v PEG-PQ-PEG + 2 mM PEG-RGDS hydrogels (n = 9 fields). Right: representative image of live (green) and dead (red) cells. Scale bar = 100 μm. (B) Confocal z-stack of mitral VICs in PEG hydrogels, stained with DAPI (blue) and phalloidin (cyan). Scale bar = 100 μm; image dimensions = 314 μm by 314 μm by 50 μm. (C) Compressive moduli of PEG hydrogels alone (n = 3 samples), PEG hydrogels with encapsulated cells (n = 3 samples), mitral valve anterior leaflet rough zone (n = 6 samples), and mitral valve posterior leaflet (n = 5 samples).

3.2. TNF-α and IL-1β decrease activation-related gene expression in 3D-cultured mitral VICs

Next, we quantified the effects of TNF-α, IL-1β, and the NF-κB inhibitor PS-1145 on the RNA expression of the activated VIC markers αSMA and SM22α (Fig. 2). The expression of ACTA2, which encodes for αSMA, was 68 ± 9% lower (p = 0.03 vs control) in samples treated with 10 ng/ml TNF-α for 2 days compared to vehicle-treated controls, and 98.4 ± 0.3% lower (p = 0.03) in samples treated with 10 ng/ml IL-1β. Similarly, the expression of TAGLN, which encodes for SM22α, was 65 ± 6% lower (p = 0.03) in TNF-α-treated samples and 89 ± 1% lower (p = 0.03) in IL-1β-treated samples versus controls. These trends were reversed by PS-1145. In samples treated with PS-1145 alone, ACTA2 was upregulated by 122 ± 42% (p = 0.03) and TAGLN was upregulated by 105 ± 20% (p = 0.03). Co-treatment with both TNF-α and PS-1145 restored ACTA2 expression to 53 ± 14% above control (p = 0.03 vs TNF-α alone) and TAGLN expression to 122 ± 3% above control (p = 0.03). Co-treatment with IL-1β and PS-1145 resulted in a less-prominent reversal of ACTA2 (61 ± 6% below control, p = 0.03 vs IL-1β alone) and TAGLN (10 ± 4% above control, p = 0.03) expression.

Fig. 2.

TNF-α and IL-1β decrease activation-related gene expression in 3D-cultured mitral VICs. (A) RNA expression of ACTA2, which encodes the activated VIC marker αSMA, in 3D-cultured mitral VICs treated with 10 ng/ml TNF-α or IL-1β for 2 days in the presence or absence of 40 μM PS-1145 (“PS”). (B) RNA expression of TAGLN, which encodes the activated VIC marker SM22α, under the same treatment conditions. ^∗ p < 0.05 (n = 4 samples per group). Relative expression levels are normalized to GAPDH.

3.3. TNF-α and IL-1β decrease αSMA protein expression in 3D-cultured mitral VICs

Using western blotting, we confirmed the effects of TNF-α, IL-1β, and PS-1145 on αSMA protein expression (Fig. 3). TNF-α and IL-1β downregulated αSMA protein expression by 77 ± 11% (p = 0.1) and 76 ± 5% (p = 0.1) respectively. PS-1145 treatment alone slightly upregulated αSMA by 31 ± 30% (p = 0.2 vs control), while co-treatment with both TNF-α and PS-1145 restored protein expression to 26 ± 20% below control (p = 0.1 vs TNF-α alone), and co-treatment with both IL-1β and PS-1145 restored protein expression to 54 ± 16% below control (p = 0.1 vs IL-1β alone).

Fig. 3.

TNF-α and IL-1β decrease αSMA protein expression in 3D-cultured mitral VICs. (A) Relative αSMA expression measured by western blotting (n = 3 samples per group) in 3D-cultured mitral VICs treated with 10 ng/ml TNF-α or IL-1β for 2 days in the presence or absence of 40 μM PS-1145 (“PS”). (B) Representative western blot image. Relative expression levels are normalized to GAPDH.

3.4. TNF-α and IL-1β decrease the proportion of activated mitral VICs in 3D cultures

In addition, we performed immunofluorescent imaging to measure how TNF-α, IL-1β, and PS-1145 affect the percentage of activated (αSMA positive) mitral VICs (Fig. 4). Whereas 48 ± 9% of vehicle-treated VICs were activated, only 28 ± 9% of TNF-α-treated cells (p = 0.0005 vs control) and 28 ± 16% of IL-1β-treated cells (p = 0.01 vs control) were activated. PS-1145 treatment alone resulted in 58 ± 12% activated cells, slightly higher than vehicle (p = 0.1 vs control). PS-1145 did not significantly reverse the effects of TNF-α (TNF-α + PS-1145: 33 ± 13% activated, p = 0.55 vs TNF-α) or IL-1β (IL-1β + PS-1145: 31 ± 19% activated, p = 1 vs IL-1β) on αSMA positivity.

Fig. 4.

TNF-α and IL-1β decrease mitral VIC activation. (A) Quantification of αSMA positivity in 3D-cultured mitral VICs treated with 10 ng/ml TNF-α or IL-1β for 2 days in the presence or absence of 40 μM PS-1145 (“PS”). (B) Representative immunofluorescence images (blue = DAPI, green = αSMA, red = phalloidin). Scale bar = 100 μm. ^∗ p < 0.05; ^∗∗∗ p < 0.001 (n = 9 fields per group).

3.5. TNF-α and IL-1β decrease mitral VIC contractility

Because increased cell-mediated contractility is a functional hallmark of myofibroblast activation, we performed a collagen gel contraction assay to determine how TNF-α, IL-1β, and PS-1145 affect mitral VIC contractility (Fig. 5). Vehicle-treated collagen gels contracted by 62 ± 5% of their original area, while TNF-α-treated gels contracted by 52 ± 5% (p = 0.009 vs control) and IL-1β-treated gels contracted by 14 ± 10% (p = 0.002) of their original area. Samples treated with PS-1145 alone contracted by 62 ± 7%, similar to vehicle-treated samples (p = 1 vs control). Interestingly, PS-1145 partially reversed the effects of IL-1β on contractility (IL-1β + PS-1145: 56 ± 4% contraction, p = 0.005 vs IL-1β), but did not significantly reverse the effects of TNF-α (TNF-α + PS-1145: 52 ± 7% contraction, p = 1 vs TNF-α).

Fig. 5.

TNF-α and IL-1β decrease mitral VIC contractility. (A) Quantification of collagen gel contraction by mitral VICs treated with 10 ng/ml TNF-α or IL-1β for 2 days in the presence or absence of 40 μM PS-1145 (“PS”). (B) Representative images of collagen gel contraction. ^∗∗ p < 0.01 (n = 6 samples per group).

3.6. TNF-α and IL-1β decrease the gene expression of transforming growth factor beta and collagens I and III

Finally, to investigate how TNF-α, IL-1β, and PS-1145 affect downstream markers of fibrotic ECM remodeling, we measured the RNA expression of TGFB1, which encodes for the pro-fibrotic cytokine TGF-β1, along with COL1A1 and COL3A1, which encode for collagen I and III respectively (Fig. 6). TNF-α and IL-1β decreased the expression of TGFB1 by 84 ± 2% (p = 0.03) and 89.5 ± 0.3% (p = 0.03) respectively. While PS-1145 treatment alone did not significantly affect TGFB1 expression (PS-1145: 10 ± 10% below control, p = 0.19), this inhibitor modestly rescued TGFB1 expression in samples treated with TNF-α (TNF-α + PS-1145: 67 ± 5% below control, p = 0.03 vs TNF-α) and IL-1β (IL-1β + PS-1145: 78 ± 1% below control, p = 0.03 vs IL-1β).

Fig. 6.

TNF-α and IL-1β decrease gene expression of transforming growth factor beta and collagens I and III. (A) RNA expression of TGFB1, which encodes the profibrotic cytokine TGF-β1, in 3D-cultured mitral VICs treated with 10 ng/ml TNF-α or IL-1β for 2 days in the presence or absence of 40 μM PS-1145 (“PS”). (B) RNA expression of COL1A1, which encodes collagen type I. (C) RNA expression of COL3A1, which encodes collagen type III. ^∗ p < 0.05 (n = 4 samples per group). Relative expression levels are normalized to GAPDH.

COL1A1 and COL3A1 expression displayed similar responses to these cytokines and inhibitors. COL1A1 expression was downregulated by TNF-α and IL-1β (TNF-α: 62 ± 3% below control, p = 0.03; IL-1β: 68 ± 2% below control, p = 0.03) and partially rescued by PS-1145 (TNF-α + PS-1145: 48 ± 5% below control, p = 0.03 vs TNF-α; IL-1β + PS-1145: 58 ± 2% below control, p = 0.03 vs IL-1β), while PS-1145 alone had no significant effect (7 ± 18% below control, p = 0.49 vs control). COL3A1 expression was also downregulated by TNF-α and IL-1β (TNF-α: 44 ± 4% below control, p = 0.03; IL-1β: 56 ± 5% below control, p = 0.03), and was not significantly affected by PS-1145 alone (10 ± 19% below control, p = 0.38). Co-treatment with TNF-α and PS-1145 did not rescue COL3A1 expression (TNF-α + PS-1145: 42 ± 5% below control, p = 0.88 vs TNF-α), while co-treatment with IL-1β and PS-1145 resulted in a small but statistically significant increase in COL3A1 expression (IL-1β + PS-1145: 46 ± 3% below control, p = 0.03 vs IL-1β).

4. Discussion

Fibrosis is a maladaptive, myofibroblast-driven ECM remodeling process that contributes to many heart valve diseases, including rheumatic heart disease and functional mitral regurgitation. However, the factors that influence myofibroblast activation and fibrosis, particularly in the mitral valve, are poorly understood. Three-dimensional culture platforms enable investigation of these questions in an environment that mimics the mechanics and dimensionality of native tissue, while eliminating the confounding effects of other cell types that are present in vivo. These design considerations are important for in vitro models of fibrosis because plastic culture substrates induce spontaneous myofibroblast activation. In this study, we used biomimetic PEG- and collagen-based 3D cultures to investigate the effects of TNF-α and IL-1β on mitral VIC activation.

We showed that TNF-α and IL-1β profoundly downregulated the myofibroblast markers αSMA and SM22α in 3D-cultured mitral VICs, and significantly attenuated VIC-mediated contraction of collagen gels. Together, these findings imply that the TNF-α and IL-1β pathways suppress fibrosis in the mitral valve by de-activating mitral VICs. Although the role of inflammatory signaling in valve disease is controversial, similar results have recently been reported in aortic valves. Using proteomic screening techniques, Aguado et al. identified several inflammatory factors that are differentially expressed before and after transcatheter aortic valve repair surgery and investigated their effects on aortic VICs using a PEG culture model. Among the cytokines that were upregulated after valve surgery, TNF-α and IL-1β de-activated aortic VICs, while other inflammatory signals such as BMP-6, CXCL9, and IFN-γ promoted aortic VIC activation [31]. Our study builds upon this work by proposing that TNF-α and IL-1β may play a similar role in the mitral valve by de-activating mitral VICs, thereby limiting excessive fibrosis under inflammatory conditions. These findings reveal targetable pathways that could lead to new pharmaceutical strategies for alleviating mitral valve fibrosis.

We also found that TNF-α and IL-1β downregulated the pro-fibrotic cytokine TGF-β1 in mitral VICs. Antagonism between the pro-inflammatory TNF-α and IL-1β pathways and the pro-fibrotic TGF-β pathway occurs in other tissue-specific fibroblasts through various molecular mechanisms. For example, TNF-α suppressed TGF-β-induced upregulation of αSMA in dermal fibroblasts by both inhibiting Smad3 phosphorylation and destabilizing ACTA2 RNA [24]. Similarly, IL-1β abrogated TGF-β-induced cardiac fibroblast activation by downregulating the RNA expression of TGF-β receptor type 1 (TGFBR1) [26]. In our study, TNF-α and IL-1β negatively regulated TGF-β signaling in mitral VICs by inhibiting the autocrine expression of TGFB1 RNA. These mechanisms are likely to be neither mutually exclusive nor collectively exhaustive, and the crosstalk between TNF-α, IL-1β, and TGF-β signaling pathways in mitral VICs warrants deeper investigation.

The expression levels of COL1A1 and COL3A1 were also downregulated by TNF-α and IL-1β treatment. Collagen type I and collagen type III are structural ECM components that are overexpressed by fibroblasts during fibrotic ECM remodeling [52]. Since VICs are responsible for the active turnover of ECM in heart valves, the downregulation of these fibrillar collagens in mitral VICs after 2 days of TNF-α and IL-1β treatment could be an early indicator of a shift in the valve ECM away from fibrosis. Future long-term studies could investigate the dynamics of collagen gene expression and protein deposition in response to either a single dose or continuous administration of these inflammatory cytokines.

Furthermore, the pharmacological NF-κB inhibitor PS-1145 reversed the inhibition of αSMA and SM22α expression by TNF-α and IL-1β, indicating that the anti-activation effects of these inflammatory cytokines are mediated by canonical NF-κB signaling in mitral VICs. To our knowledge, the effects of NF-κB signaling on fibroblast activation have not been characterized in either aortic or mitral valves. However, this pathway affects other types of fibroblasts in a heterogeneous manner. For example, one study found that NF-κB activation by TNF-α induced myofibroblast differentiation in lung mesenchymal stem cells, exacerbating pulmonary fibrosis [53]. In contrast, another study found that NF-κB activation by IL-1β inhibited lung myofibroblast activation [54]. Additionally, the relationship between NF-κB and αSMA is likely to be governed by bidirectional feedback mechanisms. Consistent with this paradigm, aortic smooth muscle cells from ACTA2 knockout mice display increased baseline levels of NF-κB activity due to elevated production of reactive oxygen species [55].

Although NF-κB inhibition by PS-1145 strongly reversed the inhibitory effects of TNF-α and IL-1β on αSMA and SM22α expression, it only weakly rescued TGF-β and collagen expression. This suggests that the inhibition of these fibrotic genes may be mediated by other signaling effectors that interact with inflammatory cytokine signaling, such as Jun N-terminal kinase (JNK) and p38 mitogen-activated protein kinase (p38 MAPK). Furthermore, the involvement of these alternative pathways may depend on context. Using JNK- and NF-κB-deficient mouse embryonic fibroblasts, one group found that NF-κB, but not JNK, was necessary for TNF-α to suppress baseline α2 collagen type I (COL1A2) expression, whereas JNK, but not NF-κB, was necessary for TNF-α to antagonize TGF-β-induced COL1A2 upregulation [56]. Other studies have shown that TNF-α stimulates the production of CCAAT-enhancer-binding proteins (C/EBPs) that directly repress the COL1A2 promoter [57]. In addition to collagens, myofibroblast-related genes may also be regulated by JNK signaling: in dermal fibroblasts, JNK inhibition, but not p38 MAPK inhibition, reversed the inhibition of αSMA expression by TNF-α [24]. Further research is necessary to fully characterize the mechanisms by which TNF-α and IL-1β suppress fibrosis-related gene expression in mitral VICs.

Interestingly, IL-1β induced more prominent effects on activation-related gene expression, protein expression, and contractility compared to the same concentration (10 ng/ml) of TNF-α. This discrepancy could be explained by the fact that TNF-α binds to both the pro-fibrotic receptor TNFR1 and the pro-regenerative receptor TNFR2. Although TNFR2 was previously thought to be expressed only in select populations of immune and endothelial cells, at least one study has reported that TNFR2 is expressed by aortic VICs and regulates the progression of aortic valve stenosis [58]. It is possible that TNF-α exerts opposing effects on αSMA expression and contractility in mitral VICs by activating both TNFR1 and TNFR2 in parallel, which would reduce the overall magnitude of its effects compared to IL-1β. Targeted blockade of the type 1 and type 2 TNF receptors in mitral VICs via pharmacological antagonists or small interfering RNA could shed light on this question.

Biomimetic 3D culture platforms, such as the PEG and collagen hydrogel scaffolds employed in this study, are valuable tools for studying the molecular mechanisms of heart valve disease. Transcriptomic studies by Mabry et al. have demonstrated the validity of these models by revealing that aortic VICs cultured in 3D within biofunctionalized PEG hydrogels closely resemble freshly-isolated VICs, while cells cultured in 2D on the surfaces of either polystyrene plates or the same PEG hydrogels exhibit significant alterations in gene expression [59]. Previously, PEG hydrogel culture models have been used to investigate how aortic VICs respond to stimuli such as TGF-β, ascorbic acid, and paracrine signals [40–42]. Although mitral valve disease is relatively understudied compared to aortic valve disease, several groups have begun using hydrogel culture platforms to answer fundamental questions about mitral VIC biology. For example, Waxman et al. encapsulated mitral VICs in collagen gels to investigate the synergistic effects of TGF-β and cyclic stretch during myxomatous mitral valve disease [60]. Similarly, Stephens et al. cultured mitral VICs on the surface of PEG hydrogels to examine age-related and region-specific differences in their response to substrate stiffness [61]. Our work builds upon these models and opens new opportunities to study the complex role of inflammatory signaling in mitral valve disease.

Although PEG hydrogel cultures enable us to study VIC biology in a physiologically relevant 3D microenvironment, one limitation of our model is that it contains only a single cell type. For this study, we chose to focus on VIC-specific effects because activated VICs are the primary drivers of maladaptive ECM remodeling in diseased valves [62]. We observed downregulation of VIC activation markers (αSMA, SM22α, and contractility), pro-fibrotic signals (TGF-β), and ECM proteins (collagen I and III) by TNF-α and IL-1β, collectively suggesting that these pro-inflammatory cytokines inhibit VIC-mediated fibrotic ECM remodeling. Furthermore, we chose to examine fibrotic rather than osteogenic remodeling in this study because mitral valve calcification, unlike aortic valve calcification, is relatively rare [33]. Nevertheless, it is important to note that the balance between mitral valve homeostasis and fibrosis in vivo is influenced by other cell types, including valvular endothelial cells and macrophages.

Valvular endothelial cells (VECs), which line the surface of valve leaflets, can transdifferentiate into VICs in a process known as endothelial-mesenchymal transition (EndMT). TNF-α has been shown to induce EndMT in aortic VECs, although only a subpopulation of cells appear to be susceptible to its effects [63]. EndMT has been observed in animal models of functional mitral regurgitation [15], and it is believed that this process contributes to fibrotic maladaptation by increasing the population of activated VICs. In this way, inflammatory cytokines such as TNF-α may conceivably have both pro-fibrotic and anti-fibrotic effects in vivo by promoting EndMT in mitral VECs even as they suppress myofibroblast activation in mitral VICs. Further complicating the matter, VIC activation and VEC EndMT are influenced by reciprocal paracrine signaling between the two cell types [64]. In future studies, structured 3D co-culture methods that have been used for aortic valve disease research [41,65] could be adapted for mitral valves, to study how VIC activation, VEC EndMT, and paracrine signaling between mitral VICs and VECs are influenced by inflammatory stimuli.

Macrophages are a heterogeneous population of immune cells that infiltrate valve tissues during both development and disease [66]. In particular, pro-inflammatory M1 macrophages infiltrate aortic valves and promote osteogenic differentiation in calcific aortic valve disease [67–69]. Although mitral valve disease typically involves myxomatous or fibrotic remodeling rather than calcification, there is evidence that macrophages may be important in mitral valve disease as well. Increased macrophage infiltration into valve leaflets has been reported in several mouse models of myxomatous mitral valve disease [70,71]. Given that proinflammatory macrophages are likely the primary source of TNF-α and IL-1β in the valve microenvironment, our results suggest that macrophage infiltration may play a protective role against fibrotic mitral valve remodeling. In the future, co-culture models containing macrophages and 3D-cultured VICs could be employed to explore the spatial and temporal dynamics of macrophage invasion, cytokine secretion, and myofibroblast activation during valve disease.

Finally, although mitral VICs from male and female animals were pooled together for this study, microarray studies in aortic VICs have identified sex-specific differences in genes related to inflammation and ECM remodeling. [72] Future work could use 3D culture methods to investigate sex-specific differences in mitral VICs in order to understand the disparities in the clinical presentation of mitral valve disease between men and women [73].

In conclusion, we have shown that TNF-α and IL-1β influence fibrosis in the mitral valve by directly inhibiting myofibroblast activation via their downstream effector NF-κB. These inflammatory cytokine pathways may play a more complicated role in valve fibrosis than previously believed. Bioengineered 3D culture platforms, such as the PEG- and collagen-based models in this study, are valuable tools for improving our mechanistic understanding of valve fibrosis. The insights gained from these models will enable the development of new pharmaceutical therapies for heart valve disease.

Statement of significance.

Mitral valve disease is a common cardiovascular condition that is often accompanied by fibrotic tissue remodeling. Valvular interstitial cells (VICs), the fibroblast-like cells that reside in heart valve leaflets, are thought to drive fibrosis during valve disease by differentiating into activated myofibroblasts. However, the signaling pathways that regulate this process in the mitral valve are not fully understood. In the present study, we cultured mitral VICs in collagen and poly(ethylene glycol) scaffolds designed to mimic the heart valve microenvironment and treated the cell-seeded scaffolds with cytokines. Using these 3D culture models, we found that the pro-inflammatory cytokines TNF-α and IL-1β downregulate myofibroblast and fibrosis markers in mitral VICs via the canonical NF-κB signaling pathway.