Fibroblast contributions to ischemic cardiac remodeling

Abstract

The heart can respond to increased pathophysiological demand through alterations in tissue structure and function ¹. This process, called cardiac remodeling, is particularly evident following myocardial infarction (MI), where the blockage of a coronary artery leads to widespread death of cardiac muscle. Following MI, necrotic tissue is replaced with extracellular matrix (ECM), and the remaining viable cardiomyocytes (CMs) undergo hypertrophic growth. ECM deposition and cardiac hypertrophy are thought to represent an adaptive response to increase structural integrity and prevent cardiac rupture. However, sustained ECM deposition leads to the formation of a fibrotic scar that impedes cardiac compliance and can induce lethal arrhythmias. Resident cardiac fibroblasts (CFs) are considered the primary source of ECM molecules such as collagens and fibronectin, particularly after becoming activated by pathologic signals. CFs contribute to multiple phases of post-MI heart repair and remodeling, including the initial response to CM death, immune cell (IC) recruitment, and fibrotic scar formation. The goal of this review is to describe how resident fibroblasts contribute to the healing and remodeling that occurs after MI, with an emphasis on how fibroblasts communicate with other cell types in the healing infarct scar ^1–6.

Introduction

The heart can respond to increased pathophysiological demand through alterations in tissue structure and function ¹. This process, called cardiac remodeling, is particularly evident following myocardial infarction (MI), where the blockage of a coronary artery leads to widespread death of cardiac muscle. Following MI, necrotic tissue is replaced with extracellular matrix (ECM), and the remaining viable cardiomyocytes (CMs) undergo hypertrophic growth. ECM deposition and cardiac hypertrophy are thought to represent an adaptive response to increase structural integrity and prevent cardiac rupture. However, sustained ECM deposition leads to the formation of a fibrotic scar that impedes cardiac compliance and can induce lethal arrhythmias. Resident cardiac fibroblasts (CFs) are considered the primary source of ECM molecules such as collagens and fibronectin, particularly after becoming activated by pathologic signals. CFs contribute to multiple phases of post-MI heart repair and remodeling, including the initial response to CM death, immune cell (IC) recruitment, and fibrotic scar formation. The goal of this review is to describe how resident fibroblasts contribute to the healing and remodeling that occurs after MI, with an emphasis on how fibroblasts communicate with other cell types in the healing infarct scar ^1–6.

Pathological and physiological cardiac remodeling

The heart is capable of remarkable phenotypic plasticity, which allows for the accommodation of alterations in pathophysiological demand. Increased demand can arise due to acute and chronic disease states (ischemic or non-ischemic pathological cardiac remodeling), or in response to pregnancy or extreme aerobic exercise (physiological cardiac remodeling) (see Fig. 1).

Figure 1: Stress-induced cardiovascular changes.

Physiological stress via exercise causes global hypertrophy that does not impede cardiovascular function or result in fibrosis. In exercise models, CM length and width is increased proportionally by longitudinal hypertrophic growth. Pathological stress can induce concentric hypertrophy where myocyte width increases greater than myocyte length (PO) or can induce eccentric hypertrophy (dilation) where myocytes are stretched, and their length increases much greater than their width (MI or HF). Healthy or exercised hearts lack fibrosis, but in PO models, interstitial and perivascular fibrosis occur throughout the heart while in MI models, large areas of CM death are repaired by replacement fibrosis. MI, myocardial infarction; PO, pressure overload.

Ischemic cardiac remodeling

Ischemic remodeling is often caused by a MI originating from an atherosclerotic plaque or thrombotic blockage in the coronary vasculature, which leads to localized ischemia and widespread CM death. Through well-classified phases of healing, the necrotic myocardial tissue within the infarct zone is replaced with a fibrotic scar called replacement fibrosis. Replacement fibrosis may provide structural support to the remaining healthy myocardium in the remote zone ⁷. Due to localized death of CMs, the development of a fibrotic infarct zone is associated with dilative remodeling and a decrease in heart function. ECM deposition results in stiffening of the ventricle and impedes pump function, leading to decreased heart function. Classical features of ischemic remodeling include the increase of CM length at the expense of width, resulting in thinned, elongated ventricles with larger surface area (termed eccentric remodeling) ^{7, 8}. Development of cardiac arrhythmia, which is a common cause of sudden cardiac death, is also a consequence of replacement fibrosis due to loss of conductance of the electrical signals that govern proper heart contraction. The deleterious effects of this natural fibrotic remodeling response (which is strictly required to prevent rupture and maintain the structural integrity of the heart following the development of an infarct) dictate the necessity of developing strategies to ameliorate, but not eliminate, the fibrotic response to MI.

Ventricular pressure overload induced cardiac remodeling

Left ventricle pressure overload (PO) is a form of non-ischemic remodeling characterized by CM hypertrophy, loss in cardiac function, and development of fibrosis. PO is often initiated by chronic hypertension, in which increased systemic blood pressure generates a greater resistance to the flow of blood from the heart. In response to hypertension, left ventricle hypertrophic remodeling occurs to compensate for increased wall stress. Hypertrophy in response to PO causes CMs to disproportionately increase in width relative to length, resulting in a universal thickening of the cardiac wall (termed concentric remodeling) ^{8, 9}. Increased pressure within coronary arteries also adds stress to surrounding myocardial tissue; therefore, fibrosis initially develops in patches in perivascular regions to increase wall support of large and small vessels. As PO increases and the CMs reach a maximum width-to-length ratio, CM apoptosis can occur that precipitates a more widespread replacement of myocardium with interstitial collagenous networks intended to strengthen the cardiac wall (termed interstitial fibrosis) ^{1, 6, 8, 9}. Over time, the thickened, increasingly fibrotic myocardium is unable to relax and expand sufficiently to allow for proper diastolic filling, leading to decreased heart function in a similar fashion as described for MI. In contrast to MI, there is currently less evidence for distinct inflammatory and resolution phases in PO^{10, 11}. Also, in contrast to MI, the development of early fibrosis is not localized, but diffused throughout the myocardium in predominantly vascular areas. Work from our group also showed that CFs from MI hearts and PO hearts have distinct genetic expression signatures not only from sedentary control animals, but also from each other - strongly suggesting that development of anti-fibrotic strategies in PO might take different forms than those most effective in MI ^{12, 13}.

Physiological cardiac remodeling

In contrast to pathological cardiac remodeling, volume overload in physiological remodeling in response to exercise or pregnancy is generally correlated preservation of heart function, or even improved cardiac output ^{14, 15}. To accommodate the increased demand for oxygen and nutrients typical in sustained endurance exercise, CM size is increased by longitudinal hypertrophic growth (increase in width-to-length ratio). While this can lead to concentric hypertrophy, physiological CM hypertrophy is not typically accompanied by fibrosis ^{8, 12, 16, 17}, and the cardioprotective benefits of exercise are reported to ameliorate the progression of pathologic cardiac remodeling ^18–22. Multiple studies have shown that chronic exercise induces transcriptional signatures in the myocardium that are consistent with improved cardiovascular function ^23–26.

Timeline of ischemic cardiac remodeling

Pathological remodeling occurs as a continuum, which has been described in detail following MI (depicted in Fig. 2). The phases of post-MI cardiac repair are broadly defined by: 1) cell death and the initiation of an inflammatory response; 2) resolution of inflammation and stimulation of fibroblast proliferation; and 3) Scar formation and maturation. These major phases are comprised of various partially overlapping and interconnected processes, which will be the focus of the remainder of this review, with a focus on fibroblast contributions to the timing and extent of inflammation and scar formation in ischemic-injury.

Figure 2: Time-course of healing post-MI in mice.

Immediately following an occlusion of a coronary blood vessel, the Necrotic Phase occurs with downstream CM necrosis in response to lack of nutrients. Next, the Inflammatory Phase occurs with a surplus of invading ICs to clear cellular debris and promote pro-inflammatory signaling. During this phase, angiogenesis also begins. CFs begin to proliferate during the Reparative Phase, which also induces fibroblast activation and a shift from pro-inflammatory to pro-reparative IC interactions. Lastly, the Maturation Phase is identified by large amounts of ECM deposits, collagen cross-linking, and the transition of AFs to a matrifibrocyte state. AF, activated fibroblast; CF, cardiac fibroblast; CM, cardiomyocyte; EC, endothelial cell; ECM, extracellular matrix; IC, immune cell.

Defining the cardiac fibroblast

CFs originate from a population of progenitor cells that reside within the epicardium, a single cell layer of mesothelium that lines the heart ²⁷. Epicardium-derived progenitor cells undergo epithelial-to-mesenchymal transition mid-gestation, and differentiate into various cardiac interstitial lineages, including CFs, which colonize the cardiac interstitium and provide structural support for the growing myocardium ^{28, 29}. It is generally accepted that the primary cellular source of ECM in the heart is the resident cardiac fibroblast. Historically, the fibroblast response to cardiac insult has been described as a phenotypic switch, where quiescent resident fibroblasts “transform” into activated myofibroblasts that become contractile and secrete copious amounts of ECM. However, recent transcriptional analyses of non-CM cells in the interstitium of injured or uninjured hearts has identified various distinct CF subtypes and challenged the binary definition of fibroblasts as quiescent or activated. It is becoming more evident that fibroblasts may exist as a heterogeneous and fluid population of cells that contribute varied roles to post-MI injury repair ^30–33. One recent study using single cell RNA-sequencing identified at least 10 CF populations, some of which expressed clearly identifiable gene expression profiles and some that were less distinct suggesting various intermediate phenotypes. This study described a population of collagen triple helix repeat containing 1 (Cthrc1)-expressing AF that emerges after MI, which the authors propose is especially important for the healing process ³¹. Another study using single cell RNA-sequencing identified 11 subpopulations of CFs; in addition to the standard “myofibroblast” definition, they identified 3 transcriptionally different populations (MYO-1, MYO-2, and MYO-3). MYO-2 and MYO-3 supported pro-fibrotic programs while MYO-1 was anti-fibrotic challenging the notion that all myofibroblasts promote pathological fibrosis ³⁰. Finally, to expand on the heterogeneity of myofibroblasts, Ceccato et al. identified different secretory profiles from myofibroblasts depending on the method of activation (physical or biochemical) ³². In vivo, it’s likely that these signals, in addition to others, converge to result in various additional secretory profiles. For the simplicity of this review, we will identify injury-responsive fibroblasts as activated fibroblasts (AFs).

While no single marker is considered fibroblast-specific, CFs are defined by the expression of various ECM molecules including collagens and fibronectin, and the expression of cell surface markers, including platelet-derived growth factor alpha (See Table 1). Recent advances in genome editing have also enabled the generation of genetic lineage tracing strategies based upon the expression of Cre-recombinase from promoters that drive fibroblast-enriched gene expression. Several of the most efficacious lineage tracing strategies make use of Cre knockin at genes that are enriched in the epicardium that restrict to fibroblasts later in life, including at the Tcf21, Wilms Tumor protein 1 (Wt1), and Tbx18 alleles; additional transgenic strategies include Cre driven by a fragment of the Col1a2 promoter in all CFs, or the labeling of AFs using a tamoxifen inducible Postn-Cre mouse line ^34–37. The identification of various CF-specific markers has significantly accelerated the study of CF contributions to fibrotic remodeling, as well as the understanding of the genetic processes that underlie fibroblast activation. We will now explore in more depth how CFs behave and respond to environmental ques throughout phases of healing post-MI.

Table 1:

Genetic markers of CF that have been used to restrict desired expression of lineage tracing reporters or create tissue-specific knockout/overexpression models. For a more comprehensive list of markers that is not limited to CF, the subject is thoroughly reviewed in ³⁷.

Marker	Name	Role
Col1a1	Collagen type 1, alpha 1	Structural component of type I collagen, major component of connective tissue and ECM 38.
Col1a2	Collagen type 1, alpha 2	Structural component of type I collagen, major component of connective tissue and ECM 39.
FSP-1 (S100A4)	S100 calcium-binding protein A4	Associated with filaments, no longer considered specific for CF 40, 41.
PDGFRα (CD140a)	Platelet-derived growth factor receptor alpha	Receptor tyrosine kinase, transduces survival, proliferation, and migration responses. Expression in heart is specific to CF – high levels in quiescent CF, lower in AF 38.
Postn	Periostin	Matricellular molecule (see text). Expression in heart is specific to AF – very low levels or absent in quiescent CF 35.
Tbx18	T-box transcription factor 18	Transcription factor. Early marker of epicardium, Tbx18-expressing progenitor cells give rise to subset of CF as well as myocardial lineages 42, 43.
Tcf21	Transcription factor 21	Transcription factor, relevant to development in heart, lung, kidney, and spleen. Expression in heart is specific to CF – high levels in quiescent CF, lower in AF 27, 35, 44.
Tie2 (CD202b)	Angiopoietin-1 receptor, TEK	Tyrosine kinase receptor, also expressed on endothelial cells and monocytes. Tie2-expressing endocardial cells yield a specific, small subset of CF during development but do not contribute to remodeling in adult phases or injury 42.
Wt1	Wilms’ tumor protein	Transcription factor. Regulates heart development and epithelial-mesenchymal transition of epicardial cells, some of which contribute to resident CF population 45.

General cardiac fibroblast contributions to post-MI remodeling

In the adult heart, CFs exhibit considerable phenotypic plasticity, which allows for a regulated response to various pathophysiologic cues^{34, 36, 46, 47}. Following MI, CFs deposit an ECM rich scar as part of the healing process and also express a number of secreted cytokines, chemokines, and growth factors ^{48, 49}. While the canonical role of CFs remains the generation of ECM and subsequent contributions to scar formation and healing, fibroblasts have also been called sentinels of tissue injury due to a unique capacity to communicate with and alter the function of neighboring cells through both direct interactions and paracrine signaling (Fig. 3) ⁵⁰.

Figure 3: Summary of CF interactions post-MI in mice.

Throughout the healing continuum post-MI, CFs interact with CMs, ECs, and ICs. In each phase of healing, different interactions become prominent that contribute to the microenvironment and signaling from CFs to other cardiac cells. During the necrotic and inflammatory phases, DAMPs that originate from dying CM signal the danger response to surrounding cells. Pro-inflammatory signaling is a major component of CFs and ICs interactions during the early phases of healing, and other pro-angiogenic and cell adhesion proteins are also secreted. During the proliferative and reparative phases, enriched gene expression programs include pro-resolution cytokines and matrix building molecules. Positive feedback from ECM tension and growth factors contributes to the progression of fibroblast activation and pathological fibrosis. Lastly, during the maturation phase, collagen cross-linking and modifying factors and chondrocyte factors are secreted from a specialized fibroblast called the matrifibrocyte. ANGPT1, angiopoietin 1; CCL5, C-C motif chemokine ligand 5; CM, cardiomyocyte; CHAD, chondroadherin; CLIP1, cartilage intermediate layer protein; COMP, cartilage oligomeric matrix protein; DAMPs, damage associated molecular patterns; EC, endothelial cell; ECM, extracellular matrix; FN, fibronectin; GM-CSF, granulocyte-macrophage colony-stimulating factor; IC, immune cell; IFNγ, interferon gamma; IL, interleukin; LOX, lysyl oxidase; MFGE8, milk fat globule epidermal growth factor 8; MIF, macrophage migration inhibitory factor; MMPs, matrix metalloproteinases; POSTN, periostin; TGFβ−1, transforming growth factor β−1; TIMPs, tissue inhibitors of MMPs; TNFα, tumor necrosis factor α; TSP, thrombospondin; VEGF, vascular endothelial growth factor.

Scar formation –

Increased levels of anti-inflammatory/pro-fibrotic factors such as TGF-β1, and changes in the biomechanical properties of the cardiac muscle, promote the activation of CFs into AFs, characterized by the robust expression of genes that encode contractile proteins such as (e.g. Acta2 and Tagln), and ECM components (e.g. Postn, Fn, and Col1a1).

Angiogenesis

Specifically, fibroblasts are reported to stimulate endothelial cells (ECs) to promote angiogenesis and revascularization through their secretions of angiopoietin 1 (ANGPT1) ⁵¹ and vascular endothelial growth factor (VEGF) ⁵². ANGPT1 has been reported as an essential factor in regulating the diameter of developing vessels ⁵³. Mesenchymal-endothelial-transition (MEndoT) has also been described suggesting that CFs are active participants in neoangiogenesis, but their overall contribution may be limited ^{54, 55}.

Immune cell polarization

ICs are recruited and polarized in the early phases of MI healing. Monocytes and macrophages respond to the initial ischemic injury by adopting a pro-inflammatory phenotype, sometimes referred to as M1 ⁵⁶. But as healing occurs, ICs can transition through a phagocytotic, proliferative phenotype and later adopt a pro-reparative profile (also called M2) that induces fibroblast activation via secretion of TGF-β1 ⁵⁶. These transitions can be triggered by cytokines (such as IL-1β) that are secreted by CFs, which are the predominant cell type responsible for activation of the inflammasome ⁵⁷.

Immune resolution

CFs promote resolution of the immune response by secreting anti-inflammatory prostaglandins in the later phases of remodeling post-MI ^{58, 59}. Additionally, increased secretions of IL-10 from CFs during the pro-resolving phase of MI healing can alter macrophage phenotypes to have a pro-resolution signature ⁶⁰.

Timeline of CF response following MI

Necrotic Phase – Death Sets the Stage

Following occlusion of a coronary artery, the necrotic phase immediately commences as ischemic myocardium begins to die, and lasts about a day in mouse MI models ⁷. Loss of cardiomyocyte membrane integrity during cell starvation and death leads to release of cytoplasmic content, such as heat shock proteins, high mobility group box 1 protein (HMGB1), histones, S100 proteins, ATP, and mitochondrial DNA, into the interstitial space. The inappropriate localization of these intracellular molecules, or the secretion of certain ECM proteins such as tenascin C and Fibronectin (FN)-extra domain A (EDA), serve as danger signals that are collectively called damage associated molecular patterns (DAMPs) or alarmins (reviewed in ⁶¹). For example, HMGB1 has been shown to be released from the chromatin in cells undergoing necrotic cell death. Once released, extracellular HMGB1 interacts with antigen presenting cells via pattern recognition receptors (PRRs), such as dendritic cells, causing them to mature and promote sterile inflammation ^62–64. PRRs are a diverse group recognizing many DAMP signaling molecules and encompassing Toll-like receptors (TLRs), nucleotide-binding oligomerization domain leucine rich repeat and pyrin domain-containing proteins (NLRPs), P2X/P2Y purinoreceptors, and interleukin (IL) receptors, amongst others. These processes couple the necrotic death of CMs in response to infarction to the activation of the inflammasome, which is the multiprotein organization system responsible for the coordinated induction of the innate immune response. One of the major consequences of inflammasome activation is the proteolytic cleavage and secretion of ILs-1α/β and −18, which is associated with the initiation of the pro-inflammatory pyroptosis programmed cell death pathway ^65–68. In the context of MI, this implies that DAMP release from damaged CMs propagates death signaling in neighboring cells via paracrine signals. Indeed, gene deletion and pharmacological inhibition studies generally indicate that induction of PRRs and the inflammasome leads to maladaptive remodeling after MI ^69–71.

PRRs are also expressed on the surface of CF, which contribute to the pro-inflammatory response and the initiation of fibrosis downstream of DAMP-dependent signaling ^72,73. Cleaved and activated IL-1α/β released from membrane pores in dying CMs interacts with and activates the extracellular domains of IL-1 receptors (IL-1Rs) on CF ⁷⁴. Upon activation, IL-1Rs induce downstream myeloid differentiation primary response 88 (MyD88)-dependent signaling that in turn signal through the IL-1R-associated kinases (IRAKs) and leads to the activation of the transcription factor NF-κB. IL-1Rs are not the only proteins that make use of this signaling module – MyD88 is obligately used by virtually all PRRs in mice, though there is evidence that redundancy exists for MyD88 in the human immune response ^{75, 76}. In addition to IL-1α/β, HMGB1, as mentioned above, is an example of a nuclear DAMP secreted from necrotic CMs that signals through PRRs. Both of these examples of CM DAMP secretion result in induction of chemokine synthesis in the CF via the activation of NF-κB-dependent transcription, shifting the CF from quiescence to active participation in the inflammatory phase, described in more detail below.

Inflammatory Phase – Recruitment and Cleanup

The inflammatory phase begins a day after the initial infarction and lasts until roughly day 4 ⁷. At this point, the chance of survival for tissue experiencing prolonged oxygen and nutrient deprivation is minimal. Neutrophils are the first responders in the necrotic and early inflammatory phase, quickly infiltrating the infarct zone to clear cellular debris and propagate pro-inflammatory signals via degranulation, which releases appropriate chemical cues to promote the coming inflammation phase in a spatiotemporally controlled manner ^{5, 10, 77}. The inflammatory phase is also characterized by the acute infiltration of pro-inflammatory monocytes and subsequent activation of tissue macrophages, which contribute to the clearance of cellular debris, as well as alerting cells in more remote areas, including CFs, to the presence of damage.

Once CFs develop a pro-inflammatory phenotype, they produce and secrete molecules that contribute to the pro-inflammatory microenvironment around the infarct. The post-MI secretome is comprised of cytokines such as interleukins (IL-1β, IL-6, IL-9, IL-11, and IL-12), tumor necrosis factor (TNF)-α, interferon (IFN)-γ, and granulocyte-macrophage colony-stimulating factor (GM-CSF); as well as chemokines including the classically pro-inflammatory CCL2 and CCL5 ^78–82. All of these factors are reported to be upregulated in CF in the inflammatory response ³⁴. In fact, autocrine feedback of IL-11 dependent signaling in CF is an essential step in the synthesis of matricellular proteins ⁸⁰. Deletion of the IL-11 receptor (IL-11RA) in mice ameliorated fibrosis in hypertension and PO-induced cardiac remodeling; future studies are required to evaluate the efficacy of targeting IL-11 in ischemic remodeling. Additionally, macrophage migration inhibitory factor (MIF) has also been shown to be preferentially upregulated in and secreted from CFs in the ischemic heart ⁸³. MIF is an inflammatory cytokine that interacts with the membrane-bound antigen-presenting protein CD74, triggering pro-survival and pro-inflammatory pathways that are in part mediated by ERK signaling^{84, 85}. Interestingly, MIF can act via paracrine signaling to amplify MIF secretion rate in adjacent CFs, inducing the expression of genes encoding pro-adrenomedullin and CCL2 - thereby propagating CF-derived inflammation in addition to promoting angiogenesis and vasodilation via interaction with ECs ⁸³.

Another example of a CF-secreted factor that acts preferentially on other CFs is milk fat globule epidermal growth factor 8 (MFGE8 or lactadherin). MFGE8 is a cell adhesion protein that was first identified as a connective factor in arteries between the smooth muscle cells and elastin ⁸⁶. Known properties of MFGE8 signaling involve a phosphatidylserine-binding domain, which allows it to bind to and opsonize apoptotic cells, as well as an integrin-binding motif that facilitates angiogenesis ^{87, 88}. MFGE8 was found to be induced in certain AF populations between 1 and 3 days post-MI, and was found to have a paracrine effect on surrounding CFs that significantly increased their native ability to perform apoptotic cell engulfment - normally the milieu of macrophages ⁸⁹; however, considering the abundance of professional phagocytes in the infarcted heart, the relative contributions of cell engulfment by CFs is most likely minimal. This property of MFGE8 signaling between CFs is particularly notable, because it most likely helps the transition from the inflammatory phase into the pro-resolution phase through increasing cell engulfment and reducing inflammatory signals.

During the initial injury response, CFs have been shown to contribute to angiogenesis and neovascularization ^{54, 82}. CFs at 3 days post-MI are reported to adopt a pro-angiogenic gene expression profile, including the significant upregulation of VEGF, which corresponds to the timing of neovascularization ^{82, 90}. In addition to pro-angiogenic secreted factors at early timepoints following MI, CFs may also undergo MEndoT. At 3 days after ischemic injury, vascular identifiers (such as VECAD, eNOS, claudin-5, and occludin) are upregulated in a subset of CFs, and incorporation of Col1a2-lineage traced CFs into vascularization in the border zones was observed ⁵⁴. However, the relative contribution of CFs into newly formed coronary arteries by MEndoT is debated, and more recent work has shown that the majority of new vessels are constructed from pre-existing endothelial cells ⁵⁵.

Proliferative phase - Cardiac fibroblasts accumulate

The early phase of a cardiac insult is generally characterized by a robust increase in CF numbers. This proliferative phase overlaps with both the inflammatory and reparative phases of post-MI remodeling (described below) before the cell cycle slows^{13, 91}. Interestingly, a recent study suggests that CF proliferation precedes and is distinct in time from fibroblast activation and subsequent ECM deposition ^{13, 92}. Our recent work complements this finding, indicating that increased CF proliferation observed upon Sprr2b/MDM2 dependent p53 degradation does not drive the AF phenotype ¹³. Regardless, inhibition of CF proliferation in transgenic mice via cell-specific induction of Cdkn1a/p21 ameliorated scar formation following MI, likely by limiting the number of CFs that secrete ECM ⁹³. A better understanding of the mechanisms that control the timing and extent of CF accumulation, and whether proliferation and fibroblast activation are coordinated through complementary signal transduction pathways, may uncover novel anti-fibrotic strategies.

Immune Resolution / Reparative Phase – Building a Scar

The immune resolution/reparative phase, identified by the presence of pro-resolution monocytes, peaks around 7 days post-MI and continues until about day 15 ⁹⁴. This phase is marked by the subsidence of pro-inflammatory factors and the emergence of anti-inflammatory, pro-reparative, and pro-fibrotic factors - many of the latter being matricellular proteins that are incorporated into the surrounding matrix to promote cardiac structural integrity. During the course of this phase, pro-inflammatory ICs are superceded by the emergence of pro-resolution ICs, while CFs transition into a contractile AF phenotype and begin to synthesize ECM in response to signaling cues from their environment (both directly and indirectly via interaction with other cell types). For instance, macrophages are an abundant source of anti-inflammatory, resolution-promoting cytokines such as IL-10. IL-10 infusion has been shown in vivo to have a beneficial effect on the post-MI remodeling process; however, it was also discovered that the effects IL-10 has on fibrosis are not from direct interactions, but rather from macrophages alternatively polarized by IL-10 that then interact with CFs ⁶⁰.

CF activation

During the resolution phase, some of the most prominent factors that directly alter CF phenotype belong to the transforming growth factor beta (TGF-β) superfamily. TGF-β superfamily members are secreted from most cardiac cells, but the prominent source post-MI in mouse models has been found to be from macrophages ⁹⁵. Interestingly, while members of the TGF-β superfamily (including TGF-β1/2/3, bone morphogenetic proteins or BMPs, growth differentiation factors or GDFs, activins, and inhibins) make use of myriad serine/threonine kinase receptors and adaptor signal transduction proteins, their signaling primarily converges downstream to activate SMAD proteins and translocate them to the nucleus. As a canonical example, extracellular TGF-β1 binds to TGFBR-II, which causes phosphorylation of the paired TGFBR-I. A zinc double finger FYVE domain-containing adaptor protein (e.g. SARA) then facilitates recruitment of one of the receptor-regulated SMADs (the specific species is ligand-dependent, and for TGF-β1 are SMAD2/3) to the activated TGFBR-I, which then phosphorylates the SMAD on a serine residue. Finally, this phosphorylated SMAD binds a co-SMAD (for TGF-β1, SMAD4), which moves to the nucleus to interact with DNA and influence the transcription of genes whose promoters contain SMAD-responsive elements. Several key factors in the transition from quiescent CF to AF phenotype are regulated by SMAD signaling: Acta2, Tgfb1, Pak2, Col1a1/1a2, Col3a1, Mmp3/14, Timp1, Rhoa, Serpine1, Itga1/2/5, Itgb3/8, Dsp, Scx (itself a transcriptional factor that interacts with and augments SMAD signaling), and Postn serve as salient examples of the diversity of SMAD signaling targets, which encompass (but are not limited to) intracellular kinases and cell cycle control proteins, secreted circulating factors, contractile and cytoskeletal proteins, cell-cell adhesion mediators, and ECM components and modulators ^96–101. Speaking to this high complexity, recent efforts continue to focus on building a comprehensive database of SMAD signaling targets using high-throughput strategies ¹⁰².

While TGF-β superfamily/SMAD signaling is a major driver of the contractile and secretory AF phenotype, it is not the only transcriptional pathway governing this process. First, non-canonical TGF-β signaling contributes to the development of fibrosis after ischemic injury in mice, evidenced by the amelioration of scar formation in mice lacking p38 (mitogen-activated kinase 14) in fibroblasts ¹⁰³. Permanent ligation of mice lacking p38 in fibroblasts led to 100% ventricular rupture, presumably due to insufficient replacement fibrosis. In addition to SMAD and p38 dependent signaling, TGF-β also induces the contractile phenotype via the activation of the serum response factor (SRF) pathway, which acts as a key determinant of differentiation and proliferation responses to stress stimuli ^{104, 105}. A prime example of the cooperativity between TGF-β and biomechanical gene regulation is the Acta2 promoter, which contains both SMAD-binding and SRF-binding elements. Alterations in biomechanical tension that lead to filamentous (F) actin polymerization facilitates the nuclear accumulation of myocardin-related transcription factor (MRTF)-A, which interacts with SRF and promotes transcription of the contractile gene program, including Acta2 ^{49, 106, 107}. Therefore, the importance of MRTF-A is revealed by studies that demonstrate an abrogation of AFs and fibrosis following MI or in aging ^{106, 108, 109}.

Collagen and elastin deposition

Accompanying the assumption of the contractile phenotype in AFs is a marked and sustained increase in transcription of genes coding for collagens, other ECM-associated structural proteins, matrix metalloproteinases (MMPs) and tissue inhibitors of matrix metalloproteinases (TIMPs). The synthesis of new matrix materials can be attributed primarily to AFs during the resolution phase, with limited evidence that other cell types are major sources of ECM (covered in more detail later) ^{7, 42, 101}. Collagens, representing at least 42 genes coding for 28 protein products, represent possibly the best-known and most diverse family of ECM protein products synthesized by AFs ¹¹⁰. Broadly, collagens can be classified as fibrillar or non-fibrillar based on their molecular structure. Fibrillar collagens, such as the ubiquitous COL1A1, are associated with integumentary, skeletal, and organ structure, while non-fibrillar collagens tend to provide parallel structural support to fibrillary collagens or (in the case of COL4A) are synthesized in sheet structures to create the foundation for the basal lamina ¹¹¹. Fibrillar collagens, which make up at least 90% of the total collagen mass in mammals, are initially synthesized as single strands that self-assemble into triple helical structures termed procollagens ¹¹². Procollagens are then hydroxylated on proline and lysine residues in a vitamin C-dependent manner by prolyl-4-hydroxylases (P4HA/B) or lysyl oxidase (LOX), either intracellularly or extracellularly, and subsequent processing steps lead to the formation of the mature collagen fibril ^113–115. The measurement of collagen (or its characteristic component hydroxyproline) has long been used as a surrogate for the degree of fibrotic remodeling in the heart, and circulating collagen type I and III propeptides have been shown to be a biomarker of multiple types of cardiac pathology in both mice and human patients ^116–119. Elastin (ELN) is a highly conserved ECM protein that is in fact absent from MI scar tissue in either its molecular or compound elastic fiber format, but has been shown to be advantageous to ventricular functional recovery post-MI in a rat model if introduced via gene therapy – likely via increasing the elasticity of the rigid MI scar and improving mechanical properties ^120–122.

Proteoglycans

While collagens certainly are the main component of the AF-derived ECM, other proteins and macromolecules synthesized by AFs are equally important to ECM biology and the biology of a scar during post-MI remodeling. Proteoglycans such as heparan sulfate are also crucial to the post-MI scar formation process, as well as ventricular functional recovery ¹²³. Heparan sulfate chains bind angiogenic growth factors present in the ECM such as VEGF-A, and their ability to retain these factors is governed at least in part by a pair of endosulfatases (Sulf1/2) that regulate how much VEGF-A is available in the ECM to promote angiogenesis – a crucial step in the revascularization and recovery of the post-MI myocardium ¹²⁴. Interestingly, this study showed AFs preferentially expressed Sulf1 – implicating AFs not only in heparan sulfate synthesis, but also in its post-translational regulation. Another chief component of the cardiac ECM synthesized by AFs is the glycosaminoglycan hyaluronic acid, which is the product of three hyaluronan synthases (Has1/2/3) located on the cell membrane ¹²⁵. Hyaluronic acid is synthesized in a wide variety of molecular sizes, which have been demonstrated to have diverse properties and participate in both inflammation and resolution – as well as aiding in the recovery from MI both natively and when implanted as a biomaterial scaffold ^126–130. It is interesting to speculate that strategies to promote or recapitulate the native ability of CFs to synthesize larger amounts of these potentially protective matrix molecules may be cardioprotective in the context of MI.

Other matricellular proteins

CFs and other cell types also interact with the ECM via myriad non-collagenous matricellular proteins - of which fibronectin (FN), periostin (POSTN), and thrombospondins (TSPs) are salient examples in MI biology. FN is a large glycoprotein that interacts specifically with membrane-bound integrins as well as collagen and heparan sulfate ¹³¹. FN has widespread contributions within cardiac ECM: integrin binding (with the classical example being α₅β₁, also termed the FN receptor), collagen binding, heparin binding, and interaction with fibrin (a major initial component of the MI scar) ^{132, 133}. FN is found in one of three isoforms based on alternative splicing, and the extra domain A/B-containing isoforms that predominate during fetal development are re-expressed following MI ^134–137. Interestingly, FN extra domain A (EDA) is thought to localize TGF-β by acting with the latent TGF-β-binding protein-1 (Ltbp1) in vitro ¹³⁸. FN-EDA has been shown to stimulate fibroblast activation and collagen production in cutaneous fibrosis ¹³⁹, and genetic deletion of FN-EDA or pharmacological inhibition of FN polymerization, reduces ischemia-reperfusion injury ^140–142.

Another matrix molecule with similarly complicated biology in the heart is the matricellular protein, POSTN, which has been shown to be highly expressed in the embryonic mouse heart but is largely absent from the uninjured post-natal myocardium (reviewed in depth in ¹⁴³). POSTN has a number of interacting roles in the heart that recall those played by FN: namely integrin binding (primarily a_vb₃ and a_vb₅, the canonical vitronectin receptors), fibrillary collagen binding, and FN binding – consistent with a role in promotion of cell motility by coupling intracellular integrin signaling to changes in the ECM ^{144, 145}. Postn is perhaps best known in recent literature as a reliable, specific marker of AFs – adult resident quiescent CFs express virtually no Postn, but post-injury the CF to AF transition is associated with a significant and persistent increase in Postn expression, which is recapitulated in a transgenic mouse line expressing tamoxifen-inducible Cre recombinase at the Postn locus ^{12, 13, 35, 92}. In the mouse MI model, gain of function in Postn did not cause noticeable increases in fibrosis, but Postn-knockout mice were more likely to experience cardiac rupture due to inferior scar formation during the resolution period ^146–148. Although Postn is a recently well-accepted marker of AFs, literature has suggested that it may only be expressed by a subset of CFs marked by Tcf21 expression ⁸². Additionally, POSTN could certainly be derived from other cell types after cardiac insult, such as smooth muscle cells or macrophages ^{56, 82}.

The final family of non-collagenous matricellular proteins to be discussed here are the TSPs, a five-member family of matrix-binding proteins, which have been thoroughly reviewed elsewhere ^{149,9, 143, 150}. While TSP-1/2 have been shown in multiple studies to discourage angiogenesis, more recently TSP-4 has been implicated in the promotion of neovascularization ^151–154. TSP-4 was also shown to directly protect the heart by promoting the nuclear accumulation of ATF6α, an ER stress response transcription factor ¹⁵⁵. Furthermore, TSP-3 (once thought to be absent from the heart) was recently found to destabilize sarcolemmal structure and inhibit integrin attachment complex formation due to a difference in molecular structure from the closely related TSPs-4/5 ¹⁵⁶. Interestingly, by introducing a mutation to incorporate the integrin-binding domain contained in TSPs-4/5 into the TSP-3 protein, the negative effect of TSP-3 during myocardial remodeling was completely reversed. Based on these studies, the TSP family is a particularly attractive candidate for targeted therapies due to the diverse roles played in the post-MI heart.

Matrix metalloproteinases

Discussion of the remodeling ECM is not limited to the components of its synthesis – the ECM, particularly in the response to cardiac insult, is a dynamic environment that is constantly being synthesized and degraded. To this end, one of the most relevant families of matrix-modifying proteins are the MMPs, a large group of calcium-dependent zinc-incorporating enzymes that were first characterized as degrading agents of ECM molecules but have also been implicated in a number of other cellular functions such as cytokine processing and inactivation as well as receptor cleavage ^{157, 158}. MMPs are initially synthesized in a latent form, but are activated by cleavage due to other proteases (including MMPs), by structural changes in the pro-domain, or by reactive oxygen species altering the cysteine residue that maintains the latent configuration ^{159, 160}. As befits such a diverse family of proteins, roles played by MMPs are context-dependent and are not limited to the extracellular space ¹⁶¹. As reviewed recently, MMPs serve in both their native enzymatic capacity and as serum biomarkers in heart failure patients ¹⁶². MMPs are primarily responsible for degradation of the ECM, which in MI has been shown to promote chamber dilation ^{163, 164}. Of note, the well-characterized MMPs-2 and −9 (which degrade both gelatinous and fibrillary collagens) have clinically relevant polymorphisms that are associated with poor cardiovascular outcomes across a variety of ethnic backgrounds – suggesting a central role in cardiovascular health ^165–173. These proteins are synthesized by both CFs and macrophages – and in at least one study, constitutive production of macrophage MMP-9 improved outcomes for aged mice after MI, making their role in remodeling potentially even more complex ¹⁷⁴. Closely related to these proteins are the four TIMPs, which are produced by both CFs and CMs and act to stabilize MMP pro-forms and limit MMP activity ¹⁷⁵. TIMP expression has been shown to be generally decreased in the failing heart, in direct contrast to the increased prevalence of MMPs ¹⁷⁶. At least one study has also shown non-MMP-related effects of TIMPs in the rat MI model –TIMPs-1 and −3 were found to promote beneficial remodeling post-MI and also were associated with increased CM resistance to apoptosis both in vivo and in vitro ¹⁷⁷. This evidence suggests that balancing MMP and TIMP expression in the post-MI heart might be a key to translational development of therapies that target these ECM-modifying molecules.

Maturation Phase – Maintaining and Modifying a Scar

Once the scar has been synthesized during the pro-reparative phase, it begins a long process of dynamic remodeling known as maturation, which is a matter of weeks in small animal models and months in humans ^178–181. This scar is distinct from the early fibrous scar in its complexity – LOX facilitates collagen cross-linking, and collagen-binding factors such as decorin and perlecan reinforce the scar and alter its mechanical properties on both micro and macro scales ^{2, 182–186}. Changes occur in resident cell types as well; the AFs that have synthesized the scar now enter a phase marked by limited apoptosis, increased senescence, and irreversible exit from the cell cycle ^{92, 187–189}. Lineage tracing in a recent study showed that AFs residing in the maturing scar lose Acta2 expression as well as the ability to divide, resulting in a distinct state that the authors termed the ‘matrifibrocyte’ ⁹². These cells retained a number of lineage-specific fibroblast genes but gained expression of bone/cartilage markers (cartilage oligomeric matrix protein – Comp or Tsp5, chondroadherin – Chad, and Cartilage intermediate layer protein 1 – Cilp1) and apoptosis resistance genes, while silencing the expression of Col1 and Col3, and some matrix remodeling factors (which peak during the resolution/reparative phases). Surviving CMs have been shown to maintain a network of connections that allow for limited electrical conduction, even in the infarcted zone ¹⁹⁰, and electrical coupling is also accomplished via fibroblasts in the scar ^191–194. Over time, infarct thinning and compaction, in which the infarct responds to mechanical stress by altering its thickness, has been noted in both animal models and in human patients, suggesting dynamic alterations in scar properties long after the initial insult ^195–200.

Conclusions:

The CF has recently emerged as one of the most phenotypically plastic cells in the heart, with distinct transcriptional and secretory programs that regulate their essential contributions to pathophysiological cardiac remodeling. It is now evident that CFs are not inactive bystanders, but are dynamic contributors to the divergent responses that characterize cardiac adaptations following MI. The enigmatic and adaptable nature of CFs, once an impediment to translational research studies, may now prove to be a powerful target in the development of therapies aimed at limiting fibrosis and inflammation, which underlie pathologic remodeling in the failing heart.

Highlights.

• Ischemic cardiac tissue is replaced with a fibrotic scar

• The classical role of resident cardiac fibroblasts is the deposition of scar tissue

• Fibroblasts play a wide variety of roles in ischemic cardiac remodeling

• Fibroblasts interact with other cardiac cell types in the ischemic heart

• Fibroblasts contribute to each of the various phases of the injury response