The interplay between HIF-1α and noncoding RNAs in cancer

Recently, the number of HRNs that have been identified is expanding rapidly, illustrating the complexity of hypoxia-responsive gene reprogramming and the importance of reconsidering the involvement of the noncoding genome in this adaption [30, 31]. MiRNAs are the most studied subgroup of ncRNAs, and hypoxia-responsive miRNAs (HRMs) have exhibited promising oncogenic and/or tumor-suppressive functions in the oncogenesis and development of cancers [32]. In this section, we systematically discuss the regulatory mechanism of ncRNAs by HIF-1α. As a result, we summarize the functions of miR-210 in tumors in Table 1 as well as other HRMs and their roles in Table 2.

It is well appreciated that the HIF complex is a crucial transcription factor coordinating the cellular transcriptional response under hypoxic stress. According to their interplay with the HIF complex, hypoxia-responsive lncRNAs (HRLs) can be categorized into HIF-dependent and HIF-independent. We summarize the regulatory mechanisms underlying the HIF-1α-altered expression of HRLs in Table 3.

Although belonging to the lncRNA family, circRNAs are always discussed separately due to their unique structure with a covalently closed continuous loop. In an experiment on breast cancer cells in a hypoxic environment, researchers found that circZNF292, circDENND4C, and circSRSF4 were upregulated after hypoxia treatment, while among these, only circDENND4C was demonstrated to be activated by the induction of HIF-1α [94]. CircDENND2A was predicted to be an HRC in glioma via bioinformatic analysis. Hypoxia-induced overexpression of circDENND2A promotes the migration and invasion of glioma cells by sponging miR-625-5p [95]. In addition, more HRCs, including circRNA_403658, circDENND4C, and circRNA_0000977, have been identified to participate in cancer progression by sponging corresponding miRNAs [29, 96, 97]. Although limited research has uncovered the role of HRCs, promising functions of circRNAs in human cancers have been preliminarily established, and we believe that HRCs will be the next hotspot in the research field of hypoxia-induced cancer progression.

MiRNAs play significant regulatory roles in eukaryotes by binding to the 3’-UTRs of corresponding mRNA transcripts, leading to silencing of the target gene at the posttranscriptional level. A large number of studies have confirmed the existence of the direct interplay between miRNAs and the 3’-UTR of HIF-1α [101‐121]. Although the classic mechanism is widespread and important in tumors, we do not describe it in detail in the section due to the simplicity of the interaction.

Based on the previous notion that HIF-1α is a target of miR-138 [120], Cai et al. proposed that lncRNA LINC00152 functions as an miRNA sponge for miR-138 through a direct interaction to abrogate the suppressive effect of miR-138 on the expression of HIF-1α [122]. Intriguingly, an almost identical role of lncRNA PVT1 acting as ceRNA for miR-199a-5p in non-small-cell lung cancer under hypoxia was later verified [123]. Additionally, the ceRNA roles of lncRNA HOTAIR [124], Linc ROR [125], lncRNA NEAT1 [126], lncRNA UCA1 [127], and lncRNA PVT1 [128] for their respective miRNAs in cancer progression have also been demonstrated. In nasopharyngeal carcinoma, regulation at the posttranscriptional level has been further extended. To be more specific, lncRNA DANCR was found to directly interact with the ILF3/ILF2 complex, and interleukin enhancer binding factor 3 (ILF3), as the most enriched DANCR-binding protein, is a double-stranded RNA-binding protein and can complex with ILF2 to stabilize mRNA and regulate gene expression, subsequently stabilizing the HIF-1α mRNA and leading to nasopharyngeal carcinoma metastasis [129].

Similar to the classic mechanism by which lncRNAs participate in cancer prognosis, the most common mechanism by which circRNAs regulate biological processes is also related to the HIF-1α model. This mechanism mainly involves three kinds of RNAs, including mRNAs, pseudogene transcripts and lncRNAs, but circRNAs have followed lncRNAs in becoming a novel hotspot of research on the ceRNA family. Research conducted by Chi et al. suggested that circRNA circPIP5K1A functions as a miR-600 sponge to inhibit miR-600 to disrupt the interaction at the 3’-UTR between HIF-1α and miR-600 to promote HIF-1α posttranscriptional expression, as well as proliferation and metastasis of non-small-cell lung cancer [130]. In addition, in hepatocellular carcinoma, circRNA_0046600 could upregulate HIF-1α by sponging miR-640 to promote cancer progression [131]. CircRNAs are a novel research focus,, so no additional studies on the regulatory roles of circRNAs in HIF-1α expression are currently available. Given the significant role of circRNAs in regulating target gene expression, we speculate that circRNAs should be the next focus in the field of ncRNA-mediated regulation of HIF-1α expression.

In addition to the basic interaction between miRNAs and the 3’-UTR of HIF-1α, miRNA-mediated transcriptional regulation of HIF-1α expression is a common mechanism in cancer progression. MiR-214 upregulates HIF-1α and VEGFA with the suppression of ING4 to promote the invasion, proliferation and migration of non-small-cell lung cancer cells [132], and a possible mechanism is that ING4, which is recruited by egl-9 family hypoxia-inducible factor 1 (EGLN1), unexpectedly has no effect on HIF-1α degeneration but acts as an adapter protein to recruit transcriptional repressors to regulate HIF activity [157]. MiR-206 can attenuate the growth and angiogenesis of non-small-cell lung cancer cells through the 14-3-3z/STAT3/HIF-1α/VEGF pathway. In particular, 14-3-3ζ binds to p-STAT3 (Ser727) and increases its activation. Knockdown of STAT3 blocks the 14-3-3ζ-induced increase in HIF-1α mRNA expression and attenuates the 14-3-3ζ-induced binding of HIF-1α to the VEGF promoter [133]. In addition, Dico et al. reported that miR-675-5p interacts with the RNA binding protein HuR to stabilize the mRNA of HIF-1α, along with its additional inhibitory effect on VHL [134].

Moreover, at the transcription level of HIF-1α expression, experimental evidence of lncRNA-mediated regulation already exists. Wang et al. suggested that lncRNA CPS1-IT1 could serve as an Hsp90 cochaperone, and this interaction in turn reduces the binding affinity between Hsp90 and HIF-1α, leading to transcriptional inactivation of HIF-1α and diminished EMT of hepatocellular carcinoma cells [135]. In addition, the lncRNA-mediated regulation of the mTOR/HIF-1α/P-gp signaling pathway marked by increased HIF-1a mRNA levels in gastric cancer cells might also suggest the alteration of HIF-1α transcriptional activity [136]. Although the function of lncRNAs as transcriptional regulators has been widely explored, the mechanisms underlying these functions remain poorly understood and require further investigation.

MiR-128, which is regulated by snail family zinc finger 1 (SNAIL) transcriptionally, in turn modulates the expression of ribosomal protein S6 kinase, polypeptide 1 (RPS6KB1), also known as p70S6K, and afterwards disrupts downstream HIF-1α at the translational level and consequently suppresses pyruvate kinase 2 (PKM2) expression to inhibit the growth and metabolism of prostate cancer cells [137], which expands the interplay between HIF-1α and miRNA at the translational level.

As for the translational activity of HIF-1α, lncRNA MEG3 was found to be decreased after nickel exposure, which triggers downstream c-Jun/ PH domain and leucine rich repeat protein phosphatase 1 (PHLPP1) to activate the Akt/p70S6K/S6 axis. Enhanced phosphorylation at Ser235/236 of the 40S ribosomal protein S6 therefore boosts HIF-1α translation in the nickel-induced malignant transformation of human bronchial epithelial cells [138]. In hepatocellular carcinoma cells, overexpressed lncRNA UBE2CP3 enhances human umbilical vein endothelial cell proliferation, migration and angiogenesis, which is attributed to the ERK/p70S6K/HIF-1α/VEGFA signaling axis activated by lncRNA expression deviating from normal status [139]. Distinctly, lncRNAs are defined as ncRNAs without translational function. However, during HIF-1α translation, lncRNAs play indispensable roles.

Complexes formed between HIF the coactivators CBP/p300 are essential for HIF transcriptional activation. FIH1, which blocks the interaction between HIF-1α and CBP/p300, is validated to be downregulated because of a corresponding miRNA deficiency in tumors, consequently suppressing the tumor hypoxia response and angiogenesis by suppressing HIF-1α transcription and VEGF production [140]. Similar mechanisms of miR-135b, miR-182, and miR-31 have been confirmed in head and neck squamous cell carcinoma [141], non-small-cell lung cancer [142] and colorectal cancer [143], respectively.

The stability of HIF-1α is a critical factor in its action on relevant gene expression, and WD repeat and SOCS box containing 1 (WSB1) has been reported to enhance the HIF-1α protein stability derived from the abnormally low expression of miR-592 in hepatocellular carcinoma cells with enhanced glycolysis and proliferation [144]. In osteosarcoma cells, which have a high energy demand but low ATP-generating efficiency, increasing miR-543 targets the 3’-UTR of protein arginine methyltransferase 9 (PRMT9) to decrease PRMT9-induced HIF-1α instability; thereafter, elevated HIF-1α boosts glycolysis and proliferation of osteosarcoma cells [145]. As an indispensable molecule in the degradation of HIF-1α, the role of PHD in HIF-1α stabilization should not be ignored. Indeed, Tanaka et al. indicated that upregulated miR-183 in glioma was able to inhibit isocitrate dehydrogenase 2 (IDH2) levels, which elevated HIF-1α levels by reducing the cellular levels of α-KG, a substrate of PHD [146]. In glioma, the targeted inhibitory effect of increasing miR-23b on VHL unsurprisingly activates HIF-1α/VEGF signaling to promote tumor progression [147].

Proteasomal degradation is often regulated by phosphorylation [158], and blocked activation of the Akt and ERK1/2 pathways caused by miR-145-mediated N-RAS and insulin receptor substrate 1 (IRS1) expression inhibition was confirmed to suppress the expression of HIF-1α and downstream VEGF in restricted colorectal cancer growth, which is speculated to depend on its interference with the normal HIF-1α protein degradation process [148]; in addition, almost the same signaling initiated by miR-30e can be seen in breast cancer [149]. Analogously, the PIK3C2α/AKT/HIF-1α/VEGFA pathway regulated by miR-26a plays a role in inhibiting angiogenesis in hepatocellular carcinoma [150]. Because of its important role in the PI3K/Akt/mTOR signaling pathway [159], mTOR and downstream HIF-1α have been experimentally suggested to be inhibited by miR-99a, which reverses the breast cancer stem cell malignant phenotype [151].

LncRNAs also play critical roles in the posttranslational regulation of HIF-1α expression. Osteosarcoma amplified 9 (OS9) has an overall effect on the degradation of HIF-1α, including hydroxylation, VHL binding, and proteasomal degradation, by interacting with both HIF-1α and PHDs [160], and lncRNA ENST00000480739 contributes to metastasis and progression of pancreatic ductal adenocarcinoma by targeting and upregulating HIF-1α [152]. Whether other forms of lncRNA-related posttranslational regulation are essential for HIF-1α needs to be further explored.

The nuclear transfer of HIF-1α is also affected by miRNAs. Importin 7 (IPO7) is a mediator specifically related to HIF-1α nuclear translocation [161], while in chronic myelogenous leukemia cells under curcumin treatment, there is a curcumin-induced downregulation of IPO7 expression caused by miR-22 activation, which further elicits blocked cytoplasm-to-nucleus shuttling of HIF-1α to restrain the glycolytic enzyme profile [153].

Similar to miRNAs, lncRNA H19 has been confirmed to positively participate in HIF-1α nuclear translocation to drive multiple myeloma cell dissemination, although the specific molecules responsible for this procedure are unknown [154]. As a transcription factor, HIF-1α plays an essential role in the nucleus. Thus, the regulation of HIF-1α nuclear transfer by ncRNAs is a promising regulatory mechanism to block the oncogenic function of HIF-1α in cancer progression.

The direct interaction between HIF-1α and lncRNAs is not confined to the 3’-UTR. Shih et al. have demonstrated an extremely important role of lncRNA MIR31HG, which acts as a co-activator and complexes with HIF-1α to facilitate the recruitment of the HIF-1 complex, augmenting the HIF-1 transcriptional network essential for oral cancer progression and leading to metabolic reprogramming, increased sphere-forming ability and metastasis [155]. However, lncRNA NDRG1-OT1 was reported to act as a scaffold for recruiting HIF-1α via its third-quarter fragment, rather than whole molecule, to increase the expression of the downstream gene N-myc downstream regulated gene 1 (NDRG1) in breast cancer cells under hypoxia, along with the different effects of the remaining fragments on the same target gene [156].

In addition to the positive feedback loop that causes continuous activation of pathway components, a negative feedback loop between HIF-1α and ncRNAs leading to the restriction of molecular members has also been confirmed by some researchers. In human umbilical vein endothelial cells, there is a negative regulatory loop containing miR-439 and HIF-1α in which HIF-1α induces miR-439 to bind to and destabilize HIF-1α mRNA, hence reducing the activity of HIF-1α in turn. Moreover, confirmation of this mechanism in HeLa cells further exhibited its significance in cancer therapeutics [180]. Similarly, based on this negative loop, in pancreatic cancer, HIF-1α-induced miR-646 expression was shown to target migration and invasion inhibitory protein (MIIP) to inhibit the deacetylation ability of HDAC6, which eventually promoted the acetylation and proteasomal degradation of HIF-1α [181].

Collectively, it seems quite feasible that ncRNAs, HIF-1α and other co-operators would eventually intertwine to form mutually reciprocal feedback loops in both positive and negative manners. We summarize these reciprocal feedback loops in Fig. 2. In these loops, any alteration in the expression level of any member would disturb the overall balance of the network, resulting in a shift to transcriptional reprogramming, posttranscriptional regulation or translational stability.

Several kinds of HRNs have shown unique value in the diagnosis of various tumors. In pancreatic cancer, plasma profiling of four miRNAs, including hypoxia-sensitive miR-210, and determination of their sensitivity and specificity values promises to generate feasible blood-based biomarkers for the early detection of pancreatic cancer [182], while the significantly increased expression of miR-107 seen in both tumor tissues and serum and its correlation with HIF-1α expression suggest the practicality of using miR-107 as a biomarker for the detection of gastric cancer and tumor hypoxia [64]. In colorectal carcinoma, circulating miR-210, miR-21 and miR-126 present high value as noninvasive markers for early diagnosis, screening, and prognosis [183].

HRNs are of great significance in evaluating the prognosis of tumors. In pancreatic cancer, the expression of miR-646 [181] and miR-548 [67] is correlated with clinicopathological indicators such as TNM stage and overall survival (OS), and hypoxia-induced lncRNA NUTF2P3-001 overexpression also indicates advanced TNM stage and shorter survival time of patients [88]. Both Low expression of miR-592 [144] and high expression of miR-130b [184] can bring about poorer OS in hepatocellular carcinoma patients. For gastric cancer, it has been demonstrated that miR-421 regulated by HIF-1α not only causes longer OS, but also can shorten the time to relapse of patients [185], and lncRNA BC005927 induced by hypoxia is also frequently upregulated in gastric cancer samples, showing adverse effects on a series of prognostic parameters, such as TNM stage, lymph node metastasis, and survival time [81]. Not surprisingly, scholars have revealed that aberrant expression of lncRNA H19 [92] and miR-215 [186] in glioblastoma confers a poor prognosis for patients. With regard to triple-negative breast cancers, a type of breast cancer with poor prognosis, patients with relatively low expression of miR-210 fortunately experienced significantly better disease-free and overall survival than those with high expression of miR-210 in a study in Japanese patients [187]. In addition, a strong correlation between high lncRNA EFNA3 expression and shorter metastasis-free survival was found in breast cancer patients [188], undoubtedly enriching the prognostic value of lncRNAs in this prevalent cancer. Innovative extraction and identification of circulating exosomal miR-21 from the serum of patients with oral squamous cell carcinoma and its close affinity with T stage, lymph node metastasis, and HIF-1α expression further supported its prognostic value, as well as the therapeutic value of inhibiting exosomes in the niche [63]. In addition, miR-210 overexpression was reported to play a potential prognostic role in upper tract urothelial carcinoma [189] and oropharyngeal squamous cell carcinomas [190].

In addition, the expression of circFAM120A was significantly downregulated in both hypoxic lung adenocarcinoma cells and cancer tissue from patients with lymph node metastasis, implying its potential to be a new biomarker of lung adenocarcinoma hypoxia [28]. Moreover, circRNAs lack a 5’ cap and 3’ ends, endowing them with more stable properties than parent linear RNAs [191]. Together with their abundant and conserved characteristics, these properties make circRNAs a remarkable candidate biomarker for neoplastic diseases.

The current practical applications related to regulatory mechanisms shared between HIF-1α and ncRNAs are relatively scarce but inspiring. For instance, most clear cell renal cell carcinomas are marked by the loss of VHL tumor suppressor gene function, continuous expression of HIF-1/2α, and maladjusted expression of oncogenic miRNAs. Rustum et al. found that the levels of specific biomarkers associated with drug resistance in clear cell renal cell carcinoma, such as HIFs, oncogenic miR-155 and miR-210, and VEGF, could be selectively downregulated by methylselenocysteine or seleno-L-methionine in a dose- and time-dependent manner, which conferred existing anticancer therapies with enhanced therapeutic efficacy and selectivity [192]. Similarly, the antitumor effect of a novel synthetic derivative of curcumin treatment seen in pancreatic cancer was partially attributed to its inhibition of the expression of miR-21, miR-210, and HIF-1α, which are aberrantly upregulated under hypoxic conditions [193]. Additionally, Isanejad et al. reported that combination hormone therapy with 5-week interval exercise training could inhibit tumor angiogenesis in a mouse model of breast cancer, and the underlying mechanism could be partially explained by the suppressive effect of this combination therapy on the miR-21/HIF-1α signaling pathway [194]. Xu et al. suggested that targeting carcinostatic miR-338-3p/HIF-1α axis was conducive to sensitizing hepatocarcinoma cells to sorafenib [102], and Bertozzi et al. found that miR-17-5p and miR-155 were involved in camptothecin-induced HIF-1α reduction in human cancer cells due to their specific targeting of HIF-1α mRNA [195].

Encouragingly, ncRNAs have been increasingly considered as potential cancer therapeutic targets owing to their tissue specificity, high expression levels and crucial roles in tumor growth and progression. To date, the development of RNA-targeting methods has provided tremendous opportunities to modulate ncRNAs for cancer therapy [196, 197]. Most excitingly, novel classes of RNA-based therapeutics show great potential to modulate ncRNA activity in diverse ways [198]. Although most ncRNA-targeted treatments remain in the early stages of development, future technical innovations will provide new opportunities, and better insights into the associations between HIF-1α and ncRNAs in cancer biology will lay wide theoretical foundations for ncRNA-related targeted therapies.