The role of hypoxia in intestinal inflammation

The major transcription factor, which mediates the cellular response to hypoxia, is hypoxia-inducible factor (HIF). HIFs are highly conserved transcription factors that are present in all metazoans. During the divergence of vertebrates, gene duplication led to the emergence of additional α subunits. Mammals contain HIF-1α, HIF-2α, and HIF-3α. HIF-α subunits are regulated through a well-characterized hydroxylation-induced proteasome-mediated degradation during normal cellular oxygen tension. Degradation is mediated by prolyl hydroxylase domain enzymes (PHDs). Mammalian cells contain three PHDs, EGL nine homolog (EGLN)1, EGLN2, and EGLN3. PHDs are α-ketoglutarate dependent dioxygenases that hydroxylate HIF-α subunit on two proline residues. PHD specificity for HIF-α subunits is not clear, but disruption of all three genes is needed for robust HIF-1α and HIF-2α activation in the intestine [4]. Hydroxylation serves as a recognition motif for the von Hippel-Lindau tumor suppressor protein, which recruits an E3 ubiquitin ligase complex leading to rapid proteasomal degradation of HIF-α subunit (Fig. 1). PHDs use molecular oxygen for the hydroxylation reaction, and therefore, in a limited oxygen environment, HIF-α subunit is not hydroxylated leading to stabilization. HIF-α subunit binds to an obligate heterodimer binding partner HIF-1β also called aryl hydrocarbon receptor nuclear translocator (ARNT) and subsequent recruitment to HIF response elements (HREs) present in the promoter of HIF target genes (Fig. 1). It is estimated that in a specific cell type, HIFs bind to roughly 500 sites following hypoxic activation [5].

Using staining techniques to visualize hypoxic foci, a robust increase in hypoxia is observed in mouse models of colitis [2]. Physiological hypoxia as mentioned above is localized to epithelial cells adjacent to the lumen. In colitis, hypoxic staining is observed throughout the mucosa. The precise mechanism for the increase in hypoxia is not clear, but it is likely due to several factors. Inflammation leads to enhanced oxygen consumption of intestinal epithelial cells. Inflammation increases local vasculitis and thus decreasing the oxygen availability to inflamed areas [6]. Recently, it was shown that transmigrating neutrophils can consume local oxygen, thereby enhancing hypoxia in colitis [7]. In addition to hypoxic staining in mouse models, HIF-1α and HIF-2α are highly increased in epithelial cells in UC and CD patients [8]. Currently, the expression and function of HIF-3α have not been thoroughly assessed.

HIF-1α and HIF-2α can bind to the same canonical HREs. However, through mouse models and cell studies, it is clear that HIF-1α and HIF-2α regulate distinct subset of genes. HIF-1α in intestinal epithelial cells is widely recognized as a major protective factor in IBD. IBD results in a dysregulation of a very complex and intricate network of tight junctions, which are critical in maintaining a barrier that is needed to separate commensal microbiota from the mucosal immune cells. HIF-1α directly regulates several barrier protective genes during injury. Moreover, HIF-1α activation can decrease cytokines and leads to an increase in β-defensins, a critical antimicrobial protein (Fig. 2) [2, 9‐13]. PHDs inhibitors, which activate HIF-1α and HIF-2α, are protective in acute colitis models through a HIF-1α-dependent mechanism [12, 14, 15]. However, chronic activation of HIF-2α in intestinal epithelial cells leads to a robust spontaneous intestinal inflammation in a dose-dependent manner [8]. Using mouse models in which HIF-1α and HIF-2α are overexpressed in intestinal epithelial cells demonstrate a distinct function for these transcription factors in IBD. Moderate overexpression of HIF-2α in intestinal epithelial cells does not result in any basal intestinal injury. However, expression of pro-inflammatory mediators is significantly increased, and the mice are highly susceptible to inflammatory injury in mouse models of colitis. Highly overexpressing HIF-2α in intestinal epithelial cells leads to spontaneous colitis, and the mice die at 35 days old from massive intestinal inflammatory disorder [8]. HIF-2α directly regulates a number of pro-inflammatory cytokines including tumor necrosis factor-α, which is essential for HIF-2α-induced inflammation [8]. Moreover, recent work has demonstrated that HIF-2α is essential in barrier function [16, 17]. A chronic increase in HIF-2α leads to high turnover of the tight junction protein occludin, leading to a decrease in barrier integrity [17]. Using similar mouse models, moderate or high overexpression of HIF-1α leads to a decrease in intestinal damage in a colitis model, as expected [18]. Interestingly, activation of HIF-1α does not result in increased tumorigenesis in a colitis-associated colon cancer model, further suggesting that HIF-1α is a good target for colitis [18]. Activation of HIF-2α not only leads to activation of the inflammatory response but there is an increase in tumor number, tumor size, and tumor progression in mouse models of colon cancer [19]. In addition to directly regulating pro-inflammatory mediators and barrier function, HIF-2α is important in the wounding response and proliferation following injury (Fig. 2) [8, 17, 19, 20]. Interestingly, overexpression of both HIF-1α and HIF-2α also leads to heightened inflammatory response suggesting that activation of HIF-1α does not protect the pro-inflammatory response of HIF-2α [8]. This data contradicts the battery of literature showing a protective function of PHD inhibitors in colitis. There maybe several reason for this discrepancy. Chemical inhibition of PHDs leads to a more robust HIF-1α activation rather than HIF-2α activation. This is indeed true for dimethyloxaloylglycine, a commonly used PHD inhibitor, where doses enough for HIF-1α activation do not lead to significant increase in HIF-2α-specific targets [8]. Also, PHD inhibitors may lead to more pulsatile activation of HIF-1α and HIF-2α rather than chronic high increase in HIF-2α, which leads to inflammatory injury. Lastly, the temporal regulation of HIF-1α and HIF-2α has not been critically assessed in chronic models of colitis. HIF-1α is regulated in a cyclical manner through a negative feedback loop mediated by mir-155. HIF-1α increases the expression of miR-155, which in turn represses HIF-1α expression following sustained hypoxia [21]. Therefore, in IBD, chronic activation of HIF-2α with low HIF-1α expression may promote the pro-inflammatory response and decrease the intestinal barrier integrity leading to heightened inflammation and injury.

Currently, the pan-PHD inhibitors dimethyloxaloylglycine, FG-4497, and TRC160334 are protective in mouse models of colitis [14, 15, 22]. However, HIF-2α may increase the inflammatory response, and therefore, optimal HIF-based therapies would be pharmacological agents that can specifically increase HIF-1α. AKB-4924 is a PHD inhibitor that results in modest activation of HIF-2α but robustly activates HIF-1α [23]. AKB-4924 increases the antimicrobial response and protective innate immune response. AKB-4924 treatment improves the intestinal barrier integrity and reduces the pro-inflammatory response. The beneficial effects of AKB-4924 were due to intestinal HIF-1α, as disruption of HIF-1α attenuated the protective role [12]. Moreover, HIF-2α inhibitors have been recently identified. HIF-2α (but not HIF-1α) contains ligand-binding cavity, although endogenous substrates have not been identified [24]. This cavity has been targeted for drug development, and several promising highly specific small-molecule inhibitors are identified [25]. Currently, these drugs have not been assessed in mouse models of colitis, but the data suggest that disruption of intestinal epithelial HIF-2α decreases the inflammatory response in colitis [8].