Fibroblasts orchestrate cellular crosstalk in the heart through the ECM

Abstract

Cell communication is needed for organ function and stress responses, especially in the heart. Cardiac fibroblasts, cardiomyocytes, immune cells, and endothelial cells comprise the major cell types in ventricular myocardium that together coordinate all functional processes. Critical to this cellular network is the non-cellular extracellular matrix (ECM) that provides structure and harbors growth factors and other signaling proteins that affect cell behavior. The ECM is not only produced and modified by cells within the myocardium, largely cardiac fibroblasts, it also acts as an avenue for communication among all myocardial cells. In this Review, we discuss how the development of therapeutics to combat cardiac diseases, specifically fibrosis, relies on a deeper understanding of how the cardiac ECM is intertwined with signaling processes that underlie cellular activation and behavior.

Introduction

The cardiac extracellular matrix (ECM) is a dynamic protein network that integrates foundational cellular processes in the heart. It not only provides structural support to cells within the myocardium, but also functions as a reservoir for growth factors and other proteins that underlie cell-cell communication. Beginning in early development, dynamic changes in the composition of the ECM network facilitate proper cell behavior and ventricular development as the heart matures to achieve full contractile rigor^1–3. ECM composition is also critical for maintaining heart function after injury or disease in adults, as the fully mature myocardium lacks regenerative healing capacity. However, adult ECM reorganization is frequently associated with chronic injury or pathology that impedes cardiac function^4,5.

Cardiac fibroblasts (CFs) can activate to a contractile cell type known as myofibroblasts, which are responsible for the bulk of ECM production, maintenance and remodeling in the heart⁶. A healthy adult heart is largely devoid of myofibroblasts, but once activated they facilitate cardiac wound healing and ventricular remodeling^7–9. Myofibroblast formation and function can be beneficial in terms of providing mechanical support to an acutely injured tissue, but their sustained activity and proliferation contributes to pathological fibrosis and tissue stiffening^9–11.

Developing therapeutics to combat cardiac diseases, specifically fibrosis, relies on a deeper understanding of how the cardiac ECM is intertwined with signaling processes for cellular activation and behavior. In this Review, we highlight the concept that the composition and dynamic state of the cardiac ECM is tied to fibroblast activation, which is also regulated by signals from cardiomyocytes and immune cells, illustrating a dynamic feedback system among myocardial cells and their surrounding ECM (Figure 1).

Figure 1. The ECM facilitates communication among fibroblasts, cardiomyocytes and macrophages in the heart.

Fibroblasts are activated by cues such as mechanical tension and inflammatory cytokines and chemokines that are typical of injury. Activated fibroblasts (myofibroblasts) modify the ECM, which in turn regulates behaviors of myocardial cells such as myocytes and macrophages. The ECM affects these cell populations through changes in mechanical tension, as well as availability of growth factors and other matricellular proteins. It remains unclear exactly how macrophages and cardiomyocytes might exert influences on the ECM, although growth factor signaling is likely key. It is also unknown the degree to which myofibroblasts convert back to a quiescent/homeostatic phenotype. Abbreviations: GF, growth factors; ECM, extracellular matrix.

The cardiac ECM

The cardiac ECM facilitates force transmission and impacts cardiac ventricular wall stiffness, and also plays a critical role in transducing molecular signals and mediating cell-cell crosstalk⁴. The ECM network harbors latent growth factors and proteins that are rapidly released during injury or ECM remodeling⁴. Active proteins can then engage cell surface receptors on multiple cardiac cell types that initiate intracellular signal transduction cascades to alter cell behavior¹². Prime examples of proteins that localize to the ECM matrix are transforming growth factor-β (TGFβ), platelet-derived growth factor (PDGF) and fibroblast growth factor (FGF). Alterations in the conformation and organization of the ECM can affect the amount and accessibility of these growth factors to cells within the myocardium.

The protein composition of the ECM, activity of crosslinking enzymes, and connections between ECM and cells all contribute to the overall stiffness of a tissue. Assessing stiffness in the heart is not trivial, as has been recently well-reviewed¹³. While some ECM proteins have been characterized, such as fibrillar collagens, a catalogue for the spatiotemporal expression of specific ECM proteins in homeostasis and stages of disease progression has not been established. Ongoing research and refinement of specialized techniques¹³ are starting to untangle how ECM composition and stiffness affect heart function.

Fibronectin is a foundational ECM protein that is laid down during early heart development, which is replaced with “priming” collagens in the neonatal period (P0 to P7), and by early adulthood >80% of the cardiac ECM is collagen type I and 11% is type III^1–3. With aging, the adult mammalian heart becomes stiffer over time in coordination with increased hypertrophy^1,14. Experimental evidence suggests that increased collagen I levels in the adult heart generates a stiffer environment for supporting greater contractility but with aging this might also impair relaxation, leading to the observed hypertrophy¹⁴. In an analysis of CFs and ECM from human heart failure patients, Perestrelo et al., described differential ECM organization that accompanied altered tissue stiffness and CF activation¹⁵. Variations in collagen organization and alignment have been shown to regulate CF differentiation in vitro¹⁶.

With respect to the adult heart and predisposition to disease, studies have investigated the contributions of a variety of ECM components to cardiac stiffness and cardiomyocyte biology. Proper fibronectin polymerization is required for cardiac fibrosis post ischemia/reperfusion injury³³, although mechanisms for this improvement are unclear. Loss of α2 fibrils of type I collagen led to decreased overall collagen I content, a more fragile collagen fiber and thus a more compliant ventricle in response to passive inflation¹⁷. While proper secretion of type I collagen fibrils is critical, crosslinking these fibrils to form a mature collagen network is also required for augmenting cardiac stiffness in response to hypertension^18,19. The essential role of fibrillar collagen in cardiac stiffness has been supported by observations in human patients with severe aortic stenosis²⁰. Furthermore, overexpression of the collagen reinforcing protein periostin in the mouse heart promoted greater pressure overload-induced hypertrophy with increased ECM content²¹.

Since fibrillar collagen crosslinking is essential for ECM maturation and cardiac stiffness, it is reasonable to hypothesize that proteins regulating ECM crosslinking could also affect myocardial organization and the cellular response to injury. For example, the lysyl oxidase family of essential collagen and elastin crosslinking proteins are upregulated in cardiac hypertrophy in both human disease and mouse models^22–24. Overexpression of lysyl oxidase-like 1 (Loxl1) from murine cardiomyocytes results in myocyte hypertrophy, increased ventricular wall thickness, and interstitial fibrosis²⁵; interestingly, cardiac function was preserved. On the other hand, pharmacologic inhibition or genetic deletion of lysyl oxidase-like protein 2 (Loxl2) significantly reduced levels of cardiac fibrosis in response to overload hypertrophy in mice, which correlated with better cardiac compliance and improved cardiac function²⁶.

Various matricellular proteins are also involved with the regulation of ECM stiffness, but their involvement in cardiac fibrosis and myocyte regulation is less straightforward. For example, both thrombospondin (Thbs)-1-²⁷ and Thbs-4-²⁸ null mice showed augmented cardiac hypertrophy after aortic transaortic constriction (TAC), which was associated with increased cardiac fibrosis, although it remains unclear how the Thbs1 or 4 proteins might regulate interstitial fibrosis in the heart. Deletion of Thbs-3, however, protected mice from TAC-induced cardiac hypertrophy through maintaining cardiomyocyte integrin expression and sarcolemmal stability.²⁹ Deletion of tenascin-C, which regulates ECM quality and cardiomyocyte attachment, reduced cardiac fibrosis and decreased cardiomyocyte hypertrophy after pressure overload³⁰. SPARC-null mice developed less fibrosis but without changes in cardiomyocyte size after pressure overload stimulation³¹. Furthermore, deletion of transglutaminase-2 led to reduced collagen content after injury to the heart but augmented cardiac hypertrophy³². These studies suggest that select matricellular proteins can regulate ECM properties and remodeling directly, which secondarily impacts cardiomyocyte hypertrophic potential.

Fibroblast-derived ECM during cardiac development

One of the most impactful scientific advancements over the past several years has been single-cell RNA (scRNA) sequencing, which allows for precise characterization of cell populations within a heterogeneous tissue³⁴. A number of scRNA-seq methodologies have been optimized for analysis of cells within the heart, and their applications and limitations have recently been discussed^35,36. Using such sequencing approaches have suggested ways in which fibroblast activation and/or differentiation might impact cardiomyocyte biology, although additional mechanistic research is needed.

Fibroblast heterogeneity has long been appreciated³⁷, but recent RNA sequencing studies have provided a higher resolution of the gene expression differences between similar cell populations, including cardiac fibroblasts. Wang et al., uncovered five distinct fibroblast gene expression states during postnatal cardiac development, some of which were identified as responsible for ECM modifications that supported cardiomyocyte growth and maturation³⁸. Farbehi et al., examined mouse hearts three and seven days after myocardial infarction and identified 11 subpopulations within platelet-derived growth factor receptor (PDGFR)-α expressing fibroblasts, distinguishable by variable expression of smooth muscle α-actin (αSMA), Wnt pathway genes, and ECM modifiers³⁹, which are characteristic indicators of fibroblast activation. In addition to fibroblasts with unique gene expression signatures, subsets of cardiomyocytes with transcription signatures specific to regions of the heart have also been identified⁴⁰, which are speculated to underlie neonatal regenerative responses⁴¹. It is not yet clear whether these cellular "subtypes" represent bonafide stable cell types or are a snapshot of a cells transitioning from one differentiated state to another^42,43.

Studies have elucidated cardiomyocyte differentiation during neonatal development⁴⁴, but the field is only now starting to uncover how the ECM and cellular microenvironment affects cardiac myocyte maturation (Figure 2). Wang et al., performed scRNA-seq on mouse hearts from several developmental stages to analyze regulatory signaling networks of different cell populations and how they might interact with one another³⁸. Their work demonstrated that CFs drive cardiomyocyte maturation through changes in their transcriptional program that includes ECM maturation. Importantly, adult CFs uniquely contributed to induced pluripotent stem cell-derived cardiomyocyte maturation that was not achieved with neonatal CFs³⁸. This has profound implications not only for congenital diseases, but also for research aimed at inducing cardiomyocyte regeneration after injury. Clearly the conditioning or maturation of the ECM by CFs can directly impact cardiomyocyte differentiation and disease responsiveness.

Figure 2. Cardiomyocytes and fibroblasts sense each other through the dynamic ECM environment.

CFs are the main producers of core ECM proteins during cardiac development, and thus affect cardiomyocyte differentiation and maturation. During cardiac injury or chronic disease conditions, the ECM is altered, and fibroblasts interpret changes in mechanical tension and cytokine/chemokine signaling. They then differentiate to myofibroblasts that secrete the bulk of ECM proteins during adult cardiac wound repair and remodeling. This increased matrix deposition is sensed by cardiomyocytes, causing changes in cell phenotype such as hypertrophy. Abbreviations: GF, growth factors; ECM, extracellular matrix; TGFβ, transforming growth factor-β; IL-6, interleukin-6; PDGF, platelet derived growth factor.

While scRNA sequencing highlights cellular gene expression heterogeneity within a sample, one drawback of these studies is the loss of information about regional tissue heterogeneity. To this end, Lacraz et al., used cryosections to assess transcriptional differences (RNA tomography) in cells across different regions of the heart following ischemia/reperfusion injury⁴⁵. By specifically analyzing transcripts isolated from infarcted tissue versus regions remote to injury, they uncovered Sox9 as a regulator of fibrotic gene expression, including collagen and other ECM proteins⁴⁵. Likewise, Mantri et al., used chick embryo hearts to combine scRNA-seq with spatial RNA-seq to show that developmental gene expression programs are region-specific within the heart across select cell types, demonstrating that cellular differentiation corresponds with morphologic changes within a tissue⁴⁶. For example, migration of epicardial-derived progenitor cells into the myocardium coincided with expression of ECM factors known to induce cell migration⁴⁶. In another study, Liu et al., utilized fate-mapping studies of cell cycle activity to show regional differences in cardiomyocyte regulation of cell division and response to injury⁴⁷, corroborating reports of heterogeneous cardiomyocyte populations. Although not tested directly, the authors proposed that changes in ECM composition could be a driver of heterogenic cell behavior in this setting.

Periostin is a matricellular protein involved with collagen fiber maturation and is a known marker for CF activation after injury²¹. However, a periostin-expressing fibroblast lineage was recently found to exist transiently during postnatal heart growth, which was distinct from the Tcf21⁺ fibroblast population that arises from the epicardium⁴⁸. Ablating periostin-expressing fibroblasts had detrimental effects on cardiomyocyte maturation and sympathetic innervation in the first seven days after birth, which coincided with the appearance of a more mature ECM network. Further investigation into the temporal expression of integrins, adhesion proteins, and other cellular machinery related to mechanosensitivity in fibroblasts is needed for full appreciation of the signaling networks that balance normal cardiac growth in coordination with the ECM.

Activated fibroblasts and cardiac fibrosis

As we have discussed, the process of ECM maturation and remodeling is primarily regulated by CFs, which directly secrete ECM components during development and in the adult heart^49,50. When stimulated by mechanical stress or secreted factors, CFs activate into a cell type known as a myofibroblast, which functions as a hybrid fibroblast/smooth muscle cell that expresses αSMA and other contractile proteins ^10,51,52. Myofibroblasts synthesize and secrete high levels of ECM and matricellular proteins, including collagens type I and III, fibronectin splice variants (e.g., Fn-EDa), periostin, tenascin C and matrix metalloproteases^52,53. Mechanical and secreted signals activate fibroblasts to remodel their surrounding matrix in three main ways: secreting ECM components, degrading existing ECM structures, and reorganizing and refining newly formed ECM networks⁴. First, fibroblasts actively synthesize and secrete fibrillar ECM proteins, which are the basic components of cardiac ECM^9,51,54,55. Secondly, fibroblasts also secrete metalloproteinases, which include proteins in the matrix metalloprotease (MMPs) and a disintegrin and metalloproteinase with thrombospondin motifs (ADAMTS) families, all of which are critical for ECM degradation and remodeling⁵⁶. Lastly, fibroblasts secrete proteins that refine and crosslink collagen, such as SPARC, osteopontin, tenascin C, periostin, and lysyl oxidase¹². Collectively, CFs can modify the ECM in a variety of ways to impact cells throughout the myocardium, including cardiomyocytes and resident/activated immune cells.

For an in-depth discussion of the process of cardiac fibrosis and mechanisms of cell activation, we refer the reader to other reviews ^{10,11,57–59}. Regarding the present Review, cardiac fibrosis occurs when myofibroblasts increase the total net deposition of ECM proteins in the myocardium. Fibrotic ECM is stiffer and less compliant than normal matrix, and results in altered biophysical dynamics within the myocardial wall. As discussed, fibroblast activation can occur after an overt injury such as MI, or in response to mechanosensitive protein stimulation, inflammatory cytokines, or a neuroendocrine/chemical stimulus^11,49,58. In response to these signals, CFs activate and proliferate, depositing collagens and other matrix proteins to alter ECM support function and ventricular wall dynamics. Importantly, fibroblast-generated new ECM deposition prevents cardiac wall rupture and maintenance of chamber geometries after MI^60–62. We recently explored the temporal dynamics of fibroblast activation and persistence in the heart after MI injury in mice⁶³. Fibroblasts within the injured area of a mouse heart are initially depleted after MI, but by three days activated fibroblasts proliferate and migrate into the injured myocardium to restore total fibroblast cellular content⁶³. This cellular reorganization is accompanied by the compaction of remaining basal lamina structures along with new ECM formation, leading to scar formation⁶³. Interestingly, the myofibroblast cell state is transient in the acutely injured heart as these cells are either lost after scar formation or they change into a more differentiated state, known as a matrifibrocyte, which expresses markers of bone and cartilage⁶³. Thus, the initially adaptative function of CF activation leads to deposition of permanent scar tissue and its initial remodeling, which is then maintained by a particular subpopulation of more highly differentiated fibroblasts (matrifibrocytes).

Using cell-specific cre-LoxP genetic systems in the mouse, a variety of genes have been examined for their contribution to cardiac fibrosis and adaptive remodeling. Cardiomyocyte-specific deletion of the TGFβ receptors 1 and 2^64,65, endothelial cell-specific deletion of endothelin-1⁶⁶, and macrophage-specific deletion of collagen IV⁶⁷ have all resulted in alterations in the ECM and changes in cardiac function at baseline or with injury. These data suggest that non-fibroblast cell types can influence the extracellular environment, although evidence has shown the primacy of the CF or myofibroblast in directly regulating ECM content in the heart. For example, fibroblast- or myofibroblast-specific deletion of Smad2/3⁶⁸, TGFβ receptors 1 and 2^68–70, mitogen-activated protein kinase (MAPK) p38⁶⁰ and β-catenin⁷¹ each showed compromised ECM production in response to cardiac injuries, which secondarily resulted in less cardiomyocyte hypertrophy. Indeed, we showed that fibroblast-specific deletion of the collagen chaperone protein heat shock protein-47 (HSP47) resulted in significantly less cardiac fibrosis and collagen type I, III and V deposition in the cardiac ECM with pressure overload stimulation, which compromised cardiomyocyte hypertrophy⁶. Importantly, deletion of Hsp47 from cardiomyocytes or endothelial cells did not reduce the induction of collagen expression and interstitial fibrosis after pressure overload⁶. Collectively, these results indicate that fibroblasts or myofibroblasts are the primary cell type in the heart responsible for fibrillar collagen production and augmentation of ECM after injury, which is required for effective cardiomyocyte hypertrophy with pressure overload and other injury-based stimuli (Figure 2).

An important point of emphasis is that cardiac fibrosis does not arise from a common etiologic origin. It is a heterogeneous process that manifests from a variety of pathologic stimuli, making it unlikely that a single therapeutic intervention will alleviate all fibrotic conditions of the heart. For example, “replacement fibrosis” is defined by the acute formation of collagenous fibrotic material, or a scar, that replaces dead cardiomyocytes, and this process is physiologically important in maintaining ventricular wall integrity and geometry^59,72. The pathways driving replacement fibrosis in myofibroblasts seem to be more robust and less therapeutically amenable to inhibition. By comparison, "interstitial fibrosis” is a thickening or expansion of interstitial ECM that is not due to myocardial cell death, but instead due to inflammation, pathological neuroendocrine signaling or diverse types of cardiomyopathies^59,72. Many types of interstitial fibrosis progress slowly and are likely driven by select signaling pathways, such as the local renin-angiotensin system in the heart⁷³, sustained β-adrenergic signaling and catecholamine overload^74,75, and inflammation^76,77. These diverse yet more singular etiologies underlying progressive interstitial fibrosis are arguably more therapeutically tangible than replacement fibrosis. “Perivascular fibrosis” occurs around the vascular adventitia throughout the heart, and is frequently observed in hypertensive disease where it can impair coronary blood flow⁷⁸. Studies are attempting to focus on context-dependent treatments for cardiovascular diseases with unique fibrotic signatures, which accompanies progress in imaging cardiac fibrosis⁷⁹, drug delivery strategies⁸⁰ and use of circulating biomarkers⁸¹ to enable personalized therapeutic strategies.

TGFβ signaling in cardiac fibroblast activation

A variety of growth factors and cytokines induce fibroblast activation, including angiotensin II, endothelin-1, TGFβ, FGF, PDGF, Interleukin (IL)-6, IL-4, IL-1^82–85. However, each of these factors converge on some aspect of the TGFβ signaling pathway to affect CF activity (reviewed in⁸⁶). In fact, TGFβ-induced fibroblast activation has been shown to override activation induced by changes in matrix stiffness⁸⁷. These data, among other reports, have implicated TGFβ as a central factor in the activation and regulation of fibroblasts that leads to the development of cardiac fibrosis.

The precursors of TGFβ ligands (1, 2 and 3) are sequestered in the cardiac ECM network as a latent complex. Cardiac stress, inflammation, or changes in mechanical stretch leads to the release of the ligands where they can signal to CFs and other myocardial cells. While there is speculation that cardiomyocytes might be the major source of TGFβ ligands during cardiac remodeling⁶⁵, it is still unclear which cell type(s) within the heart produce TGFβ that resides within the ECM. A neutralizing antibody treatment that primarily targeted interstitial TGFβ was shown to reduce cardiac fibrosis in an HCM model⁶⁵ and following pressure overload injury⁸⁸. However, the antibody treatment failed to rescue adverse cardiac remodeling in response to pressure overload⁸⁸, suggesting that interstitial cells are the main effector cell type of TGFβ signaling underlying cardiac fibrosis. This was further supported by studies that specifically ablated TGFβ receptors 1 and 2 in CFs, which also attenuated cardiac fibrosis and improved heart function after injury^68–70.

Unfortunately, therapies that target TGFβ signaling to combat cardiovascular disease have not been promising. In several preclinical animal models, systemic administration of both small molecule inhibitors and neutralizing antibodies to TGFβ isoforms or their receptors have resulted in toxicities that include hemorrhage, heart valve degeneration, and histopathologic changes in various organs^89–91. In addition, pirfenidone is an FDA-approved treatment for patients with idiopathic pulmonary fibrosis that is thought to largely act through inhibition of TGFβ signaling and has been shown to decrease fibrosis in several pre-clinical models of cardiac injury^92,93. However, in a recent clinical trial to assess whether it could improve outcomes in a cohort of HFpEF patients⁹⁴, pirfenidone treatment significantly decreased cardiac fibrosis as measured by MRI but did not improve diastolic dysfunction. While TGFβ signaling is undoubtedly critical to the development of cardiac fibrosis and myofibroblast formation in the heart, current clinical studies collectively illustrate a need for alternate targets, or at least more selective therapeutic targets within the TGFβ pathway.

One molecular target downstream of TGFβ activation is MAPK p38, which is involved with CF differentiation^16,60. Indeed, use of select p38 inhibitors have shown reduced cardiac fibrosis and even reduced skeletal muscle fibrosis with disease, suggesting a viable treatment approach in chronic and progressive cardiac fibrotic diseases^95,96. Recently, a study showed that treating mice with the small molecule salinomycin, which is a mesenchymal cancer cell inhibitor, diminished injury following MI and reverted angiotensin II-induced fibrosis⁹⁷. The same study showed that salinomycin decreased activation of human CFs in culture, which are also mesenchymal cells. Other studies have shown that the expression of IL-11 also occurs downstream of TGFβ activation and appears to be specifically related to fibroblast activation, which makes for an attractive therapeutic target^98,99. In addition, the ST2/IL-33 signaling axis has been implicated in cardiac fibrosis through direct signaling at the level of the myofibroblast^100,101. Additional studies are needed to ascertain whether systemic IL-11 inhibition, ST2 inhibition or salinomycin treatment will have anti-fibrotic effects, either with or without TGFβ inhibition. As IL-11 and IL-33 are also implicated in the pathobiology of renal failure, their regulation presents an attractive possibility for treating patients with multi-organ disease such as cardiorenal and chronic kidney diseases⁹⁹.

Mechanical signals in cardiac fibroblast activation

CFs not only process signals from growth factors and cytokines, but also sense and respond to mechanical signals (Figure 1)^52,102. It is well established in vitro that culture substrate stiffness affects the CF transition to myofibroblasts^55,103. While CFs maintain a quiescent phenotype when cultured on a hyaluronic acid (HA) gels that have a physiologically-relevant stiffness, they are activated and express a full spectrum of ECM proteins when cultured on stiffer HA gels that mimic a pathological environment, such as the scar region following MI¹⁰³. Similarly, fibroblasts can be activated through sensed changes in mechanical stretch from their surrounding environment¹⁰³. These in vitro findings match changes of in vivo gene expression profiles post-MI discussed above⁶³.

Several molecular signaling pathways are involved in the ability of CFs to sense and interpret mechanical changes in their environment. For an in-depth discussion, we refer the reader to a recent review¹⁰⁴. Briefly, these pathways include: 1) cell surface integrin complexes that bind specific ECM proteins; 2) regulation of actin and other cytoskeletal proteins, including those within focal adhesions that directly respond to changes in cell shape; 3) ion channels such as TRPV4 and TRPC6 that are critical for the activation of downstream signaling pathways and intracellular kinases; and 4) the nuclear mechano-sensing complexes such as the LINC complex, which directly affects gene expression in response to mechanical stimuli¹⁰⁴.

While changes in ECM stiffness and tension are known to activate fibroblasts through signaling cascades initiated by integrin receptor activation¹⁰⁵, only recently have researchers determined that the Hippo signaling pathway is also essential for the mechanosensing ability of CFs during development and disease conditions¹⁰². Using an epicardial-specific Wt-1-CreERT2 mouse line, deletion of YAP/TAZ¹⁰⁶ or LATS1/2¹⁰⁷ led to defects in the coronary vasculature during development, illustrating that the Hippo pathway is critical for epicardial cell differentiation during development, from which cardiac interstitial fibroblasts are derived. Indeed, Hippo signaling is required for CFs to maintain a quiescent stage in the postnatal heart. Lats1/2 deletion in Tcf21-expressing fibroblasts led to their spontaneous transition to a myofibroblast state, leading to cardiac fibrosis without injury¹⁰⁸. Furthermore, deletion of YAP using the same Tcf21-MerCreMer system attenuated injury-induced cardiac fibrosis either post-MI or in response to chronic AngII/PE treatment¹⁰⁹, while YAP overexpression with adeno-associated virus infection in mice augmented stress-induced fibrosis in vivo¹¹⁰. Collectively, studies suggest that identifying critical components of the Hippo signaling pathway may offer more specific therapeutic targets to quell cardiac fibrosis.

Cardiac ECM directly affects cardiomyocytes in disease

While some evidence suggests that cardiomyocytes and fibroblasts can directly contact each other, which would have to occur through the restrictive basal lamina that surrounds each cardiomyocyte¹¹¹, the majority of studies have focused on the interactions between these two cell types through paracrine signaling or ECM modifications. One primary means of such signaling could be related to the composition of the ECM itself. As discussed earlier, defective collagen production by CFs due to deletion of Hsp47 resulted in less cardiomyocyte hypertrophy with pressure overload⁶. We also previously showed that Postn-null mice have altered ECM characteristics and reduced ability to expand the ECM with pressure overload, which similarly reduced cardiomyocyte hypertrophy with disease-inducing stimulation²¹. Cardiomyocytes directly attach to ECM generated by the fibroblasts through both integrins and the dystroglycan-sarcoglycan complex located within the cell membrane, which provides cytoskeletal support and allows for stress and mechanical sensing¹¹². Indeed, deletion of β1 integrin from the heart results in fibrosis and dilated cardiomyopathy, and these mice were intolerant to pressure overload stimulation¹¹³, similar to mice in which the integrin-linking proteins Talin-1/2 were deleted¹¹⁴. In contrast, overexpression of α7/β1d integrin in the hearts of transgenic mice was protective and reduced the degree of cardiomyopathy following MI injury¹¹⁵. Interestingly, a recent study demonstrated that mechanical cues perceived by cardiomyocytes lead to altered nuclear localization and chromatin reorganization, affecting cell fate¹¹⁶. Thus, cardiomyocytes sense the underlying ECM and its integrity in formulating a disease response, and it is reasonable to speculate that cardiomyocytes and fibroblasts communicate to each other through this dynamic ECM environment and its mechanical properties (Figure 2).

Macrophages in fibroblast activation and ECM composition

While CFs sense cell-ECM interactions as part of their activation cues, they are also activated by inflammatory cytokines and chemokines that are typical of the diseased heart. Several excellent reviews have covered the ways in which immune responses affect fibroblast activation and disease progression^{77,117,118,119}. Immune cells are among the first responders to an injury where they actively mediate clearance of dead cells and initiate inflammatory processes that underlie tissue healing and ECM remodeling (reviewed in¹¹⁹). Among other cell types, macrophages, neutrophils, T cells, B cells, mast cells, and dendritic cells all contribute to these processes^120,121. However, sustained inflammatory states and prolonged activity of select immune cell types continually activate fibroblasts, exacerbating pathological fibrosis and cardiomyopathy. Thus, regulating the inflammatory response and activity of immune cell subtypes has been another therapeutic strategy in attempting to better control cardiac wound healing and pathological fibrosis^77,117,118. In this section, we focus on a discussion of macrophage communication with fibroblasts and cardiomyocytes, and macrophage-derived ECM modifications (Figure 3).

Figure 3. Promising therapeutic targets to mediate CF activation and fibrosis.

Based on recent studies, there are several avenues that can be further explored for their potential to alleviate cardiac fibrosis. It is likely that a combination of these approaches will efficiently combat the progression of fibrosis after injury or disease. 1) The extracellular matrix (ECM) can be targeted with inhibitors of crosslinking and fibrillogenesis, or indirectly through changing cell-derived expression of ECM factors. 2) Cell surface receptors and downstream signaling cascades can be used for targeted pathway inhibition, including signaling cascades like transforming growth factor-beta, p38/MAPK (mitogen-activated protein kinase) and Yap/Taz. 3) Cell surface markers can be used to identify and target unwanted cells, such as FAP recognition of CAR (chimeric antigen receptor) T cells. This approach may also be utilized to target macrophage subpopulations. 4) Post-transcriptional modifiers, such as muscleblind-like 1 (MBNL1) and histone deacetylases (HDACs), could be exploited to alter RNA transcript stability and subsequent cell activation. 5) Cell activation may be modified by regulating epigenetic modifications and transcription factor expression (HMGA1, high-mobility group A1).

Like fibroblasts, cardiac macrophage populations are heterogeneous during development, homeostasis and disease^122–124. The traditional definitions of “M1” and “M2” macrophages as pro- or anti-inflammatory are evolving to include specific cell surface markers that more closely define a particular subtype of macrophage and its functions¹²⁵. For example, cardiac tissue resident macrophages that are of embryonic origin and are CCR2^-, MHCII^low , TIMD4⁺, and CX3CR1⁺ are generally protective to the heart, while recruited circulating monocyte-derived macrophages that are CCR2⁺ (MHCII^high, Ly6C^high) are generally inflammatory and detrimental to the long term healing response^126–129. Different types of macrophages and their cell surface receptors provide unique targets for modulating tissue responses and the activity of fibroblasts.

Depletion of CD206⁺ macrophages attenuated injury after MI in mice, which contributed to altered fibroblast activation (through macrophage-derived IL-1α and osteopontin) and decreased collagen deposition¹³⁰. Yan et al., classified monocytes and macrophage populations based on their level of expression of Ly6C, CD11b and F4/80 each, observing differential upregulation of genes involved with ECM remodeling, including collagen organization¹³¹. Recently, our lab showed that CCR2⁺ monocytes and macrophages are recruited and expand in number within the central core of the MI area of injury in the heart, while the tissue resident CCR2^- CX3CR1⁺ macrophages expand most notably around the periphery of the injury area¹³². Mechanistically, CFs co-cultured with MI-derived CCR2⁺ cardiac macrophages showed induction of markers of myofibroblast formation and activation, whereas co-culturing with cardiac-derived CX3CR1⁺ macrophages did not elicit such an activation response¹³². These results corroborate studies showing divergent functions for different macrophage sub-populations in the heart. For example, tissue resident (CCR2^-) macrophages, previously denoted as “M2”, are associated with enhanced cardiac healing after injury and less fibrosis, while CCR2⁺ macrophages, previously termed “M1” macrophages, are typically pro-inflammatory and promote fibrosis and maladaptation^126,127. Studies suggest that CCR2⁺ macrophages secrete a large amount of pro-inflammatory cytokines, including TGFβ, IL-1β, IFN and IL-6^133,134, which act upon local/recruited CFs. In addition to cytokine and growth factor secretion, macrophage phagocytosis of debris and ECM is critical to the healing process, paving the way for remodeling of the myocardium with angiogenesis¹²⁵. An in-depth discussion of the function of different populations of macrophages in the heart was recently published, and provides additional considerations for ways in which they may affect fibroblast function after injury¹³⁵.

There are also data linking immune cell regulation of the fibroblast and its control of ECM composition to the regenerative response of cardiomyocytes^136–138. This again suggests that myocytes and fibroblasts communicate in the heart through the ECM, but that macrophages are important regulators of fibroblast activity during this process. Indeed, macrophages are known to be required for neonatal heart regeneration¹³⁶. Godwin et al., used clodronate-loaded liposomes to deplete macrophages after cryo injury of axolotl hearts in a model of adult heart regeneration¹³⁷. In this model, depletion of macrophages accelerated CF activation and decreased tissue regeneration, with altered expression of matrix metalloproteases and lysyl oxidase enzymes that was associated with poor ECM architecture¹³⁷. Furthermore, Simoes et al., recently demonstrated that cardiac macrophages themselves may be able to deposit several collagen isoforms, as well as proteins that bind and modify collagen⁶⁷. Collectively, these studies suggest that macrophages are critical to the initial wound healing response through modifying fibroblast activation, impacting ECM composition directly (Figure 1). However, it is still unclear in any of the discussed models whether macrophage-derived ECM deposition and modifications provide long-lasting contributions to the matrix directly, or if they set up the tissue microenvironment to properly attract and activate CFs and other cells.

Emerging therapies to treat cardiac fibrosis

Anti-fibrotic therapies for cardiovascular disease have remained largely unsuccessful. Given ongoing discoveries about the temporal regulation and heterogeneity of cell populations in the heart, successful modulation of fibrosis will likely require a combination of approaches. Figure 3 provides a summary of therapeutic targets of fibroblast activation that are discussed in this Review.

It is reasonable to speculate that modifying the ECM directly may attenuate disease progression. To this point, studies have targeted myofibroblasts or ECM proteins to decrease cardiac fibrosis. Genetically blocking myofibroblast-specific TGFβ signaling during fibrotic progression led to partial injury resolution in a chronic HCM mouse model⁷⁰, suggesting that cardiac fibrosis is reversable under certain conditions. Lineage tracing studies revealed that periostin-expressing myofibroblasts could convert back to a quiescent fibroblast state after cessation of AngII/PE treatment⁶¹. Furthermore, directly targeting fibronectin fibrillogenesis with a small molecule inhibitor, pUR4, attenuated the development of fibrosis and improved heart function following ischemia/reperfusion injury³³. Even considering results from these studies, it is still unclear the extent to which ECM modifications can truly be reversed, which depends on other aspects of the injury or disease. For example, if fibrosis develops to replace dead cardiomyocytes, removing fibrotic tissue without replenishing myocardial cell populations would be detrimental to heart function and could lead to ventricular wall rupture. Likewise, if interstitial fibrosis is providing structural support to an ailing heart, alleviating the fibrosis without providing another source of structure would impact whole organ function. In this regard, future studies should investigate a multifaceted therapeutic approach that addresses cell and organ function in coordination with beneficial ECM composition.

In the past several years, molecular targets have been identified that affect fibroblast activity in the heart. A non-histone chromatin-binding protein, high-mobility group A1 (HMGA1), was shown to regulate CF activation through modulation of FOXO1¹³⁹, which was previously connected to the development of cardiac fibrosis^140,141. Overexpression of HMGA1 in periostin-expressing fibroblasts increased, and knockdown diminished, isoproterenol- and angiotensin II- induced cardiac fibrosis and dysfunction in vivo, and directly modulated CF activation in vitro¹³⁹. Concomitant inhibition of FOXO1 reversed the effects of HMGA1 overexpression, without obvious drawbacks in vitro or in vivo. However, it is not entirely clear how HMGA1 might regulate other essential biologic and cellular processes, including inflammatory signaling in sepsis-induced cardiac injury¹⁴². In humans, HMGA1 was identified as a candidate gene for susceptibility to MI¹⁴³ and as a genetic variant in some patients with type 2 diabetes¹⁴⁴, providing rationale for further investigation into its therapeutic potential.

Another emerging target of CF activation is the family of histone deacetylases (HDACs), which are classically known to modulate gene expression and extranuclear signaling transduction cascades¹⁴⁵. HDAC inhibition with Givinostat improved heart function in several models of diastolic dysfunction^146,147. Recently, Travers et al., showed that Givinostat seems to work by directly modulating CF activation¹⁴⁸. Interestingly, although overt fibrosis did not manifest in their model of HFpEF, fibroblast activation and expression of ECM modifying genes were repressed by HDAC inhibition, leading to improvement in cardiac function. These studies not only point to possible benefits of targeting HDACs to regulate CF phenotypes, but also illustrate advantages to monitoring the status of the cardiac ECM, even outside of overt fibrotic disease.

The BET family of transcriptional coactivators have also been identified as mediators of cardiac pathology. Using small molecule inhibitors to interfere with their function have lessened heart failure in mouse injury models^149,150. Alexanian et al., recently used the BET inhibitor JQ1 to reversibly mediate CF activation in response to pressure overload hypertrophy¹⁵¹. Their experiments showed that JQ1 treatment mitigated cardiac fibrosis and improved survival by decreasing Meox1-mediated CF activation¹⁵¹. Interestingly, the researchers observed phenotypic reversion upon removal of BET inhibition¹⁵¹, again demonstrating the persistence of fibrotic signals independent of traditional CF activation states.

Another promising avenue for controlling CF activation is through transcriptome modulation. Muscleblind-like 1 (MBNL1) is an RNA binding protein that can affect RNA transcript stability and localization, and was recently shown to modulate the CF differentiation in vivo¹⁵². Fibroblast-specific knockout of MBNL1 significantly reduced the fibroblast transition to myofibroblasts after injury, as well as further progression to the matrifibrocyte phenotype that resides within a mature scar^63,152. Interestingly, the authors found that this effect was, at least in part, due to Sox9 transcript stability by MBNL1¹⁵². Furthermore, they again observed a disconnect between absolute gene expression and the progression of fibrosis, as knocking out MBNL1 in periostin-expressing fibroblasts did not alter αSMA expression, but still protected against injury-induced cardiac dysfunction¹⁵². While more research is needed, post-transcriptional modulation of RNA translation provides another targeted approach to curbing fibroblast activation.

A major recent advance in cancer therapeutics is engineering chimeric antigen receptor (CAR)-cytotoxic T cells that cause them to recognize proteins on the surface of cancer cells to effectively target and eliminate them¹⁵³. Aghajanian et al., treated mice with CAR T cells targeted to a subpopulation of CFs that express fibroblast activation protein (FAP), thereby decreasing fibrosis and maladaptive tissue remodeling in response to injury¹⁵⁴. More recently, the same group used CD5-targeted lipid nanoparticles to generate FAP-specific CAR T cells in vivo¹⁵⁵, successfully reducing angiotensin II-induced hypertrophy in mice. While clinical data are needed, a major advantage to treating human patients would be the temporal control of targeting a specific cell population after injury. Additional studies are needed to determine: a) if FAP is indeed the best cellular marker for targeting human CFs with chronic disease; b) the temporal regulation of any CAR T cell target of interest, keeping in mind the heterogeneity of each cell population; and c) full assessment of toxicities associated with depleting activated fibroblasts in other organs^154,155. The development of these and other techniques for reprogramming and drug delivery can easily be applied to different macrophage populations in the heart as well. These exciting advancements in targeting a specific cell population at a specific time provide favorable avenues to alter the course of pathologic tissue scarring and heart repair in humans.

Concluding remarks

The ECM is critical for proper cell communication in the heart. Different cell populations, largely CFs, condition the ECM to provide structural support, growth factors and other signaling proteins that regulate the behavior of all cells within the heart. Because of the inherent heterogeneity in tissue architecture, it is becoming clear that a much larger systems biology approach is necessary to understand both homeostasis and disease progression in the heart. Untangling the multitude of interactions among the ECM and myocardial cell populations will prove beneficial in our efforts to advance therapeutic options to address pathological remodeling of the heart that occurs with disease and aging.