β_IV-Spectrin/STAT3 complex regulates fibroblast phenotype, fibrosis, and cardiac function

Cardiovascular disease (CVD) remains a leading cause of death worldwide (1). A common finding across multiple forms of CVD is increased cardiac fibrosis, leading to cardiac dysfunction and life-threatening arrhythmia (2–4). Cardiac fibroblasts (CFs) are collagen-producing cells distributed throughout the myocardium that form a fibrous structure to support organ function (2, 5). In response to injury, quiescent CFs undergo a remarkable activation process, characterized by phenotypic transition to a cell type with increased collagen production, proliferation, and contractility (e.g., myofibroblast) (4–6). Although a myriad of neurohumoral and biomechanical factors that promote CF activation have been identified, the precise mechanisms underlying phenotypic transition and activation of these critical cells in vivo remain unclear (4, 5, 7).

The spectrin-based cytoskeleton is an essential structure for the mechanical integrity of metazoan cell membranes (8, 9). The spectrin molecule forms as a flexible heterotetramer of α- and β-subunits with broad distribution of spectrin isoforms in mammalian tissues (2 α- and 5 β-spectrin isoforms expressed in humans) (9, 10). Mounting data support an expanded role for spectrins in not only providing mechanical support for membranes but also in coordination of critical cell signaling pathways (11). In particular, β_IV-spectrin has emerged as a dynamic scaffold that organizes local signaling domains for the regulation of ion channels and membrane excitability in multiple cell types (12–15). Recently, our group identified a role for β_IV-spectrin in control of gene transcription in the heart via association with signal transducer and activation of transcription 3 (STAT3) (16), a multifunctional transcriptional regulator involved in a variety of cellular responses including hypertrophy, proliferation, and survival (17–20). Importantly, loss of β_IV-spectrin in response to chronic pressure overload promotes dysregulation of STAT3-dependent gene expression, increased fibrosis, and cardiac dysfunction in mice (16).

Here, we used genetic and acquired mouse models of β_IV-spectrin deficiency, including an inducible fibroblast-specific β_IV-KO mouse, to test the hypothesis that the β_IV-spectrin/STAT3 complex is essential for maintenance of a quiescent phenotype in CFs. We propose that dysregulation of the spectrin-based complex promotes CF activation, fibrosis, and cardiac dysfunction. Our findings point to a previously unidentified pathway for controlling fibroblast phenotype and fibrosis in response to chronic stress in vivo.

Increased fibrosis and cardiac dysfunction in the setting of β_IV-spectrin deficiency. Cardiac structure and function were first analyzed in 2 different models of β_IV-spectrin deficiency: (a) qv^4J mice that carry a spontaneous mutation in Spnb4 resulting in expression of truncated β_IV-spectrin lacking STAT3 binding (16, 21); and (b) WT mice subjected to 6 weeks of transaortic constriction (TAC) to induce heart failure (acquired β_IV-spectrin deficiency) (16). Echocardiography revealed a significant decrease in ejection fraction in qv^4J animals compared with WT, although not to the level observed in WT TAC mice (Figure 1, A and B). In addition, qv^4J mice also showed decreased fractional shortening and left ventricular (LV) dilation, relative to WT baseline consistent with remodeling changes observed in WT TAC mice (Supplemental Table 1; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.131046DS1). In addition to cardiac dysfunction, β_IV-spectrin–deficient animals displayed defects in cardiac electrical impulse propagation. Specifically, conduction velocity of electrical impulse propagation (evaluated in longitudinal and transverse directions) was reduced in both qv^4J and WT TAC hearts whereas inducibility of ventricular tachycardia (VT) was increased compared with WT baseline (Figure 1, C–F). Analysis of isolated myocytes revealed a highly eccentric growth in qv^4J myocytes with increased length-to-width ratio (aspect ratio) compared with WT (Supplemental Figure 1). Masson’s trichrome staining of ventricular sections showed a marked increase in fibrosis in qv^4J hearts compared with WT, similar to the level observed in WT TAC hearts (Figure 2, TAC fibrosis levels similar to previous reports, ref. 16). Together these data define a wide array of cardiac remodeling changes, including mechanical and electrical dysfunction, linked to β_IV-spectrin deficiency.

Figure 1

Cardiac mechanical and electrical dysfunction in the setting of β_IV-spectrin deficiency. (A) Representative echocardiograms from WT and qv^4J at baseline and from WT following 6 weeks of transaortic constriction (TAC) animals. (B) Summary data (mean ± SEM) for echocardiographic features. n = 5 animals for all groups; *P < 0.05 by 1-way ANOVA and Holm-Sidak post hoc pairwise comparison. (C) Representative activation maps of isolated, Langendorff perfused hearts during S1 pacing (cycle length = 150 ms). (D) Summary data (mean ± SEM) of conduction velocity in longitudinal (long) and transverse (trans) directions and provided in units of m/s. n = 10 hearts for WT and qv^4J and n = 6 for WT TAC; *P < 0.05 by 1-way ANOVA and Holm-Sidak post hoc pairwise comparison. (E) Optically recorded action potentials during programmed stimulation (S1S2 protocol) in isolated, Langendorff perfused hearts in WT baseline, qv^4J baseline, and WT TAC (Scale bar: 200 ms). Red arrow indicates initiation of S2 stimulus. Representative traces are shown for each heart at the longest S1S2 interval that resulted in tachycardia (demonstrated for qv^4J and WT TAC) or shortest S1S2 interval with successful capture (demonstrated for WT). (F) Summary data of ventricular tachycardia (VT) incidence following S1S2 pacing in WT baseline, qv^4J baseline, and WT TAC hearts. Inducibility was assessed over a range of S1S2 intervals, as described in Methods. n = 10 hearts for WT and qv^4J and n = 6 for WT TAC; *P < 0.05 by χ² test.

Figure 2

Increased fibrosis in mouse models of β_IV-spectrin deficiency. (A) Cardiac fibrosis was assessed by Masson’s trichrome staining of ventricular heart sections (collagen labeled blue) from WT and qv^4J animals at baseline and from WT animals following 6 weeks of transaortic constriction (TAC). Original magnification, ×20 shown in zoom image with scale bars: 200 μm. (B) Summary data (mean ± SEM) showing fibrosis as percentage of tissue area. n = 5 where N is the number of hearts analyzed for each group (4 longitudinal sections of entire ventricles analyzed per heart). *P < 0.05 by 1-way ANOVA and Holm-Sidak post hoc pairwise comparison.

β_IV-spectrin deficiency promotes nuclear accumulation of STAT3 in cardiac fibroblasts. The observation of increased fibrosis in β_IV-spectrin–deficient mice led us to ask whether CF function is directly affected by loss of β_IV-spectrin. Although our previous work has elucidated a role for β_IV-spectrin in regulating STAT3 signaling in myocytes (16), nothing is known about expression or function of the β_IV-spectrin/STAT3 complex in CFs. Therefore, expression of the complex was analyzed in isolated primary CFs from adult WT (baseline and TAC) and qv^4J hearts by immunostaining. Successful isolation of fibroblasts was verified by positive vimentin expression by immunoblotting and immunohistochemistry (Supplemental Figure 2). Prominent β_IV-spectrin immunoreactive signal was detectable in WT but not in qv^4J or WT TAC CFs (C-terminal antibody epitope and STAT3 binding motif absent in qv^4J allele, while total β_IV-spectrin expression is decreased in TAC; ref. 16) (Figure 3). Interestingly, STAT3 was found concentrated with β_IV-spectrin mostly along the membrane periphery in WT CFs (Figure 3A). In contrast, CFs from qv^4J and WT TAC hearts showed abnormal STAT3 cytoplasmic localization, with a substantial increase in STAT3 nuclear distribution compared with WT baseline (Figure 3, A and B). Immunoblotting showed no difference in the level of total STAT3 between the 3 groups (Supplemental Figure 2).

Figure 3

β_IV-spectrin deficiency promotes STAT3 relocalization in cardiac fibroblasts. (A) Representative confocal microscopy images (original magnification, ×10) of permeabilized adult WT, qv^4J, and WT TAC cardiac fibroblasts (CFs) immunostained for β_IV-spectrin (red), STAT3 (green; yellow arrow indicates nucleus border), phalloidin (gray in merged image), and DAPI (blue in merged image). White asterisk in merged images indicates region of zoom shown in far right panel. Yellow arrow in zoom image indicates nucleus border. Scale bar: 20 μm. (B) Summary data (mean ± SEM) of STAT3 nuclear localization in WT, qv^4J, and WT TAC CFs. n = 5 per group where N is the number of hearts (5 cells analyzed per heart); *P < 0.05 by 1-way ANOVA and Holm-Sidak post hoc pairwise comparison.

β_IV-spectrin deficiency alters cardiac fibroblast gene expression and activity. We next sought to determine whether loss of β_IV-spectrin affected CF phenotype (gene expression, activation state). STAT3 transcriptional activity was first assessed in WT and qv^4J CFs using a LUC reporter assay. Consistent with the increase in STAT3 nuclear localization, qv^4J CFs showed a marked increase in STAT3 transcriptional activity compared with WT (Figure 4A). In parallel, RNA sequencing and pathway analysis was performed on isolated WT and qv^4J CFs and revealed dramatic transcriptional changes induced by β_IV-spectrin deficiency with enrichment of genes involved in cell proliferation, motility, and differentiation (Supplemental Table 2). STAT3 was predicted to be a likely upstream mediator of observed transcriptional changes (P = 1.45e-12). Transcriptional changes in select genes were confirmed by quantitative PCR (qPCR) (Figure 4B). Additional genes previously identified as markers for an activated myofibroblast phenotype (22) were also assessed. Significant differences were observed between qv^4J and WT CFs in fibrotic genes Col1a1 and Vim previously identified as regulated by STAT3 (16, 19, 23), but not other myofibroblast markers including Acta2 (Supplemental Figure 3), suggesting a partial reprogramming process perhaps analogous to an immature proto-myofibroblast state (7). Fibroblast proliferation was assessed to test for functional consequences of altered gene expression in the setting of β_IV-spectrin deficiency. A significant increase in proliferation (measured as BrdU intensity) was observed in qv^4J CFs at 6 and 12 hours following BrdU treatment compared with WT (Figure 4C). BrdU results were confirmed by manual counting of cells at the same time points (Supplemental Figure 2). Furthermore, qv^4J CFs showed a significant increase in contractility, assessed by collagen gel volume compaction rate, relative to WT (Figure 4, D and E). Together, these data demonstrate that loss of β_IV-spectrin alters CF gene expression and phenotype, producing a more active cell type. TGF-β levels are elevated in heart failure and have been linked to increased fibrosis (24). Therefore, the effect of TGF-β on β_IV-spectrin and related genes was assessed in isolated CFs. Interestingly, CFs treated with TGF-β showed a significant decrease in β_IV-spectrin together with changes in select genes differentially regulated in qv^4J CFs compared with WT (Supplemental Figure 4, compare to Figure 4B). As a first step in determining potential relevance of findings to humans, we performed acute β_IV-spectrin knockdown in human adult ventricular fibroblasts. An acute time course was used for optimal knockdown of β_IV-spectrin (72-hour transfection time per manufacturer’s instructions) and confirmed by decreased β_IV-spectrin expression by qPCR and loss of β_IV-spectrin immunoreactive signal in human CFs transfected with β_IV-spectrin siRNA (Supplemental Figure 5). Expression of select genes differentially expressed in qv^4J fibroblasts was assessed. A significant increase in SERPINA3 was detected in β_IV-spectrin–deficient human CFs compared with control (scrambled siRNA) (Supplemental Figure 5), consistent with observations in the qv^4J mouse (other genes showed a nonsignificant trend).

Figure 4

β_IV-spectrin deficiency alters STAT3 transcriptional activity, gene expression, and activity of cardiac fibroblasts. (A) Summary data (mean ± SEM) of STAT3 transcriptional activity in WT and qv^4J cardiac fibroblasts (CFs) transfected with STAT3 luciferase reporter. n = 3 per group where N is the number of independent preparations for each genotype; *P < 0.05 by 2-tailed t test. (B) Expression of select genes (relative to Rpl7) determined by qPCR in WT and qv^4J CFs. Data presented as mean ± SEM; n = 3 for WT and n = 4 for qv^4J where N is the number of independent preparations; *P < 0.05 by 2-tailed t test. (C) Summary data (mean ± SEM) of BrdU intensity measured at 6 and 12 hours following BrdU treatment in WT and qv^4J CFs as a measure of proliferation (levels normalized to WT at 12 hours). n = 4 independent preparations for each genotype; *P < 0.05 by 2-tailed t test. (D) Representative images of collagen gels seeded with WT or qv^4J CFs over 72-hour time course. Collagen gels without seeded cells (control) are shown for comparison. (E) Summary data (mean ± SEM) of change in collagen gel volume over the 72-hour time course in WT and qv^4J CFs. n = 9 for WT and n = 14 for qv^4J, where N represents the number of independent preparations; *P < 0.05 by 2-tailed t test.

Rescue of quiescent phenotype in qv^4J fibroblasts. To directly test the role of β_IV-spectrin in observed changes in qv^4J fibroblasts, qv^4J CFs were transfected with a β_IV-spectrin construct comprised of repeats 10 through the C-terminus (β_IV,_10-C, containing putative STAT3 binding motif in repeat 15; ref. 16). Six days following transfection, qv^4J CFs expressing β_IV,_10-C showed extranuclear STAT3 distribution, in contrast with the mostly nuclear localization observed in qv^4J expressing empty vector and WT (Figure 5, A and B, and Supplemental Figure 6). Consistent with the redistribution of STAT3 out of the nucleus, qv^4J CFs expressing β_IV,_10-C displayed a decrease in proliferation as assessed by BrdU assay (Figure 5C). At the same time, qv^4J CFs treated with the STAT3 inhibitor S3I-201 showed a similar decrease in proliferation compared with vehicle control, supporting a role for STAT3 dysregulation in CF activation in the setting of β_IV-spectrin deficiency (Figure 5D).

Figure 5

Rescue of quiescent fibroblast phenotype in β_IV-spectrin–deficient fibroblasts. (A) Representative confocal microscopy images (original magnification, ×40) of permeabilized adult qv^4J cardiac fibroblasts (CFs) immunostained for β_IV-spectrin (red), STAT3 (green), and DAPI (blue in merged image) following transfection of empty vector plasmid (control) or β_IV,10-C rescue construct (scale bar: 5 μm). (B) Summary data (mean ± SEM) of STAT3 nuclear localization in WT and qv^4J CFs transfected with empty vector or β_IV,10-C. *P < 0.05 by 1-way ANOVA with Holm-Sidak post hoc pairwise comparison; n = 5 per group where N is the number of independent preparations (5 cells analyzed per preparation). (C and D) Summary data (mean ± SEM) of BrdU intensity measured at 6 and 12 hours following BrdU treatment in WT and qv^4J CFs either (C) transfected with empty vector or β_IV,10-C, or (D) treated with STAT3 inhibitor S3I-201 (100 μM) or vehicle (3% DMSO in PBS). n = 6 independent preparations for each group; data normalized to WT control (empty vector or vehicle) at 12 hours; *P < 0.05 by 1-way ANOVA with Holm-Sidak post hoc pairwise comparison.

STAT3 inhibition abrogates cardiac fibrosis and dysfunction in the setting of β_IV-spectrin deficiency. To determine whether rescue of the β_IV-spectrin–deficient CF phenotype with STAT3 inhibition translated to functional improvement in vivo, we treated qv^4J and WT animals with S3I-201 (20 mg/kg i.p. injection daily) or vehicle (3% DMSO in PBS) for 2 weeks. Echocardiography revealed significant improvement in ejection fraction in qv^4J treated with S3I-201 compared with WT (Figure 6, A and B). In addition, Masson’s trichrome staining revealed a significant decrease in fibrosis in qv^4J mice treated with S3I-201 compared with vehicle (Figure 6, C and D).

Figure 6

STAT3 inhibition normalizes fibrosis and cardiac function in β_IV-spectrin–deficient mice. (A) Representative echocardiograms from WT and qv^4J animals following 2 weeks of treatment with the STAT3 inhibitor S3I-201 (20 mg/kg i.p. injection daily) or vehicle (3% DMSO in PBS). (B) Summary data (mean ± SEM) of the percent change in ejection fraction in WT and qv^4J animals following 2 weeks of S3I-201 or vehicle treatment. n = 10 animals for each group; *P < 0.05 by 1-way ANOVA with Holm-Sidak post hoc pairwise comparison. (C) Representative Masson’s trichrome–stained ventricular sections (collagen labeled blue, original magnification, ×20) from WT and qv^4J animals after 2 weeks of S3I-201 or vehicle treatment. Scale bar: 200 μm. (D) Summary data (mean ± SEM) of the percent change in fibrosis in WT and qv^4J animals following 2 weeks of S3I-201 or vehicle treatment. n = 10 for both genotypes with S31-201; n = 5 for both genotypes with vehicle where N is the number of hearts (4 longitudinal sections of entire ventricles analyzed per heart); *P < 0.05 by 1-way ANOVA with Holm-Sidak post hoc pairwise comparison.

Fibroblast-specific KO of β_IV-spectrin induces fibrosis and cardiac dysfunction. To determine the specific contribution of fibroblasts to cardiac dysfunction observed in our in vivo models of global β_IV-spectrin deficiency, we generated a tamoxifen-inducible fibroblast-specific β_IV-spectrin–KO mouse (β_IVifKO) by cross breeding periostin^MerCreMer with β_IV-spectrin–floxed mice (β_IVifKO, Figure 7A). It has been shown that periostin is specifically expressed in activated CFs (22) and that treatment with angiotensin II increases periostin expression in the heart (ref. 25 and Supplemental Figure 7). Therefore, β_IV-ifKO and control (β_IV-floxed, Cre-) mice were treated for 2 weeks with tamoxifen/angiotensin II to ensure robust activation of the periostin promoter, MerCreMer, and subsequent β_IV-spectrin–KO in CFs. β_IV-ifKO fibroblasts showed similar transcriptional changes to those observed in qv^4J (compare Figure 7B and Figure 4B) together with a loss of β_IV-spectrin and STAT3 mislocalization, without altering myocyte β_IV-spectrin expression (Figure 7, C–E and Supplemental Figure 8). In parallel, β_IV-ifKO mice showed increased fibrosis (Figure 7, F and G) and decreased cardiac function (Figure 7, H and I and Supplemental Table 3) compared with control mice, supporting a role for fibroblast reprogramming in cardiac dysfunction in the setting of β_IV-spectrin deficiency.

Figure 7

Fibroblast-specific deletion of β_IV-spectrin leads to altered gene expression and STAT3 dysregulation in cardiac fibroblasts (CFs) together with decreased cardiac function and fibrosis. (A) The β_IV-spectrin floxed mouse was cross-bred with the periostin^MerCreMer mouse to generate an inducible fibroblast-specific β_IV-spectrin KO mouse (β_IV-ifKO). Two-week treatment of animals with angiotensin II (2.16 mg/kg i.p. injection daily) together with tamoxifen (75 mg/kg i.p. injection daily) (+Tam/Ang II) was used to ensure robust activity of periostin^MerCreMer (periostin highly expressed in activated fibroblasts) and effective β_IV-spectrin KO in cardiac fibroblasts (22). (B) Expression of select fibrotic genes (relative to Rpl7) determined by qPCR in CFs isolated from control (β_IV-spectrin floxed, Cre-) and β_IV-ifKO following 2 weeks of tamoxifen/angiotensin II treatment (+tam/AngII). Data presented as mean ± SEM; n = 3 for control and n = 4 for β_IV-ifKO where N is the number of independent preparations; *P < 0.05 by 2-tailed t test. (C) Representative confocal microscopy images (original magnification, ×10) of permeabilized adult control and β_IV-ifKO CFs immunostained for β_IV-spectrin (red), STAT3 (green), phalloidin (gray in merged image), and DAPI (blue in merged image). White asterisk in merged images indicates region of zoom shown in far right panel. Scale bar: 20 mm. (D and E) Summary data (mean ± SEM) of β_IV-spectrin immunoreactive signal (normalized to DAPI) and STAT3 nuclear localization in control and β_IV-ifKO CFs. n = 4 per group where N is the number of independent preparations (5 fields analyzed per preparation for β_IV-spectrin and 5 cells analyzed per preparation for STAT3 nuclear localization); *P < 0.05 by 2-tailed t test. (F) Masson’s trichrome–stained ventricular heart sections (collagen labeled blue, original magnification, ×20). Scale bars: 200 μm. (G) Summary data (mean ± SEM) showing fibrosis as percentage of tissue area. n = 3 different hearts for each group (4 longitudinal sections of entire ventricles analyzed per animal); *P < 0.05 by 1-way ANOVA and Holm-Sidak post hoc pairwise comparison. (H) Representative echocardiograms and (I) summary data (mean ± SEM) for ejection fraction from control and β_IV-ifKO animals at baseline and following 2 weeks of tamoxifen/angiotensin II treatment. n = 7 for control baseline; n = 6 for control+tam/AngII; n = 10 for β_IV-ifKO baseline; n = 8 for β_IV-ifKO+tam/AngII where N represents the number of animals; *P < 0.05 by 1-way ANOVA and Holm-Sidak post hoc pairwise comparison.

Here we define a role for β_IV-spectrin in tuning the phenotype of CFs via changes in STAT3 subcellular distribution. Mechanistically, we demonstrated that loss of β_IV-spectrin, as observed in TAC, promotes nuclear accumulation of STAT3 and CF activation resulting in a more proliferative, contractile cell type. Ultimately, these changes at the cellular level result in increased fibrosis, cardiac remodeling, and defects in electrical impulse propagation. Based on our findings, we anticipate that the β_IV-spectrin/STAT3 complex will prove an effective target for improving both cardiac mechanical and electrical function in the setting of complex disease (e.g., heart failure).

STAT3 has emerged as an important multifunctional transcriptional regulator of fibrosis, proliferation, ECM synthesis, survival, and growth (17). Previous studies have demonstrated that STAT3 is dysregulated in systemic sclerosis and serves as a crucial checkpoint for fibroblast activation and tissue fibrosis in the skin (19). In CFs, it has been shown that miR-21 induces STAT3 activation resulting in increased proliferation with upregulation of α-smooth muscle actin, collagen I, and collagen III promoting atrial fibrillation in mice, whereas S31-201 promoted downregulation of miR-21 abrogating myofibroblast activation, fibrosis, and atrial arrhythmogenesis (18). At the same time, other studies have shown that constitutive STAT3 activation via IL-6 promotes and enhances CF collagen biosynthesis and activation in vitro, whereas STAT3 suppression abrogates deposition of extracellular matrix in mice following myocardial infarction (26). Similarly, IL-6 deletion has been shown to prevent STAT3 phosphorylation, abrogate fibrosis, and improve cardiac function following TAC (27). STAT3 activation has also been identified downstream of excess TGF-β, leading to increased CF extracellular matrix production (20). Our findings add to the growing literature supporting STAT3 as a critical regulator of cardiac fibrosis. Importantly, our results demonstrate that fibroblast-specific dysregulation of the STAT3 signaling is sufficient to induce cardiac fibrosis. Moreover, our data identify an alternative pathway for noncanonical activation of STAT3 downstream of β_IV-spectrin dysfunction.

Although we demonstrated that STAT3 inhibition abrogates CF phenotype changes and remodeling observed in β_IV-spectrin–deficient mice, we cannot rule out involvement of other signaling pathways. In this manner, it is interesting to consider how the β_IV-spectrin/STAT3 axis explored here interacts with other signaling pathways known to be involved in fibroblast activation and fibrosis. For example, the TGF-β/SMAD signaling pathway is well established as an important node for regulating fibroblast gene expression and phenotype (5, 28, 29). Interestingly, mice deficient in embryonic liver fodrin (ELF, also referred to as β_II-spectrin) show mislocalization of SMAD3/4 and loss of TGF-β–dependent transcriptional response in mouse embryonic fibroblasts (30). More recently, TGF-β has been shown to induce degradation of α_II- and β_II-spectrin subunits in dermal fibroblasts (31), although the authors did not identify an involvement in downstream effects (β_IV-spectrin isoform was not studied). Our findings show that TGF-β promotes loss of β_IV-spectrin in CFs (Supplemental Figure 4). Based on this previous work and our findings, it will be interesting to determine whether there is a functional link between TGF-β/SMAD and β_IV-spectrin/STAT3 pathways in controlling CF phenotype or whether they represent parallel networks that respond to distinct stress stimuli. Furthermore, these data, together with our previous observation that CaMKII activation induces downregulation of β_IV-spectrin (16), suggest that β_IV-spectrin integrates several upstream pathological signals to induce cardiac remodeling.

Recent lineage tracing studies have shown that myocardial infarction induces activation of quiescent, resident CFs in the infarct border region, which then infiltrate the central infarct zone to facilitate scar generation and maintenance (6). Although our studies focused on reactive fibrosis in the setting of chronic pressure overload, we anticipate that a similar dysfunction occurs in the infarct border zone where Ca²⁺/calmodulin kinase II is activated to potentially promote β_IV-spectrin degradation (16, 32, 33). Based on these findings, it is interesting to consider the possibility that loss of β_IV-spectrin is a critical step in CF activation important for myocardial repair following myocardial infarction.

Although this work focuses on the contribution of fibroblasts to dysfunction induced by spectrin deficiency, we have found in our previous work that cardiomyocyte-specific deletion of β_IV-spectrin results in fibrosis (16). It will be interesting to determine how myocytes, fibroblasts, and other cells communicate in vivo to coordinate the organ level response to stress. Going forward, it will be equally important to determine how our findings in the mouse translate to human disease. Recently, several loss-of-function variants in β_IV-spectrin have been identified in human patients with neuropathy, myopathy, and congenital deafness (34, 35). We anticipate that our findings will have important implications for human patients with both acquired and inherited forms of disease, given: (a) conservation of the β_IV-spectrin/STAT3 complex across mouse and humans; (b) our previous studies showing loss of β_IV-spectrin in human heart failure (36); and (c) results from this study showing transcriptional changes in human fibroblasts with acute spectrin knockdown (Supplemental Figure 5).

In this study, we sought to determine whether fibroblasts have a role in cardiac remodeling induced by β_IV-spectrin deficiency using a β_IV-ifKO mouse model. Although these studies indicate that fibroblasts contribute directly to dysfunction, it is important to note that the model uses angiotensin II to increase expression of periostin, thereby driving fibroblast-specific β_IV-spectrin deletion. Angiotensin II has been shown to induce fibroblast activation and fibrosis (37). However, this confounding variable is mitigated by the observation that angiotensin II-treated β_IV-ifKO mice show a greater extent of fibrosis and cardiac dysfunction compared with angiotensin II-treated WT mice. Finally, although our studies focus on fibrosis as a driving factor for cardiac dysfunction, it is likely that coordinated changes in other factors (e.g., hypertrophy, angiogenesis) contribute to the overall phenotype in the setting of β_IV-spectrin deficiency. Notably, qv^4J myocytes show a highly eccentric myocyte growth (Supplemental Figure 1) that would be expected to drive LV dilation and impaired systolic function. It will be exciting for future studies to carefully dissect the interplay of these different tissue-level remodeling changes.

Mouse models. Adult (2–4 months) C57BL/6J male and female wildtype (WT, control) and truncated β_IV-spectrin (qv^4J) littermate mice were used. The qv^4J mice were obtained from Jackson Laboratory and express a Sptnb4 allele with a spontaneous insertion point mutation at C4234T (Q1358→Stop) resulting in a premature stop codon proximal to the STAT3 binding region in β_IV-spectrin (21, 36). periostin^MerCreMer mice (22) were cross-bred with β_IV-spectrin floxed mice (16) to obtain tamoxifen-inducible β_IV-ifKO mice. Adult male and female β_IV-ifKO or control (β_IV-spectrin floxed, Cre-) mice were fed a diet containing 400 mg/kg tamoxifen citrate (Envigo) and treated with tamoxifen (MilliporeSigma) dissolved in corn oil (75 mg/kg i.p. injection daily) and angiotensin II (2.16 mg/kg i.p. injection daily) for 2 weeks to ensure robust activity of MerCreCre under Postn promoter (6, 38).

Optical mapping of ex vivo heart preparations. Optical mapping techniques were used to measure the electrical impulse propagation in the anterior portion of the isolated heart, as described (39). Briefly, hearts were perfused with oxygenated Tyrode’s solution (at 37°C and pH of 7.4) via the aorta using a rolling pump at 2 mL/min. Tyrode’s solution was comprised of 140 mM NaCl, 1.0 mM MgCl₂*6H₂O, 1.2 mM NaH₂PO₄, 4.0 mM KCl, 1.8 mM CaCl₂*2H₂O, 5.6 mM glucose, and 10 mM HEPES sodium salt. The motion uncoupler, Blebbistatin (from stock solution of 5 mg/mL in DMSO), was perfused for 15 minutes to obtain clear signal production with minimal alteration to electrophysiology. The voltage-sensitive dye, di-4-ANEPPS (4 μM) was subsequently perfused for 15 minutes. Images were obtained to verify that the dye successfully reached the extremities of the tissue. The LEX2-LZ4-G light source (SciMedia) with a 531/40 filter was used to excite the dye. The MiCAM05 camera with a 600-mn long-pass filter was used to capture optical signals at 1000 Hz (1 ms/frame). Resulting images had an area of 1.06 × 1.06 cm². Custom-designed Zang Analysis MATLAB program was executed to obtain values from optical signals measured during programmed stimulation (40). AP duration at 80% repolarization (APD₈₀) was measured from optical signals throughout the mapped heart. Conduction velocity was measured and analyzed in transverse and longitudinal directions. Activation maps of stimulus 1 pacing were generated in RHYTHM software via MATLAB program (41). The following S1S2 protocol was used to assess inducibility of VT: 6 s of S1 pacing at a cycle length of 150 ms was followed by a S2 stimulus at a S1S2 interval which decreased from 100 ms in increments of 10 ms until VT (defined as 3 or more successive beats) was observed or 1:1 capture failed.

Isolation of primary mouse and human ventricular cardiac fibroblasts. Mouse CFs were isolated from left and right ventricles under sterile conditions, as described (42). Human CFs were enzymatically isolated from nonfailing, LV tissue, as described (43). Briefly, mouse or human tissue was minced in 2 mg/mL collagenase II (Worthington) dissolved in 1× Ham’s F-10 buffer (Corning). After digestion, the extract was filtered and centrifuged. The supernatant was discarded and cells were resuspended in DMEM; 1×, supplemented with 10% FBS, 1% L-glutamine, and 1% Pen/Strep. Cells were allowed to adhere to culture plates for approximately 4 to 5 hours prior to media removal containing nonadherent cells (e.g., endothelial, myocytes). Fresh feeding media was replenished, and cells were grown for 5 to 7 days to 80% to 100% confluency at 37°C in 5% CO₂. A subset of WT CFs was stimulated with TGF-β1 (10 ng/mL, R&D Systems) or vehicle control (1× PBS) for 48 hours, following 24 hours of serum starvation. All cell experiments were conducted at passage one (P1) unless otherwise noted.

Transfection of primary mouse and human cardiac fibroblasts. Isolated mouse CFs were transfected with β_IV,10-C plasmid (1500 mg) or empty vector (1500 mg, control). Plasmids were generated as previously described (16). For transfection, lipofectamine 2000 (Invitrogen) was mixed with β_IV,10-C plasmid and added to CFs for a 4-hour transfection incubation with serum starvation. CFs were then cultured for 6 days with normal feeding media. In parallel experiments, WT and qv^4J CFs were treated for 48 hours with S3I-201 (100 μM) or vehicle control (3% DMSO in 1× PBS) for 48 hours. Acute knockdown of β_IV-spectrin was achieved in human CFs by transfecting cells for 4 hours under serum starvation with lipofectamine RNAiMAX (Invitrogen) mixed with human siRNA SASI_Hs02_00354399 (10 μM) and SASI_Hs02_00357989 (10 μM), or SIC001-MISSION siRNA universal negative control (10 μM) (Sigma-Aldrich). CFs were then cultured for 72 hours with normal feeding media.

Proliferation assay. CFs were seeded into 96-well culture-treated plates, as described (44). Briefly, cells were adhered for 24 hours with serum starvation. Proliferation was then assessed for 12 hours through incorporation of BrdU, 5x. BrdU proliferation assay (Cell Signaling) was used to quantify proliferation rates (via colorimetric assessment) at desired time points of 6 and 12 hours based on manufacturer’s protocol. Results were validated in a subset of experiments by manual cell counting using a hemocytometer. Experiments were conducted in technical triplicates.

Collagen gel formation and macroscopic gel contraction measurements. Type I rat-collagen gels (2 mg/mL) were prepared by mixing 10x PBS, sterile H₂O, acidic rat tail collagen, and 1M NaOH. Cells were added (500,000 cells/mL) and mixed before gelation. Cell-collagen mixtures were cast into 24-well culture plates and incubated at 37°C in 5% CO₂ for 1 hour. After incubation, 1 mL of culture feeding media was added, and the gels were released from wells to subject for floating. Collagen gels were photographed at various time points for 72 hours. Photographs were obtained prior to the start of the experiment and at regular intervals thereafter. Photographs were analyzed using NIH ImageJ software. Specifically, the diameter of each gel was measured in perpendicular directions and then averaged. As previously described, isotropic compaction was assumed to measure the volume ratio of gels before and after compaction (45). Gels without cells served as controls. Experiments were conducted in technical triplicates.

STAT3 reporter assay. STAT3 transcriptional activity in CFs was evaluated via a STAT3 plasmid (pGL4.47, Promega, 10 μg). Specifically, lipofectamine (Invitrogen) was mixed with the plasmid and added to CFs for a 4-hour transfection incubation with serum starvation. CFs were then cultured for 48 hours with normal feeding media. After 48 hours of culture, cells were lysed, and LUC activity was measured using Promega Dual LUC Reporter Assay according to manufacturer’s protocol. Experiments were conducted in technical triplicates in 96-well plate. Relative LUC activity was quantitatively measured as the ratio of firefly (reporter) to Renilla (normalization, control) activity.

Statistics. SigmaPlot 14.0 was used for statistical analysis. A 2-tailed t test was used to determine P values for single comparisons. For multiple comparisons, a 1-way ANOVA with Holm-Sidak post hoc test was used (data presented as mean ± SEM). The null hypothesis was rejected for P value < 0.05.

Study approval. Animal studies were conducted in accordance with the Guide for the Care and Use of Laboratory Animals published by the National Institutes of Health (NIH) following protocols that were reviewed and approved by the IACUC at The Ohio State University. Nonfailing hearts were obtained through the Lifeline of Ohio Project. The local institutional review board approved the use of human subject tissue. This investigation conforms to the principles outlined in the Declaration of Helsinki.

See Supplemental Material for additional methods and primer sequences for qPCR (Supplemental Tables 4 and 5).