Mild Chronic Hypoxia-Induced HIF-2α Interacts with c-MYC Through Competition with HIF-1α in Hepatocellular Carcinoma Proliferation

Background: Hepatocellular carcinoma (HCC) has emerged as a major cause of cancer deaths globally, in which hypoxia and activated hypoxia-inducible factors (HIFs) play an important role. The sibling rivalry between HIF-1α and HIF-2α in hypoxic tumor growth and progression is still debated, including in HCC. This may be associated with the regulation of unique target genes, like c-MYC and mTOR. Methods: In the current study, twenty-six corresponding tumor- and non-tumor tissues, taken from patients with HCC, who underwent liver resection, were analyzed. In vitro, co-immunoprecipitation (Co-IP), Western blot, MTT assay, colony formation assay and Annexin V-FITC/PI staining apoptosis Assay were used to elucidate the relationship between HIF-1α and HIF-2α in hypoxic HCC cell proliferation and involved mechanism. Results: HIF-2α, but not HIF-1α, has a positive correlation with the expression of c-MYC in tumor tissues. In vitro, rapid HCC cell proliferation and increased interaction of HIF-2α/c-MYC were observed in mild chronic hypoxia. It was conrmed in situ proximity ligation assay and co-immunoprecipitation assay that although mild hypoxia up-regulated all HIF-1α, HIF-2α, and c-MYC, mTORC2-regulated HIF-2α competed with HIF-1α to bind c-MYC. HIF-2α knockdown, but not HIF-1α knockdown, decreased the expression of downstream c-MYC and mTORC1, suppressed hypoxic cell proliferation, and induced more cell apoptosis. Moreover, the inhibition of upstream protein PI3K by inhibitor Apitolisib counteracted this mechanism of adaptation to mild hypoxia hereby inducing cancer cells to apoptosis in HCC. Conclusions: study cell proliferation during mild chronic hypoxia response. PI3K/mTORC2/ HIF-2α/c-MYC axis plays a key role in this process. PI3K inhibitor Apitolisib is suggested as a potential treatment option for HCC, especially for rapidly growing HCC in mild chronic hypoxia.


Introduction
Hepatocellular carcinoma (HCC) has emerged as a major cause of cancer deaths globally and numbers are increasing (1,2). Although anti-HCC treatments developed quickly in recent years, HCC behaves highly refractory to most anti-cancer therapies (3). Five-year survival of HCC patients remains dismal (4). To understand the mechanism of tumorigenesis and tumor resistance may provide a new therapeutic approach in HCC.
Recent studies indicated that rapid growth of tumors companied with intense metabolic activity, faulty vascularization, multiple nodules, and necrosis consumed a substantial amount of oxygen and therefore often caused a hypoxic microenvironment in HCC (5,6). Hypoxia, triggering a lot of molecular and cellular responses, is an important in uencing factor for tumor aggressiveness and therapeutic response (7). In addition to transcatheter arterial (chemo) embolization (TAE/TACE), other anti-tumor treatments, including Sorafenib, were con rmed to increase Hypoxia in HCC, which is involved in tumor resistance and progression (8,9).
The overexpression of HIF-1α or HIF-2α has been detected in HCC patients and is closely associated with poor clinical outcome (6). HIF-1α and HIF-2α share high sequence homology, but they regulate downstream genes independently and have different regulations in response to hypoxia in tumors (11)(12)(13).
The two HIFα isoforms even appear to oppose each other on tumor cell proliferation (14). The sibling rivalry in hypoxic tumor growth and progression is mediated in part through the regulation of unique target genes, as well as interactions with important cellular pathways, including c-MYC and mTOR (15,16). c-MYC, an important role in cellular growth, differentiation, apoptosis, and metabolism, is indicated as an integral part of the HIF-α mediated hypoxic response (15,17). Classic, HIF-1α regulates cell cycle and DNA repair genes through inhibiting of c-MYC activities; HIF-2α, in contrast, promotes cell proliferation by enhancing c-MYC activities (16). However, recent studies demonstrated that HIF-1α can cooperate with c-MYC to induce the expression of speci c target genes, including those encoding the glycolytic enzyme hexokinase 2 (HK2), pyruvate dehydrogenase kinase 1 (PDK1), and vascular endothelial growth factor A (VEGFA) (18,19). Moreover, c-MYC induces the expression of HIF-1α Posttranscriptionally in solid tumors, further proving that Hif-1α is downstream of c-MYC (19)(20)(21). Although HIF-1α is ubiquitously expressed, HIF-2α limited to endothelium, kidney, heart, lungs, small intestine, and liver (22,23). Therefore, the relationship of HIF-1/2α with c-MYC in HCC is not yet clear.
This oppose effect between HIF-1α and HIF-2α is also shown in regulating the mechanistic target of rapamycin (mTOR) pathway, which plays a central role in physiology, metabolism, the aging process, and common diseases (24,25). The mTOR kinase forms a complex with the cofactors and depending on whether it binds Raptor or Rictor, it forms mTORC1 or mTORC2, respectively (25). HIF-1α translation can be driven by growth factor-activated PI3K and mTOR1 (26,27). Overexpressed HIF-1α may, in turn, have a negative feedback on mTORC1 activity. However, HIF-2α, being regulated by PI3K/mTORC2 but not PI3K/mTORC1, was con rmed to stimulate the expression of mTORC1 under hypoxic situation (12,28,29). Additionally, it has been indicated the indispensable effect of either mTORC1 or mTORC2 on c-MYC activity (30,31). An analysis of the mechanism that HIF-1α and HIF-2α antagonize one another to balance c-MYC in HCC responding to hypoxia may, therefore, provide insight into its rapid growth in hypoxic microenvironment; those insights could then lead to a new therapeutic approach for anti-tumor treatment of HCC.

Page 4/43
Twenty-six corresponding tumor-and non-tumor tissues from patients with HCC who underwent liver resection were analyzed in the current study. This study was carried out with the patients' informed consent and approval from the local ethics committee. This approval is followed the guidelines stated in the Declaration of Helsinki. The clinical characteristics for all patients are provided in Table 1.  Colony-formation assay Cells (400 cells/well) were plated in 6-well plates. After treatment conditions for 14 days, cells were xed and stained using crystal violet (20% ethanol and 0.5% crystal violet in water). The number of colonies, de ned as > 50 cells/colony, were counted.
In situ Proximity Ligation Assay (PLA assay) All cells were pipetted into 8-well culture slides. After treatments, cells were xed, permeabilized, and blocked. Primary antibodies were incubated with the cells overnight at 4°C. Afterward, PLA MINUS and PLA PLUS probes were added. The ligation solvent and ligase were mixed and then incubated before the nal ampli cation solution was added into cells. Samples were detected using a uorescence microscope and counted with BolbFinder Software (http://www.cb.uu.se/~amin/BlobFinder/).

Co-Immunoprecipitation (Co-IP)
Cells were transfected with the plasmid of Flag-HIF-1α and Flag-HIF-2α (Sino Biological, Wayne, PA, USA). After 48 hours of transfection, stably transfected cells were selected and treated. Afterward, the following steps were carried out by the manufacturer's protocol of the Pierce TM Co-Immunoprecipitation-Kit (Thermo Scienti c, Waltham, MA, USA). Cell lysates were immunoprecipitated with Flag or c-MYC antibody and the expression of Flag and c-MYC in the eluate were detected by western blot analysis.

Western blotting
Cells were lysed using RIPA buffer (Sigma-Aldrich, St. Louis, MO, USA) for 10 min on ice, followed by centrifugation. The supernatant was collected, and the protein concentration was determined using a BCA™ Protein Assay Kit (Thermo Scienti c, Rockford, IL, USA). Whole-cell extracts (20μg) were heated with LDS sample buffer (Invitrogen, Carlsbad, CA, USA) at 70 °C for 10 min, separated using SDS polyacrylamide gel electrophoresis in 4-12 % Bis-Tris gel (Invitrogen, Carlsbad, CA, USA), and transferred to the nitrocellulose membrane (Bio-Rad, Hercules, CA, USA). After blocking for 1 h, the membrane was incubated overnight with primary antibodies at 4 °C. Subsequently, the membrane was probed using IRDye® 680RD Goat anti-Mouse/Rabbit IgG secondary antibody (LI-COR Biosciences, Lincoln, Nebraska, USA) for 1 h at room temperature. The membranes were visualized using an Odyssey CLx scanner (LI-COR Bioscience, Lincoln, Nebraska, USA) and the results were obtained using Image Studio software (LI-COR Bioscience, Lincoln, Nebraska, USA).

Short hairpin RNA (shRNA) Transfection
Cells were seeded at a density of 1.5 x 10 5 -2.5 x 10 5 in 6-well plates and incubated overnight with normal growth medium without antibiotics. Subsequent detailed steps were operated by the protocol from Santa Cruz Biotechnology (Santa Cruz, CA, USA). 48 hours after transfection, the medium was aspirated and replaced with fresh medium containing 1μg/ml puroMYCin every 3 days to select stably transfected cells.

Apoptosis
Cells (1 x 10 6 cells/well) were seeded in six-well plates and treated. The treated cells were harvested onto ice and xed. Then the cells were incubated with PI stock solution in the dark for 30 min at room temperature. the sub-G1 cell fraction of activated cells was detected by FACS Calibur (BD Biosciences, San Jose, CA, USA) and the data was analyzed with Flowjo Software (www. owjo.com).

Scratch Wound Assay
During cells were treated, the time-lapse images for each position in the wells were analyzed and the wound areas were measured to determine the percentage of wound closure at several time points throughout the course of the assay. To ensure that the similar wound areas were compared, the produced wound area was traced and measured by using ImageJ software(http://imagej.nih.gov/ij/).

Statistical Analysis
The IC50, t-test, and one-way analysis of variance were calculated, and the gures were plotted using GraphPad Prism 7.0. The values obtained are the mean (± SEM) of three replicates. Statistical difference was considered signi cant when p < 0.05.

HIF-2α, but not HIF-1α,has a positive correlation with the expression of c-MYC in HCC patients
To evaluate the correlation between HIF-1/2α and c-MYC, the expression of HIF-1α, HIF-2α, c-MYC, Rictor and Raptor were detected in 26 patients diagnosed with HCC ( Fig. 1A-B). The highly expressed targets above in tumor tissues compared to corresponding non-tumor tissues were con rmed to be associated with the degree of tumor stage (Fig. 1C). Especially, the average expression of c-MYC in tumor tissues was up-regulated at all UICC stages aberrantly. In correlation analysis, the positive relationships between HIF-2α and c-MYC, HIF-2α and Rictor, Rictor and c-MYC, HIF-1α and Raptor were con rmed (score >0.5, Fig. 1D). Moreover, the linear correlation between HIF-2α and c-MYC was double-check (r = 0.7772, P < 0.0001, Fig. 1E). This suggests that HIF-2α, not HIF-1α, correlates its expression with c-MYC.
In the analysis of the expressions of targets at each UICC stage, unlike other proteins, the increasing number of patients with high expressed HIF-2α was tumor stage-dependent ( Fig. 1F-G, s Fig. 1A-B). Kaplan-Meier survival analysis from HCC public gene database further showed that high HIF-2α, compared to other targets, is more closely related to a poor overall survival (Fig. 1H). These data point to the important role of HIF-2α, but not HIF-1α, in tumorigenesis and correlations with c -MYC.

Rapid cell proliferation and increased interaction of HIF-2α/c-MYC in HCC under mild and chronic hypoxia
In vitro, Cocl 2 was introduced into HCC cell lines Huh7 and HepG2 to create a hypoxic environment. The and HepG2 cells upon incubation with Cocl 2 low . Cocl 2 IC50 inhibited the total clone formation rate remarkably (Fig. 2D).
To investigate the relationship between activated HIF-1/2α and highly expressed c-MYC, a Co-IP assay was applied and indicated that the HIF-2α/c-MYC interaction signi cantly increased under mild hypoxia compared to the HIF-1α/c-MYC interaction (Fig. 2E). Likewise, active interaction of HIF-2α/c-MYC and noninteraction of HIF-1α/c-MYC was con rmed in PLA assay analysis (Fig. 2F). These results indicate that although hypoxia condition up-regulates all HIF-1α, HIF-2α, and c-MYC, HIF-2α seems to compete with HIF-1α to interact with c-MYC.
HIF-2α enhances c-MYC and mTORC1 activities and its key role in hypoxic cell proliferation To study further this "competition" between HIF-1α and HIF-2α, shRNA transfection targeting HIF-1α and HIF-2α (Gene: EPAS1) was conducted in Huh7 and HepG2 cell lines (s Fig. 2A). Unlike that HIF-1α silencing did not affect the protein levels of HIF-2α, c-MYC, Rictor, Raptor and phosphorylation of mTOR, HIF-2α knockdown signi cantly decreased the expression of c-MYC, Raptor, and phosphorylation of mTOR, but not Rictor and HIF-1α ( Fig. 3A-B). Two non-redundant complexes of mTOR have been described: mTORC1 containing Raptor and mTORC2 containing Rictor (Fig. 3C) (25). These results above explained that up-regulated expressions of c-MYC and Raptor by hypoxia can be inhibited when HIF-2α was knockdown, suggesting that c-MYC and mTORC1 are the downstream of HIF-2α. Furthermore, more interaction of HIF-1α/c-MYC was observed in shHIF-2α transfected cells under hypoxia condition (Fig. 3D-E). On the contrary, HIF-1α knockdown promoted the protein binding between HIF-2α and c-MYC. This demonstrated that highly expressed HIF-2α counteracts HIF-1α to interact with c-MYC in hypoxia.
To con rm that HIF-2α is a key in hypoxic cell proliferation, the cell viability, migration, and apoptosis of shHIF-1/2α transfected cells under low hypoxia conditions were evaluated. Compared to HIF-1α knockdown, HIF-2α knockdown inhibited hypoxic cell growth more signi cantly (Fig. 4A). Also, shHIF-2α cancer cells had low cellular migration and more apoptosis (Fig. 4B-E). The induced apoptosis in hypoxic cells through silencing HIF-2α was also con rmed by up-regulated protein levels of apoptotic proteins: FADD, cleaved caspase 3, cleaved caspase 8, and cleaved caspase 9 (Fig. 4F). Therefore, HIF-2α knockdown may reverse and sensitize cancer cells to hypoxia-induced apoptosis.
HIF-2α is regulated by PI3K/mTORC2 in response to hypoxia Since that HIF-2α knockdown does not affect the expression of Rictor, and recent studies indicated that PI3K/mTORC2 regulates transcription of HIF-2α/EPAS1 in neuroblastoma (32). Here, we hypothesized that mTORC2, as an upstream protein, regulates the expression of HIF-2α in HCC responding to hypoxia. As shown in results, the pathway of PI3K/AKT/mTOR was activated in cell lines Huh7 and HepG2 after exposure to Cocl 2 low treatment (Fig. 5A). shRNA targeting Rictor was applied to decrease the formation of mTOR complexes-mTORC2 (s Fig. 3A). Rictor knockdown inhibited the phosphorylation of mTOR and Akt, followed by down-regulated expression of Rictor, HIF-2α, and c-MYC in Cocl 2 low treated cells (Fig. 5B).
Furthermore, the interaction of both HIF-1α/c-MYC and HIF-2α/c-MYC was suppressed in shRictor transfected cells (Fig. 5C-D). Compared with untreated cells and hypoxic cells, cell proliferation was reduced after knockdown of Rictor (Fig. 5E). Additionally, shRictor transfected cells was shown to have a lower cell viability than shHIF-2α transfected cells in response to hypoxia. Altogether, these data support the hypothesis that mTORC2 plays a key role in promoting hypoxic cell proliferation by up-regulating the expressions and interaction of downstream HIF-2α and c-MYC in HCC.
The PI3K inhibitor Apitolisib induce hypoxic cell death in HCC The involvement of the PI3K/mTORC2 pathway in response to hypoxia suggests that blockade of this pathway may be a promising approach for the treatment of HCC. Here, Apitolisib (GDC-0980), a recently developed PI3K inhibitor for the treatment of solid tumors, was introduced and assessed its antineoplastic effect on HCC under hypoxia (33)(34)(35). Apitolisib treatment was shown to cause a dosedependent loss in cell viability in Huh7 and HepG2 (Fig. 6A). Under hypoxia condition, Apitolisib at IC50 value concentration (Apitolisib IC50 ) blocked the activities of PI3K, Raptor, Rictor, HIF-1/2α, and c-MYC; and then inhibited cell growth ( Fig. 6B-C). In FACS analysis, sub-G1 cell fraction detection was increased in Apitolisib treated hypoxic cells (Fig. 6D-E). This induction of apoptosis was also observable in DAPI stained cells, which showed apoptotic features, such as DNA fragments (Fig. 6F). These results were clearly con rmed by highly expressed apoptotic proteins: FADD, cleaved caspase 3, cleaved caspase 8, and cleaved caspase 9 in cells upon incubation with Cocl2 low and Apitolisib IC50 (Fig. 6G). Taken together, these data above indicate that PI3k inhibition by Apitolisib counteracts the mechanism of adaptation to hypoxia hereby inducing cancer cells to apoptosis in HCC.

Discussion
HIFs are broadly expressed and associated with poor prognosis in human cancers (10). There is now mounting evidence that its subunits HIF-1α and HIF-2α can promote highly opposing outcomes (15,36). Speci cally, the effect of HIF-2α on c-MYC is of interest, as it directly opposes that of HIF-1α, allowing differential control of tumorigenesis and tumor development based on the HIF-α subunit expression (37).
However, a more complex question emerged as both HIF-1α and HIF-2α protein expression are coexpressed in multiple tumor types including HCC, suggesting that understanding their relationship and the mechanism of regulating c-MYC need to be further studied (6,23).
In the current study, the correlation of HIF-2α, but not HIF-1α, with c-MYC was con rmed in tumor tissues from 25 patients with HCC. The expression of HIF-2α increased UICC stage-dependently, which suggests the role of HIF-2α in tumor development. In a previous study, high expressed HIF-2α was detected in HCC and adjacent noncancerous tissues, but not in normal liver tissues, which is positively associated with shorter overall survival (38). This is similar to our Kaplan-Meier survival analysis of the HCC gene dataset from TCGA. Moreover, increasing evidences speculate a positive regulatory role for HIF-2α in brosis, pathogenesis, metastasis, and resistance to anti-tumor treatments in HCC (22,(39)(40)(41)(42)(43)(44). Taken together, HIF-2α may be a potential target for the treatment of HCC.
The role of HIF-2α in HCC was also investigated both in vitro and in vivo that siRNA-mediated knockdown of HIF-2α in HepG2 cells impaired cell cycle progression in the presence of CoCl 2 and reduced cell proliferation (45). HCC mice lacking HIF2α in myeloid cells showed decreased in ltration of tumorassociated macrophages and delayed tumor progression (46). Here, HCC cells were shown as rapid growth in response to mild and long hypoxia, and highly expressed all HIF-1α, HIF-2α, and c-MYC. It was reported before that knockdown of HIF-1α increased HIF-2α and that knockdown of HIF-2α increased HIF-1a in HCC cells, suggesting a balance between HIF-1α and HIF-2α in HCC cells (47). In this study, HIF-2α was further con rmed to have a competed ability with HIF-1α to interact with c-MYC. More interaction of HIF-2α with c-MYC and non-interaction of HIF-1α with c-MYC was observed in hypoxic cells. This suggests that the balance between HIF-1α and HIF-2α seems to be based on interacting with c-MYC rather than simply regulating the expression of HIF-1α or HIF-2α.
In addition, it's HIF-2α knockdown that inhibited HCC cell proliferation and the expression of c-MYC under mild hypoxia remarkably, rather than HIF-1α knockdown. Similar results were to be found in renal tumor that HIF-2α actually increases c-MYC transcriptional activity to promote cell growth, HIF-1α does not (37,48,49). This may due to that HIF-2α levels may increase over time in hypoxia and play a role during mild, chronic hypoxia (50,51). Oxygenation in solid tumors varies from physiologic levels of approximately 5% to 8% O 2 to near anoxia (52). Tumor hypoxia therefore is also highly heterogeneous with both chronic and acute hypoxia (53). It was demonstrated that HIF-1α seems to have the dominant role in controlling responses to acute hypoxia, whereas HIF-2a is responsible for low or chronic hypoxia (51). The low and prolonged hypoxia increased transcription of HIF-2α consistently, whereas suppressed transcription of HIF-1α (54). Mechanistically, this may due to HIF-1a feedback regulation by prolyl-4-hydroxylase (PHD) under chronic hypoxia at 1% O 2 ; increased acetylation of the core histones H3 and H4 within proximal promoter region HIF-2α; and switch from HIF-1α to HIF-2α by hypoxia-associated factor (54)(55)(56). Increased HIF2α forms a complex with MAX, causing a dose-dependent stabilization of MYC-MAX and MYC-MAX-SP1 complexes, resulting in promoting c-MYC expression and its binding with MAX. These effects occur rapidly and can be detected after only 1-2 hours at 0.5% O 2 (15). mTORC2, activated by upstream PI3K, regulates the expression of HIF-2α in neuroblastoma (25,32). VHLde cient RCC cells that constitutively express HIF-2α have been also shown to have a separate dependency on mTORC1 and mTORC2 for HIF-1α and HIF-2α expression (57). Here, Rictor knockdown signi cantly inhibited the form of mTORC2, followed by decreasing the activities of HIF-2α and c-MYC. This shows that HIF-2α expression is regulated by hypoxia via the PI3K/ mTORC2 pathway in HCC cells.
Because the PI3K pathway appears to be a major activator of HIF2 activity and PI3K/mTOR inhibitors are in clinical use and late clinical trials (58). A new PI3K inhibitor Apitolisib was applied in this study and has been shown to suppress the PI3K/mTORC2/ HIF-2α/c-MYC axis and strongly induce apoptosis in HCC cells under mild and long hypoxia. This suggests that Apitolisib is considered as a potential treatment option for HCC, especially for rapidly growing HCC in low and chronic hypoxia.

Conclusions
In summary, this study highlights the role of HIF-2α but not HIF-1α in activating and binding c-MYC in HCC cell proliferation during mild and chronic hypoxia response. PI3K/mTORC2/ HIF-2α/c-MYC axis plays a key role in this process (Figure 7). PI3K inhibitor Apitolisib is suggested as a potential treatment option for HCC, especially for rapidly growing HCC in mild and chronic hypoxia.

AVAILABILITY OF DATA AND MATERIALS
The data used and analyzed during the current study are available from the corresponding author on reasonable request.

ETHICS APPROVAL AND CONSENT TO PARTICIPATE
All of our experiments involving human participants were approved by The Ethics Committee of Tianjin Medical University and performed in accordance with the Declaration of Helsinki. We obtained human HCC tissue and adjacent normal tissue from HCC patients at the Tianjin Medical University Cancer Institute and Hospital with all patient's informed consent.