Substrate stiffness modulates cardiac fibroblast activation, senescence, and proinflammatory secretory phenotype

Abstract

In vitro cultures of primary cardiac fibroblasts (CFs), the major extracellular matrix (ECM)-producing cells of the heart, are used to determine molecular mechanisms of cardiac fibrosis. However, the supraphysiologic stiffness of tissue culture polystyrene (TCPS) triggers the conversion of CFs into an activated myofibroblast-like state, and serial passage of the cells results in the induction of replicative senescence. These phenotypic switches confound the interpretation of experimental data obtained with cultured CFs. In an attempt to circumvent TCPS-induced activation and senescence of CFs, we used poly(ethylene glycol) (PEG) hydrogels as cell culture platforms with low and high stiffness formulations to mimic healthy and fibrotic hearts, respectively. Low hydrogel stiffness converted activated CFs into a quiescent state with a reduced abundance of α-smooth muscle actin (α-SMA)-containing stress fibers. Unexpectedly, lower substrate stiffness concomitantly augmented CF senescence, marked by elevated senescence-associated β-galactosidase (SA-β-Gal) activity and increased expression of p16 and p21, which are antiproliferative markers of senescence. Using dynamically stiffening hydrogels with phototunable cross-linking capabilities, we demonstrate that premature, substrate-induced CF senescence is partially reversible. RNA-sequencing analysis revealed widespread transcriptional reprogramming of CFs cultured on low-stiffness hydrogels, with a reduction in the expression of profibrotic genes encoding ECM proteins, and an attendant increase in expression of NF-κB-responsive inflammatory genes that typify the senescence-associated secretory phenotype (SASP). Our findings demonstrate that alterations in matrix stiffness profoundly impact CF cell state transitions, and suggest mechanisms by which CFs change phenotype in vivo depending on the stiffness of the myocardial microenvironment in which they reside.

NEW & NOTEWORTHY Our findings highlight the advantages and pitfalls associated with culturing cardiac fibroblasts on hydrogels of varying stiffness. The findings also define stiffness-dependent signaling and transcriptional networks in cardiac fibroblasts.

INTRODUCTION

The adult heart contains resident cardiac fibroblasts (CFs), which serve critical roles in maintaining tissue architecture by secreting extracellular matrix (ECM) components (1, 2). In response to stress, resident CFs transition to activated fibroblasts (3), often referred to as myofibroblasts, which express the marker protein α-smooth muscle actin (α-SMA), a component of intracellular stress fibers (4, 5). CF activation is associated with transcriptional reprogramming that results in enhanced production and secretion of ECM proteins, resulting in pathological fibrosis of the heart. A causal role for activated fibroblasts in cardiac fibrosis was established using mice harboring a tamoxifen-inducible Cre cassette. These mice were used to selectively deplete activated CFs using diphtheria toxin, which resulted in blunted cardiac fibrosis in response to angiotensin II infusion or myocardial infarction (5). Stress-induced fibrotic remodeling of the heart results in a variety of pathogenic outcomes, including diastolic dysfunction, a contributor to the development of heart failure with preserved ejection fraction (HFpEF; 6).

Mechanistic studies of cardiac fibrosis often necessitate the use of cultured primary CFs. However, serial passage of fibroblasts results in replicative senescence (7), and exposure of CFs to the supraphysiologic stiffness of tissue culture polystyrene (TCPS) triggers spontaneous conversion of the cells into myofibroblasts (8). Together, these phenotypic switches limit the ability to obtain large numbers of quiescent CFs required for many biochemical studies, and confound interpretation of data acquired with cultured CFs. In an attempt to overcome these drawbacks, fibroblasts have been cultured on substrates with a lower elastic modulus, such as polyacrylamide gels (9). Furthermore, we have developed poly(ethylene glycol) (PEG) hydrogels that can be tuned to various stiffnesses (10, 11) and have shown that soft hydrogels prevent CF activation (12, 13).

Here, we characterized the impact of serial passage and culture substrate stiffness on primary CF phenotype. Culture of early and late passage adult rat ventricular fibroblasts (ARVFs) on soft hydrogel is sufficient to reverse their activated state, as evidenced by reduced α-SMA staining and downregulation of expression of a multitude of ECM-encoding genes. Conversely, low-stiffness hydrogels failed to diminish replicative senescence of ARVFs and instead exacerbated the senescent phenotype of the CFs through a partially reversible mechanism. Low stiffness hydrogel-induced CF senescence was also characterized by increased expression of NF-κB-dependent genes that encode secreted, proinflammatory factors that are indicative of the so-called senescence-associated secretory phenotype (SASP). Our findings further highlight the profound impact of matrix stiffness on CF phenotype, provide guidelines for culturing primary CFs in a manner that minimizes spontaneous activation and senescence, and suggest mechanisms by which CF function changes in vivo contingent on the stiffness of the myocardial microenvironment.

MATERIALS AND METHODS

ARVF Isolation and Culture

Right and left heart ventricles from ∼14-wk-old female Sprague–Dawley rats were used to isolate ARVFs. Tissue was minced with scissors and subjected to two consecutive enzymatic digestions (1 mg/mL collagenase type 2; Worthington Biochemical). Cells were collected by gentle centrifugation, as previously described (14). Cells were maintained on plastic tissue culture dishes in DMEM-F12 media (Corning 10-092-CV) supplemented with 20% BenchMark fetal bovine serum (FBS; Gemini BioProducts 100–106), 1% penicillin-streptomycin-l-glutamine (Corning 30-009-Cl) and 1 µmol/L ascorbic acid for a certain number of passages until being plated appropriately for downstream assays. Animal studies were conducted using a protocol approved by the Institutional Animal Care and Use Committee of the University of Colorado Anschutz Medical Campus, following appropriate guidelines and in accordance with the United States Public Health Service Policy on Humane Care and Use of Laboratory Animals.

PEG Norbornene Hydrogels

PEG norbornene (PEG-Nb) hydrogels were prepared as previously described (15). Eight-arm, 40-kDa amine-functionalized PEG (JenKem) was used to synthesize norbornene-functionalized PEG molecules using previously published synthesis methods (10). PEG-Nb hydrogel precursor solutions (4% wt/vol and 10% wt/vol formulations) were used in this study to fabricate soft and stiff hydrogels by mixing 8-arm 40-kDa PEG-Nb with 5-kDa PEG-dithiol cross linker (JenKem) and 2 mM CRGDS cell adhesive peptide (Bachem) at a 0.99:1 thiol-to-ene ratio in PBS. Lithium phenyl-2,4,6-trimethylbenzoylphosphinate (LAP, 1.7 mM) was used as a photoinitiator to photopolymerize the hydrogel precursor solution using ultraviolet (UV) light (365 nm) at 4 mW/cm² for 3 min. Thiol-functionalized coverslips (12 mm) were used to make gels for immunostaining experiments (final gel thickness, 130 μm), and 25 mm coverslips were used to make gels for RNA isolation (final gel thickness, 150 μm). Gels were sterilized with a 5% isopropyl alcohol solution in PBS for 30 min, followed by three washes with PBS, and allowed to swell overnight in ARVF cell culture media at 37°C and 5% CO₂ before cell culture.

Phototunable PEG-Anthracene Hydrogels

For the phototunable study, anthracene-functionalized PEG precursors (PEG-Ant) and anthracene functionalized GRGDS were synthesized as previously described (16). Eight-arm, 20-kDa PEG-Ant hydrogel precursor solutions (4% wt/vol, 8% wt/vol, and 13% wt/vol) were used in this study to fabricate static soft, static stiff, and dynamic stiffening gels with 1 mM anthracene-functionalized GRGDS. Hydrogel precursor solutions were polymerized on 12 mm anthracene-functionalized coverslips to make gels for cell culture (final gel thickness, 100 μm). For photostiffening studies, 13% PEG-Ant hydrogel precursor solutions were first photopolymerized to a soft (E ∼ 4 kPa) stiffness at 4.4 mW/cm² UV light for 4 min and 27 s. ARVFs cultured up until passage 3 or passage 6 in plastic tissue culture dishes were transferred to soft PEG-anthracene hydrogels (100,000 cells/well) and cultured for 2 days. Hydrogels with ARVFs were next stiffened during a second photopolymerization step for 80 s at 10 mW/cm² to achieve a final stiffness of E ∼ 25 kPa.

Rheology Characterization

Synthesis and rheology characterizations of PEG-Nb and PEG-Ant hydrogels were performed as previously described (15, 16). For hydrogel measurements, PEG solutions were UV polymerized on the rheometer stage and characterized using a parallel plate (8 mm diameter). Storage modulus (G′) and loss modulus (G′′) measurements were made using oscillatory shear rheology with an amplitude of 1% and frequency of 1 Hz. The storage modulus values were converted to elastic modulus (E) measurements using the formula E = 2 G′(1 + ν) with a Poisson’s ratio of ν = 0.5, assuming G′ ⋙ G′′. The final 10 elastic modulus plateau measurements were averaged to report a final stiffness measurement for a sample.

SA-β-Galactosidase Staining

AVRFs were fixed and stained to detect SA-β-Gal activity (Cell Signaling Technology Kit No. 9860), following manufacturer’s instructions. Briefly, after fixation, the reagent mix was adjusted to pH 6.0 and incubated for 16 h at 37°C. Images were obtained using an Olympus BX63 automated fluorescence microscope, and the percentage of senescent cells was manually calculated as a ratio of total cells using ImageJ software.

Immunofluorescence Imaging

For α-SMA and F-actin staining, ARVFs cultured on two-dimensional (2-D) soft or stiff PEG-Nb hydrogels (250,000 cells/well), or TCPS in six-well dishes (50,000 cells/well) were fixed in 4% EM-grade paraformaldehyde (Electron Microscopy Sciences 15714-S) in PBS for 10 min. Fixed cells were permeabilized with 0.025% Triton X-100 in PBS. Cells were then incubated with 100 nM Acti-stain 555 phalloidin (Cytoskeleton; PHDH1) for 30 min in the dark to detect F-actin. Cells were subsequently incubated with the primary antibody for α-SMA (Abcam; ab7817; 1:100 dilution) overnight at 4°C. The following day, cells were washed and incubated with a fluorescein-conjugated secondary antibody (ImmunoResearch Laboratories; 715-095-150; 1:100 dilution) for 1 h in the dark. Images were acquired using an EVOS cell imaging system microscope (ThermoFisher Sci).

Immunoblotting

Total protein was isolated from ARVFs using radioimmunoprecipitation assay (RIPA) lysis buffer, consisting of 25 mM Tris (pH 7.6), 150 mM NaCl, 1% NP-40, 1% sodium deoxycholate, and 0.1% SDS, supplemented with Halt proteinase and phosphatase inhibitor cocktail (Thermo Scientific 1861280). Protein lysates were quantified using the PierceBCA protein assay kit (Thermo Scientific 23225). Proteins were separated using SDS-PAGE and transferred onto 0.2-μm PVDF membrane (Bio-Rad 162-0177). After blocking in intercept blocking buffer (LI-COR 927-60001), blots were probed with antibodies directed against α-SMA (Sigma A2547; 1:1,000) and GAPDH (R&D Systems AF5718; 1 μg/mL) overnight at 4°C in intercept blocking buffer with 0.2% Tween-20. Following incubation with primary antibodies, blots were rinsed in Tris-buffered saline-Tween 20 (TBST) and probed sequentially with donkey anti-goat IRDye 800CW (LI-COR D20119-12; 1:10,000) and goat anti-mouse IRDye 680RD (LI-COR D10902-01; 1:10,000) in intercept blocking buffer with 0.2% Tween-20 and 0.2% SDS. Protein bands were visualized using the LI-COR Odyssey XF Imager (LI-COR Model 2802) and quantified using LI-COR image studio software.

Quantitative Real-Time PCR

Total RNA was obtained from ARVFs using QIAzol lysis reagent (Qiagen 79306) and purified using the BCP phase separation reagent (Molecular Research Center BP 151). Following precipitation using 2-propanol, total RNA was washed in 80% ethanol and reconstituted in water. Complementary DNA (cDNA) was prepared from 500 ng of total RNA using the Verso cDNA synthesis kit (Life Technologies AB1453). Gene expression analysis was performed using a StepOnePlus Real-Time PCR system (Applied Biosystems) and PowerUp SYBR Green Master Mix (Applied Biosystems A25743). Amplicon abundance was quantified using the ΔΔCt method. Primer sequences are listed in Supplemental Table S1 (Note: Supplemental material may be found at https://doi.org/10.6084/m9.figshare.23826030.v1).

RNA-Sequencing

Total RNA was collected using the Direct Nb-zol RNA Miniprep Kit (Zymo Research R2051) from three biological replicates of ARVFs (obtained from independent rat hearts). Cells were cultured on TCSP or transferred to 2-D soft PEG-Nb hydrogel at passage 2 or passage 5 for an additional 48 h; once transferred, the cells were considered to be P3 and P6, respectively. The UC Denver Genomics and Microarray core facility performed mRNA selection and cDNA library preparation followed by 2 × 150 base pair fragment sequencing on an Illumina NovaSeq 6000 sequencing system.

RNA-Seq Processing and Differential Expression Analysis

Reads were trimmed and verified for quality with Trim Galore! (v.0.6.4_dev, https://www.bioinformatics.babraham.ac.uk/projects/trim_galore/). Trimmed reads were aligned with STAR (v.2.7.9a) to the rat genome (Rnor_6.0) with Ensembl (v.104, https://ensembl.org) annotations in a single pass using default settings for gene counting [quantMode GeneCounts (17)]. Reads were loaded into R (v.3.3.3) for processing (18). Transcripts with <1 read per million mapped in all samples (71%) were removed before differential expression (DE) analysis with DESeq2 (v.1.14; 19). For comparisons between supports (P3 soft gel vs. P3 plastic; P6 soft gel vs. P6 plastic), the processing date was added as a blocking variable. For passage comparisons (soft gel P6 vs. soft gel P3; plastic P6 vs. plastic P3) no blocking variable was used. Genes were deemed to be differentially expressed with adjusted P value < 0.05 and absolute log2 fold change > 1. All plots (heat map, volcano, Venn) were generated using R. Heat maps show row-normalized regularized log2 counts per million and were generated with pheatmap (20). Venn diagrams were generated using Vennerable (21). RNA-seq data are deposited in GEO (GSE244614).

RNA-Seq Pathway Analysis

Following differential gene expression analysis of RNA-seq data, core analyses were accomplished using ingenuity pathway analysis (IPA) to identify affected candidate pathways and upstream regulators by comparing gene expression, upregulated or downregulated, in P6 versus P3 ARVFs grown on TCPS and P6 ARVFs grown on soft hydrogel versus TCPS (FC > +1.2 or FC < −1.2, FDR < 0.05). For the analysis of P6 versus P3 ARVFs grown on TCPS, the top 10 pathways that were significantly altered are reported, rank-ordered by significance, and shown as −Log10 P value. For pathway analysis of both P3 and P6 ARVFs grown on soft hydrogel versus the ARVFs grown on TCPS, we selected the top five pathways and also performed analysis of “upstream regulators” in P6 ARVFs, comparing the two conditions. For illustration purposes, the mechanistic networks for TGF-β and NF-κB are displayed. Interaction analysis of the selected pathways was performed with Cytoscape using String and MCODE plugins to discover gene interactions and key regulator genes. Heat maps were generated using heatmapper (http://www.heatmapper.ca/).

Statistical Analysis

Statistical analyses were completed by ANOVA, followed by post hoc testing (Tukey’s test) or by unpaired t test, using GraphPad Prism software. Statistical significance, defined as P < 0.05, is reported.

RESULTS

Serial Passage Promotes Cardiac Fibroblast Senescence

First, ARVFs were cultured to address the susceptibility of serially passaged CFs to replicative senescence. Upon isolation, passage 0 (P0) ARVFs were plated on TCPS, representing (P1), and over the course of the next 14 days, the cells were passaged up to five times (Fig. 1A). Quantitative PCR analysis was performed to assess abundance of the cyclin-dependent kinase inhibitor 2 A (CDKN2A) mRNA transcript, which encodes p16, a prototypical marker of senescence (22). Elevated p16 expression was observed at P4 and continued to increase at P5 and P6 (Fig. 1B). Consistent with this, another readout of senescence, senescence-associate β-galactosidase (SA-β-Gal) activity (23), was increased in ARVFs from P4 through P6 (Fig. 1, C and D). ARVFs also became progressively larger with increased passage (Fig. 1, E and F), which is another hallmark feature of cellular senescence (24).

Figure 1.

Serial passage promotes cardiac fibroblast senescence. A: schematic representation of the experiment to address the impact of passage number (P) on adult rat ventricular fibroblast (ARVF) phenotype; P0 = ARVFs that were initially isolated and plated, before trypsinization and passaging. B: quantitative real-time polymerase chain reaction (RT-PCR) analysis of p16 mRNA expression. C: staining for endogenous senescence-associated β-galactosidase (SA-β-Gal) activity. D: quantification of the percentage of ARVFs that are SA-β-Gal positive. E: bright-field images of ARVFs; scale bar = 100 μm. F: quantification of the size of ARVFs. G: immunoblot analysis of α-smooth muscle actin (α-SMA) and GAPDH, with n = 2 technical replicates of ARVFs per passage. H: relative α-SMA protein expression was quantified using LI-COR Image Studio software and was normalized to GAPDH levels. I: quantitative RT-PCR analysis of Acta2/α-SMA mRNA expression. B, D, F, H, and I: data are presented as means +SE, with statistical analysis performed using one-way ANOVA with Tukey’s multiple comparisons test; *P < 0.05. Each point represents data value from a technical replicate. TCPS, tissue culture polystyrene.

The supraphysiologic stiffness of TCPS causes rampant myofibroblast activation marked by the presence of α-SMA-containing stress fibers (25). Myofibroblast activation of ARVFs, as evidenced by readily detectable α-SMA protein expression, was noted at P2 and persisted out to P6 (Fig. 1, G and H). When compared with P2–P5, there was a modest increase in expression of the mRNA for α-SMA (ACTA2) at P6 (Fig. 1I).

Softer Matrices Reverse Cardiac Fibroblast Activation but Concomitantly Trigger Cellular Senescence

We previously demonstrated that culture of ARVFs on hydrogels with elastic moduli lower than TCPS leads to deactivation of the cells (12). To determine if this approach could be used to diminish or reverse ARVF senescence, synthetic PEG hydrogels were generated using photoinitiated thiol-ene click chemistry (10, 11). By altering the final PEG polymer weight percentage, we characterized hydrogels with elastic moduli (E) of ∼6 kPa (soft) and ∼41 kPa (stiff) using rheology, which reflect the approximate stiffness of healthy and fibrotic hearts, respectively (Supplemental Fig. S1) (26, 27). TCPS has a tensile strength of ∼3 GPa.

ARVFs were cultured on TCPS to P2 or P5 over the course of ∼5 to 21 days, respectively, and were subsequently trypsinized and replated on TCPS or stiff or soft hydrogels for an additional 48 h. Both hydrogels reduced α-SMA and F-actin staining, indicating ARVF deactivation; for P5 cells, deactivation was most evident with soft hydrogel (Fig. 2, A and B). Surprisingly, both soft and stiff hydrogels significantly promoted premature senescence of P2 ARVFs, as revealed by SA-β-Gal staining, with the most dramatic induction of senescence occurring on soft hydrogel (Fig. 2, C and D). Although p16 expression was not increased by upon culture of P2 ARVFs on soft hydrogel, mRNA abundance of a related marker of senescence, cyclin-dependent kinase inhibitor 1 A (CDKN1A), which encodes p21, was augmented (Fig. 2, E and F). For P5 ARVFs, culture on soft matrix was insufficient to further increase SA-β-Gal expression (Fig. 2, G and H). However, exposure of these cells to soft hydrogel did lead to significant elevations in expression of the p16 and p21 senescence marker genes (Fig. 2, I and J). Prior culture on high tensile strength substrate appeared to be required for senescence activation of ARVFs, since plating P0 cells on soft hydrogel for 48 h failed to induce SA-β-Gal or p16 and p21 expression (Fig. 2, K–N); P0 is used to define freshly isolated cells before they have been subjected to culture.

Figure 2.

Softer matrices reverse cardiac fibroblast activation but concomitantly trigger cellular senescence. Passages 2 (P2; A) or 5 (P5; B) adult rat ventricular fibroblasts (ARVFs) were trypsinized and replated on tissue culture polystyrene (TCPS) or the indicated hydrogels. After an additional 48 h, cells were fixed and stained for α-smooth muscle actin (α-SMA) and F-actin; scale bar = 100 μm. C: P2 ARVFs were replated as indicated and stained for senescence-associated β-galactosidase (SA-β-Gal) activity; scale bar = 50 μm. D: quantification of SA-β-Gal-positive ARVFs. Quantitative real-time polymerase chain reaction (RT-PCR) analysis of p16 (E) and p21 (F) mRNA expression. G: SA-β-Gal staining of P5 ARVFs replated as indicated; scale bar = 50 μm. H: quantification of SA-β-Gal-positive ARVFs. Quantitative RT-PCR analysis of p16 (I) and p21 (J) mRNA expression. K: SA-β-Gal staining of P0 ARVFs replated as indicated; scale bar = 50 μm. L: quantification of SA-β-Gal-positive ARVFs. Quantitative RT-PCR analysis of p16 (M) and p21 (N) mRNA expression. For D–F, H–J, and L–N, data are presented as means +SE, with statistical analysis performed using one-way ANOVA with Tukey’s multiple comparisons test (3 groups) or t test (2 groups); *P < 0.05. Each point represents data value from a technical replicate.

Soft Matrix-Induced Cardiac Fibroblast Senescence Is Partially Reversible

To address whether matrix-induced CF senescence is reversible, a phototunable soft hydrogel was used. Leveraging the UV-mediated photodimerization of anthracene groups (16), we synthesized PEG polymers functionalized with anthracene to enable polymer cross-linking with cytocompatible UV light in the presence of ARVFs. Rheological measurements of static PEG-Ant gels reveal elastic moduli of 6.4 ± 2.6 kPa for soft gel formulations and 36.7 ± 9.0 kPa for stiff formulations. For dynamic stiffening hydrogels, gel precursor solutions were first stiffened to a soft elastic modulus of 4.2 ± 0.6 kPa, ARVFs were plated, and then gels were further stiffened to a final elastic modulus of 25.6 ± 4.0 kPa (Supplemental Fig. S2). Replating P2 ARVFs on dynamic stiffening soft hydrogel for 2 days led to senescence activation (Fig. 3A, left). Remarkably, when the gel was stiffened via a pulse of UV exposure, and the cells were maintained for an additional 2 days in culture, there was a significant reduction in SA-β-Gal-positive ARVFs, indicating a reversal of cellular senescence (Fig. 3A, middle). In contrast, when ARVFs were cultured on nonphototunable hydrogel and expose to UV, senescence continued to progress over the next 2 days, indicating that the observed reversal was due to substrate stiffening as opposed to an indirect effect triggered by UV light (Fig. 3A, right). Evidence of reversal of CF senescence after increasing hydrogel stiffness was also observed when examining p16 and p21 mRNA expression, although only the reduction in p21 expression reached statistical significance (Fig. 3, C and D). Two days of increased substrate stiffness was not sufficient to promote α-SMA mRNA expression in ARVFs; cells cultured on soft hydrogel for 4 days had the lowest level of α-SMA mRNA expression (Fig. 3E).

Figure 3.

Soft matrix-induced cardiac fibroblast senescence is partially reversible. A: passage 2 (P2) adult rat ventricular fibroblasts (ARVFs) were trypsinized and replated on phototunable or nonphototunable hydrogels. After 2 days, the indicated cells were exposed to ultraviolet (UV) for 80 s and retained and cultured for a subsequent 2 days before fixation and staining for endogenous senescence-associated β-galactosidase (SA-β-Gal) activity; scale bar = 50 μm. B: quantification of SA-β-Gal-positive ARVFs. Quantitative real-time polymerase chain reaction (RT-PCR) analysis of p16 (C), p21 (D), and Acta2/α-SMA (E) mRNA expression. F: P5 ARVFs were trypsinized and replated on phototunable or nonphototunable hydrogels. After 2 days, the indicated cells were exposed to UV for 80 s and retained and cultured for a subsequent 2 days before fixation and staining for endogenous SA-β-Gal activity; scale bar = 50 μm. G: quantification of SA-β-Gal-positive ARVFs. Quantitative RT-PCR analysis of p16 (H), p21 (I), and Acta2/α-SMA (J) mRNA expression. For B–E and G–J, data are presented as means +SE, with statistical analysis performed using one-way ANOVA with Tukey’s multiple comparisons test; *P < 0.05. Each point represents a data value from a technical replicate.

An identical experiment was performed with P5 ARVFs replated on phototunable soft hydrogel. With these later passage cells, substrate stiffening with UV light failed to reverse senescence and actually appeared to exacerbate senescence induction at the level of SA-β-Gal and marker gene expression (Fig. 3, F–I). Thus, under the described conditions, reversal of senescence can be achieved with early, but not late, passage CFs. Surprisingly, 2 days of increased substrate stiffness reduced α-SMA mRNA expression in ARVFs; cells cultured on soft hydrogel for 4 days also had a lower level of α-SMA mRNA expression (Fig. 3J).

Serial Passage and Substrate Stiffness Elicit Distinct Transcriptional Programs in Cardiac Fibroblasts

To address the impact of passage number and substrate stiffness on gene expression in CFs, transcriptomic profiling by RNA sequencing (RNA-seq) was performed using RNA from P2 and P5 ARVFs that were trypsinized and subsequently replated on TCPS or soft hydrogel for an additional 48 h (n = 3 for each condition; once replated, the cells were considered to be P3 and P6, respectively). Principal component analysis of differentially expressed genes revealed clear segregations of the transcriptomes of ARVFs in each condition (Fig. 4A). Initially focusing on P3 and P6 cells grown on TCPS, using a fold expression change >1.5× and adjusted P value of <0.05 as cutoffs, differential expression (DE) analysis showed that 993 genes were upregulated and 784 genes downregulated at P6 compared with P3; the top 15 changes in each direction are labeled (Fig. 4B; see Excel File 1 for full DE results). Ingenuity pathway analysis (IPA) of transcripts that were upregulated in P6 compared with P3 showed a strong enrichment for endoplasmic reticulum (ER) and Golgi apparatus function (Fig. 4C). Conversely, genes downregulated in P6 versus P3 ARVFs grown on TCPS were predominantly associated with protein translation (Fig. 4D), which is noteworthy since senescent cells have been shown to exhibit decreased ribosome biogenesis coupled with reduced protein synthesis (28).

Figure 4.

Serial passage of cardiac fibroblasts elicits senescence-related transcriptional responses. A: passages 2 (P2) and 5 (P5) adult rat ventricular fibroblasts (ARVFs) were trypsinized and replated on tissue culture polystyrene (TCPS) or soft hydrogel for an additional 48 h before harvesting for RNA-seq analysis. Principal component analysis (PCA) of gene expression clearly segregated each treatment group, with 47% of variance accounted for by principal component 1 (PC1, x-axis). B: volcano plot indicating mRNA transcripts that are upregulated (red) or downregulated (blue) in P6 ARVFs on TCPS compared with P3 ARVFs on TCPS; top 15 changes in each direction are labeled. Ingenuity pathway analysis was used to reveal the top 10 canonical pathways that were significantly increased (C) or decreased (D) in P6 ARVFs on TCPS vs. P3 ARVFs on TCPS, rank-ordered by significance and shown as −Log10 P value.

Analyses were next performed to address how exposure of late passage CFs to soft hydrogel influenced cell state. Comparing P6 ARVFs cultured on soft matrix to those maintained on TCPS, using a fold expression change of >1.5× and an adjusted P value of <0.05 as cutoffs, DE analysis showed that 992 genes were upregulated and 858 genes downregulated; the top 15 changes in each direction are labeled (Fig. 5A). Decreased genes were predominantly associated with ECM production, which is consistent with the ability of soft hydrogels to reverse CF activation (12). In contrast, genes upregulated in P6 ARVFs exposed to soft hydrogel versus TCPS were highly enriched for proinflammatory mediators (Fig. 5C). Further analysis, focusing on predicted upstream regulators that were inhibited or activated upon exposure of P6 ARVFs to soft hydrogel, strongly suggested that the lower tensile strength matrix activated profibrotic TGF-β signaling, and concomitantly activating proinflammatory NF-κB-dependent transcription (Fig. 5, D and E).

Figure 5.

Soft hydrogel concomitantly decreases profibrotic gene expression and induces proinflammatory gene expression in cardiac fibroblasts. A: volcano plot indicating mRNA transcripts that are upregulated (red) and downregulated (blue) in passage 6 (P6) adult rat ventricular fibroblasts (ARVFs) cultured on soft hydrogel vs. tissue culture polystyrene (TCPS); top 15 changes in each direction are labeled. Ingenuity pathway analysis (IPA) was used to reveal the top 5 canonical pathways that were significantly decreased (B) or increased (C) in P6 ARVFs cultured on soft hydrogel vs. TCPS, rank-ordered by significance and shown as −Log10 P value. IPA was also performed to identify the top “upstream regulators” that were decreased or increased in P6 ARVFs cultured on soft hydrogel compared with TCPS, which uncovered downregulation of profibrotic TGF-β-dependent gene expression (D) and concomitant upregulation of proinflammatory NF-κB target gene expression (E), IPA was used to reveal the top 5 canonical pathways that were significantly decreased (F) or increased (G) in P3 ARVFs cultured on soft hydrogel vs. TCPS, rank-ordered by significance and shown as −Log10 P value.

Analysis of earlier passage cells revealed similar effects. In P3 ARVFs cultured on a soft matrix compared with TCPS, there was reduced expression of genes involved in ECM production, with an attendant increase in proinflammatory gene expression (Fig. 5, F and G). Activation of proinflammatory gene expression in CFs cultured on soft hydrogel is reminiscent of induction of the SASP (29).

String analysis of transcript expression in P6 ARVFs cultured on soft hydrogel versus TCPS suggested that proteins encoded by genes of the modulated senescence pathway are interactive with proteins involved in the NF-κB signaling and ECM pathways (Supplemental Fig. S3A). Furthermore, MCODE analysis indicated the main regulatory genes as Myl12a, H-Ras, TGF-β1, ITGB1, and Col6a1 (Supplemental Fig. S3B), and specified TGF-β1 as the key regulator of the senescence pathway given the number of proposed interactions between this cytokine and senescence-associated genes (Supplemental Fig. S3C). MCODE analysis also suggested that H-Ras is the key hub gene connecting the senescence pathway with the NF-κB and ECM networks (Supplemental Fig. S3D).

Given the key roles of CFs in homeostatic control of the nonremodeling heart through regulation of ECM turnover (30), we also examined the impact of passage number and substrate stiffness on the expression of transcripts encoding matrix-degrading enzymes. The only clear pattern to emerge from this analysis was a general reduction in expression of a disintegrin and metalloproteinase (ADAM), ADAM with thrombospondin motifs (ADAMTS), ADAMTS-like (ADAMTSL), matrix metalloproteinase (MMP), and tissue inhibitor of MMP (TIMP) in P6 ARVFs cultured on soft hydrogel compared with TCPS (Supplemental Fig. S4). Interestingly, when comparing P3 ARVFs cultured on soft gel versus TCPS, few metalloproteinase genes were found to be downregulated; however, pairwise comparison of P6 ARVFs cultured on soft gel and TCPS revealed nearly all these genes to be downregulated or not significantly modulated. Notably, when comparing P3 and P6 ARVFs on soft gels, there was a more marked downregulation of TGF-β signaling and upregulation of NF-κB pathways, as was demonstrated in Fig. 5, D and E.

Soft Hydrogel Promotes NF-κB Activation in Cardiac Fibroblasts

Given the increase in expression of NF-κB target genes in P6 ARVFs cultured on soft hydrogel, we next sought to determine if this proinflammatory transcription factor is indeed activated in these CFs. Canonical NF-κB activation involves nuclear translocation of the p50:p65 heterodimer into the nucleus (31). As such, indirect immunofluorescence was performed to determine the subcellular distribution of the p65 subunit of NF-κB, also known as RelA, in P6 ARVFs after replating on TCPS or soft hydrogel; cells were costained to α-SMA to reveal myofibroblasts. As shown in Fig. 6A, robust nuclear staining of p65 was observed in ARVFs replated on soft hydrogel, whereas the transcription factor was undetectable in CFs maintained on TCPS, presumably because of diffuse cytoplasmic localization. Consistent with these findings, induction of the proinflammatory markers CXCL1 and CX3CL1 in ARVFs replated on soft hydrogel was blocked upon treatment of the cells with the NF-κB inhibitor, BAY11-7082. These findings demonstrate that induction of proinflammatory gene expression in CFs cultured on soft hydrogel is due, in part, to NF-κB activation.

Figure 6.

Soft hydrogel promotes NF-κB activation in cardiac fibroblasts. A: passage 5 (P5) adult rat ventricular fibroblasts (ARVFs) were trypsinized and replated on tissue culture polystyrene (TCPS) or soft hydrogel for 3 days before fixation and staining for the p65 subunit of NF-κB or α-smooth muscle actin [α-SMA; scale bar = 10 μm. P5 ARVFs were trypsinized and replated on TCPS or soft hydrogel for 3 days. The NF-κB inhibitor, BAY11-7082 (5 μM), or DMSO vehicle control (0.1% final concentration)], was added to the cells for 24 h before harvesting RNA for quantitative real-time polymerase chain reaction (RT-PCR) analysis of CXCL1 (B) and CX3CL1 (C) mRNA transcripts. Data are presented as means + SE, with statistical analysis performed using one-way ANOVA with Tukey’s multiple comparisons test; *P < 0.05. Each point represents data from a technical replicate. D: model for substrate stiffness-mediated regulation of TGF-β- and NF-κB-dependent gene expression programs in cardiac fibroblasts. A portion of the model was created with a licensed version of BioRender.com.

DISCUSSION

Studies of the mechanisms controlling activation/deactivation of CFs hold promise for uncovering novel therapeutic targets for cardiac fibrosis. Here, we provide a comprehensive evaluation of the impact of serial passage and substrate stiffness on the phenotype of cultured primary CFs. Serial passage of ARVFs on TCPS (∼3 GPa) results in progressive and robust induction of cellular senescence, marked by elevated antiproliferative marker gene expression, expression of SA-β-Gal, and a profound increase in cell size. By P2, ARVFs are highly activated, as evidenced by α-SMA expression. Replating CFs on stiff (∼41 kPa) or soft (∼6 kPa) hydrogels is sufficient to reverse activation of the ARVFs, but culminates in further induction of senescence, especially when the CFs are cultured on soft hydrogel. At the molecular level, we provide evidence to support a model in which culture of CFs on TCPS triggers profibrotic and proliferative TGF-β signaling, which is reversed when the cells are placed on lower tensile strength substrates (Fig. 6D). However, reducing substrate stiffness also results in augmented expression of antiproliferative, senescence-associated genes, and activation of proinflammatory NF-κB-dependent transcription (Fig. 6D). These findings highlight the extreme mechanosensitivity of CFs and underscore the importance of carefully monitoring cellular activation and senescence when using cultured primary CFs for mechanistic studies.

Senescence is generally regarded as a state of permanent cell cycle arrest, in part because of retinoblastoma (Rb)-mediated chromatin remodeling and suppression of E2F target gene expression (32). In addition, the DNA damage response promotes induction of p16 and p21, which inhibit proproliferative CDK signaling (33). However, there is evidence to suggest that cell cycle exit in senescence is reversible in some instances, for example when p16 levels are low (34). This has led to the proposed existence of “light/reversible” and “deep/irreversible” senescent states, which are governed by factors such as p16 level and epigenomic modifications (35). Using phototunable hydrogels, we found that soft substrate-induced CF senescence is partially reversible in P2 ARVFs, but not P5 ARVFs. Thus, we posit that early passage CFs exhibit light senescence that can be circumvented, whereas late passage CFs are epigenetically locked into an irreversible, deep mode of senescence. Consistent with a role for epigenetic events in controlling substrate stiffness-mediated CF senescence, our prior studies of mesenchymal stem cells (MSCs) and vascular interstitial cells clearly established that mechanical stimuli influence chromatin remodeling (36–38).

Cell cycle withdrawal is only one element of senescence. For instance, senescent cells also often exhibit profound changes in the amount and composition of the proteins they secrete. This altered secretome is referred to as the SASP, components of which include cytokines, chemokines, and proteases (39). Thus, senescent cells are not simply dormant but instead can influence tissue structure and function via paracrine signaling. We provide evidence to suggest that the culture of CFs on soft hydrogel promotes SASP induction, with the inflammatory component of the program being driven by the NF-κB transcription factor. The mechanism by which soft matrix triggers NF-κB nuclear import and stimulation of downstream target genes remains unknown, but could be related to the recent demonstration that culturing MSCs on soft substrate (∼2 kPa) led to increased clustering of cell surface tumor necrosis factor (TNF) receptors, and thereby enhanced NF-κB activation in response to TNF-α treatment (40).

How might alterations in substrate stiffness influence CF phenotypes in vivo? The elastic modulus of the soft hydrogel used in our studies is roughly equivalent to that of healthy adult heart tissue (26). In this regard, it is important to note that induction of CF senescence by the soft substrate required prior exposure of the cells to TCPS. The degree of elevated stiffness needed to prime the cells for subsequent soft substrate-induced senescence remains unknown. Nonetheless, we propose that CFs in a healthy heart, which has a stiffness of ∼8 kPa, are quiescent but not senescent. In response to stress, these cells proliferate and become activated to produce large amounts of ECM, thereby stiffening their microenvironment. As activated CFs migrate away from fibrotic lesions, the progressive decrease in stiffness to which the cells are exposed likely triggers senescence as a means of limiting further expansion of the CFs and preventing additional fibrosis of the heart. As a reinforcing mechanism, the less stiff microenvironment also reduces TGF-β signaling and consequent profibrotic gene activation. If the senescent CFs express components of the SASP, such as cytokines and chemokines, these cells could also impact immune cell interactions with the heart to influence the magnitude and/or duration of the fibrotic response.

In summary, cultured primary CFs exhibit tremendous plasticity in response to passaging and alterations in substrate stiffness, both of which influence profibrotic “myofibroblast activation” of the cells, as well as their ability to transform into a senescent state. Our findings highlight distinct molecular mechanisms that govern CF cell state transitions, which are likely relevant to fibrotic remodeling of the heart in vivo, and provide guidelines for working with cultured primary CFs for mechanistic studies.

Limitations

The present study focused on understanding the impact of serial passage and substrate stiffness on the phenotype of primary CFs from female rats. Future investigation with CFs from male rats will be needed to address the potential for sexually dimorphic responses at the level of senescence activation and reversal and NF-κB-dependent proinflammatory secretome production. Sex differences in CF ECM production and responsiveness to therapy have been noted in aged mice (41), but whether this extends to CF senescence phenotypes is, to our knowledge, unknown. Furthermore, studies of the consequences of altered substrate stiffness on nonfailing and failing human CFs will ascertain the translational significance of the current findings. These studies will need to include an assessment of the stiffness that is required to prime CFs for subsequent induction of senescence following plating on soft hydrogel. In particular, rather than TCPS, it will be important to evaluate CFs grown on a substrate with an elastic modulus that approximates a stable scar or a fibrotic myocardium and then replated on a substrate with an elastic modulus similar to a healthy heart. Regarding our experimental system, we prepare 12-mm-diameter hydrogels on round glass coverslips. Due to the limited growth area, the protein yield following cell homogenization is often below the threshold required for analysis of expression or posttranslational modifications by immunoblotting.

Moving forward, researchers will need to balance practicality and cost with culturing CFs for next generation in vitro disease modeling. TCPS will remain a standard for primary cell expansion, and based on our data, we recommend culturing and using ARVFs for experiments at or before P3 on TCPS. Future methods where ARVFs are expanded directly on hydrogel-coated plates with larger surface areas, which can be engineered in-house and/or purchased commercially, would need to be evaluated to determine if cell senescence can be avoided with long-term culture of large batches of ARVFs.

DATA AVAILABILITY

Data are available upon request from the authors.

SUPPLEMENTAL DATA

Supplemental Fig. S1–S4 and Supplemental Table S1 may be found at https://doi.org/10.6084/m9.figshare.23826030.v1.

GRANTS

T.A.M. received funding from National Institute of Health (NIH) Grants HL116848, HL147558, DK119594, HL127240, and HL150225. B.A.A. received funding from American Heart Association Grant 942253, NIH Grants HL148542 and DP2-HL173948, and the Chan Zuckerberg Initiative science diversity leadership grant. M.B.F. and M.R. were supported by American Heart Association Grants 34380603 and 35210627. J.G.T. received funding from the NIH Grant K99HL166708. T.A.M. and K.S.A. were supported by a University of Colorado AB Nexus grant.

DISCLOSURES

T.A.M. is on the scientific advisory boards of Artemes Bio and Eikonizo Therapeutics, is cofounder of Myracle Therapeutics, and has a subcontract from Eikonizo Therapeutics related to an SBIR grant from the National Institutes of Health (HL154959). None of the other authors have any conflicts of interest, financial or otherwise, to disclose.

AUTHOR CONTRIBUTIONS

M.B.F., M.R., K.S.A., B.A.A., and T.A.M. conceived and designed research; M.B.F., M.R., J.G.T., K.B.S., and B.A.A. performed experiments; M.B.F., M.R., M.E.L., B.A.A., and T.A.M. analyzed data; M.B.F., M.R., K.B.S., K.S.A., B.A.A., and T.A.M. interpreted results of experiments; M.B.F., M.R., M.E.L., B.A.A., and T.A.M. prepared figures; M.B.F. and T.A.M. drafted manuscript; M.B.F., M.R., M.E.L., K.S.A., B.A.A., and T.A.M. edited and revised manuscript; M.B.F., M.R., J.G.T., K.B.S., M.E.L., K.S.A., B.A.A., and T.A.M. approved final version of manuscript.