Myofibroblast Differentiation During Fibrosis: Role of NAD(P)H Oxidases

Abstract

Progression of fibrosis involves interstitial hypercellularity, matrix accumulation and atrophy of epithelial structures, resulting in the loss of normal function and ultimately organ failure. There is common agreement that the fibroblast/myofibroblast is the cell type most responsible for interstitial matrix accumulation and consequent structural deformations associated with fibrosis. During wound healing and progressive fibrotic events fibroblasts transform into myofibroblasts acquiring smooth muscle features, most notably the expression of alpha-smooth muscle actin (α-SMA) and synthesis of mesenchymal cell related matrix proteins. In renal disease, glomerular mesangial cells also acquire a myofibroblast phenotype and synthesize the same matrix proteins. The origin of interstitial myofibroblasts during fibrosis is a matter of debate where the cells are proposed to derive from resident fibroblasts, pericytes, perivascular adventitial, epithelial, and/or endothelial sources. Regardless of the origin of the cells, TGF-β1 is the principal growth factor responsible for myofibroblast differentiation to a profibrotic phenotype and exerts its effects via Smad signaling pathways involving MAPK and Akt/PKB. Additionally, reactive oxygen species (ROS) play important roles in progression of fibrosis. ROS are derived from a variety of enzyme sources of which the NAD(P)H oxidase family has been identified as a major source of superoxide and hydrogen peroxide generation in the cardiovasculature and the kidney during health and disease. Recent evidence indicates that the NAD(P)H oxidase homolog Nox4 is most accountable for ROS-induced fibroblast and mesangial cell activation where it has an essential role in TGF-β1 signaling of fibroblast activation and differentiation into a profibrotic myofibroblast phenotype and matrix production. Information on the role of ROS in mesangial cell and fibroblast signaling is incomplete and further research on myofibroblast differentiation during fibrosis is warranted.

Introduction

Progression of fibrosis is remarkably similar in most organs involving pathogenic processes of interstitial hypercellularity, matrix accumulation and atrophy of epithelial structures that lead to loss of normal function and organ failure. Three phases of fibrogenesis have been described in the kidney [1] that predictably apply to other organ systems as well. There is an induction phase characterized by infiltration of inflammatory cells, principally mononuclear cells or macrophages, that release profibrogenic cytokines and growth factors. This initial event is followed by the activation of fibroblasts to transition into myofibroblasts that also secrete biological active products and matrix proteins. The third phase is ongoing synthesis and accumulation of matrix despite resolution of the primary stimulus. Central to the activation of fibroblast to a profibrotic myofibrobroblast is transforming growth factor beta (TGF-β1) that can be released through paracrine and autocrine pathways. The roles inflammatory, tubular and vascular cells play in fibrosis is complex and multifactorial, a topic adroitly covered in several recent reviews [2–4]. In the kidney, interstitial fibrosis is a common pathway of progressive renal diseases leading to end stage renal disease regardless of the eitiology [5]. Also, interstitial fibrosis is the strongest morphologic predictor of clinical outcome and is most tightly linked to progression of disease [6]. Despite primary glomerular origin of most renal diseases, interstitial involvement indicates a more ominous outcome [7]. With increasing morbidity associated with fibrosis related to obesity, diabetes, and heart disease there is imminent importance to understand the pathophysiology of the myofibroblast for the design of future therapeutics. This review will discuss biological factors that lead to myofibroblast differentiation and the origin of these cells in the interstitium. Attention is placed on TGF-β and reactive oxygen species (ROS) as principal mediators of fibrosis with special emphasis on the NAD(P)H oxidase of the Nox family (particularly Nox4) as requisite elements in signaling of TGF-β-induced myofibroblast differentiation.

Myofibroblasts in Fibrosis

Activation and transformation of fibroblasts into myofibroblasts was first described by Gabbiani showing that after injury, during wound healing, or progressive fibrotic events, activated cells acquired smooth muscle features including expression of desmin, caldesmon, and SM-myosin heavy chains and an actin isoform (alpha-smooth muscle actin, α-SMA) [8;9]. Consequently, α-SMA positive myofibroblasts were identified as the primary cell type responsible for interstitial matrix accumulation in fibrotic diseases including the kidney [8–13]. Moreover, an α-SMA phenotype is considered a useful marker for myofibroblast differentiation in a number of disease settings [8–27]. Similarly, an α-SMA + phenotype in intrinsic mesenchymal cells from a variety of organs can be induced by prolonged culture, exposure to cytokines such as interleukin (IL)-1, IL-6, and tumor necrosis factor (TNF-α) and growth factors including platelet-derived growth factor (PDGF), angiotensin II, and TGF–β. These include liver fat storing cells (Ito cells), periportal fibroblasts, pancreatic stellate cells, breast stromal cells, lung fibroblasts, and renal mesangial cells and kidney fibroblasts [28–35]. These cells express little to no α-SMA in their resting state, however, upon activation in vitro or during disease, they highly express this protein along with extracellular mesenchymal matrix proteins collagen I, collagen III, and cellular fibronectin. More recently, epithelial cells have been reported to transition into matrix producing fibroblasts and α-SMA + fibroblasts (myofibroblasts), see section on EMT below. Collectively, these cells are referred to as “myofibroblasts” despite their origin.

Control of the transition of fibroblasts to myofibroblasts is complex and involves a variety of chemical factors, extracellular matrix proteins, and mechanical microenvironment [8;13], see below. Hinz [13] categorized the differentiation of fibroblasts into myofibroblasts functionally, by a two-step process where fibroblasts are activated to form proto-myofibroblasts with contractile stress fibers in response to locally released cytokines or changes in the composition, organization and mechanical property of the extracellular matrix (ECM). Proto-myofibroblasts further differentiate into myofibroblasts by neo-expressing α-SMA [13]. These steps are believed to be responsible for migration and re-population of damaged tissues followed by matrix synthesis and contraction of connective tissue during remodeling. Prolonged exposure to an insult results in inappropriate matrix expansion by myofibroblasts leading fibrosis and loss of tissue function [36].

Activation of Mesangial Cells and Interstitial Fibroblasts in Renal Disease

The bulk of excess extracellular matrix synthesis during the progression of chronic renal disease can be attributed to two similar mesenchymal cell types, the glomerular mesangial cell and the interstitial fibroblast. Mesangial cells have been suggested to be specialized pericytes [37] juxtaposed to the capillaries within the renal glomerulus and fibroblast occupy the peritubular interstitial space. Other than their location, the two cell types are quite similar in their response to profibrotic stimuli during disease by acquiring an α-SMA positive myofibroblast phenotype and the subsequent synthesis of matrix proteins including fibronectin, laminins and type IV collagen. When activated, both cell types also acquire the expression of type I and type III collagens [10–12;16;21;23;38–40] and activation of both cell types upregulates an alternatively spliced isoform of fibronectin containing extra domain EIIIA (equivalent to ED-A in human tissues) that plays an essential role in myofibroblast differentiation (see below). Also, α-SMA deficient mice, show enhanced cell proliferation and type-1 procollagen expression in the interstitium after UUO and in mesangioproliferative glomerulonephritis, suggesting that the function of α-SMA in these two mesenchymal cell types is similar (41). The interpretation of these seemingly paradoxical results was that the presence or absence of an α-SMA phenotype have opposing functions of maintaining a cell in a contractile or productive state, respectively. Thus, the relative presence of α-SMA may provide suppressing or accentuating roles in fibrosis progression (41).

The Origin(s) of Myofibroblasts

The origin(s) of interstitial myofibroblasts in fibrosis is currently under extensive investigation and scientific reports in kidney, liver and lung indicate the cells may be derived from one or more sources (Figure 1). Originally, myofibroblasts were thought to be derived from intrinsic fibroblasts that reside within the interstitium, but this simple concept has evolved into a much more complex and controversial area of investigation. The origins of interstitial myofibroblasts now include activation of resident fibroblasts or pericytes [19;20;22–26;42], expansion of perivascular fibroblasts [19;20;22;43–48], infiltration of circulating bone marrow-derived fibrocytes [49–51], epithelial to mesenchymal transition (EMT, see below) and/or endothelial-mesenchymal (EndoMT) transition [52;53]. Each mode of interstitial encroachment results by myofibroblast migration, proliferation, and ultimately matrix expansion within the interstitial space. EMT has become the preeminent hypothesis of renal fibroblasts and myofibroblast encroachment into the peritubular interstitium during kidney fibrosis (reviewed in [54–58]. EMT is based on the reversal of an essential developmental process, known as mesenchymal to epithelial transition (MET) whereby embryonic mesenchymal cells differentiate into the tubular epithelium [59]. During the process of EMT, mechanisms are believed to exist to reverse the MET and epithelial cells lose specialized differentiation markers such as E-cadherin and cytokeratins, and acquire mesenchymal proteins including α-SMA, vimentin, and fibroblast specific protein-1 (FSP-1) that has been associated with cell migration and invasiveness. According to this hypothesis, the transformed epithelial cells synthesize matrix proteinases that digest basement membrane proteins, thereby allowing their migration into the interstitial space where they proliferate and synthesize extracellular matrix proteins commonly associated with mesenchymal cells [54–58]. Such local conversion of epithelium was reported by Strutz et al after the discovery the expression of FSP-1 by tubular epithelial cells during tubulointerstitial nephritis [60]. Since this original description, expression of FSP-1 in injured proximal tubules, with or without co-localization of mesenchymal markers, has been taken as evidence of EMT as the major contributor to fibrosis in numerous models of renal fibrosis including ureteral obstruction, renal ablation, diabetic nephropathy, nephrotoxic serum nephritis, and polycystic kidney disease [54–58, see listings in 56;61;62]. A role for EMT in liver and lung fibrosis has subsequently been described [63;64].

Figure 1. Proposed origin(s) of interstitial myofibroblasts during fibrosis.

Proposed origin(s) of interstitial myofibroblasts during fibrosis. Myofibroblasts responsible for interstitial expansion and structural damage during progressive organ injury have been proposed to be derived from one or more sources: (1) activation of resident fibroblasts or pericytes, (2) infiltration of circulating bone marrow-derived fibrocytes, (3) expansion of perivascular adventitial fibroblasts, (4) endothelial–mesenchymal (EndoMT) transition, and/or (5) epithelial-to-mesenchymal transition (EMT). Transforming growth factor-beta (TGF-β) released via paracrine or autocrine pathways induces myofibroblast differentiation expressed by the acquisition of an alpha-smooth muscle actin (α-SMA) phenotype and consequent synthesis of mesenchymal matrix proteins collagen type I (Col I), collagen type III (Col III), and fibronectin EIIIA (FN). Nox NAD(P)H oxidase (Nox4)-derived reactive oxygen species (ROS) have a central role in TGF-β signaling of myofibroblast differentiation.

The existence of EMT as a source of interstitial fibroblast (myofibroblast) encroachment has recently become a matter of debate (61;62;65). A large number of studies postdating the inception of the EMT hypothesis reported that myofibroblasts and their secreted matrix proteins are located exclusively in the interstitial space during experimental fibrosis without evidence of epithelial locations [17–26;66–74]. Moreover, recent studies specifically designed to determine the origin of myofibroblasts by utilizing state-of-the-art protein marker, gene reporter or epithelial tracking strategies observe myofibroblasts exclusively within the peritubular interstitium in the absence of evidence for EMT in a variety of models of renal fibrosis [19;20;22–26;42;66;72–75]. The above studies are in contrast to a number of reporter studies illustrating a tubular origin of myofibroblasts or α-SMA negative collagen producing fibroblasts that migrate into the interstitial space (76, see the following reviews for a comprehensive discussion 56;61;62). Also, elimination of cells expressing FSP-1 was reported to reduce fibrosis (77).

Much of the controversy over the origin of interstitial cells rests on selection and specificity of markers for cells destined to become interstitial matrix producing (myo)fibroblasts. Arguments have been made that FSP-1, also known as S100A4 is a superior and more consistent marker of tissue fibroblasts because α-SMA does not define the “universe of fibroblasts” in that not all fibroblasts express α-SMA” [60;76] and α-SMA is not expressed in all FSP-1 positive fibroblasts [54;78;79]. Indeed, the presence of FSP-1 in epithelial cells is now used to define EMT [61]. On the other hand, FSP-1, also known as S100A4 and a variety of other names, is not only observed in fibroblasts as the name implies but can be expressed by a wide variety of cell types [61,80–82], including resident fibroblasts, endothelial cells, monocytes, or macrophages residing in or near the renal interstitial space during fibrosis [19;20;25;26;66;83]. Conversely, others show no cross-reactivity of FSP-1 with macrophages in renal interstitial fibrosis (76;84;85). An ongoing debate exists over the selection of antibodies and materials used to detect S100A4, macrophages, and other cell types and whether “preferred” reagents are being used (25;66;84). S100A4 is also associated with cellular behaviors other than migration or invasiveness including cell stress, proliferation and apoptosis [80;81;86] that are present during tubular injury [42;69;87]. Thus, expression of FSP-1 may also be viewed as a marker for cell activation and reflect a generalized response to cell injury as well as cell migration and invasiveness. Varying viewpoints exist for and against the use of either α-SMA or S100A4 as suitable markers of interstitial matrix producing cells. Answers to this debate will require additional experimentation or the discovery of specific markers to determine the contributions of the various aforementioned cell types in renal interstitial fibrosis.

We are of the opinion that interstitial myofibroblasts are derived from local fibroblasts, pericytes, or perivascular cells [19;20;22;72,73]. A cellular location of myofibroblasts in the perivascular space also supports the histological descriptions that mesenchymal cell-derived collagen type I or type III, and/or fibronectin are first expressed in the perivascular spaces of arteries or arterioles in fibrosis related to cardiovascular [88], liver [89], lung [90] and kidney [19;22;43–48] disease. A perivascular source of myofibroblast encroachment becomes noteworthy because myofibroblasts in the adventitium is a major site of reactive oxygen species (ROS) in the aorta during vascular disease. The adventitia layer is less structured in smaller arteries and arterioles, but like the aortic adventitia, it may be the source of perivascular myofibroblasts during renal disease.

Myofibroblast Differentiation: Roles of TGF-β and Alternatively Spliced Fibronectin EIIIA

A number of growth factors are associated with myofibroblast differentiation including PDGF, angiotensin II, CTGF, and TGF-β1 [13]. TGF-β1 is frequently associated with a myofibroblast α-SMA phenotype in liver, lung, and kidney disease [8;91–93] and has been determined to be the preeminent growth factor responsible for fibroblast activation and matrix synthesis in vitro and during vascular disease and fibrosis [93–95]. Also, TGF-β1 plays a pivotal role in fibrogenesis in which a number of growth factors, including PDGF and angiotensin II, exert their effects by directly stimulating TGF-β1 production [92–98]. Additionally, TGF-β1 directly promotes myofibroblast development by inducing expression of α-SMA phenotype [34;35;95;99;100]. Similarly, TGF-β1 is considered the foremost growth factor for induction of EMT in culture and in vivo [101].

TGF-β1 contributes to fibrosis by the direct activation of myofibroblast synthesis of fibronectin, laminins collagen types I, III, IV and VI. Of these matrix proteins, collagen type VI and fibronectin (EIIIA) are closely associated with myofibroblast differentiation in vitro and in disease [13]. Fibronectin EIIIA is abundantly expressed during embryogenesis [21;102] and at the margins of healing wounds [103]; whereas, this domain is spliced out of plasma fibronectin (hepatocyte-derived) and many tissue-specific cells in adult organs [104], suggesting that this matrix protein has important and specialized functions in tissue remodeling [105]. α-SMA myofibroblasts and fibronectin EIIIA frequently co-localize in fibrotic disease as well as in glomerular and interstitial lesions in kidney diseases [21;22;38;39;106;107].

TGF-β1 differentially regulates the expression of fibronectin EIIIA [108] and induces expression of α-SMA in a variety of mesenchymal cells in culture including renal fibroblasts and mesangial cells [34;100;109]. Moreover, fibronectin EIIIA is mandatory for TGF-β1 induction of myofibroblast differentiation and α-SMA expression [109;110]. The close proximity of the EIIIA splice site to the Arg-Gly-Asp-Ser (RGDS) cell binding domain suggests that this variant has specific functional roles [105]. Certainly, fibronectin EIIIA is closely associated with activated cells undergoing high rates of migration, proliferation, and differentiation [21;38;39;102;103;105] and fibronectin EIIIA is considered the most reliable marker of myofibroblast-derived extracellular matrix [13].

Reactive Oxygen Species (ROS) in Fibroblast and Vascular Pericyte Activation

Accumulating evidence also indicates that oxidative stress resulting in generation of ROS, mainly in the form of superoxide, and hydrogen peroxide play a significant role in the initiation and progression of cardiovascular and renal disease [111–113]. Superoxide and its dismuted derivative hydrogen peroxide are formed by the univalent reduction of oxygen, generally mediated by several enzyme systems such as xanthine oxidase, uncoupled nitric oxide synthase, myeloperoxidase, mitochondrial respiratory oxidases, lipoxygenases, and the NAD(P)H oxidases of the Nox family. Exacerbated production of ROS may directly cause macromolecular damage or function as signaling molecules that induce cellular damage or fibrogenic responses through stress-sensitive pathways. It is important to mention that there is growing evidence indicating that ROS are not always damaging, especially at low levels, and in fact may act as physiological signaling molecules.

Examples of ROS-induced fibrogenic responses include distinct cell functions such as hypertrophy, migration, proliferation, apoptosis and regulation of extracellular matrix [34;106;111;112;113]. Perivascular adventitial fibroblasts in the rabbit and rat aorta [112;114–116], pericytes from the microvasculature [117], and renal myofibroblasts [34] are a major source of superoxide. These observations become highly relevant to fibrogenesis because not only are fibroblast responsible for ROS generation, but ROS are directly linked to transmodulation of fibroblasts to α-SMA expressing myofibroblasts [34;35;115;118;119]. Moreover, the perivascular space is noticeably reactive and is the location where myofibroblasts have been shown to first appear during the course of renal fibrosis (see above).

A role for ROS in the pathogenesis of chronic renal disease has been observed in a number of diverse models including diabetic nephropathy, proliferative glomerulonephritis; IgA nephropathy, hypertensive renal disease, and fibrosis due to transient ischemic injury, renal ablation, and ureteral obstruction (thoroughly referenced in [113;120]). Importantly, the NAD(P)H oxidases of the Nox family have recently gained heightened attention as principal mediators of injury associated with vascular and renal disease [111;112;120–123]. Moreover, NAD(P)H oxidase has been identified as the enzyme system most responsible for superoxide generation by adventitial fibroblasts after vascular injury of the aorta [112;116;124], and is now recognized as a key mediator of cell proliferation and matrix accumulation in renal disease [106;120;125–127]; a topic that will be reviewed in depth below.

NAD(P)H Oxidases of the Nox Family as Major Sources of ROS in the Cardiovasculature and the Kidney

Early studies on NAD(P)H oxidases were performed in neutrophils and phagocytic cells, investigating the respiratory burst NADPH oxidase system [128]. The phagocyte respiratory burst oxidase or phagocyte NADPH oxidase catalyzes the NADPH-dependent reduction of molecular oxygen to generate superoxide anion which is dismuted to hydrogen peroxide [128;129]. The phagocyte oxidase consists of two plasma membrane-associated proteins, gp91^phox (the catalytic subunit) and p22^phox which comprise flavocytochrome b₅₅₈, as well as cytosolic factors, p47^phox, p67^phox, p40^phox and the small GTPase Rac. Homologues of gp91^phox, termed Nox (for NAD(P)H oxidase) proteins, have been found in all vascular and renal cells [111;1112;120;121;130–134] (Figure 2). To date, the Nox family comprises seven members: Nox1 to -5, of which Nox2 is gp91^phox, and the dual oxidases Duox1 and -2 [122;130;135;136] (Figure 2). Nox1, gp91^phox/Nox2 and Nox4 are the NAD(P)H oxidase isoforms that are predominantly expressed in the cardiovascular and cardiorenal systems. Note that the calcium-dependent isoform Nox5 is also found in the human vasculature but the fact that the enzyme is not present in rodents has hampered the investigation of its role in cardiovascular pathologies [121;134]. Moreover, no data related to Nox5 expression in the human kidney are yet available.

Figure 2. Structure and molecular organization of the cardiorenal nicotinamide adenine dinucleotide phosphate (NAD(P)H) oxidases of the Nox family.

The top right panel illustrates the topology and the enzymatic reaction catalyzed by the Nox enzymes. The other panels represent the molecular structure of the different isoforms of Nox oxidases predominantly expressed in the cardiorenal system, gp91phox/Nox2, Nox1, and Nox4. All cardiorenal Nox proteins form a complex with p22phox, but the cytosolic subunits differ from the Nox oxidase isoforms. FAD, flavin adenine dinucleotide; H2O2, hydrogen peroxide; O₂^•−, superoxide.

While the mechanism by which the activity of the Nox enzymes is regulated in cardiovascular or renal cells and how they generate reactive oxygen species (ROS) is not fully understood, a number of regulatory subunits have been identified. Importantly, the isoforms Nox1, gp91^phox/Nox2 and Nox4 (but not Nox5) appear to require p22^phox as a docking subunit [130;135;136]. Activation mechanisms for Nox1 are similar to that of gp91^phox/Nox2, and involve complex formation with regulatory cytosolic subunits upon agonist stimulation. Although Nox1 seems to primarily interact with the p47phox homologue NoxO1 (Nox organizer 1), the p67phox homologue NoxA1 (Nox activator 1) and Rac upon activation, it was reported that p47^phox and p67^phox can partially supplant NoxO1 and NoxA1, respectively [121;122;130;1132;134–136] (Figure 2).

The most abundant Nox isoform in the cardiorenal system is Nox4 [120;130;131;135]. The isoform Nox4/Renox was cloned from the kidney and is highly expressed in renal tubules, fibroblasts, glomerular mesangial cells and podocytes [34;106;137–141]. Nox4 is a 578 amino acid protein that exhibits 39 % identity to the phagocyte gp91^phox/Nox2 with special conservation in the six membrane-spanning regions and binding sites for NADPH, FAD and heme, the electron transfer centers that are required to pass electrons from NAD(P)H to oxygen to form superoxide and hydrogen peroxide [121;130;133;135–138]. Evidence to date suggest that Nox4 heterodimerization with p22^phox is sufficient to enhance the enzyme activity and it does not require cytosolic subunits that are essential for other Nox isoforms [130;135;136;142;143]. Moreover, it was shown that the Nox4 dehydrogenase domain exists in a conformation that allows the spontaneous transfer of electrons from NADPH to FAD, a property explaining the constitutive activity of the enzyme [144]. Hence, Nox4 has been referred to as a “constitutively active” enzyme that is regulated primarily at the level of its expression in response to various stimuli [130;133;136]. As a corollary, the overall ROS output of Nox4 may be directly Nox4 governed by its expression level. Interestingly, it seems that transcriptional events are involved in the chronic control of Nox4 and protein expression [118;141;145–153] whereas Nox4 is acutely regulated through translational mechanisms without change in its mRNA levels [34;154–157]. It was suggested that stimulation with certain agonists cause early Nox4 protein accumulation independent of mRNA transcription by promoting translation of existing mRNA copies [154;156;157]. These types of translational-mediated regulation mechanisms were also described for other proteins in the kidney and implicated in the pathogenesis of renal disease [154;158–160]. Importantly, the fact that release of ROS by Nox4 does not appear to be triggered acutely after recruitment of cytosolic subunits upon agonist stimulation as for gp91^phox/Nox2 and Nox1 does not mean that oxidant production by Nox4 is not controlled by activation of receptors by their ligands (i.e. agonist stimulation). Indeed, it has been reported that Nox4 expression is rapidly upregulated by binding of agonists such as TGF-β [34;118], angiotensin II [154], insulin or insulin-like growth factor [156;161] to their membrane receptors. Since the enzyme is constitutively active, receptor activation by these ligands results in increased Nox4-dependent ROS production. Finally, it should be noted that even if Nox4 regulation may occur primarily at the expression level, Nox4 regulatory proteins that enhance its activity named Poldip2 and NoxR1 have been recently identified [162;163]. A potential implication of Rac in the control of Nox4 function was suggested in endothelial cells and mesangial cells [139;152;164] but it remains a contentious topic.

Nox4 complexity is further illustrated by the difficulty to identify the type of ROS (superoxide or hydrogen peroxide) produced by the enzyme. While it is clear that gp91^phox/Nox2 and Nox1 release primarily superoxide, the nature of the ROS produced by Nox4 in cells or tissues is controversial. It was documented that Nox4 generates mostly hydrogen peroxide in vascular smooth muscle cells or heterologous overexpression systems [130;133;145;165;166] while other studies in vascular smooth cells as well as in cardiac or renal cells and tissue were able to detect Nox4-dependent superoxide and hydrogen peroxide production [34;106;118;141;155;167–169]. It was proposed that Nox4 differs from other Nox enzymes because the superoxide produced by Nox4 is rapidly converted to hydrogen peroxide, thereby rendering superoxide release from the enzyme practically undetectable [133;165]. It is of interest to note that Nox4 has been shown to be associated to intracellular compartments or organelles such as endoplasmic reticulum, mitochondria or nucleus [121;130;133;143;147;155;170]. Although effective probes designed to measure intracellular superoxide exists, the particular subcellular localization of Nox4 may render more difficult the detection of the superoxide generated by the oxidase. In addition, negatively charged superoxide anion does not permeate biological membranes whereas hydrogen peroxide obtained after superoxide dismutation is readily permeable [130;145].

Nox-Derived ROS as Mediators of Cardiovascular and Kidney Disease. Role of Nox4 in Myofibroblast and Mesangial Cell Activation

A growing array of evidence suggests that Nox enzymes contribute to the pathogenesis of cardiovascular and renal disease, including hypertension, atherosclerosis, renal or cardiac fibrosis as well as diabetic nephropathy or cardiomyopathy [106;111;120;121;130–134;136;163;171–173]. This is due to the fact that multiple stimuli and agonists implicated in these pathologies like TGF-β, PDGF, angiotensin II, hyperglycemia, thrombin, urotensin, oxidized LDL, IGF-I, vascular endothelial growth factor (VEGF), and aldosterone have been shown to alter the activity or expression of the Nox proteins and subunits, and ultimately the amount of ROS produced [34;106;114;118;120;121;130;141;147;148;151;152;154–156;174–180]. For instance, upregulation of gp91^phox/Nox2, Nox1, Nox4 and p22^phox (mRNA and protein) together with increased superoxide or hydrogen peroxide generation has been reported in response to angiotensin II in the vasculature and the kidney in vitro as well as in vivo [111;112;120;121;148;150;154;165;174;181–183]. Similar to angiotensin II, enzyme activity or gp91^phox/Nox2, Nox1, Nox4 and p22^phox expression are upregulated in response to hyperglycemia in vascular and renal cells as well as in experimental models of diabetes [106;112;141;148;151;152;155;171;181–192]. Although most of the vascular and renal Nox oxidases are regulated by pro-fibrotic stimuli such as TGF-β, angiotensin II, or hyperglycemia, it is important to point out that Nox4 appears to play a predominant role in the activation of fibroblasts into the myofibroblast phenotype and the subsequent fibrotic processes taking place in vitro as well as in vivo. Interestingly, it is established that TGF-β, the most potent pro-fibrotic factor, specifically increases the expression of Nox4 and ROS production in a myriad of cell types including smooth muscle cells, endothelial cells, hepatocytes and fibroblasts [34;35;118;169;193–196]. Moreover, we have documented that increased Nox4 expression accounts for TGF-β-, angiotensin II- and high glucose-mediated oxidative stress and renal cell activation that is accompanied by increased extracellular matrix protein and α-SMA expression [34;106;154;155]. More specifically, Nox4 was clearly identified as a critical mediator of high glucose- or angiotensin II-induced mesangial cell activation [106;154;155] as well as of the effects of TGF-β on kidney fibroblast differentiation [34]. However, it should be noted that it has not been directly demonstrated that the oxidase is implicated in TGF-β signaling in mesangial cells. Similar observations were made in cardiac fibroblasts where it was shown that the induction of Nox4-dependent ROS production by TGF-β is required for the conversion of fibroblasts into myofibroblasts, which in turn produce a large amount of extracellular matrix [118]. Nox4 is also closely linked to TGF-β1-induced cytoskeletal changes including filipodia formation and F-actin assembly in human vascular endothelial cells [196] and with stress fibers in differentiated vascular smooth muscle cells that constitutively express α-SMA [167], indicating that Nox-derived ROS are essential to the maintenance of cell shape and differentiation.

The fact that these factors are known as critical mediators of vascular, cardiac or renal fibrosis further supports the concept that Nox4 is the principal source of ROS that promote oxidants-mediated tissue injury and fibrotic processes associated to cardiovascular diseases. A causative relationship between Nox4-derived ROS and fibrogenic responses to renal injury was demonstrated in vivo by showing that treatment of type 1 diabetic rats with Nox4 antisense oligonucleotides reduced ROS production and prevented fibronectin EIIIA accumulation in the kidney [106]. Similarly, genetic silencing of Nox4 with small interfering RNA abrogates myofibroblast differentiation and fibrogenesis in a murine model of lung fibrosis [35]. Given that TGF-β-dependent myofibroblast or modified pericyte activation is a critical component of the fibrogenic processes leading to tissue injury in these diseases, it is reasonable to speculate that the deleterious effects of TGF-β may be also mediated by Nox4-derived ROS in vivo. However, despite that it is established that Nox4 is a target of TGF-β in cultured renal cells, direct in vivo evidence of Nox4 recruitment as an intermediate of the pathway linking TGF-β to glomerular and tubulointerstitial fibrosis is not yet available.

TGF-β1 and Redox Signaling in Fibroblast and Vascular Pericyte Activation

Smad/ERK Transcriptional Pathway in Mesangial Cell and Fibroblast Signaling

TGFβ-1 exerts its effects on the cell surface by binding TGFβ receptor type II TGFR-II and subsequent phosphorylation of a Type 1 receptor subunit forming a heterodimeric complex. Both TGFR-I and TGFR-II possess tyrosine kinase activity that subsequently signal through a canonical pathway involving the Smad family of transcriptional activators [197–200]. In fibroblasts, TGFβ-1 regulation of α-SMA transcription and myofibroblast differentiation is mediated via TGFR1 phosphorylation of Smad2/3 that subsequently complexes with Smad4 and translocates to the nucleus where the dimer binds to the promoter region of the α-SMA gene [13]. Transcriptional control is regulated by a variety of transcription factors and the upregulation of an inhibitory Smad7 [13]. Phenotypic and functional changes associated with TGF-β1-induced fibroblast terminal differentiation are differentially regulated by Smad proteins [200]. In some cell types, ERK is required for activation of the Smad pathway [201]. In mesangial cells and fibroblasts, TGF-β/Smad signaling (Smad 2/3) is tightly controlled by MAP kinase (Ras/MEK/ERK) signaling cascades [202;203]. Also, ERK and Akt/PKB act as alternative pathways in TGFβ-1 signaling of matrix proteins [203–208], making these three signaling proteins the predominant transduction pathways by this growth factor. Additionally, TGF-β signaling of tubular epithelial cells in EMT may involve both Smad dependent through integrin-linked kinase and Smad independent signal transduction pathways such as p38 mitogen-activated protein kinase (MAPK) [199;209;210].

Role of ROS and Nox4 in Mesangial Cell and Fibroblast Signaling (Figure 3)

Figure 3.

Pathways of Nox-dependent signal transductions implicated in mesangial cells (MCs) (left panel) and renal fibroblasts (right panel) activation. See text for detail. Ang II, angiotensin II; ERK, extracellular receptor kinase; H₂O₂, hydrogen peroxide; MEK, MAP kinase kinase; NFAT, nuclear factor of activated T cell; O₂^•−, superoxide; PKB, protein kinase B; ROS, reactive oxygen species; α-SMA, alpha-smooth muscle actin; TGF-β, transforming growth factor-beta.

The observations that both TGF-β1 and ROS activate fibroblasts transition to an α-SMA myofibroblast and enhanced matrix synthesis [34;35] indicate that these two molecules are interrelated and may share signaling pathways in myofibroblast differentiation during fibrosis.

ROS are known to activate many tyrosine and serine threonine kinases [133;211;212], indicating that products of oxidation play a central role in cell signaling. Indeed, a number of hormones, growth factors, and cytokines induce the production of hydrogen peroxide in a variety of target cells [211;212]. A regulatory role for ROS in PDGF- and angiotensin II-induced signal transduction is well documented [213;214]. Also, TGF-β1-induced fibronectin synthesis by mesangial cells occurs via calcium mobilization and calmodulin/calcineurin-dependent activation of transcription factor NFAT in a ROS-dependent manner [215]. Similar results were observed for TGF-β1 induced tubule epithelial cells and renal fibroblasts [215].

NAD(P)H oxidases of the Nox family have received heightened interest as source of ROS signaling in a number of cell types. [120;121;133;173;178;216]. Receptor binding of PDGF, angiotensin II, TGF-β and/or TNF-α, rapidly activates NAD(P)H oxidase followed by intracellular ROS generation and activation of protein tyrosine kinases, serine/threonine kinases, phospholipases (e.g. phospholipases C and A₂), and calcium-dependent pathways [120;121;133;173;178;216]. NAD(P)H oxidase signaling pathways can be complex and diverse often with conflicting data among cell type, signaling pathway and cell behavior examined [120;121;133;173;178;216]. Accumulating evidence points toward the NAD(P)H oxidase of the Nox family and particularly Nox4 as the predominant enzyme source for ROS generation in fibrotic disease, placing this enzyme family in a pivotal position in cell signaling (see above). Studies by our group indicate that angiotensin II-induced ROS generation in mesangial cells acts principally through Nox4 as an upstream activator of ERK1/ERK2 [140], Pyk-2/Src/PDK-1 [154], Akt/PKB and/or p70^S6K [139;154;217;218]; pathways that lead to cell hypertrophy and increased protein synthesis and/or fibronectin expression. (Figure 3). More specifically, Pyk-2 appears to act as a molecular scaffold, binding to both PDK-1 and Src, thereby allowing Src to tyrosine-phosphorylate and activate PDK-1, that in turn activate its downstream effectors, Akt/PKB and p70^S6K [154]. Interestingly, Nox4 and p22^phox both contribute to angiotensin II-dependent oxidative stress and fibronectin accumulation in mesangial cells [139;154;218]. Because it is known that p22^phox interacts with Nox4 and enhance its activity (see above), it is reasonable to think that in these cells, Nox4 and p22^phox may form a complex that accounts for angiotensin II-induced ROS generation and the subsequent fibrotic response. However, given that gp91^phox/Nox2 and Nox1 also interact with p22^phox, their role cannot be rule out in conditions where these subunits are also expressed in the cells. These in vitro studies were extrapolated in vivo to type 1 diabetic nephropathy showing co-localization of Nox4 with α-SMA positive mesangial cells and that antisense oligonucleotides to Nox4 reduced ERK1/2 and Akt/PKB phosphorylation and inhibited glomerular hypertrophy and fibronectin EIIIA expression [106]. Interestingly, other studies with mesangial cells exposed to advanced oxidation protein products place a NAD(P)H oxidase likely to be Nox4 distal to PKC activation, resulting in ECM overproduction and upregulation of TGF-β1 [219], raising the possibility that Nox enzymes may also function as upstream modulators of TGF-β1 synthesis. As mentioned earlier, we would like to point out that the role of Nox4 in TGF-β1-mediated ROS generation and matrix protein accumulation has not been directly established in mesangial cells. However, it was documented that p22^phox is required for TGF-β1-induced ROS production [220] and that Src, PDK-1 or Akt/PKB, targets of Nox4-derived ROS, play a role in TGF-β1-dependent matrix accumulation [205;221] in mesangial cells. Thus, given that TGF-β1 is a major regulator of Nox4 in most of the cell type examined (see above), it is tempting to speculate that as for angiotensin II the p22^phox/Nox4 complex may also mediates TGF-β1 fibrotic effects in mesangial cells. It is appealing to consider that a Nox4- and p22^phox-containing oxidase may be a pivotal signal transducer commonly shared by both fibrotic stimuli in mesangial cells. This is highly relevant for the design of therapeutic strategies to prevent the initiation or progression of fibrotic renal diseases. Whether TGF-β1-induced mesangial cell activation share portions of these ROS signaling cascades in Smad-related cellular events remains to be elucidated.

Studies with fibroblasts derived from heart, lung, and kidney also indicate that Nox4 is central to TGF-β1-induced ROS generation and myofibroblast differentiation to a profibrotic phenotype. For example, TGF-β1-induced transition of cardiac or pulmonary fibroblasts to α-SMA positive myofibroblasts is dependent on Nox4 regulation of Smad2/3 [35;118]. Nox4 small interfering RNA or dominant negative Smad3 significantly reduced these effects in vitro [35;195] as well as myofibroblast expansion and fibronectin synthesis in bleomycin induced fibrosis in vivo [35]. We also showed that Nox4 is the predominant Nox oxidase homolog in kidney myofibroblast differentiation and expression of fibronectin (EIIIA) [34]. However, unlike cardiac and pulmonary myofibroblasts, our results place Nox4 downstream of Smad3 and proximal to ERK (Figure 3). Chemical inhibitors of NAD(P)H oxidase and Nox4 siRNA, specifically, had no effect on Smad3 phosphorylation, but blocked downstream ERK phosphorylation and subsequent α-SMA expression and fibronectin EIIIA synthesis (Figure 3). These results follow a similar course described in pulmonary vascular smooth muscle cells where TGF-β-induced proliferation occurs through a Nox4-dependent pathway downstream of Smad3 [195]. Moreover, recent studies by Haurani et al indicate that Nox4 causes feedback inhibition of its own expression and regulates migration in adventitial fibroblast [114], although the signaling pathway of this loop remains to be elucidated. TGF-β1 also, induces PAI-1 expression in human lung fibroblasts where nuclear MAPK phosphatase MKP-1 is a molecular target of Nox4-generated ROS leading to a sustained activation of JNK and p38 MAPKs [169]. The location of these signaling molecules with respect to Smad has not been determined. Taken together, the data summarized above indicate that Nox4 is essential in TGF-β-induced myofibroblast differentiation and profibrotic cell behaviors; however, the position of Nox4 derived ROS in the various signaling pathways varies among fibroblast organ type.

Concluding remarks

The role of the myofibroblast in progression of fibrosis cannot be overstated. Despite the controversy over the origin of the interstitial myofibroblast during fibrosis, there is common agreement that it is responsible for much of the accumulated matrix and ultimate organ damage. Similarly, the mesangial cell, when activated, acquires myofibroblast features and is most responsible for matrix synthesis in the glomerulus during disease. Angiotensin II and TGF-β are the growth factors with the greatest impact on myofibroblast differentiation and synthesis of mesenchymal matrix proteins. The signaling pathways used by mesangial cells and fibroblasts are diverse depending on the cell type and agonist they are exposed to; however recent studies indicate that NAD(P)H oxidase-generated ROS are common to cell activation. It seems also that Nox4 may be the isoform that that is most responsible for growth factor-induced ROS generation and occupies a central role in mesangial cell and myofibroblast differentiation and matrix synthesis. Information on the role of ROS in angiotensin II- and TGF-β-induced mesangial cell and fibroblast signaling is incomplete, however recent evidence indicates they most likely share similar pathways. Further research in this area will shed important information on myofibroblast differentiation and aid in development of therapeutic approaches to fibrosis.