Pericytes in Tissue Fibrosis

Abstract

Pericytes are mural cells, embedded within the microvascular basement membrane and primarily involved in preservation of vessel integrity and in regulation of vascular permeability and blood flow. Their study poses major challenges due to the absence of specific and reliable markers for their identification. Emerging evidence suggests that, in addition to their involvement in regulation of microvascular responses, pericytes may also play a central role in repair, inflammation and fibrosis in many different organs. Following injury, pericytes may dissociate from endothelial cells, acquiring inflammatory and pro-fibrotic phenotypes. Fibrogenic activation of pericytes has been reported in many different pathologic conditions and may involve stimulation by inflammatory cytokines, Transforming Growth Factor-β or Platelet-Derived Growth Factor-BB. Activated pericytes may stimulate fibrosis by secreting fibroblast-activating growth factors, by producing proteins involved in extracellular matrix remodeling and by depositing structural and matricellular matrix proteins. Conflicting findings have been reported on the phenotypic plasticity of pericytes and their capacity to convert to fibroblasts and myofibroblasts. Organ-specific differences in pericyte populations and differences in sensitivity and specificity of the pericyte fate mapping and fibroblast identification strategies may account for the conflicting observations reported in various studies. This review manuscript deals with the fate, role and mechanisms of activation of pericytes in tissue fibrosis. We discuss both the general mechanisms of pericyte activation and the organ-specific roles of pericytes in fibrotic conditions involving the kidney, liver, lung, heart and central nervous system. Understanding the role of pericytes is important to develop effective therapies for fibrotic conditions.

Graphical Abstract.

Mechanisms mediating the fibrogenic actions of pericytes. In fibrotic conditions pericytes become activated and disassociate from endothelial cells. Activated pericytes stimulate fibrosis through several different mechanisms: a) paracrine effects on fibroblasts mediated through secretion of fibrogenic growth factors, such as Transforming Growth Factor (TGF)-β and Platelet-Derived Growth Factors (PDGFs), b) stimulation of macrophage recruitment through secretion of cytokines and chemokines, c) release of matrix metalloproteinases (MMPs) involved in extracellular matrix (ECM) remodeling, d) production of ECM macromolecules, including matricellular proteins and structural collagens. In addition, pericytes have been suggested to convert to fibroblasts and α-smooth muscle actin (α-SMA)-expressing myofibroblasts in several different models of fibrosis. Symbols: SPARC, Secreted Protein Acidic and Rich in Cysteine; TIMPs, Tissue Inhibitors of Metalloproteinases; CCN2, Cellular Communication Network Factor 2; TGFbR, TGF-β receptor.

1. Introduction

Pericytes are mural cells that tightly surround capillaries, arterioles and venules, and are embedded within the microvascular basement membrane. Their main functions are to preserve the integrity of the vessels and to regulate microvascular permeability and blood flow (21). Additional vessel-independent roles have been demonstrated for organ-specific populations of pericytes, such as vitamin A storage in the liver (107),(22), or erythropoietin production in the kidney (31),(147). In addition to their homeostatic functions, pericytes have also been suggested to play an important role in a broad range of neoplastic, inflammatory and fibrotic diseases (136),(161),(61),(109).

Fibrosis is a pathologic response characterized by excessive accumulation of extracellular matrix (ECM) components in the interstitium and results in structural perturbations and functional impairment of the affected organs (89),(80). Although fibroblasts, the main matrix-producing cells, are the central cellular effectors of fibrosis, several other cell types can contribute, either by secreting fibrogenic growth factors and matricellular proteins, or by transdifferentiating into matrix-producing fibroblast-like cells. A large body of evidence implicates pericytes in the pathogenesis of fibrotic remodeling, in several different organs, including the liver, the kidney, the lung and the heart. The underlying mechanisms remain controversial, as some studies have suggested that pericytes exhibit remarkable plasticity and can contribute to fibrosis by transdifferentiating into matrix-producing fibroblasts, while other investigations suggested that pericytes maintain their mural cell characteristics and can contribute to fibrosis mostly through paracrine activation of fibroblasts. This review manuscript will discuss the role of pericytes in organ fibrosis and the molecular mechanisms implicated in fibrogenic pericyte activation.

2. Definition and heterogeneity of pericytes

Pericytes were first described in the second half of the 19^th century by Eberth (52) and Rouget (173) as contractile cells located near the capillary endothelium. The term “pericyte” was coined by Zimmerman in 1923 in his study on the fine structure of the microcirculation (224). 150 years after their initial discovery, pericytes remain enigmatic and their identification poses significant challenges. Pericytes are defined as mural cells embedded within a basement membrane that is continuous with the endothelial basement membrane. In addition to the requirement for electron microscopic analysis, which is often impractical, this definition becomes particularly problematic in pathologic conditions associated with angiogenesis, in which pericytes may dissociate from microvessels. Thus, in most cases, pericyte identification is based on a combination of topographic and morphological criteria, coupled with expression of pericyte-associated genes and proteins.

Pericytes are abundant in blood microvessels, but are virtually absent in lymphatic capillaries (178). Morphologically, pericytes are comprised of the soma, a protruding cell body that contains the nucleus, and elongated processes that can be defined as primary (extending along the vessel axis) or secondary (wrapping around the vessel circumference) (Figure 1). Within the same organ pericytes are heterogeneous, exhibiting a gradient of transitional phenotypes that have been well-characterized in the mouse brain (204), (76) but remain poorly defined in other species and tissues. Thus, the morphological characteristics of the pericytes within the same organ are determined by their localization within the vascular tree. Pre-capillary pericytes have short primary processes, while their secondary processes are wide and may fully encircle the vessel (44),(10). In contrast, pericytes enwrapping capillaries have very long primary processes and short secondary processes that only partially enwrap the vessel (84) (Figure 1A).

Figure 1: Pericytes in the microvasculature in normal tissues.

A: A continuum of pericyte phenotypes has been described in normal tissues. Pre-capillary pericytes have short primary processes and wide secondary processes that may fully encircle the vessel. Capillary pericytes, on the other hand, have long primary processes that run on the endothelium in the length of the capillary, and short secondary processes that only partially enwrap the vessel. This model is based on studies on the cerebral microvasculature; pericytes in different organs have distinct morphological characteristics. B: In most tissues, mature pericytes are separated from endothelial cells by a basement membrane. The two cell types communicate through tight junctions, adherens junctions, gap junctions, or adhesion plaques. Most endothelial-pericyte contacts are of the peg-and-socket type. The majority of these contacts involve cytoplasmic finger-like protrusions of the pericytes that protrude through a hole in the basement membrane into an endothelial pocket-like invagination.

Ιn most tissues, mature pericytes are separated from endothelial cells by the microvascular basement membrane; however, the membrane is discontinuous and the two cell types come into contact at discrete points through basement membrane gaps. Most endothelial-pericyte contacts are of the peg-and-socket type, characterized by the insertion of cytoplasmic finger-like protrusions of the pericytes into an endothelial pocket-like invagination (47). Communication between endothelial cells and pericytes involves several different types of intercellular junctions, including tight junctions, adherens junctions, gap junctions and adhesion plaques (Figure 1B).

Pericytes exhibit remarkable inter-organ variability with phenotypes ranging from the brain pericytes, which are tightly associated with endothelial cells and serve a barrier function, to the pericytes of the liver (also known as hepatic stellate cells (HSCs)), which lack a basement membrane and are loosely associated with the endothelium, or the glomerular mesangial cells of the kidney, which are rounded and compact, and contact a minimal abluminal microvascular area (7). The extent of pericyte coverage varies significantly across different organs. The brain and the retinal microvasculature display the highest levels of pericyte enwrapment, whereas other tissues, such as the skeletal muscle, exhibit a much lower pericyte-to-endothelial cell ratio (181),(185). Pericyte coverage and morphology are closely linked to function. High pericyte density and extensive microvascular coverage are associated with endothelial barrier properties and with reduced endothelial cell turnover (9),(8). For example, in the cerebral circulation, tight pericyte coverage contributes to the integrity of the blood-brain barrier, while in the kidney the looser arrangement of pericyte-like mesangial cells around microvessels facilitates efficient fluid filtration.

3. Pericyte markers

The absence of specific and reliable markers of pericytes has hampered progress in understanding their role in homeostasis and in pathologic conditions. Table 1 summarizes information on the sensitivity and specificity of the most commonly used markers in pericyte identification. Cspg4/neural-glial 2 (NG2) and Platelet-Derived Growth Factor Receptor β (PDGFRβ) have been extensively used to label and track pericytes in many different tissues. Transgenic NG2^dsred mice and inducible NG2-CreER drivers have acceptable sensitivity and specificity and have been used for pericyte labeling, lineage tracing and targeting in several different organs (223)^,(18)^,(20, 87). However, several other cell types, including glial cells (90), a subset of vascular smooth muscle cells (VSMCs), and embryonic cardiomyocytes also express NG2 (146). PDGFRβ has lower specificity than NG2 and exhibits high level expression in a large subpopulation of resident fibroblasts (78),(179),(2). Thus, the value of PDGFRβ as a pericyte marker in fibrotic conditions is limited.

Table 1.

Sensitivity and specificity of commonly used pericyte markers.

Pericyte marker	Specificity	Sensitivity	References
NG2 (Neuron-glial antigen 2)/CSPG4 (chondroitin sulfate proteoglycan 4)	• Relatively high specificity for pericytes in several different adult organs. • Glial cells, a subset of VSMCs, embryonic cardiomyocytes, and mesangial cells also express NG2.	• Relatively high sensitivity in labeling pericytes in several different organs. • Some pericyte subpopulations, (e.g. placental pericytes) express low levels of NG2. Not all brain pericytes express NG2.	(223),(18),(20, 87), (90), (146), (121), (69), (73)
PDGFRβ (Platelet-derived growth factor receptor β)	• Less specific for pericytes than NG2. • A large fraction of interstitial fibroblasts, mesangial cells and VSMCs also express PDGFRβ.	• Relatively high sensitivity, however some subsets of pericytes, (e.g. placental pericytes) express low levels of PDGFRβ.	(78),(179), (2), (34), (69)
PDGFRα (Platelet-derived growth factor receptor α)	• Low specificity. Expressed specifically in fibroblasts.	• Low sensitivity. PDGFRα expression has been reported only in subsets of pericytes (e.g. in the lung).	(34), (92), (67), (49)
Tbx18 (T-Box Transcription Factor 18)	• Relatively specific for mural cells (both pericytes and VSMCs in adult mammalian tissues. • In developing embryonic hearts, also expressed by fibroblast-like cells and cardiomyocytes.	• Relatively low sensitivity as Tbx18 expression may be limited to specific subpopulations of pericytes (Tbx18 labels a smaller percentage of pericytes than NG2).	(73), (125), (158), (126).
RGS5 (Regulator of G-protein signaling 5)	• Although suggested to be specific in some single cell transcriptomic studies, other investigations suggest limited specificity. Rgs5 expression has been reported not only in pericytes, but also in VSCMs, cardiomyocytes, and bone progenitors.	• Relatively low sensitivity. It has been suggested that RGS5 may be a pericyte activation marker in response to hypoxia. • Low RGS5 expression in pericytes from aging hearts.	(24), (118), (157), (170), (194)
Vimentin	• Very low specificity. Highly expressed by all mesenchymal cells, including fibroblasts. Also detected on Kupffer cells and endothelial cells.	• Although expressed by most pericytes, levels vary in different subpopulations.	(15), (200), (103)
Nestin	• Relatively low specificity. In addition to pericytes, various types of progenitor cells and VSMCs have been reported to express nestin.	• Low sensitivity. Expressed by a subpopulation of pericytes.	(117),(174),(20), (143)
Desmin	• Very low specificity. Broadly expressed by muscle cells, including cardiomyocytes, skeletal myocytes and VSMCs, but also by myofibroblast subsets.	• Low sensitivity. Only subpopulations of brain and liver pericytes express desmin.	(221), (14), (13), (54), (127)
α-SMA (α-smooth muscle actin)	• Low specificity. Significantly more abundant in VSMCs and myofibroblasts.	• Relatively low sensitivity. α-SMA expression in pericytes depends on their location and activation state.	(43)
CD146/ MCAM (melanoma cell adhesion molecule)	• Low specificity. Besides pericytes, CD146 also labels endothelial cells.	• Although most pericytes express CD146, subpopulations lacking CD146 expression have been reported.	(33), (208),(163),(6), (131)
Higd1b (Hypoxia Inducible Domain Family Member 1B)	• Higd1b has been suggested to be specific for pericytes in the lung and heart. However, its specificity has not been systematically studied.	• Suggested to be a sensitive marker of pericytes in the lung and heart. However, its sensitivity has not been systematically studied.	(103)
SLC6A12 (Solute Carrier Family 6 Member 12) and SLC19A1 (Solute Carrier Family 19 Member 1)	• Have been suggested to be specific markers of brain pericytes.	• Very low sensitivity with no expression by pericytes in other human organs, such as the kidney, lung, liver and the muscle.	(191)
Atp13a5 (ATPase 13A5)	• Relatively specific for CNS pericytes.	• Very low sensitivity. No labeling of pericytes in other organs, such as the heart, kidney, and the liver.	(74)
HMW-MAA (High molecular weight-melanoma-associated antigen)	• Limited specificity. HMW-MAA can be also expressed by melanoma cells, and dermal fibroblasts.	• Relatively low sensitivity with variable expression levels in different pericyte subpopulations.	(159), (96), (190), (26)
CD73 (also referred to as ecto-5′-nucleotidase)	• Low specificity. CD73 may be also expressed by perivascular fibroblasts and endothelial cells.	• Relatively low sensitivity with variable expression levels depending on topographical location of the pericytes.	(91), (189), (151).
CD13 (also referred to as aminopeptidase N)	• Low specificity. CD13 can also be expressed on inflammatory and endothelial cells.	• Low sensitivity labeling only a subset of pericytes.	(67), (57),(127), (141)
GFAP (Glial fibrillary acidic protein)	• Lacks specificity in many organs. Although expressed by hepatic stellate cells (HSCs), in the brain it labels predominantly astrocytes.	• Low sensitivity in most tissues. Even in the liver, GFAP labels only a fraction of HSCs.	(140), (142), (134)

Several other markers have been used to identify pericytes but lack specificity. Intermediate filament proteins, such as vimentin and desmin are broadly expressed by several different cell types. Vimentin is highly expressed by all mesenchymal cells, including fibroblasts (15). Desmin is localized in all muscle cells, including cardiomyocytes, skeletal myocytes and VSMCs. Although some studies have used α-smooth muscle actin (α-SMA) as a pericyte marker, α-SMA expression in pericytes is inconsistent and depends on the topographic location and state of activation of the cells. Moreover, α-SMA expression levels are much higher in VSMCs than in pericytes and are also upregulated in fibroblasts upon myofibroblast conversion (43). CD146/melanoma cell adhesion molecule (MCAM) labels pericytes (33), but is also expressed by endothelial cells (208),(163),(6). The transcription factor Tbx18 was found to label pericytes and VSMCs in several different issues (73); however, its expression may be limited to specific subpopulations of pericytes (125). RGS5 has been suggested to be “pericyte-specific” (24); however, it is also highly expressed in VSMCs (118), cardiomyocytes (157) and bone progenitors (170). Higd1b encoding Hypoxia Inducible Domain Family Member 1B was found to label pericytes in the lung, heart and several other organs and was more specific for pulmonary pericytes than NG2 and PDGFRβ (103). However, its value as a reliable and specific marker of pericytes in various tissues has not been sufficiently studied.

Several other markers may label organ-specific pericyte populations. For example, analysis of scRNA-seq data followed by immunofluorescence studies showed that the transporter proteins SLC6A12 and SLC19A1 are relatively specific markers of brain pericytes but are not expressed by pericytes in other human organs, such as the kidney, lung, liver and muscle (191). Moreover, Atp13a5 was found to be a reliable marker of pericytes in the Central Nervous System (CNS); however, the inducible Atp13a5^CreERT2 system did not label pericytes in other organs, such as the heart, kidney, and liver (74). Despite the widespread use of single cell transcriptomics to study the diverse phenotypes of pericytes in normal and diseased tissues, no single transcript has emerged as a specific pan-pericyte marker (137).

4. Tissue Fibrosis

Fibrosis is characterized by increased deposition of ECM proteins and is a common pathophysiologic companion of many diseases affecting virtually any tissue (80). Fibrotic remodeling often reflects unrestrained activation of a reparative program in response to injury. Reparative interstitial cells, mostly identified as fibroblasts, typically respond to injury by increasing their expression of ECM proteins. This reparative response serves a protective role, by preserving tissue architecture and by transducing matrix-driven signals that support survival and function of parenchymal cells. Depending on the type, severity, and duration of the injurious insult, and on the regenerative capacity of the affected organ, fibrosis can become a progressive, chronic and dominant pathologic response that contributes to organ dysfunction. The role of chronic fibrotic remodeling in disease progression is well-established in a broad range of primary fibrotic conditions, such as idiopathic pulmonary fibrosis (97) and scleroderma (133). In other diseases, the role of fibrosis in organ dysfunction and in mediating the manifestations of the disease is more nuanced. For example, in the infarcted heart, formation of a fibrous scar in the absence of robust regenerative capacity is critical to protect the ventricle from catastrophic complications, such as left ventricular rupture (60). In both reparative and maladaptive fibrotic responses, fibroblast-like cells serve as the main cellular effectors of fibrosis, producing large amounts of ECM proteins. Other cell types, including macrophages, lymphocytes and vascular cells can contribute to the fibrotic response mostly by producing fibroblast-activating mediators. In addition to their role as regulators of fibroblast activity, several different cell types have been suggested to undergo fibroblast conversion, thus directly contributing to the fibrotic response.

5. Fibrogenic activation of pericytes

Strategically located along vessels, pericytes sense injury and undergo activation, regulating inflammatory and fibrogenic responses. Although fibroblasts and myofibroblasts are the main sources of structural ECM proteins in fibrotic tissues, pericytes can also play important roles in fibrosis through several different mechanisms (Graphical Abstract). First, pericytes may be an important source of fibrogenic cytokines and growth factors, thus activating fibroblasts located in close proximity to microvessels, and contributing to interstitial and perivascular fibrosis. Second, pericytes can act as inflammatory cells, stimulating recruitment of fibrogenic immune cells, such as macrophages and lymphocytes. Third, in chronic fibrotic conditions, activated pericytes may promote stimulation of a fibrogenic program in endothelial cells. Fourth, pericytes can regulate matrix metabolism, either by producing ECM proteins or by secreting a broad range of proteases and anti-proteases. Finally, it has been suggested that under certain conditions, pericytes can convert to fibroblasts and myofibroblasts, thus directly promoting fibrosis.

5.1. Fibrogenic effects of pericytes mediated through cytokine secretion and subsequent paracrine activation of fibroblasts or macrophages.

Upon stimulation, pericytes can produce a broad range of inflammatory mediators that contribute to fibrotic tissue remodeling by activating fibroblasts, or by recruiting macrophages that serve as a source of fibrogenic mediators. Hypoxia, activation of Toll-Like Receptor (TLR)-mediated pathways and cytokine stimulation induce a pro-inflammatory and fibrogenic phenotype in pericytes. In isolated pericytes, TLR4 activation increases expression of a broad range of cytokines (such as IL-1β and IL-6) and chemokines (including CCL2, CCL20, CXCL1, CXCL2 and CXCL10) (72). In vivo, tissue injury stimulates a TLR-dependent Myd88-mediated pro-inflammatory program in pericytes that ultimately results in activation of the NLRP3 inflammasome and subsequent IL-1β and IL-18 secretion (115). On the other hand, hypoxia increases pericyte expression of a broad range of fibrogenic mediators, including Transforming Growth Factor (TGF)-β1, PDGF-BB and IL-6 (32). Stimulation with pro-inflammatory cytokines, such as Tumor Necrosis Factor (TNF)-α, IL-1β and IFN-γ, induces inflammatory activation of brain pericytes, activating Nuclear Factor (NF)-κB and STAT1 cascades, whereas PDGF-BB treatment suppresses inflammatory activation (30). Although increased expression of inflammatory and fibrogenic mediators by activated pericytes is well-documented, both in vitro (65) and in vivo, the relative contribution of the pericytes vs other cell types as a source of cytokines and growth factors is unclear.

5.2. Fibrogenic actions of pericytes may involve recruitment and activation of macrophages and other immune cell populations.

Macrophages are critical effector cells in tissue fibrosis (212), acting predominantly through secretion of fibroblast-activating growth factors and matricellular proteins (119). Given their close spatial relationship with macrophages, and their capacity to produce macrophage-activating mediators, pericytes may promote fibrosis, at least in part, through modulation of macrophage phenotype. Although this is a plausible concept, in vivo evidence demonstrating pericyte-mediated activation of a fibrogenic macrophage phenotype in fibrotic conditions is lacking. However, in other diseases, such as cancer, crosstalk between pericytes and macrophages has been demonstrated and involves secretion of pericyte-derived, macrophage-activating cytokines (206),(215). It should also be noted that macrophages are heterogeneous and functionally diverse and do not exert unidimensional pro-fibrotic actions but can also promote repair. In fibrotic conditions, macrophage populations with anti-fibrotic, reparative and regenerative properties have been identified and characterized (166),(149) Thus, pericyte-mediated macrophage recruitment may also stimulate repair (138),(183).

In addition to their effects on macrophages, pericytes can also regulate recruitment, phenotype and function of lymphocytes. It has been suggested that pericytes may act as non-professional antigen presenting cells, thus modulating T cell activation (105),(123). T lymphocytes have been implicated in activation of fibroblasts and may contribute to fibrotic remodeling in many different tissues (25). Thus, the fibrogenic effects of pericytes may also involve, at least in part, recruitment and activation of lymphocytes.

5.3. Direct effects of pericytes on ECM composition and remodeling.

Although fibroblasts are the main source of structural collagens in most fibrotic conditions (100), other cell types can directly regulate matrix composition by secreting specialized matrix proteins (matricellular glycoproteins) and by producing proteases involved in matrix metabolism. Activated pericytes express collagen in an organ-dependent manner. In the liver, pericyte-like HSCs are a major source of collagens in fibrotic conditions. Moreover, descriptive immunohistochemical studies suggested that in organ-specific fibrotic conditions, collagen is localized in lung pericytes, but not in cardiac or renal pericytes (20).

Pericytes are also an important source of fibronectin, thus contributing to the formation of the provisional matrix network that facilitates migration of leukocytes and interstitial cells (175). Moreover, pericytes have been reported to produce significant amounts of bioactive matricellular proteins, such as periostin (218), SPARC (Secreted Protein Acidic and Cysteine-rich) (11), tenascin-C and thrombospondin-1 (4). Following secretion, these mediators can amplify fibrotic responses by activating growth factors and by regulating matrix-degrading proteases (59).

In addition to their role as a source of matrix macromolecules, pericytes can also regulate the intricate balance between matrix synthesis and degradation by producing proteases and their inhibitors. Localized activation of matrix metalloproteinases (MMPs) has been documented in pericytes following tissue injury (203), and may initially contribute to basement membrane degradation (113) and disruption of the microvascular network (202).

5.4. Do pericytes convert to fibroblasts and myofibroblasts?

Although some studies have suggested that pericytes can serve as multipotent progenitor cells, capable of differentiating into a broad range of lineages, including adipocytes, follicular dendritic cells, skeletal muscle cells and fibroblasts (39),(42),(121), this concept is controversial (73). Several lineage tracing studies have suggested that pericytes undergo fibroblast conversion in animal models of kidney and lung fibrosis. In renal and pulmonary fibrosis, fate mapping experiments and single cell transcriptomics support the notion that pericytes are an important source of activated fibroblasts and myofibroblasts (121),(91),(180),(111),(92). TGF-β, the matricellular protein CCN2, integrin β1 (ITGB1) activation (219), STAT3 (1) and the SOX9-NAV3-YAP1 axis (162) have been implicated in pericyte to myofibroblast conversion. In contrast, experiments using an inducible Tbx18 Cre driver to track mural cells showed no significant pericyte to fibroblast conversion in several models of organ fibrosis (73). A subset of pericytes exhibiting high expression of collagen was identified; however, these cells retained high expression of pericyte markers, such as CD146.

The conflicting findings likely reflect the challenges in labeling and identifying fibroblasts and pericytes in injured tissues. In the absence of specific and reliable pericyte markers, various studies have used different Cre drivers for fate mapping of pericytes in experimental models of fibrosis (Table 2). Inducible Gli1, Pdgfrb, Foxd1, Tbx18 and NG2/Cspg4 Cre drivers have distinct and poorly characterized specificity and sensitivity profiles and label different populations of perivascular cells that do not always meet strict criteria for their identification as pericytes. On the other hand, fibroblasts exhibit remarkable heterogeneity and diversity and are defined on the basis of functional and morphological criteria, rather than through the expression of a single specific marker (155). In most studies documenting large scale pericyte to fibroblast conversion, documentation of fibroblast identity is based on immunofluorescent staining for a single, relatively non-specific marker (Table 2). In a recent study, we studied the fate of pericytes in a mouse model of cardiac fibrosis caused by acute myocardial infarction, using a combination of lineage tracing and scRNA-seq approaches. Lineage tracing of pericytes using the inducible NG2/Cspg4 Cre driver coupled with the PDGFRa^EGFP reporter system to reliably label fibroblasts showed that only 4% of infarct fibroblasts are derived from pericytes (4). High resolution scRNA-seq of the NG2 lineage cells showed that infarct pericytes diversify and expand after infarction, increasing their expression of ECM genes. A cluster of NG2 lineage cells with expression of fibroblast identity genes (such as Pdgfra and Tcf21) emerged during the proliferative phase of infarct healing, supporting the notion that a subpopulation of pericytes undergoes fibroblast conversion. We suggest that a similar combination of fate mapping and single cell approaches is needed for rigorous in vivo documentation of pericyte to fibroblast transdifferentiation.

Table 2.

Studies examining whether pericytes convert to fibroblasts and myofibroblasts in various pathophysiologic conditions using lineage tracing models.

Model of organ injury	Cre driver	Fibroblast identification	Findings	Ref
Kidney: Unilateral ureteric obstruction (UUO) and ischemia/reperfusion injury (IRI)	FoxD1-CreERT2	Interstitial α-SMA+ cells	In both models FoxD1-derived cells expanded and differentiated into α-SMA+ myofibroblasts, accounting for a large majority of myofibroblasts.	(91)
Kidney: UUO and IRI	FoxD1-Cre	Interstitial α-SMA+ cells	In both models FoxD1-derived cells differentiated into α-SMA+ myofibroblasts.	(31)
Kidney: UUO and IRI	Gli1-CreERT2	Interstitial α-SMA+ cells	10d post UUO: almost all Gli1-derived cells expressed α-SMA. ~45% of all α-SMA+ myofibroblasts were derived from Gli1+ progenitors.	(110)
Liver: Toxic (CCl4), biliary (TAA), and fatty (cholestasis) injury	Lrat-Cre	Interstitial α-SMA, Col-GFP	In all models, the vast majority of Lrat-derived HSCs demonstrated an overlap with α-SMA and Col-GFP.	(134)
Liver: CCl4 injury	Gli1-CreERT2	Interstitial α-SMA+ cells	28d post injury: almost all Gli1-derived cells expressed α-SMA. ~39% of all α-SMA+ myofibroblasts were derived from Gli1+ progenitors.	(110)
Liver: CCl4 injury	PDGFRβ-Cre	Interstitial α-SMA+ cells	6 weeks post injury: most PDGFRβ-derived cells co-expressed α-SMA.	(79)
Lung: Bleomycin-induced injury	Foxd1-CreER	Interstitial α-SMA+ cells, Col1a1-GFP	45% (7d post injury) and 68% (14d post injury) of αSMA+ myofibroblasts were derived from Foxd1+ cells and their progeny. More than 30% of Foxd1-derived cells also expressed GFP transgene. Conversely, 47% of Coll-GFP+ cells were derived from Foxd1+ cells and their progeny.	(92)
Lung: Bleomycin-induced injury	NG2-CreER, FoxJ1-CreER	Interstitial α-SMA+ cells	14 and 21d post injury: no relevant expression of α-SMA by NG2- or FoxJ1-derived cells.	(167)
Lung: Bleomycin-induced injury	Gli1-CreERT2	Interstitial α-SMA+ cells	14d post injury: majority of Gli1-derived cells expressed α-SMA. ~37% of all α-SMA+ myofibroblasts were derived from Gli1+ progenitors.	(110)
Lung: Bleomycin-induced injury	ABCG2-CreERT2	Interstitial α-SMA+ cells	14d post injury: colocalization of ABCG2-eGFP and α-SMA signaling in vivo. In vitro: ABCG2-derived cells expressed α-SMA only after stimulation with TGF-β (not with bleomycin).	(132)
Lung: Bleomycin-induced injury	PDGFRβ-Cre	Interstitial α-SMA+ cells	14 and 28d post injury: most PDGFRβ-derived cells co-expressed α-SMA	(79)
Heart: Myocardial infarction (MI)	NG2-CreER	PDGFRα-GFP Sc-RNAseq of NG2 lineage cells: Pdgfra, Itga11, Tcf21 as fibroblast identity genes.	7d post MI: about 4% of all PDGFRα-GFP+ fibroblasts were pericyte-derived, sc-RNA-seq showed emergence of a new cluster 7 (~14.5% of all NG2 lineage cells in the infarct) with pericyte-derived cells which acquired fibroblast identity genes (Pdgfra, Itga11 and Tcf21).	(4)
Heart: MI	NG2-CreERT2 Tbx18-CreERT2	PDGFRα scRNAseq: Pdgfra, Col1a1, S100a4, Ddr2	2,4,7,14d post MI: no colocalization of NG2-derived cells with PDGFRα. 4,7,14d post MI (NG2): only a minor cluster corresponded to fibroblasts (Pdgfra, Col1a1, S100a4, Ddr2). 7d post MI (Tbx18): cluster corresponded to fibroblasts was larger.	(158)
Heart: MI	Tbx18-CreER	Fibroblast activation genes (Postn)	7d post MI: colocalization of Tbx18-derived pericytes with Postn.	(153)
Heart: Angiotensin II(AngII)- induced myocardial fibrosis, ascending aortic constriction (AAC)	Gli1-CreERT2	Interstitial and periadventitial α-SMA+ cells	28d post AngII infusion and 56d post AAC: almost all Gli1-derived cells exhibited α-SMA expression. ~60% of all α-SMA+ myofibroblasts were derived from Gli1+ progenitors.	(110)
Heart: Transverse aortic constriction (TAC)	Tbx18-CreERT2	CD140a/PDGFRα, Col1a1-GFP	The vast majority of interstitial Tbx18-derived cells were negative for the fibroblast marker PDGFRα, with the exception of a very small subset of Tbx18-derived cells, mainly located close to large vessels.	(73)
Heart: TAC	NG2-CreER, PDGFRβ-CreERT2	Sc-RNAseq and RNA in situ hybridization: Dcn, Col1a1, Pdgfrα	14d post TAC: no significant pericyte to fibroblast conversion.	(150)
Heart: Diabetic cardiomyopathy	NG2-CreER	PDGFRα-EGFP	No significant pericyte to fibroblast conversion.	(3)
CNS: Cortical stab wounds	Tbx18-CreERT2	CD140a/PDGFRα, Col1a1-GFP	The vast majority of interstitial Tbx18-derived cells was negative for the fibroblast marker CD140a/PDGFRα, with the exception of a very small subset of Tbx18-derived cells, mainly in the vicinity of large vessels. 4d post injury: Collal-GFP+ fibroblasts were not derived from Tbx18+ cells and their progeny.	(73)
CNS: Ischemic stroke	Tbx18-CreER	Genes related to fibrosis (col1a1)	7d post stroke: colocalization of Tbx18-derived pericytes with col1a1.	(153)
CNS: Spinal cord injury	NG2-CreER	col1a1-GFP	7 and 14d post injury: NG2-derived pericytes and col1a1-GFP+ fibroblasts are two distinct populations.	(186)
CNS: Spinal cord injury	GLAST-CreERT2	α-SMA, vimentin	5 and 14d post injury: GLAST-derived pericytes expressed α-SMA and vimentin. These cells were embedded in fibronectin- and collagen I- rich ECM.	(46)
CNS: Spinal cord injury	GLAST-CreER	Fibronectin, α-SMA	5d post injury: GLAST-derived pericytes expressed fibronectin and transiently α-SMA.	(67)

It should be emphasized that the controversy regarding pericyte to fibroblast conversion may be more semantic than substantive. Given the functional basis of the definition of fibroblasts as connective tissue-producing cells, their phenotypic heterogeneity, and the absence of specific markers for their identification, activated fibrogenic pericytes may overlap with fibroblast populations. Moreover, pericyte activation to increase production of ECM proteins has similar functional implications regardless of the expression of fibroblast identify markers. ScRNA-seq of human samples has produced evidence that fibrogenic activation of pericytes is prominent in fibrotic conditions. In patients with Dupuytren’s disease, a fibroproliferative condition involving the palms, scRNA-seq identified a population of pericytes with characteristics of myofibroblast precursors (114). Moreover, in human chronic kidney disease, pseudotime trajectory analysis of single cell transcriptomic data was consistent with the notion that a cluster of NOTCH3+/RGS5+/PDGFRa- pericytes may be one of the sources of activated myofibroblasts (along with MEG3+/PDGFRa+ and COLEC11+/CXCL12+ fibroblast subpopulations) (111).

6. Pericytes in organ fibrosis

6.1. Pericytes in renal fibrosis.

Chronic renal injury generates a fibrogenic milieu leading to deposition of excessive ECM within the renal parenchyma, resulting in glomerular and tubulointerstitial fibrosis. Progressive tubulointerstitial fibrosis is the final common pathway for all kidney diseases that lead to chronic renal failure (222). In patients with chronic kidney disease, tubulointerstitial fibrosis is correlated with the decline in kidney function and is a strong predictor of disease progression (165).

The kidney contains a significant population of pericytes, which play important roles in homeostasis, maintaining the integrity of the microvasculature and contributing to blood pressure regulation (195),(180). Functionally distinct subsets have been identified, including Gli1+ perivascular cells (that may overlap with fibroblast-like cells), and a renin-producing pericyte population in juxtaglomerular arterioles (188). Some studies have suggested that mesangial cells may be identified as pericytes (180); however, single cell transcriptomics showed that mesangial cells are heterogeneous and possess a complex phenotype with features of pericytes, VSMCs and fibroblasts (77).

The role of pericytes in renal fibrosis remains controversial. Studies in 2 different models of renal injury, unilateral ureteric obstruction (UUO) and ischemia/reperfusion injury, suggested that pericytes initially detach from the capillaries, proliferate and acquire expression of the myofibroblast marker α-SMA, accompanied by high levels of collagen synthesis (91), (121), (211), (35), (29). Associative studies have suggested fibrogenic activation of pericytes in several other models of chronic renal injury, including salt-sensitive hypertension (95) and aging (101). Lineage tracing experiments showed that FoxD1-derived pericytes were the predominant cell type contributing to the α-SMA+ myofibroblast pool (91). In contrast, another study using lineage tracing and pericyte depletion experiments showed that NG2+/PDGFRβ+ pericytes do not significantly contribute to the myofibroblast pool and do not play an important role in renal fibrosis (116). The majority of myofibroblasts in this study appeared to originate from resident fibroblasts with smaller contributions of bone marrow-derived cells, endothelial cells and epithelial cells (116). As previously discussed, the conflicting findings between various studies likely reflect differences in the specificity and sensitivity of the Cre drivers used to track pericytes and the use of different criteria to identify fibroblasts and myofibroblasts. In many studies the use of non-specific markers, such as collagen or α-SMA does not document fibroblast or myofibroblast transdifferentiation but may simply suggest acquisition of a fibrogenic profile by cells that maintain mural cell identity.

Fibrogenic pericyte activation following renal injury appears to involve effects of growth factors, such as TGF-βs and PDGFs, secreted by endothelial and epithelial cells. Following UUO, the upregulation of TGF-β and PDGFs in injured endothelium and the elevated expressions of TGF-β receptor, TGFβRII, and its downstream effector Smad2, in both endothelial cells and pericytes preceded pericyte to myofibroblast transition and renal fibrosis (211). In vitro, TGF-β1 stimulation induces α-SMA upregulation in pericytes and stimulates production of profibrotic cytokines by renal epithelial cells which in turn enhances pericyte proliferation and their transformation into myofibroblasts (211). Similarly, PDGF-BB exerts fibrogenic actions on pericytes (205). Perturbation of the endothelial-pericyte interaction upon renal injury induces pericyte detachment from capillaries, increasing microvascular permeability and causing capillary rarefaction, associated with accentuated immune responses and acquisition of a fibrogenic pericyte profile (122). Epithelial cell injury may also contribute to pericyte-driven renal fibrosis. Injury-associated upregulation of the Hedgehog (Hh) ligands in tubular epithelial cells stimulates proliferation of pericytes and promotes their transformation into α-SMA+ myofibroblasts through the Hh effectors, Gl1 and Gli2 expressed on PDGFRβ+ interstitial pericytes and perivascular fibroblasts (55). At the molecular level, the detachment of pericytes from the endothelial layer and their migration towards the renal interstitium is facilitated by increased expression of proteases, such as ADAMTS1 (a disintegrin and metalloproteinase with thrombospondin motif-1), which is induced early after renal injury (177). ADAMTS1 cleaves the bonds between pericytes and capillary basement membrane proteins and inhibits angiogenesis (177).

The intracellular signaling cascades responsible for fibrogenic activation of renal pericytes remain poorly understood. Experiments in models of folic acid- or aristolochic acid-induced renal fibrosis demonstrated that STAT3 activation is implicated in acquisition of a fibrogenic phenotype by pericytes (1). Moreover, in the UUO model, a signaling pathway involving SOX9 and downstream activation of NAV3 and YAP1 signaling was involved in pro-fibrotic pericyte activation (162).

6.2. Hepatic stellate cells (HSCs), the pericytes of the liver, as mediators of hepatic fibrosis.

Progressive hepatic fibrosis typically follows chronic liver injury caused by a broad range of insults, including infectious (such as viral hepatitis), toxic (alcohol- or drug-induced), metabolic (related to non-alcoholic fatty liver disease/NAFLD), cholestatic (due to obstruction of bile flow) or autoimmune conditions (62). Eventually fibrous tissue accumulation in the diseased liver leads to the development of cirrhosis, an advanced stage of fibrotic remodeling, associated with the formation of regenerative nodules of hepatic parenchyma, that are separated by fibrous bands. The main cellular effectors of liver fibrosis are the HSCs, a population of pericyte-like cells that undergo fibrogenic activation upon injury and produce large amounts of ECM proteins, contributing to hepatic fibrosis. Several other cell types, including portal fibroblasts and immune cells may contribute to hepatic fibrosis either directly (by producing ECM proteins) or indirectly (by secreting fibrogenic mediators that activate HSCs) (108).

Although HSCs may lack some of the strict credentials required for pericyte identification, such as the relation with a microvascular basement membrane, they are considered pericyte-like cells on the basis of their topographic location in close association to endothelial cells, and their constitutive expression of some pericyte markers, such as CD146 and desmin (107),(94). HSCs reside in the subendothelial space of Disse between the hepatic sinusoidal endothelial cells and the parenchymal cells. In the healthy liver, they represent ~10% of all hepatic cells (197) and play a central homeostatic role by controlling storage and release of vitamin A.

Extensive evidence supports the central role of HSCs in the pathogenesis of hepatic fibrosis (Figure 2). Following a broad range of insults, HSCs proliferate and increase expression of α−SMA, acquiring a myofibroblast-like matrix-secreting phenotype, associated with loss of vitamin A-containing lipid droplets (135), (128), (88), (12), (99). Several studies have reported that the injured liver is infiltrated by transitional cells with intermediate ultrastructural features between HSCs and fibroblasts/myofibroblasts, suggesting a transformation of HSCs into fibroblasts/myofibroblasts (128), (88), (99). Lineage tracing experiments with the Lrat^Cre driver that labels ~99% of HSCs demonstrated that in a broad range of hepatic fibrotic conditions, 80–90% of matrix-secreting myofibroblasts originate from HSCs (134). Upon activation and myofibroblast conversion, HSCs upregulate mRNA for procollagens alpha1(I) (145), (199), (68), alpha1(III), alpha1(IV) (68), produce fibronectin (102), laminin and proteoglycans (70). Activated HSCs also secrete anti-proteases that inhibit matrix degradation and accentuate fibrosis, such as the tissue inhibitor of metalloproteinases (TIMP) 1 and 2 (16). Moreover, activated HSCs also increase their responsiveness to TGF-βs by upregulating both type I and II TGF-β receptors (50) and may exert autocrine pro-fibrotic effects by secreting fibrogenic growth factors, such as TGF-β (68), (199), (145), (51), PDGFs (210), and connective tissue growth factor (CTGF)/CCN2 (148).

Figure 2. The role of liver pericytes, also known as Hepatic Stellate Cells (HSCs), in hepatic fibrosis.

Although they lack a basement membrane, HSCs are located adjacent to hepatic endothelial sinusoidal cells in the space of Disse and are considered the pericytes of the liver. In most chronic liver conditions, HSCs become activated and produce large amounts of extracellular matrix proteins (ECM), thus contributing to hepatic fibrosis. Fibrogenic factors induced upon hepatic injury, such as transforming growth factor-β (TGF-β), platelet-derived growth factors (PDGFs), fibroblast growth factors (FGFs), and epidermal growth factor receptor (EGFR) ligands are released from damaged endothelial cells, hepatocytes, activated platelets, and Kupffer cells and play a major role in HSC activation. Activated HSCs express proinflammatory and profibrotic genes and increase the surface expression of growth factor receptors, including type I and type II TGF-β receptors (TGFbR). Pericyte-derived fibrogenic mediators, stimulate fibroblasts to produce ECM components and further enhance pericyte activation via autocrine signaling. A large number of activated pericytes undergo transdifferentiation into α-smooth muscle actin-positive (α-SMA⁺) myofibroblasts and become the primary ECM-producing cells. Activated HSCs can also regulate matrix turnover by releasing matrix metalloproteinases (MMPs) and tissue inhibitors of metalloproteinases (TIMPs). In addition, activated pericytes stimulate immune cells to secrete matricellular proteins and pro-inflammatory cytokines, thereby amplifying profibrotic remodeling. CCN2, Cellular Communication Network Factor 2.

The mechanism of HSC activation after hepatic injury involves paracrine actions of hepatocytes, endothelial cells, platelets and macrophages. In most forms of liver injury, hepatocytes are the main primary targets. Injured and dying hepatocytes release a broad range of Danger-Associated Molecular Patterns (DAMPs) (5) and secrete fibrogenic mediators, thus promoting fibrogenic activation of HSCs. Degradation of the native ECM and de novo deposition of specialized matrix proteins, such as fibronectin also generate a pro-inflammatory and fibrogenic milieu that stimulates HSCs. Moreover, pro-inflammatory cytokines activate endothelial cells, increasing recruitment of immune cells in the injured liver and contributing to stimulation of a pro-fibrotic program. Members of the TGF-β superfamily (64), PDGFs (104), Fibroblast Growth Factors (FGFs) (198) and Epidermal Growth Factor (EGF) receptor ligands (152) are the best studied and most potent activators of HSCs in the injured liver.

It is increasingly appreciated that HSCs exhibit significant heterogeneity (217), both under homeostatic conditions (40) and in disease (172). Studies in a mouse model of NAFLD identified 4 distinct clusters of HSCs, including myofibroblasts, intermediately activated cells, inflammatory HSCs and a subpopulation of proliferating cells (172). Moreover, a subpopulation of HSCs that expresses the pericyte marker FOXD1 has been suggested to have increased fibrogenic capacity (213).

6.3. Pericytes in pulmonary fibrosis

Focal or diffuse lung fibrosis is a feature of various pathological lung conditions such as idiopathic pulmonary fibrosis, asthma, pulmonary hypertension, or chronic obstructive pulmonary disease. The lung contains a significant population of pericyte-like cells (209),(220). In addition to their role in maintaining vessel integrity (63), pulmonary pericytes also play a role in development, contributing to lung morphogenesis and patterning (98). Although descriptive studies have documented that a broad range of lung diseases are associated with a pro-inflammatory and/or fibrogenic pericyte phenotype, studies examining the contribution of pulmonary pericytes in the pathogenesis of lung fibrosis have produced conflicting results.

Several studies have suggested that lung pericytes may undergo myofibroblast conversion in experimental models of pulmonary fibrosis. In a mouse model of bleomycin-induced lung injury, Foxd1-derived pericytes expanded and expressed structural collagens and the myofibroblast marker α-SMA. More than 50% of all myofibroblasts in fibrotic regions originated from Foxd1-derived pericytes (92). A population of resident fibroblasts also underwent activation and served as an additional major source of matrix-secreting myofibroblasts (92). Another study used lineage tracing of cells expressing the mesenchymal stem cell marker ABCG2 and identified these cells as perivascular pericyte-like cells that were distinct from NG2+ mature pericytes (132). In the bleomycin model of pulmonary fibrosis, ABCG2+ cells activated proliferative and migratory programs and transitioned to activated myofibroblasts (132).

In contrast, other studies suggested that pericytes may not play a critical role in lung fibrosis. In the model of bleomycin-induced lung fibrosis tracing of pericytes with inducible NG2-CreER and FoxJ1-CreER drivers showed no significant pericyte to myofibroblast conversion (167). Moreover, in the same model, ablation of Foxd1+ pericytes attenuated inflammation but did not significantly affect lung fibrosis (93). The conflicting findings likely reflect the use of different drivers with distinct sensitivity and specificity profiles to track pericytes and the known challenges in documentation of fibroblast identity.

Human data cannot examine whether pericytes convert to fibroblasts but provide support to the notion that pericytes acquire a pro-fibrotic profile in patients with pulmonary fibrotic conditions. In lung tissues from patients with idiopathic pulmonary fibrosis, increased numbers of PDGFRβ+ pericytes with high expression of collagen 1 and α-SMA populated areas of fibrosis (207),(214),(176).

The mediators involved in fibrogenic activation of lung pericytes remain poorly understood. Extensive in vitro evidence suggests that TGF-β may upregulate expression of collagen and fibronectin by pericytes, while promoting pericyte-to-myofibroblast transformation (176). In addition to the effects of fibrogenic growth factors, the matrix environment may also contribute to fibrogenic pericyte activation. In vitro studies showed that the altered matrix microenvironment associated with idiopathic pulmonary fibrosis induced a myofibroblast-like profile in human pericytes (176).

6.4. The role of pericytes in myocardial fibrosis.

Ultrastructural studies have identified pericytes in the heart as extensively branched cells that form an incomplete layer around the capillary endothelium (58),(82). As a highly vascular organ with an extensive microvascular network, the heart contains a large number of pericytes. In human myocardial samples, electron microscopic analysis showed that mural cells (pericytes and smooth muscle cells) account for ~22–28% of interstitial cells (156). A study in healthy adult rat hearts has suggested that pericytes may be more numerous than fibroblasts, representing the second most abundant non-cardiomyocyte cell type (only outnumbered by endothelial cells) (139), In contrast, a flow cytometric study in adult mice showed that fibroblasts are more numerous than pericytes (154). Conflicting data on relative abundance of pericytes vs fibroblasts may reflect the use of different strategies for cell identification. Cardiac pericytes have been implicated in homeostatic regulation of blood flow and microvascular permeability (41) and may contribute to a broad range of cardiac pathologies by regulating inflammation, fibrosis and vascular function (61).

As in many other organs, identification and specific labeling of cardiac pericytes is challenging due to the absence of specific and reliable markers. NG2 is expressed by embryonic cardiomyocytes (146); however, in the adult mammalian heart NG2 expression marks mural cells, both microvascular pericytes and a subset of vascular smooth muscle cells (2). PDGFRβ on the other hand is expressed by both mural cells and a large subpopulation of cardiac fibroblasts (2).

Myocardial fibrosis can be reparative or maladaptive. Because of the extremely limited regenerative capacity of the adult mammalian heart, sudden death of up to a billion cardiomyocytes in myocardial infarction results in formation of a scar which serves to protect the ventricle from catastrophic rupture (replacement fibrosis) (83). In other myocardial pathologic conditions, extracellular matrix deposition predominantly involves the interstitial space (which surrounds each cardiomyocyte) or perivascular areas and is maladaptive, promoting both systolic and diastolic dysfunction (interstitial and perivascular fibrosis) (60).

Pericytes have been implicated in the pathogenesis of reparative cardiac fibrosis through the secretion of fibroblast-activating mediators; however, their direct conversion to activated fibroblasts and myofibroblasts is relatively limited (4),(158) (Figure 3). Lineage tracing experiments using the inducible NG2^CreER driver to label mural cells, combined with scRNA-seq examining the transcriptional profile of NG2 lineage cells demonstrated that only 4–5% of PDGFRα+ fibroblasts (identified through a reporter system) were derived from pericytes (4). Fibrogenic activation of pericytes may involve actions of TGF-β (4),(158). Moreover, TGF-β-mediated activation of infarct pericytes was also involved in vascular maturation after infarction. In healing infarcts, scar maturation requires coating of infarct neovessels with mural cells (225),(48),(163), a process that involves activation of TGF-β and PDGFRβ signaling pathways (225),(4). Defective coating of the vasculature with mural cells causes perturbations in scar formation, associated with hemorrhagic foci.

Figure 3. Pericytes in cardiac fibrosis.

Pericytes are implicated in the pathogenesis of cardiac fibrosis in myocardial infarction (MI, left) and in heart failure caused by pressure overload (right). The reparative fibrotic response after MI can be divided into 3 distinct but overlapping phases: the inflammatory, proliferative, and maturation phase. In the inflammatory phase, damage-associated molecular patterns (DAMPs) released by dying cardiomyocytes stimulate a pro-inflammatory pericyte phenotype, associated with cytokine release and production of matrix-degrading enzymes, such as matrix metalloproteinases (MMPs). Activated pericytes detach from endothelial cells, increasing microvascular permeability and facilitating inflammatory cells extravasation. During the proliferative phase of infarct healing, activated pericytes participate in extracellular matrix (ECM) deposition and remodeling through several mechanisms by secreting fibroblast-activating mediators and by synthesizing collagens and matricellular proteins. Moreover, a subset of activated pericytes may undergo fibroblast conversion. In the maturation phase, pericytes stimulated by Transforming Growth Factor (TGF)-β and Platelet-Derived Growth Factors (PDGFs) contribute to maturation of the infarct vasculature by coating newly formed vessels. Information on the role of pericytes in heart failure caused by pressure overload is limited. In the pressure-overloaded heart, mechanosensitive activation and stimulation of neurohumoral cascades (such as the renin-angiotensin-aldosterone system and adrenergic pathways), along with cytokines and growth factors released from neighboring cells, activate pericytes. Upon activation, pericytes contribute to fibrotic remodeling by producing ECM proteins and by secreting fibroblast-activating mediators.

Information on the role and fate of pericytes in chronic heart failure is more limited. In heart failure patients, pericyte function was impaired; these perturbations were associated with evidence of defective mechanotransduction (168). Moreover, pericytes in cardiomyopathic hearts exhibited altered transcriptional profiles, characterized by increased expression of genes encoding matricellular proteins (106), consistent with fibrogenic activation. However, pericyte to fibroblast conversion has not been convincingly documented in chronically failing hearts. Although Gli1+ perivascular cells may account for the expansion of fibroblasts in models of cardiac fibrosis, these cells are probably not mature pericytes, but rather fibroblasts located in the adventitia (110). Moreover, an investigation using an inducible Tbx18-CreER driver to label mural cells found no significant conversion of pericytes to fibroblasts, or any other lineages, in the failing pressure-overloaded heart (73). However, a subset of pericytes exhibited high expression of collagen that was not associated with loss of pericyte marker expression, such as CD146, suggesting fibrogenic activation in the absence of fibroblast conversion. In addition, a study examining the fate of pericytes and VSMCs in the pressure-overloaded heart using NG2-CreER, PDGFRβ-CreER and Myh11-CreER drivers, combined with scRNA-seq experiments showed no significant pericyte to fibroblast conversion, but demonstrated upregulation of matrix genes in pericytes, reflecting their fibrogenic activation (150).

6.5. Pericytes in CNS fibrosis.

Adult mammals have extremely limited capacity to regenerate their CNS. Fibrous tissue formation after CNS injury is important for containment of the damage but may also hamper axonal regeneration (144). Pericytes have been suggested to serve as central cellular effectors of the fibrotic response in a broad range of CNS injury models (46). In the acute phase of cortical brain trauma, the number of PDGFRβ+ pericytes co-expressing NG2 and desmin is initially reduced due to apoptosis. This early pericyte loss is followed by expansion of the pericyte population induced through activation of a proliferative program (221),(57). Brain injury also activates pericytes resulting in upregulation of biological pathways and genes related to inflammation and fibrosis (153). Both in vitro and in vivo experiments suggest an important role for the PDGF-BB/PDGFRβ axis in expansion of the pericyte population after brain injury (182). The expansion and activation of pericytes in the injured brain may prevent microvascular injury, and mediate a more efficient fibrotic response and reparative astrogliosis, thus improving neuronal functional recovery (192),(129),(182). However, on the other hand, pericyte-driven fibrotic scarring may act as a neuronal extrinsic barrier that limits neuroplasticity, preventing axon regeneration and remodeling of neural networks (112).

It should be emphasized that interpretation of studies examining the role of pericytes in CNS scarring is challenging due to differences in the criteria used to define pericytes. Some studies use a perivascular location as synonymous to pericyte identity without considerations of ultrastructural criteria or expression of more specific mural cell markers. Experimental approaches using less specific pericyte markers are typically more likely to support a central role for pericytes in fibroblast expansion, possibly due to the inclusion of perivascular fibroblast-like populations.

Studies examining the role of pericytes in spinal cord scarring have produced conflicting results. It has been suggested that a subpopulation of pericytes (termed type A pericytes), defined by their expression of glutamate/aspartate transporter (GLAST) expand after spinal cord injury and migrate away from the vascular wall (46), expressing genes associated with inflammation and ECM synthesis (45), and contributing to inflammation, fibrosis, and extension of injury (216),(81). This pericyte subpopulation is independent of NG2+ pericytes and expresses not only PDGFRβ, but also PDGFRα and CD13 (67). Thus, these cells may represent perivascular fibroblasts rather than mature pericytes. In contrast to these observations, fate mapping experiments using NG2-CreER driver to trace mural cells along with a col1α1-GFP reporter system to identify matrix-producing fibroblasts showed no significant contribution of pericytes to the expanding fibroblast population in the injured spinal cord (186).

7. Pericytes in systemic fibrosis-associated conditions.

7.1. Pericytes in systemic sclerosis.

Systemic sclerosis (also known as scleroderma) is an autoimmune disorder of unknown etiology that affects several different organ systems, including the kidney, lung, intestine, skin and heart. Vascular dysfunction, inflammatory cell infiltration and progressive fibrosis are the major underlying pathologic abnormalities that mediate organ dysfunction in patients with systemic sclerosis (86). It has been suggested that pericytes may serve as the link between vascular dysfunction and fibrosis in scleroderma patients. Early ultrastructural studies identified profound microvascular changes in tissues from patients with scleroderma, associated with basal lamina thickening and perturbations in pericyte morphology and in the structure of endothelial-pericyte contacts (27). Subsequent studies in tissue sections from systemic sclerosis patients showed that pericytes and fibroblasts proliferated and share expression of common markers (160). These findings may suggest that fibrosis in systemic sclerosis patients may involve fibrogenic activation of pericytes that may express fibroblast-activating mediators, or even undergo fibroblast transdifferentiation (196),(171),(53),(193).

Studies using cells harvested from scleroderma patients and experiments in mouse models support the fibrogenic activation of pericytes. In cutaneous perivascular cells harvested from systemic sclerosis patients, increased expression of ADAM12 (A Disintegrin and Metalloproteinase domain-containing protein 12) was found to be associated with an activated fibrogenic phenotype (36). Moreover, a study investigating the profile of pericytes in patients with lung fibrosis due to systemic sclerosis, perturbed Prostaglandin E2 signaling was suggested to be involved in vascular dysfunction, inflammation and fibrosis (164). Evidence documenting conversion of pericytes into fibroblasts and myofibroblasts is less robust. In a study overexpressing the cytosolic scaffolding protein SARA (Smad Anchor for Receptor Activation) using a constitutive Pdgfrb-Cre driver in a mouse model of cutaneous scleroderma, inhibition of fibrosis was attributed to attenuated pericyte to fibroblast conversion (37). However, the broad expression of Pdgfrb in fibroblasts limits the specificity of the approach for pericyte targeting.

7.2. Pericytes in the fibrotic responses associated with diabetes and aging.

Diabetes and aging are associated with fibrotic remodeling of several different organs, including the heart, kidney and liver (201),(17),(28). Pericyte perturbations have been described in animal models (184) and in patients with diabetes (130),(85) and have been implicated in the pathogenesis of diabetes-associated complications (75), such as diabetic retinopathy (38), in which pericyte dropout may mediate uncontrolled endothelial cell proliferation and neovessel formation (56) contributing to progression of the disease (23).

Moreover, high glucose has profound effects on pericyte phenotype, inducing apoptosis (169), inhibiting contractility (66),(124), and stimulating collagen synthesis (120). Thus, it is tempting to hypothesize that diabetes may disrupt interactions between pericytes and endothelial cells, potentially leading to fibrogenic activation of pericytes that may promote tissue fibrosis. There is currently no in vivo evidence to support this notion. In the db/db mouse model of obesity-associated diabetic cardiomyopathy, no significant pericyte to fibroblast conversion was noted. Lineage tracing experiments using the inducible NG2-CreER driver coupled with a PDGFRa^EGFP reporter line (to identify fibroblasts) showed no evidence of pericyte to fibroblast conversion in 6 month-old db/db mouse hearts (3), Moreover, transcriptomic analysis showed no significant effects of diabetic heart disease on expression of most fibrosis-associated genes by pericytes. Db/db mouse cardiac pericytes only showed upregulation of Timp3 expression (3). In diabetic hearts, the effects of pericytes may predominantly affect the endothelium inducing growth arrest through cytokine-mediated paracrine signaling (71). It should be emphasized that studies in models of diabetes and obesity have significant limitations, related to their translatability to the human disease and to the challenges of long-term follow-up. Thus, the absence of fibrogenic activation in pericytes of relatively young 6 month-old diabetic mice does not exclude significant perturbations at later timepoints.

Interstitial fibrosis is also a common finding in aging organs and may be implicated in the development of renal fibrosis and dysfunction and in the pathogenesis of Heart Failure with preserved Ejection Fraction (HFpEF) (17). Pericyte numbers decline in aging mouse kidneys (101),(187). In the mouse heart, aging reduced pericyte coverage and attenuated expression of RGS5 by pericytes. In vitro and in vivo experiments implicated RGS5 loss in the acquisition of a fibrogenic profile by pericytes (194). Moreover, in the skeletal muscle, aging was associated with perturbed reparative function and increased fibrogenic activity of pericytes, contributing to accentuation of fibrosis (19). Thus, in aging tissues, pericytes may undergo alterations that promote their fibrogenic activation.

8. Conclusions and future directions.

Pericytes play roles that extend beyond their classical functions in regulating microvascular structure and hemodynamics. A large body of evidence suggests that in many fibrotic conditions, pericytes undergo activation and adopt a pro-inflammatory and matrix-remodeling phenotype, with the capacity to activate fibroblasts. Fate mapping studies have suggested that specific subpopulations of pericytes may also undergo transdifferentiation into fibroblasts. However, the prevalence and significance of these transitions may be organ- and disease-specific and their documentation is dependent on the criteria used to define pericytes and to confirm fibroblast identity.

A deeper understanding of pericyte biology is essential for designing therapeutic strategies aimed at attenuating ECM deposition while preserving microvascular integrity in fibrotic disorders. Looking ahead, several critical questions remain unanswered. Within the complex vascular niche, can we precisely identify pericyte subsets with fibrogenic potential? Are such pericytes universally involved in fibrotic responses across diverse organ systems, or is their pro-fibrotic role restricted to specific pathological conditions? Do the tissue-specific properties of pericytes underlie the differing capacities for repair, regeneration, and fibrosis observed among organs? Furthermore, can therapeutic strategies targeting pericytes exert broad-spectrum anti-fibrotic effects, or will their benefits be confined to disorders characterized by prominent perivascular fibrosis? It is increasingly appreciated that pericytes play a major role in pathologic responses in many different organs. Their contributions are no longer viewed as a niche interest confined to vascular biology but are now integral to broader investigations into the mechanisms of tissue injury, repair, and fibrosis across multiple organ systems.