Hypoxia. Hypoxia in the pathogenesis of systemic sclerosis.

Autoimmunity, microangiopathy and tissue fibrosis are hallmarks of systemic sclerosis (SSc). Vascular alterations and reduced capillary density decrease blood flow and impair tissue oxygenation in SSc. Oxygen supply is further reduced by accumulation of extracellular matrix (ECM), which increases diffusion distances from blood vessels to cells. Therefore, severe hypoxia is a characteristic feature of SSc and might contribute directly to the progression of the disease. Hypoxia stimulates the production of ECM proteins by SSc fibroblasts in a transforming growth factor-beta-dependent manner. The induction of ECM proteins by hypoxia is mediated via hypoxia-inducible factor-1alpha-dependent and -independent pathways. Hypoxia may also aggravate vascular disease in SSc by perturbing vascular endothelial growth factor (VEGF) receptor signalling. Hypoxia is a potent inducer of VEGF and may cause chronic VEGF over-expression in SSc. Uncontrolled over-expression of VEGF has been shown to have deleterious effects on angiogenesis because it leads to the formation of chaotic vessels with decreased blood flow. Altogether, hypoxia might play a central role in pathogenesis of SSc by augmenting vascular disease and tissue fibrosis.

Introduction

Oxygen homeostasis is a sine qua non for metazoan organisms. Reduction in physiological oxygen concentrations leads to metabolic demise because oxygen is the terminal electron acceptor during ATP formation in mitochondria and is a central substrate in many enzymatic reactions. Whereas lack of oxygen causes metabolic cell death, increased oxygen concentrations carry a risk for oxidative damage to proteins, lipids and nucleic acids, possibly initializing apoptosis or carcinogenesis. Thus, even slight changes in systemic and cellular oxygen concentrations induce a tightly regulated machinery of short-acting and long-acting response pathways to keep the supply of oxygen within the physiological range.

Molecular responses to hypoxia and endogenous hypoxia markers have been elucidated in detail during the past two decades. In this context, the molecular characterization of the transcription factor hypoxia-inducible factor (HIF)-1 and unravelling of its regulation were breakthroughs for our understanding of cellular adaptation to reduced oxygenation. HIF-1 protein accumulates under hypoxic conditions in many different cell types. It activates the transcription of genes that are of fundamental importance for oxygen homeostasis, including genes involved in energy metabolism, angiogenesis, vasomotor control, apoptosis, proliferation and matrix production [1].

Systemic sclerosis (SSc) is characterized by a triad of micro-angiopathy, activation of humoral and cellular immune responses and tissue fibrosis, affecting the skin as well as a variety of internal organs, including lung, heart and gastrointestinal tract [2]. Using nailfold capillaroscopy, alterations in the capillary network can be observed early in SSc. Vascular alterations include sac-like, giant and bushy capillaries, microhaemorrhages and a variable loss of capillaries that result in avascular areas [3]. The microangiopathy with progressive loss of capillaries leads to decreased blood flow followed by a lack of nutrients and tissue hypoxia. In advanced disease, fibrosis of the skin and of multiple internal organs, which results from excessive extracellular matrix production of activated fibroblasts, is the most obvious histopathological hallmark of SSc. Because the accumulation of extracellular matrix increases diffusion distances from blood vessels to cells, tissue malnutrition and hypoxia may be aggravated by fibrosis. In summary, severe tissue hypoxia is present in SSc and may even be involved in disease progression.

The present review presents current knowledge of molecular signalling pathways in response to hypoxia and discusses the role that hypoxia plays in the pathogenesis of SSc.

Molecular structure of hypoxia-inducible factor-1

In 1995, Wang and coworkers cloned the transcription factor HIF-1, based on its ability to bind to the 3' enhancer region of the erythropoietin gene [4]. Structural analysis revealed two subunits: HIF-1α (120 kDa) and HIF-1β (91 to 94 kDa). Both HIF-1 subunits contain a basic helix-loop-helix domain, enabling them to recognize and bind to specific DNA sequences, called HIF-1 DNA binding sites (HBSs), within the regulatory regions of hypoxia-inducible genes. Both proteins are also charactarized by two Per/ARNT/Sim (PAS) regions located at the amino-termini. Using HIF-1α deletion mutants, Jiang and coworkers [5] demonstrated that the helix-loop-helix domain and the PAS-A region of HIF-1α are sufficient for heterodimerization with HIF-1β. The most intriguing structural element of HIF-1α is the oxygen-dependent degradation domain (ODDD), which links HIF-1α to the cellular oxygen sensor. Under normoxic conditions the hydroxylation of two proline residues within the ODDD results in ubiquitinylation and degradation of HIF-1α. In contrast, hydroxylation and degradation of HIF-1α are decreased in hypoxic milieus because oxygen is the critical substrate in hydroxylation reactions. Thus, lack of oxygen leads to HIF-1α accumulation [6].

Stabilization of hypoxia-inducible factor-1α protein

In contrast to the expression of HIF-1β, that of HIF-1α is tightly controlled by cellular oxygen levels. Cellular HIF-1α is not detectable under normoxic conditions because it is rapidly degraded after translation. After exposure to low oxygen concentrations, levels of HIF-1α increase exponentially. Maximal response is usually reached at oxygen concentrations of about 0.5%.

Hydroxylation of two proline residues within the ODDD (positions 402 and 564) triggers the oxygen-dependent regulation of HIF-1α. This hydroxylation is catalyzed by a family of 2-oxoglutarate dependent dioxygenases called prolyl hydroxy-lase domains (PHDs) [7]. During the hydroxylation process, PHDs split molecular oxygen and transfer one oxygen atom to one of the proline residues. The second oxygen atom reacts with 2-oxoglutarate, generating succinate and carbon dioxide. The co-substrate ascorbic acid keeps the ferrous ion of the catalytic site in its bivalent state. The ability of PHDs to modify HIF-1α depends on the concentration of its substrate oxygen. Under normoxic conditions, PHDs hydroxylate HIF-1α efficiently, leading to the rapid degradation of the HIF-1α subunit. In contrast, the rate of hydroxylation is reduced at low oxygen levels. Thus, PHDs function as intra-cellular oxygen sensors and provide the molecular basis for the regulation of HIF-1α protein concentrations by cellular partial pressure of oxygen [8].

The hydroxylation of HIF-1α is similar to the prolyl modification of collagens [9,10]. However, collagen prolyl hydroxylases are unable to hydroxylate the proline residues of HIF-1α [9]. Three human HIF-1α dioxygenases have been identified thus far [8,11,12]: PHD3 (HPH-1/EGLN3), PHD2 (HPH-2/EGLN1) and PHD1 (HPH-3/EGLN2). All three PHDs have the potential to hydroxylate HIF-1α. Nevertheless, PHD2 exhibits the greatest prolyl hydroxylase activity in normoxic cells [13]. It is the key limiting enzyme for HIF-1α turnover and its knockdown by small interfering RNA stabilizes HIF-1α levels, whereas single knockdown of PHD1 or PHD3 has no effect on the stability of hypoxic conditions. Appelhoff and coworkers [14] demonstrated that PHD3 activity exceeded the activity of PHD2 in MCF-7 breast cancer and BXPC-3 pancreatic cancer cell lines under hypoxic conditions. Inhibition of PHD3 in hypoxic cells led to higher HIF-1α levels than inhibition of PHD2.

Recently, an endoplasmatic prolyl-4-hydroxylase (P4H) with a transmembrane domain, which is more closely related to the collagen prolyl hydroxylases, has also been shown to hydroxylate HIF-1α in vitro [15].

An additional mechanism for the regulation of HIF-1α stability was demonstrated by Jeong and coworkers [16]. Arrest defective (ARD)1, an acetyltransferase, binds directly to the ODDD of HIF-1α in the cytoplasm and acetylates a single lysine residue at position 532. Acetylation of this specific lysine residue favours the interaction of HIF-1α and the E3 ubiquitin ligase complex, and stimulates the degradation of HIF-1α. As shown by vascular endothelial growth factor (VEGF) promoter-driven luciferase reporter gene assays, ARD1 not only destabilizes HIF-1α protein, but it also downregulates its transactivation activity in ARD1-transfected HT1080 human fibrosarcoma cells under hypoxic conditions. Mutation of lysine residue 532 to arginine or application of antisense ARD1 results in stabilization of HIF-1α even under normoxic conditions [16,17]. In contrast, levels of HIF-1α decreased when deacetylation was inhibited. Finally, mRNA and protein levels of ARD-1 are diminished under hypoxia, resulting in less acetylated HIF-1α [16].

Blocking hydroxylation of proline residues 402 and 564 as well as blocking acetylation of lysine 532 have been demonstrated to prevent degradation of HIF-1α under normoxic conditions, thus abolishing the oxygen-dependent regulation of HIF-1α signalling [6,9,16]. These findings suggest that both pathways – hydroxylation and acetylation of HIF-1α – are essential for the physiological regulation of cellular responses to hypoxia.

Upregulation of prolyl hydroxylase domain activity in chronic hypoxia

Interestingly, PHD2 and PHD3 are induced by hypoxia in a HIF-1α-dependent manner, thereby creating a negative feedback loop of HIF-1α signalling [14,18]. In this context, a functional hypoxia-regulated element has been identified in the PHD3 gene, enabling direct regulation of PHD3 by HIF-1. Recently, Ginouvès and coworkers [19] reported increased PHD activity in response to chronic hypoxia. PHD2 and PHD3 protein levels reached a maximum after 24 hours of hypoxia, whereas PHD activity rose steadily for 7 days, indicating that further mechanisms besides induction of PHDs led to increased PHD activity. Consistent with these findings, PHD activity increased with prolonged hypoxia in vivo. Only low PHD activity but high HIF-1α levels were observed in mice exposed to 6 hours of hypoxia at 8% oxygen, whereas PHD activity increased markedly after 24 hours of hypoxia, resulting in a subsequent reduction in HIF-1α. After 24 hours of 8% oxygen, escalation of hypoxia to 6% oxygen concentration for another 2 hours caused a re-accumulation of HIF-1α [19]. Together these findings suggest that HIF-1α is induced in response to hypoxia, accumulates in acute hypoxia and is removed as the activity of PHDs increases in chronic hypoxia.

Ginouvès and coworkers [19] also suggested a mechanism that may lead to augmented PHD activity that is distinct from PHD gene induction. During hypoxia, HIF-1 induces pyruvate dehydrogenase kinase-1, which has been reported to decrease mitochondrial oxygen consumption by inhibiting mitochondrial respiration [20,21]. Inhibition of mitochondrial respiration may increase intracellular oxygen levels and accelerate oxygen-dependent HIF-1α hydroxylation by PHDs [19]. Therefore, augmented PHD activity in chronic hypoxia might create an effective negative feedback loop for HIF-1α signalling. Although this hypothesis must be confirmed with further experiments, separating acute from chronic hypoxia will certainly gain importance for future studies, especially when evaluating HIF-1α or PHDs as possible therapeutic targets for diseases in which hypoxia has been implicated, such as SSc.

Degradation of hypoxia-inducible factor-α

The rapid degradation of HIF-1α under normoxic conditions is mediated by the von Hippel-Lindau tumour suppressor protein (pVHL) [22]. The β-subunit of pVHL interacts directly with the ODDD of HIF-1α when proline residue(s) 402 and/or 564 are hydroxylated, but not without this modification. pVHL itself is part of the E3 ubiquitin ligase complex. Interaction of proline-hydroxylated HIF-1α with pVHL/E3 ubiquitin ligase complex activates the ubiquitination machinery, thereby promoting degradation of HIF-1α [1,9,23,24]. A similar mechanism of recognition is proposed for the acetylation of the lysine residue 532 [16]. Under hypoxic conditions, the ODDD is neither hydroxylated nor acetylated, pVHL cannot bind and HIF-1α is not ubiquitinated. Thus, degradation of HIF-1α in the proteasome is inhibited and HIF-1α protein accumulates.

Binding of HIF-1 to HIF binding sites, formation of the transcriptional complex and regulation of HIF-1 transactivation

After translocation into the nucleus HIF-1α dimerizes with ARNT/HIF-1β. The HIF-1 heterodimer then binds via its basic helix-loop-helix domain to the HBS within the hypoxia-responsive element of most hypoxia-regulated genes [25-27]. The HBS is essential but not sufficient for HIF-1 gene activation. Besides the HBS, a complete hypoxia-responsive element contains additional binding sites for transcription factors that are not sensitive to hypoxia. These co-stimulatory factors, including cAMP response element binding protein (CREB)-1 of the lactate dehydrogenase A gene [28] or activator protein-1 (AP-1) in the VEGF gene [29], are also required for efficient transcription of oxygen-sensitive genes. Multimerization of HBS can substitute for additional transcription factors in several HIF-regulated genes [30-33].

For efficient induction of HIF-1-regulated genes, HIF-1 must be activated. Simple blockade of HIF-1α degradation (for example with chemical proteasome inhibitors such as N-carbobenzoxyl-L-leucinyl-L-leucinyl-L-norvalinal) results in accumulation of HIF-1α but is often not sufficient for transactivation [34]. Two modifications of HIF-1α involved in the regulation of HIF-1α transactivation have been identified: hydroxylation of the carboxyl-terminal transactivation domain and protein phosphorylation by tyrosine kinase receptors.

At low oxygen concentrations the carboxyl-terminal transactivation domain of HIF-1α recruits several co-activators, including p300 and CREB-binding protein, which are required for HIF-1 signalling [35,36]. Under normoxic conditions the enzyme FIH-1 (factor-inhibiting HIF-1) hydroxylates an asparagine residue at position 803, thereby preventing interaction of HIF-1α with p300 with CREB-binding protein [37]. Consequently, oxygen-sensitive asparagine hydroxylation, inhibiting HIF-1 transactivation, is part of the oxygen-sensing mechanism [37,38].

Other members of the hypoxia-inducible factor family

Two proteins closely related to HIF-1α have been identified and designated HIF-2α and HIF-3α [39,40]. HIF-2α and HIF-3α are both able to dimerize with HIF-1β and bind to HBSs [41,42]. HIF-2α is similar to HIF-1α with regard to its genomic organization, protein structure, dimerization with HIF-1β, DNA binding and transactivation [22,35,43,44]. Moreover, both proteins accumulate under hypoxic conditions [45-47]. However, experiments with knockout mice revealed that HIF-1α and HIF-2α could not compensate for loss of the each other [31,48,49]. This finding suggests that the different α subunits of HIF might not be redundant and possess different biological functions.

Hypoxia in systemic sclerosis

Hypoxia and its central mediator HIF-1 control a large variety of different genes. Upregulation of HIF-1 in response to hypoxia regulates erythropoiesis, angiogenesis and glucose metabolism, as well as cell proliferation and apoptosis [1,7]. Using DNA microarray studies on primary pulmonary arterial endothelial cells, Manalo and coworkers [50] observed that a minimum of 2.6% of all human genes were regulated by hypoxia in a HIF-1-dependent manner. In theory, microangio-pathy and tissue fibrosis should result in reduced tissue oxygenation and may provoke HIF-1-dependent response to hypoxia. The reduced capillary density and vascular malformations should lead to decreased blood flow with lack of nutrients and oxygen in involved organs in SSc patients [51]. Besides the microangiopathy, tissue fibrosis might further aggravate tissue malnutrition and hypoxia. The progressive accumulation of extracellular matrix proteins such as collagens, fibronectin and glycosaminoglycans [52] increases distances between cells and their supplying vessels and may impair diffusion. Hence, lack of functional capillaries as well as impaired diffusion implicate significant tissue malnutrition and chronic hypoxia in SSc patients (Figure 1).

Figure 1

Vicious circle of hypoxia and fibrosis in the pathogenesis of SSc. To the upper left is shown nailfold capillaroscopy from a patient with systemic sclerosis (SSc) with capillary rarification and vascular alterations, including sac-like, giant and bushy capillaries. Vasculopathy leads to reduced blood flow and causes tissue hypoxia in SSc. To the upper right is shown a haematoxylin and eosin stained skin section from an experimental mouse fibrosis model with increased skin thickness due to extracellular matrix (ECM) deposition. ECM deposition increases diffusion distances from blood vessels to cells and reduces tissue oxygenation. In the 'vicious circle', shown at the bottom of the figure, tissue hypoxia leads to activation of dermal fibroblasts and upregulation of ECM production. Further ECM deposition aggravates tissue malnutrition and hypoxia. Hypoxia once again stimulates ECM production in dermal fibroblasts.

Indeed, two studies demonstrated severe hypoxia in lesional, fibrotic skin of SSc patients [53,54]. In both studies, low oxygen levels were only found in lesional skin of SSc patients, whereas the oxygen levels in nonfibrotic skin were not decreased compared with skin of healthy volunteers.

Using a noninvasive transcutaneous technique to measure oxygen levels, Silverstein and coworkers [53] showed that oxygen levels of fibrotic skin were inversely related to skin thickness. The lowest oxygen levels were measured in SSc patients with severely thickened skin. Indirect correlation of oxygen levels with dermal thickness supports the concepts of impaired diffusion due to accumulation of extracellular matrix in lesional skin of SSc patients. Patients suffering from primary Raynaud's disease did not exhibit hypoxic skin, and oxygen levels were similar to those in healthy individuals.

We quantified oxygen levels in the skin of SSc patients by applying an oxygen partial pressure (PO2) histography method, involving introduction of a small polarographic needle electrode directly into the dermis [54]. To exclude systemic influences on local oxygen levels, we determined arterial oxygen saturation, haemoglobin content, blood pressure and heart rate, and patients rested for at least 10 minutes before the experiment. For each patient about 200 single measurements of PO2 were taken at a predefined area on the dorsal forearm, and an individual PO2 mean value was determined. Average PO2 in the skin of healthy individuals was 33.6 ± 4.1 mmHg (4.4 ± 0.5% oxygen per volume), whereas involved skin of SSc patients exhibited significantly decreased oxygen levels, with a mean PO2 value of 23.7 ± 2.1 mmHg (3.1 ± 0.3%). In contrast, the average PO2 in nonfibrotic skin of SSc patients did not differ from that in healthy individuals (mean PO2 37.9 ± 8.6 mmHg, correspon-ding to 5.0 ± 1.1%).

In summary, both studies demonstrated that hypoxia is a characteristic feature of involved, fibrotic skin of SSc patients. Although cutaneous blood flow, a potential confounding factor, was not determined in any of these studies, the inverse correlation of skin thickness with cutaneous PO2 suggests that impaired oxygen diffusion due to extracellular matrix accumulation might contribute to tissue hypoxia in SSc.

Role played by hypoxia-inducible factor-1α in systemic sclerosis

Considering the presence of hypoxia, one would assume that HIF-1α is strongly upregulated in SSc [54,55]. This presumption is fortified by the fact that several cytokines and growth factors, upregulated in SSc, are able to stabilize HIF-1α under certain conditions. Examples include inter-leukin-1β, transforming growth factor (TGF)-β, platelet-derived growth factor (PDGF), fibroblast growth factor 2 and insulin-like growth factors (IGFs) [56-58].

Despite severely reduced oxygen levels and despite the over-expression of these growth factors, protein levels of HIF-1α in the skin of SSc patients were even below the levels seen in healthy control skin [54]. Skin specimens from SSc patients did not exhibit increased expression of HIF-1α protein by immunohistochemistry. HIF-1α staining was moderate to high in the epidermis of healthy individuals, whereas the expression of HIF-1α in SSc patients was restricted to single keratinocytes. HIF-1α protein was not detectable in the dermis of healthy individuals and SSc patients. Moreover, the HIF-1α expression pattern in involved skin in SSc patients did not correlate with upregulated VEGF, one of the main transcriptional targets of HIF-1α [54].

PHD-dependent HIF-1α negative feedback loops in chronic hypoxic conditions might be a plausible explanation for decreased HIF-1α levels in fibrotic skin of SSc patients. Considering the clinical course of SSc, lesional skin in SSc patients can be categorized as a chronically hypoxic tissue. In this context, low HIF-1α levels may be caused by negative HIF-1α feedback loops, even despite severe hypoxia. Increased PHD activity in response to chronic hypoxia [19] might lead to rapid HIF-1α degradation and decreased HIF-1α levels in fibrotic SSc skin. This theory is also supported by studies on the effects of prolonged hypoxia in murine organs. In mice exposed to 6% oxygen, HIF-1α protein reached maximum levels in the brain after 4 to 5 hours but declined afterward, attaining basal normoxic concentrations after 9 to 12 hours. Similar results were obtained for kidney and liver [59].

However, the low levels of HIF-1α in the skin of SSc patients per se do not argue against the persistent activation of oxygen-sensitive pathways in SSc. Marked and persistent upregulation of the oxygen-dependent gene VEGF is observed in lesional SSc skin even in late stages of SSc. Thus, the response to hypoxia appears to persist in chronic states, but might be driven by HIF-1α-independent pathways, for instance HIF-2α and HIF-3α. However, the role played by other members of the HIF family in the pathogenesis of SSc has not yet been investigated in detail.

Insufficient response to hypoxia: dysregulation of angiogenesis in systemic sclerosis

Angiogenesis and vasculogenesis are fundamental mechanisms in improving oxygenation of hypoxic tissue. HIF-1 promotes vascularization by inducing the expression of multiple angiogenic mediators such as VEGF, placental growth factor, angiopoietin 1 and 2, and PDGF-BB [60]. VEGF drives angiogenesis by activating endothelial cells in hypoxic tissue and vasculogenesis by mobilizing and recruiting endothelial progenitor cells [61-63]. In addition, VEGF exhibits synergistic angiogenic effects together with PDGF and fibroblast growth factor-2 [64].

Sufficient tissue vascularization depends on strict regulation of VEGF expression. Chronic and uncontrolled over-expression of VEGF induces the formation of chaotic vessels, characterized by glomeruloid and haemangioma-like morphology [65,66]. Dor and coworkers [67] demonstrated in pTET-VEGF165/MHCα-tTa transgenic mice, in which VEGF expression can be conditionally switched off in an organ-dependent manner by feeding tetracycline, that time-dependent regulation of VEGF expression was essential for adequate vascularization. Although short-term over-expression of VEGF induced the formation of new mature and functional vessels in adult organs, prolonged exposure to VEGF without subsequently switching off its gene expression by tetracycline resulted in the formation of irregularly shaped, sac-like vessels leading to reduced blood flow. Irregularly shaped, sac-like vessels are reminiscent of the disturbed vessel morphology in SSc [3]. Hence, the microvascular defects in SSc might partly be caused by uncontrolled over-expression of VEGF.

VEGF levels are markedly upregulated in the skin of SSc patients compared with healthy volunteers [54]. As analyzed by in situ hybridization, the mean percentage of epidermal keratinocytes expressing VEGF was significantly increased in SSc patients compared with normal individuals. These findings were consistent with dermal expression levels of VEGF. In contrast, normal individuals did not exhibit VEGF expression in the dermis. VEGF was expressed in most SSc patients in a variety of different dermal cell types, including fibroblasts, endothelial cells and leucocytes [54]. VEGF was induced in dermal SSc fibroblasts in response to hypoxia, but expression levels did not differ significantly between fibroblasts from SSc patients and those from healthy volunteers [54]. However, as the oxygen levels are significantly lower in lesional skin of SSc patients than in control individuals, the induction of VEGF by hypoxia is only operative in SSc patients, but not in normal volunteers. Both receptors for VEGF, namely VEGF receptors 1 and 2, were also over-expressed in the skin of SSc patients. Therefore, enhanced activation of the VEGF/VEGF receptor axis may lead to typical changes in SSc vascularization, causing tissue malnutrition and hypoxia [54]. Because the expression of VEGF is stimulated by hypoxia, one might speculate that the hypoxia could augment vascular disease in SSc by contributing to persistent over-expression of VEGF. However, it remains to be demonstrated that chronic hypoxia alone is indeed sufficient to cause persistent upregulation of VEGF in vivo. Alternatively, the persistent over-expression of VEGF in SSc might also be driven by cytokines. Interleukin-1β, PDGF and TGF-β are all upregulated in SSc and can stimulate the expression of VEGF [54,68,69].

Induction of fibrosis by hypoxia

Microangiopathy with impaired angiogenesis and excessive accumulation of extracellular matrix may cause severe hypoxia in SSc [53,54]. However, what is the exact role played by hypoxia in the pathogenesis of SSc? Is it just the consequence of microangiopathy and fibrosis or does it contribute to the progression of SSc?

DNA microarray studies revealed the first causal links between hypoxia and fibrosis [50]. Manalo and coworkers [50] detected a striking number of genes encoding collagens or collagen-modifying enzymes that were induced in pulmonary endothelial cells after 24 hours at 1% oxygen. These genes included collagen (COL)1A2, COL4A1, COL4A2, COL5A1, COL9A1 and COL18A1, as well as procollagen prolyl hydroxylases (P4HA1 and P4HA2), lysyl oxidase (LOX) and lysyl hydroxylases (procollagen lysyl hydroxylase and procollagen lysyl hydroxylase 2). Similar links between hypoxia and fibrosis have also been found in other models and organs, for example kidney [70,71], liver [72] and lung [73]. Together, these findings indicate that hypoxia could promote extracellular matrix production and that it may actively be involved in the pathogenesis of profibrotic disorders such as SSc.

We could demonstrate that hypoxia induced several extra-cellular matrix proteins, including fibronectin-1, thrombo-spondin-1, proα 2(I) collagen (COL1A2), IGF-binding protein 3 (IGFBP-3) and TGF-β-induced protein (TGF-βi) in cultured dermal fibroblasts [74]. Type 1 collagens and fibronectins are the major matrix proteins within fibrotic lesions [52]. Thrombospondin-1 also accumulates in SSc and modulates angiogenesis. TGF-βi is an extracellular matrix protein that is known to be highly expressed in arteriosclerotic plaques [75] and in zones of thickened extracellular matrix in the bladder [76]. IGFBP-3 directly induces the synthesis of fibronectin in lung fibroblasts [77] and protects IGF-1 from degradation. IGF-1 itself stimulates collagen synthesis and downregulates the production of collagenases in fibroblasts [77].

Induction and production of these extracellular matrix proteins in response to hypoxia was time dependent and inversely correlated with oxygen levels [74]. Most of these proteins were significantly upregulated after 24 hours of oxygen deprivation, with a further significant increase after 48 hours. The expression of fibronectin-1, thrombospondin-1, COL1A2 and IGFBP-3 was significantly enhanced at 8% oxygen concentration and increased further with lower oxygen levels, reaching a maximum at 1% oxygen. Of note, severe and chronic hypoxia, as may be found in the skin of SSc patients [54], was associated with the most marked effects on the induction of extracellular matrix proteins.

These results were confirmed in a mouse model of systemic normobaric hypoxia [74]. Consistent with the results obtained in vitro, extracellular matrix proteins were upregulated in mice exposed to hypoxia after 24 hours compared with control mice breathing air with 21% oxygen. Prolonged exposure for 48 hours resulted in further upregulation of fibronectin 1, thrombospondin 1 and COL1A2, whereas TGF-βi and IGFBP3 mRNA levels decreased slightly. Because TGF-β is a major stimulus for the induction of extracellular matrix proteins in SSc [52,78], its role for hypoxia-dependent fibrogenesis was also studied in dermal SSc fibroblasts. Neutralizing antibodies against TGF-β completely abrogated the induction of COL1A2, fibronectin 1, thrombospondin 1 and TGF-βi in SSc fibroblasts that were cultured under hypoxic conditions for 48 hours [74]. These findings suggest that inhibition of TGF-β-dependent pathways may prevent the profibrotic effects of hypoxia.

Consistent with the results on TGF-β signalling, the expression of the fibrogenic cytokine connective tissue growth factor (CTGF) was also shown to be upregulated in SSc in response to hypoxia [79]. CTGF is a critical mediator of TGF-β-induced skin fibrosis in SSc [80]. Its serum levels are elevated in SSc patients and have been suggested to correlate with skin fibrosis [81]. Hong and coworkers [79] found increased levels of CTGF mRNA and protein in fibroblasts exposed to 1% of oxygen or treated with cobalt chloride, a chemical stabilizer of HIF-1α. The induction of CTGF in response to hypoxia depended on HIF-1α [79]. Because the authors concentrated on short-term hypoxia of up to 4 hours, it remains unclear whether CTGF is also induced by chronic hypoxia and by HIF-1α-independent mechanisms in SSc.

Thus, accumulating evidence suggests that hypoxia might be actively involved in pathogenesis of SSc by stimulating the release of extracellular matrix protein. This could result in a vicious circle of hypoxia and fibrosis. Hypoxia stimulates the production and accumulation of extracellular matrix. The resulting tissue fibrosis inhibits diffusion of oxygen, causing further tissue hypoxia, which stimulates further the production of extracellular matrix (Figure 1). Activation of TGF-β-dependent pathways appears to play a central role in the induction of extracellular matrix proteins by hypoxia, and inhibition of TGF-β signalling might prevent hypoxia-induced tissue fibrosis. However, further studies are needed to characterize further the role played by hypoxia in SSc and to identify the molecular mechanisms activated by hypoxia in SSc.

Conclusion

Capillary rarification and disturbed blood flow, as well as excessive extracellular matrix accumulation, cause chronic tissue hypoxia in SSc. However, levels of HIF-1α protein are decreased, probably due to PHD-dependent negative feedback loops. Interestingly, physiological mechanisms to overcome tissue hypoxia are impaired and dysregulated in SSc. Insufficient angiogenesis and vasculogenesis cannot abolish tissue malnutrition and hypoxia. Compensatory over-expression of VEGF might even result in a futile vascular response to hypoxia, characterized by the chaotic vessel formation. Hypoxia stimulates the production of several extracellular matrix proteins in SSc fibroblasts in a time- and concentration-dependent manner. The excessive deposition of matrix might impair further the diffusion of oxygen and cause a vicious circle of hypoxia and tissue fibrosis. Currently, there are no specific modulators of HIFs or PHDs available for clinical use. Thus, it is not yet possible to target hypoxia selectively in SSc patients. However, because inhibition of TGF-β prevented the induction of extracellular matrix by hypoxia, blocking of TGF-β signalling might be one approach to target at least in part the hypoxia-induced matrix production in SSc.

Abbreviations

ARD: arrest defective; COL: collagen; CREB: cAMP response element binding protein; CTGF: connective tissue growth factor; HBS: HIF-1 DNA binding site; HIF: hypoxia-inducible factor; IGF: insulin-like growth factor; IGFBP: insulin-like growth factor binding protein; ODDD: oxygen-dependent degradation domain; PAS: Per/ARNT/Sim; P4H: prolyl-4-hydroxylase; PDGF: platelet-derived growth factor; PHD: prolyl hydroxylase domain; PO2: oxygen partial pressure; pVHL: Von Hippel-Lindau tumour suppressor protein; SSc: systemic sclerosis; TGF: trans-forming growth factor; TGF-βi: TGF-β-induced protein; VEGF: vascular endothelial growth factor.

Note

This review is part of a series on Hypoxia edited by Ewa Paleolog.

Other articles in this series can be found at

Competing interests

The authors declare that they have no competing interests.