RA is a systemic autoimmune disorder characterised by chronic inflammation of the synovial membranes of joint tissues at multiple anatomical sites which ultimately leads to localised destruction and debilitating deformity (64,65). The RA joint synovium is characterised by both inflammatory and hypoxic regions (Fig. 3), which are highly infiltrated with lymphocytes (CD4⁺ T cells, and B cells), macrophages and macrophage-like and fibroblast-like synoviocytes (66). The molecular basis of RA is still poorly understood, mainly because RA is a heterogeneous disease composed of several possible treatment responses, and clinical manifestations (67–69). These differences make RA difficult to treat, and further studies on the crosstalk between pathways involved in the disease are required.

The deregulation of several transcription factors, such as NF-κB, activator protein-1 (AP-1), and signal transducer and activator of transcription (STATs), has been strongly associated with the inflammatory setting of RA (70–72). NF-κB, in particular, has been shown to be highly activated in the RA synovium (73,74). This is exceptionally important due to the major role of NF-κB in activating inflammatory responses, such as through the activation of the pro-inflammatory cytokine, TNF-α, or the chemokine, IL-8 (75). The activation of a coordinated and complex network of pro-inflammatory cytokines, chemokines, metalloproteases (MPPs) and metabolic proteins by NF-κB, leads to the activation of a positive feedback loop, enhancing the activation of more pro-inflammatory signals that ultimately results in chronic and persistent inflammation (Fig. 3) (75,76).

The HIF family of proteins are additional transcription factors with direct relevance to RA (77,78). Recently, HIF-1α was identified as a key player in RA, and therefore as a potential therapeutic target (79). HIF is important to coordinate the hypoxia response in the synovial tissue, and the deregulation or failure of that response leads to cellular dysfunction, and can ultimately lead to cell death (80). Furthermore, the intense hypoxic region in the synovial tissue (2–4%), activates a hypoxic response through HIF, which is involved in regulating several genes involved in apoptosis, vasomotor control, energy metabolism, and importantly, angiogenesis (Fig. 3) (16,48,81–83).

Even though the role of HIF in RA has been firmly established, the contribution of each α-subunit remains poorly understood. Recently, HIF-2α was implicated as the essential catabolic regulator of inflammation in RA (78). In that study, the authors demonstrated that the overexpression of HIF-2α in joint tissues, but not HIF-1α, was sufficient to induce RA pathogenesis (78). The full contribution of the α-subunits to RA remains elusive. However, it seems clear that each α-subunit contributes differently to the progression of RA. HIF-1α plays a more anti-inflammatory role, whereas HIF-2α acts in a pro-inflammatory manner. What regulates this differential expression of the isoforms is still unknown. However, taking into consideration that NF-κB is the main activator of the HIF transcription factors, it would be interesting to understand whether NF-κB has any role in this HIF-1α to HIF-2α switch, and whether that would be dependent of the presence of hypoxia, inflammation, or both combined.

The intestinal mucosa is exposed to steep hypoxic gradients (63) and is in a constant state of controlled inflammation, which is necessary to allow tolerance to otherwise harmless ingested dietary antigens (Fig. 4) (84). This fine balance is pathologically disturbed in inflammatory bowel disease (IBD); a relapsing-remitting progressive disorder of the gastrointestinal tract that comprises both Crohn’s and ulcerative colitis. The symptoms of IBD can range from mild to severe and include abdominal pain, intestinal bleeding, weight loss, fever and diarrhoea (85). The two IBD sub-types have different distribution patterns: ulcerative colitis is restricted to the colon, whereas Crohn’s colitis can affect any part of the GI tract. Both are thought to occur when inappropriate immunological activity in the intestinal mucosa results in epithelial barrier dysfunction leading to exposure of the mucosal immune system to luminal antigenic material and further cycles of inflammation and barrier dysfunction that underlie disease progression (86,87).

Hypoxia has been found to play a role in IBD. Lower resting oxygen levels have been demonstrated in sections of IBD tissue compared to the controls using a 2-nitroimidazole based approach (63). In keeping with these observations, HIF-1α and HIF-2α activation has been associated with disease and increased vascular density in human specimens (88). The increased vascular density was subsequently demonstrated to be effected by vascular endothelial growth factor (VEGF), an established target of the HIF system (89). Compartmental analyses of the effects of hypoxia have been possible in murine models, where hypoxia has been shown to affect the epithelium, primarily during periods of inflammation (63). The colonic epithelium is the most hypoxic and HIF-active tissue layer because it is physically farthest away from the colonic vascular plexus and closest to the anoxic bowel lumen. This effect is exacerbated by oxygen consumption by luminal bacteria (90), and the presence of inflammatory mediators and lipopolysaccharide (LPS), which have been shown to regulate HIF activity (48).

In the context of IBD, HIF system activity is thought to be protective, acting through three mechanisms: i) inhibition of epithelial cell apoptosis; ii) enhanced expression of barrier-protective genes; and the iii) promotion of neutrophil apoptosis (Fig. 4) (86). Evidence of the anti-apoptotic effects of HIF has been demonstrated indirectly through experiments to investigate the role of the hydroxylase inhibitor, dimethyloxaloylglycine (DMOG), in colitis. Using a murine model of dextran sodium sulfate (DSS)-induced colitis, HIF stabilisation following treatment with DMOG has been shown to prevent apoptosis in a mechanism thought to be mediated by the anti-apoptotic protein, cIAP-2 (91). Recently, this effect has been specifically attributed to PHD1, since the homozygous loss of PHD1, but not PHD2 or PHD3, has been shown to be protective in the same murine model of DSS-induced colitis (92). This effect is most likely HIF-dependent, since the conditional knockout of HIF-1α in mouse intestinal epithelial cells has been shown to result in an enhanced susceptibility to the development of colitis (63).

In addition to its anti-apoptotic effects, the HIF system can protect against colitis through the expression of barrier-protective genes. Several HIF-dependent target genes have been proposed as mediators of this effect: CD55 (93), ecto-50 nucleotidase (94), A2B receptor (95) MUC-3 (96), intestinal trefoil factor (97), and P-glycoprotein (98) all play a role in the regulation of the intestinal mucosa barrier and have all been demonstrated to be regulated in a hypoxia-dependent manner.

There is also evidence of the differential effects of the HIF-α isoform in IBD. HIF-2α expression has been shown to be increased in colon tissues of mice after the induction of colitis. This was also observed in patients with ulcerative colitis or Crohn’s disease (62). Interestingly, in that study, while the loss of HIF-2α was associated with attenuated colonic inflammation, the overexpression of HIF-2α led to spontaneous colitis and increased inflammation.

Other transcriptional programs are active in IBD in addition to those enacted by the HIF system. IBD is primarily an inflammatory pathology and NF-κB activity has been linked to its progression (5). A high degree of NF-κB induction has been demonstrated in intestinal macrophages and epithelial cells (99). In IBD, inflammatory cytokines can drive NF-κB activation, leading to the production of more inflammatory cytokines and potentiating further NF-κB activation (Fig. 4). NF-κB-induced TNF-α expression is one example of this type of positive feedback loop (100). Interestingly, NF-κB can have a dual role in IBD, potentiating inflammation in intestinal macrophages while protecting from inflammation in mucosal epithelial cells. Sharing interesting similarities to the effects of HIF activation, NF-κB signalling in intestinal epithelial cells has been shown to be protective against the development of colitis (101). Deletion of the NF-κB pathway in intestinal epithelial cells results in decreased expression of anti-apoptotic genes, such as Bcl-xL, and leads to reduced epithelial barrier function and increased susceptibility to colitis (102). Conditional knockout of NEMO and subsequent NF-κB inhibition has been shown to result in severe epithelial inflammation in a murine model (102). Similarly, epithelial cell-specific IKKβ deletion has been shown to result in the sustained production of pro-inflammatory Th1 cytokines and increased intestinal inflammation (103). Several treatments have been proposed to target NF-κB activity in IBD, including proteasome blockade, the administration of non-coding RNAs to interfere with NF-κB-DNA binding and anti-TNF-α immunotherapy. However, all have been met with significant systemic toxicity due to the broad role of NF-κB in multiple organs.

Sharing similarities with the microenvironment of RA, in IBD both inflammation and hypoxia are present in the intestinal epithelium and contribute to disease progression (104). It is generally [but not universally (105)] understood that both NF-κB and HIF activity are protective in episodes of colitis (101). Significant crosstalk between these pathways has already been established, and it has been proposed that both pathways may act in concert to contribute to the epithelial barrier function of the colon in a process that is deregulated in IBD.

One example of this crosstalk is the regulation of apoptosis by both pathways (106,107). The caspase recruitment domain family, member 9 (CARD9) is understood to function as a molecular scaffold for the assembly of a BCL10 signalling complex that activates NF-κB (106), and has also been shown to be involved in the regulation of hypoxia-sensitive pathways (107). CARD9 therefore represents one point of crosstalk that may be important in the development of IBD as a promising target for further investigation.

Our laboratory and others have demonstrated NF-κB-dependent HIF-1α mRNA regulation (61,108). NF-κB can also regulate HIF signalling through IKKγ and HIF-2α, which increases HIF-2α transcriptional activity through interaction with cAMP response element-binding (CREB) binding protein (CBP)/p300 (109). Negative feedback through the NF-κB-dependent induction of the micro-RNA, miR-155, in response to LPS has been shown to target HIF-1α for silencing (110). Furthermore, our laboratory have recently demonstrated an evolutionarily conserved negative feedback mechanism through which HIF can regulate NF-κB in a mechanism that is dependent on the kinases, TAK-IKK and CDK6 (59).

CRC is a lethal disease affecting over 500,000 individuals annually (111). In contrast to the protective effects of HIF and NF-κB activity in IBD, both can play important roles in the development of colorectal malignancy. In CRC, the hypoxic milieu is similar to that of IBD but, critically, the cells are transformed to allow them to react differently to the activation of either system. In addition, chronic inflammation is a hallmark of cancer (112). The role of the HIF system and the role of NF-κB activity are considered below, and the significance of their crosstalk with respect to the development of CRC is examined.

The role of the HIFs in cancer progression has long been appreciated due to their ability to promote angiogenesis through one of the principally identified HIF-1α target genes, VEGFA (113). However, it is becoming more evident that hypoxia and the HIF system can affect tumour growth through modulation of proliferation, apoptosis and epithelial to mesenchymal transition (EMT) (Fig. 5). Hypoxia and the subsequent HIF activation are generally understood to be prognostically bad and lead to tumour progression (114). In CRC, HIF-1α stabilisation has been shown to lead to a poor disease outcome. Shay et al (114) demonstrated that the inhibition of HIF signalling using acriflavine halted the progression of an autochthonous model of established colitis-associated colon cancer in immuno-competent mice. In their model, treatment with acriflavine was shown to decrease tumour number, size and advancement, in an effect thought to be mediated through the inhibition of HIF-dependent targets, such as VEGFA. These data provide a direct link between HIF-1α expression and tumour progression. However, HIF isoform activation can be antagonistic in the context of tumour progression. In contrast to the effect of high HIF-1α expression, high HIF-2α expression has recently been reported to prevent CRC progression (115). The antagonistic effects of HIF-1α and HIF-2α are important for the regulation of proliferation and apoptosis in cancer biology (40,82,116). The HIF system can affect proliferation through the regulation of cMyc. HIF-1α can promote cell cycle arrest by the direct opposition of c-Myc activity and the induction of p21 in CRC (116). Conversely, HIF-2α has been shown to promote proliferation in through its augmentation of cMyc function (40).

The HIF system can also affect apoptosis through the regulation of p53. p53 stability leads to apoptosis in somatic cells and it is frequently mutated in cancers in pursuit of immortality. HIF-1α has been shown to stabilise wild-type p53 via physical interaction through its ODD (41,117). As a form of negative feedback, p53 can promote the degradation of HIF-1α (118). The negative feedback of wild-type p53 on HIF-1α could explain the increased stability of HIF-1α in tumours that express mutant p53 which is incapable of degrading HIF-1α. The net result of the p53-HIF-1α interaction is increased apoptosis in damaged cells that are exposed to hypoxia (119). HIF-2α can inhibit p53 phosphorylation, resulting in a reduction in p53 pathway activity and the prevention of apoptosis in response to damaging stimuli (120). In addition to its role in the regulation of p53, HIF has been linked to the positive regulation of apoptosis through the control of several pro-apoptotic factors, including caspase-3, Fas and Fas ligand (121).

Hypoxia is a critical determinant of the motile and invasive phenotype of cancer cells. HIF activation is also important in the regulation of genes involved in EMT, including the direct regulation of the EMT-promoting transcription factors, Snail and Twist, which have both been described as direct targets of the HIF system (122–124). EMT is a critical event in the induction of tumour metastasis (125). Notch has also been shown to mediate HIF-1α-dependent EMT (126).

The role of NF-κB in CRC is an active area of study (101,127–129). Inflammation is an important trigger in the establishment and development of CRC. Patients with long-standing IBD have an increased risk of developing CRC (127,128). In this context, NF-κB activation can promote tumourigenesis and CRC progression. In CRC, chronic inflammation results in sustained reactive oxygen species (ROS) production, leading to DNA damage (Fig. 5) (130). Treatment with non-steroidal anti-inflammatory drugs (NSAIDs) reduces the development of CRC in patients with IBD and hereditary CRC (131,132), and the inactivation of NF-κB signalling reduces the formation of inflammation-associated tumours (101,129). IL-6 has been shown to be important for the number and size of tumours formed in mice (133), and IKKβ conditional knockout mice have been shown to develop more numerous tumours (134).

As with the HIF system, the mechanism of NF-κB-induced tumourigenesis and progression can be multifactorial. The activation of the NF-κB pathway confers survival, proliferation, angiogenic and migratory advantages (Fig. 5) (112,135–138); all of which are hallmarks of cancer (112). NF-κB activation can block apoptosis by regulating the anti-apoptosis proteins, such as inhibitor of apoptotic proteins (IAPs) (139), or by the inhibition of prolonged c-Jun N-terminal kinase (JNK) signalling, modulating the accumulation of ROS (140). Alternatively, NF-κB activation can enhance IL-2 production, which can activate Janus kinase 3 (Jak3) by autophosphorylation (141). Jak3 can activate STAT3. Jak3 and STAT3 over-activation has been observed in human colon cancer in vivo and in vitro, and shown to prevent apoptosis, leading to poor prognosis (142,143). In addition, NF-κB activation can affect proliferation and cell growth through the regulation of its target genes, cyclin D1 and cMyc (144–146), and promote angiogenesis through the regulation of VEGF and IL-8 (136). Finally, NF-κB activation has been shown to affect the expression of matrix metalloprotease-9 (MMP-9), in murine colon adenocarcinoma cells (147), an important protein in the regulation of migration and invasion.

The data presented above demonstrate a clear overlap between the effectors of the HIF and NF-κB systems in the establishment and development of CRC. Solid tumours are characterised by the presence of hypoxia, as well as inflammation (6). Potential points for crosstalk include the regulation of cMyc and p53 (Fig. 5). NF-κB interacts with the co-activators, p300 and CREB-binding protein, to inhibit p53 function. This effect is reinforced by the NF-κB-dependent upregulation of the p53 inhibitor, mouse double minute 2 (MDM2) (3,148) and is similar to that exerted on p53 by HIF-1α. The expression of NF-κB, HIF, VEGF and Bcl-3 has been shown to correlate with proliferation, angiogenesis, decreased survival and a poor clinical outcome (149,150). In addition, TNF-α has been shown to stabilise Snail and β-catenin in a process that requires the downregulation of glycogen synthase kinase-3β (GSK3β) by NF-κB and the activation of Akt cascades, resulting in the promotion of EMT (151). These data are clinically important since NF-κB and Twist have been associated with lymph node metastasis in patients with CRC (152). Interestingly, HIF has been shown to interact with both Snail and Twist, making this another potential point for crosstalk between the pathways (153).

In addition to the mechanisms outlined above, there is a complex interplay between HIF, NF-κB and adenomatous polyposis coli (APC), that appears to be important in CRC (Fig. 5). One of the earliest events in the development of CRC is loss of the APC gene. Our laboratory has recently reported a functional crosstalk between HIF-1α and APC at the transcriptional level (154). HIF activation represses APC expression, acting at its promoter to result in positive activation and proliferation through the Wnt/β-catenin signalling and the TCF-LEF pathway (155), reduction in genetic and microtubule stability and reductions in cell migration (6,156).

The repression of APC by HIF-1α is complicated by the fact that medium levels of β-catenin can induce NF-κB, resulting in positive feedback, and high levels of β-catenin inhibit NF-κB, resulting in negative feedback (6). Further studies are required to determine the functional significance of this interaction in vivo. However, it represents another exciting point of crosstalk with importance for CRC disease progression.