Hypoxia and hypoxia-inducible factors in diabetes and its complications

Maintenance of oxygen homeostasis is fundamental for all cells of metazoan organisms. Hypoxia occurs when oxygen consumption exceeds oxygen supply. Complex adaptive mechanisms have developed to facilitate cell survival during hypoxia, such as metabolic reprogramming of mitochondrial oxidative phosphorylation to anaerobic glycolysis, increased erythropoiesis and angiogenesis, proliferation, differentiation and migration. Impaired adaptive responses to hypoxia contribute to the pathophysiology of many diseases [1].

Hypoxia-inducible factors (HIFs) are major regulators of adaptive responses to hypoxia. As transcription factors, HIFs directly activate the expression of several hundred target genes to maintain cellular oxygen homeostasis [1]. HIF signalling can also repress gene expression, mostly indirectly via HIF target genes such as transcriptional repressors and microRNAs [1]. HIF-induced adaptive responses to hypoxia are protective in many diseases but can be detrimental, such as when they promote cancer progression. HIF signalling also interacts with other signalling pathways, such as Notch and NF-κB, to regulate responses to hypoxia [2].

HIFs are heterodimeric proteins, consisting of an oxygen-sensitive α subunit and a constitutively expressed HIF-1β subunit. There are three isoforms of HIF-α, namely, HIF-1α, HIF-2α and HIF-3α. HIF-1α is ubiquitously expressed but HIF-2α and HIF-3α are tissue-specific. HIF-1 is induced early following the onset of hypoxia, whereas the activation of HIF-2 is usually slower and is sustained for longer [3]. Despite acting through a common hypoxia response element (HRE), HIF-1 and HIF-2 can activate distinct target genes depending on the cellular context [4]; HIF-3 is less well-studied.

The regulation of HIF signalling has been recently reviewed [5]. Although the transcription and translation of HIF genes are regulated by various mechanisms, HIF signalling is largely regulated at post-translational levels. The protein stability of HIF-α is tightly regulated by oxygen and the t_½ of HIF-1α is less than 5 min upon reoxygenation. The regulation of HIF-1α under conditions of normoxia is depicted in Fig. 1a. HIF-1α is hydroxylated in the presence of oxygen, 2-oxoglutarate and iron by prolyl hydroxylase domain proteins (PHDs) 1–4. Hydroxylation occurs on the two conserved proline residues (P402 and P564) in the oxygen-dependent degradation domain. Hydroxylated HIF-1α is ubiquitylated by the von Hippel–Lindau protein (VHL), a ubiquitin E3 ligase, before being degraded by the 26S proteasome. Upon hypoxia, HIF-α is stabilised and translocates to the nucleus, where it dimerises with HIF-1β, binds to the conserved HRE in HIF target genes, and activates gene transcription (Fig. 1b). The transactivation activity of HIF is also regulated by oxygen, mediated by factor-inhibiting HIF-1 through the hydroxylation of an asparagine residue (N803), which prevents the recruitment of coactivators CREB-binding protein (CBP) and p300. HIF-α stability can also be regulated by oxygen-independent pathways, such as heat shock protein (Hsp) 90 and receptor for activated C kinase 1, as well as p53 and glycogen synthase kinase 3β-mediated mechanisms. Other post-translational modifications, such as phosphorylation, SUMOylation, deubiquitylation and epigenetic modification, can also regulate HIF signalling (reviewed in [5]). HIF-1 signalling can also be regulated by dozens of other proteins, many of which are HIF-1 target genes and participate in feed-forward or feedback regulation of HIF-1 function [2].

Emerging evidence indicates that in diabetes, tissues such as the retina [6], kidney [7, 8], pancreatic islets [9], adipose [10], skin and wounds [11] are hypoxic, suggesting that hypoxia plays a central role in the development of diabetes and diabetes complications. However, HIF-1-mediated adaptive responses to hypoxia are impaired in diabetes, leading to cellular dysfunction. The underlying mechanisms are still not completely understood.

Several studies have shown that HIF-1 activation in diabetic tissues is submaximal for the observed degree of hypoxia, due to the repression of HIF-1 signalling. Both protein stabilisation and transactivation activity of HIF-1 are inhibited in wounds [11, 12], kidney [7] and heart [13, 14] of individuals with diabetes and in animal models of diabetes. Hyperglycaemia is, by definition of the disease, common for both type 1 and type 2 diabetes. High glucose levels inhibit the stabilisation and function of HIF-1 under hypoxic conditions in dermal fibroblasts and endothelial cells [11, 12], cardiomyocytes [13], retinal epithelial cells [15] and proximal tubular cells [16]. The inhibitory effects of hyperglycaemia on HIF-1α stability is p53-independent, since high glucose can still destabilise HIF-1α during hypoxia in p53^−/− fibroblasts [12]. However, PHD inhibition or VHL inactivation can largely rescue HIF-1α stability [11], indicating an important role for PHD and VHL in mediating the degradation of HIF-1α during hyperglycaemia (Fig. 1c). However, the underlying mechanisms facilitating PHD- and VHL-mediated HIF-1α degradation during hyperglycaemia are still not fully understood.

Hyperglycaemia also induces the accumulation of methylglyoxal, which mediates HIF-1α destabilisation in a PHD- or VHL-independent manner (Fig. 1c). Methylglyoxal increases the interaction of HIF-1α with Hsp40 and Hsp70, facilitating the ubiquitylation of HIF-1α through the E3 ligase carboxy terminus of Hsp70-interacting protein and subsequent proteasomal degradation [15].

Hyperglycaemia not only inhibits HIF-1α stability, but also represses the transactivation activity of HIF-1 (Fig. 1c). Hyperglycaemia inhibits the activity of both N-terminal and C-terminal transactivation domains of HIF-1α [11]. This is at least partially mediated by methylglyoxal modification of p300, which inhibits its recruitment to HIF-1 [17]. Methylglyoxal modification of HIF-1α can also inhibit the hetero-dimerisation of HIF-1, thus repressing HIF-1 function [18].

Although HIF-1 inhibition by hyperglycaemia has been shown in several studies, paradoxically, the activation of HIF-1 signalling by high glucose through a carbohydrate response element binding protein-mediated mechanism has been reported in glomerular mesangial cells, indicating a cell context-specific regulation of HIF-1 in diabetes [19].

Besides hyperglycaemia, hyperlipidaemia is a common feature of type 2 diabetes. During hypoxia in cardiomyocytes, high levels of fatty acids (palmitate or oleate) have been shown to inhibit succinate generation from glycolysis, which, in turn, represses HIF-1 activation by facilitating the PHD-dependent degradation of HIF-1α [14]. Whether this mechanism is specific for cardiomyocytes needs to be determined, since in most cells glycolysis is activated in diabetes. Metabolites from the citric acid cycle have been shown to regulate PHD and other 2-oxoglutarate-dependent enzymes; their potential role in mediating the dysregulation of HIF in diabetes warrants further investigation.

Accumulating evidence suggests that hypoxia and inappropriate responses to hypoxia due to dysregulated HIF-1 signalling are important pathogenic factors, occurring both in tissues central for the development of diabetes (pancreatic beta cells and adipose tissue) and in tissues susceptible to diabetes complications (nerves, retina, heart, blood vessels, kidney and wounds).

As discussed above, hypoxia and impaired adaptive responses to hypoxia secondary to insufficient HIF-1 activation in diabetic tissues are fundamental pathogenic factors for the development of diabetes and diabetes complications. Therefore, strategies designed to increase HIF-1 signalling could lead to promising therapies for the treatment of diabetes and its complications.

Pharmacological induction of HIF-1 promotes wound healing in experimental diabetes models [11, 20, 21]. Recent preclinical studies in diabetic animal models have shown that PHD inhibition can also prevent the progression of diabetic nephropathy [8, 27] and atherosclerosis [29], protect the ischaemic heart [14, 51] and peripheral neuron [52], and improve cognitive function [53]. Some studies also show that PHD inhibition is beneficial for the prevention and treatment of obesity and metabolic disorders [35, 36] and for improving beta cell function [37].

The prolyl hydroxylase inhibitor (HIF-PH inhibitor) roxadustat (FG-4592) has recently been approved for the treatment of anaemia caused by chronic kidney disease [54] and several other HIF stabilisers are undergoing clinical trials. However, the clinical therapeutic effects of PHD inhibitors on diabetes and diabetes complications needs further investigation.

While topical application of HIF inducers for diabetic foot ulcer has only minimal systemic effects, further mechanistic and translational research is required to identify the right dose, temporal window and tissue-specific application for systemic use of HIF inducers to minimise potential side effects. More efforts to decipher the regulation of HIF-1 signalling in diabetes may provide novel and more specific therapeutic targets as well as efficient biomarkers for the identification of individuals who are most likely to benefit from HIF-targeting therapy.

Diabetic tissues are hypoxic; however, adaptive responses to hypoxia are impaired due to the dysregulation of HIF-1 signalling in diabetes. This contributes to the progression of diabetes and its complications, which can potentially be prevented or treated by modulating HIF-1 expression or activity (Fig. 2). Further mechanistic, translational and clinical research is warranted to identify specific and efficient hypoxia- or HIF-targeting therapies for diabetes and its complications. A summary of the main points of this review can be found in the text box ‘Take-home messages’.

This is an emerging field with contributions from many research groups. Due to space limits, we regret that it has not been possible to cite all relevant publications in this review.