Hyperthermia combined chemotherapy regulates energy metabolism of cancer cells under hypoxic microenvironment

Objective: Oral squamous cell carcinoma (OSCC) represents one of the main types of head and neck malignant tumors with high incidence and mortality as well extremely poor prognosis. Hyperthermia (HT) shows great promises for tumor therapy. However it can promote autophagy in tumor microenvironment, which is found to serve as a surviving mechanism for cancer cells. Inhibiting autophagy has been considered as an adjuvant anti-cancer strategy. The present study investigated the role of HT-induced autophagy, while attempting to combine chemotherapy and autophagy blocking with HT in OSCC cells under hypoxia and starvation microenvironment. Materials and methods: HIF-1α and Beclin-1 expression in tissues was determined by immunohistochemistry in 80 OSCC sample pairs. The IC50 of CoCl 2 , YC-1 (an inhibitor of HIF-1α) and 3-MA (an inhibitor of autophagy) was detected by CCK-8. CoCl 2 and complete culture medium without serum were used to achieve the hypoxic and nutrient decient microenvironment, respectively. HT was performed by heating in a 42 ℃ water bath. The role of HT and YC-1,3-MA on autophagy in vitro were assessed by qRT-PCR and Western blot, and the secretion of high mobility group box1 (HMGB1) was determined by ELISA. The migration and apoptosis rates of cells were assessed by wound healing assay and ow cytometry. Results: We observed that HIF-1α and Beclin1 were highly expressed in OSCC tissues, which were correlated with more advanced malignancy features. CoCl 2 could establish hypoxia microenvironment, induce HIF-1α expression with dose-dependence as well as promote cell migration in Cal-27 and SCC-15 cells. Notably, hyperthermia and hypoxia could activate the HIF-1α/BNIP3/Beclin1 signaling pathway and promote HMGB1 secretion, which triggered cytoprotective autophagy to counteract the hypoxia and starvation cellular stresses, as indicated by downregulation of p62 and light chain 3-II (LC3 II). Furthermore, we found that hyperthermia combined YC-1 and/or 3-MA suppressed autophagy and cell migration whereas facilitated cell apoptosis. Conclusion: The present study demonstrated that combined use of YC-1 and 3-MA might increase death of tumor cells in physiological and hyperthermia conditions, which could be relevant with the inhibition of autophagy in OSCC tumor cells under hypoxia microenvironment in vitro.


Introduction
Oral squamous cell carcinoma (OSCC) is the 10th most common human malignancy worldwide and accounts for over 95% of malignant tumors in the head and neck, with high mortality [1,2]. Despite advances in diagnosis and treatment, the 5-year survival rate of OSCC is less than 50%, and the recurrence rate is about 65%, resulting in its poor prognosis and tendency of lymph node metastasis [3]. Hypoxia is a crucial microenvironment condition for solid tumor pathophysiology and tumor metastasis, where cancer cells proliferate rapidly and form large solid tumor masses, leading to obstruction and compression of the blood vessels surrounding these masses. These abnormal blood vessels do not function su ciently and generate poor O 2 supply to the central tumor regions, and the levels of oxygenation within the same tumor are highly variable from one area to another and can change over time [4,5]. As an adaptive response to hypoxic stress, hypoxic tumor cells activate several survival pathways to carry out their essential biological processes compared with normal cells. The hypoxia inducible factor-1α (HIF-1α) is one crucial facilitator for energy adaption and oxygen metabolic stress in hypoxia and nutrition de ciency tumor environment, and functions as a general regulator of tumor aggressiveness and metastasis as well [4,6,7]. Hyperthermia (HT) has been successfully used in the clinical treatment of many cancers including the head and neck cancer [8] for decades, which sensitizes tumor cells to radiation by inhibiting DNA repair and increasing the aggregation of damaged nuclear proteins [9,10]. Previous studies have shown that initiating triggers for death in heat-shocked cells include: induction of physiological cascades; thermal protein unfolding and aggregation; necrosis that occurred at extremely elevated temperatures [14,15]. Furthermore, hyperthermia-related elevated blood ow and vascular permeability in the heated tumor region also promote higher intratumor and peritumor drug concentrations to improve the e cacy of chemotherapy [10][11][12]. Therefore, hyperthermia acts as a complement of radiotherapy, chemotherapy and molecular targeted oncotherapy. However, tumor cells possess homeostatic responses to reduce heat-shock induced cell death, which involve cell cycle arrest and transient induction of the transcription of genes encoding molecular chaperones and heat shock proteins (HSPs) [13]. In short, hyperthermia induces the expression of HSPs and inhibit DNA damaged repair, whereas DNA damage, hyperthermia and HPSs evokes autophagy, which was associated with facilitated cell survival and decreased programmed cell death [13][14][15][16].
Autophagy is a cellular pathway which present only in eukaryotic cells to degrade the aged and damaged organelles, and misfolded proteins. It functions as a recycling program to provide biofuel to cells from degraded macromolecules to maintain su cient ATP production for survival [17], and is a key component in maintaining homeostasis of cellular environment. Depending on the exact cell type and conditions, it either acts as a protagonist or an antagonist of apoptosis [18]. Autophagy also plays a crucial role in cancer pathophysiology. It is believed to prevent cancer development, but can also protect cancer cells within an already established tumor from shortage of nutrients and hypoxic conditions [18]. Furthermore, cellular damage caused by heating can be repaired and reversed by autophagy, resulting in incomplete cell necrosis [16,19]. Researches have shown that high mobility group box-1 protein (HMGB1) regulates autophagy [20]. HMGB1 is a late in ammatory mediator associated with sepsis, malignancy, and immune disease [20], which is passively released by necrotic tissues or actively secreted by stressed cells [21]. Intracellularly, HMGB1 is involved in DNA repair, transcription and recombination as well in the regulation of apoptosis/autophagy balance. Once secreted, it participates in a variety of processes such as in ammation, proliferation, differentiation, migration, invasion and tissue regeneration [22]. Therefore, HSPs and autophagy are two controllers of cellular proteostasis. Under stressful cellular conditions, these two mechanisms are likely to complement each other [14]. However, whether hyperthermia-induced autophagy facilitates cell survival or accelerates cell death during the development of OSCC is still controversial. Hence, the present study aimed to investigate the relationship between hyperthermia and autophagy in hypoxia and nutritionde cient tumor microenvironment of human OSCC. Meanwhile, the underlying mechanism was examined. This study might provide a novel promising therapeutic regiment for human OSCC.

Material And Methods
Human OSCC clinical samples The study was approved by the Ethics Committee of Yantai Yuhuangding Hospital and written informed consent was provided by all patients. OSCC and adjacent normal tissues were obtained from 80 patients with primary OSCC, including 56 men and 24 women, aged 37-86 years, who underwent surgical resection the tumor at the Yantai Yuhuangding Hospital between August 2015 and April 2017. None of the patients had received any chemotherapy or radiotherapy before excision. All samples were con rmed by pathological examination. The histological grade and tumor stage were assigned according to the World Health Organization (WHO) [23] and the International Union against Cancer classi cation system [24].
The expression levels of HIF-1α and Beclin1 were analyzed by IHC. Brie y, antigen retrieval was performed by incubating the sections in 10 mM citric acid buffer (pH 6.0) at 100˚C for 15 min. Subsequently, sections were dewaxed in xylene at room temperature and rehydrated in a descending ethanol series (absolute ethanol for 5 min, 95% ethanol for 5 min, 90% ethanol for 5 min and 80% ethanol for 5 min). Following three washes with PBS-Tween (0.05% Tween-20 in PBS), the sections were blocked by 5% BSA (Sangon Biotech Co., Ltd., China) in TBS for 45 min at room temperature. The sections were subsequently incubated at 4˚C overnight with rabbit anti-HIF-1α or anti-Beclin1 polyclonal antibody diluted in 3% BSA/TBS solution (1:100). The slides were washed with PBST and incubated with a HRP-conjugated goat anti-rabbit IgG secondary antibody (1:1000) at room temperature for 45 min. The slides were subsequently stained with 3,3'-diaminobenzidine tetrahydrochloride at room temperature for 10 min, and then counterstained with 0.5% Harris' hematoxylin at room temperature for 5 min. Finally, the sections were dehydrated with ethanol (80% ethanol for 5 min, 90% ethanol for 5 min, 95% ethanol for 5 min and absolute ethanol for 5 min), dried and mounted with neutral balsam. The images were screened using a microscope (magni cation, x100 and x400). For scoring staining intensity, the expression level of HIF-1α and Beclin1 in TSCC tissues were evaluated using a numerical scale (-, negative; +, weak; ++, moderate; +++, strong; Fig. 1).
Cell cytotoxicity and cell viability assay CCK-8 assay (Sangon, Shanghai, China) was used to detect the cytotoxic effect of different drug treatment on cancer cells and the 50% inhibitory concentration (IC50) of the drug was calculated. OSCC cells were planted into 96-well plates a density of 5 × 10 3 cells/well, supplied with 100 µL complete growth medium. After 24 h incubation, cells were exposed to CoCl 2 , YC-1, and 3-MA at the concentrations of 25 µM to 200 µM, 10 µM to 100 µM, 0.5 µM to 150 µM, respectively. Untreated cells were used as control. At each time point, the cells were washed and incubated with 100 ul RPMI 1640 plus 10 µl CCK-8 solution at 37˚C for 3 h. Subsequently, the absorbance was measured at 450 nm with a microplate reader (BioTek Instruments, Inc.,USA). Each experiment was performed at least in triplicate. Dose response curves were established to determine the IC50 values for CoCl 2 , YC-1, and 3-MA in the two OSCC cell lines.

Establishment of hypoxic environment and heat treatment
Hypoxic environment was established by exposing cells to serum-free medium with IC50 of CoCl 2 . Hyperthermia treatment was performed by partially submerging cell culture ask in a thermostatically controlled circulating water bath (Shanghai Yiheng Scientic Instrument Co, LTD, China). Cells were treated at 42 ± 0.1 ℃ for 60 min and cooldown to 37 ℃ in less than 5 min.

RNA extraction and qRT-PCR
Total RNA from cells was extracted with Trizol reagent (Sangon, Shanghai, China). cDNA was synthesized by using PrimeScript™ RT Master Mix Kit (Takara, Japan). qRT-PCR was performed on a StepOne™ Real-Time PCR System (Applied Biosystems, USA) with a SYBR Premix Ex Taq Kit (TaKaRa, Japan). According to the manufacturer's instructions, the PCRs were conducted at 95 °C for 30 s, followed by 40 cycles of 95 °C for 3 s, and 60 °C for 30 s. All reactions were performed in triplicate. 2 −ΔΔCT method was applied to calculate the relative fold change of gene expression. All results were normalized to GAPDH. Primer sequences (Sangon, Shanghai, China) were listed below: HIF-1α: up: Cell migration assay Cells were seeded in 6-well plates and cultivated until 100% con uence. In the serum-free RPIM1640 medium, 100 µM CoCl 2 was added to simulate a hypoxia environment and cells were treated separately with 50 µM 3-MA and/or 25 µM YC-1 combined with heat treatment for 1 hour in a 42 °C water bath. Then the cells were scraped with a 200 µl of pipette tip and washed with PBS for three times. At 0 and 24 h after incubation in serum-free medium, the images of wound healing were captured using an inverted microscope (magni cation, × 400). The area of each wound was quanti ed using Image-Pro Plus 6.0 software (Media Cybernetics, Inc.). The cell migration rate (%)was calculated as follows: [(Area of wound at 0 h -Area of wound at 24 h)/ Area of wound at 0 h] × 100%.
Flow cytometry analysis of apoptosis Cell apoptosis were detected by ow cytometry using the FITC-AnnexinV/PI Apoptosis Assay Kit (BD, USA) following manufacturer's instructions. Brie y, the cells (5 × 10 5 cells/well) that treated with chemotherapy (YC-1 and 3-MA) and hyperthermia (42 °C heat treatment) were harvested and centrifuged at 500 × g for 5 min at room temperature, then washed twice with PBS and resuspended in 500 µl binding buffer solution at a density of 1 × 10 5 cells/ml. The cells were subsequently stained with 5 µl FITC Annexia V and 5 µl PI using at room temperature for 15 min in the darkness. Apoptotic cells were analyzed using a CytoFLEX ow cytometer and CytExpert software (version 2.0; Beckman Coulter, Inc.) within 1 h.

Enzyme-linked immunosorbent assay (ELISA)
The levels of HMGB1 in chemotherapy (YC-1 and 3-MA) and hyperthermia (42 °C heat treatment)-treated Cal-27 and SCC-15 cells that were in hypoxia and normoxia condition were determined using ELISA kits (Ilerite Biotechnology Co, China), in line with the manufacture's protocol.

Statistical analysis
Statistical analysis was performed using the GraphPad Prism version 5.0 (GraphPad Software, La Jolla, CA). Data were presented as the mean ± standard deviation, comparisons between groups were performed by using Student's ttest or a one-way ANOVA, followed by a Tukey's post hoc test for multiple comparisons. A χ 2 test was used to determine the association between the expression levels of HIF-1α, Beclin1 respectively and OSCC clinical and histopathological features. P < 0.05 was considered statistically signi cant difference.

HIF-1α and Beclin1 expression levels increased in OSCC tissues
The clinical signi cance of HIF-1α and Beclin1 in OSCC patients was evaluated by analyzing the expression level of HIF-1α and Beclin1 in 80 OSCC tissues by scoring of IHC (Fig. 1). HIF-1α mainly located in the cell nucleus while Beclin1 mainly located in the cell membrane and cytoplasm with a small amount in the cell nucleus. In this study, we observed that HIF-1α and Beclin1 were expressed both in cancer tissue and adjacent normal tissues. By multiplying the score of staining intensity and percentage of positive cells of each tissues section, we found that HIF-1α and Beclin1 were both highly expressed in OSCC tissues compared with normal tissues (P<0.05, Table 1). In addition, the high expression level of HIF-1α and Beclin1 was associated with poor cell differentiation, lymph node metastasis, advanced pathological TNM stage, and large tumor size (P<0.05, Table 2), but was not correlated with gender or age.

Establishment of hypoxic microenvironment
Low oxygen-induced hypoxia is the optimal hypoxia model. However, induction of chemical hypoxic conditions using

Hyperthermia induced autophagy of OSCC cells in hypoxic microenvironment
To clarify the relationship between hypoxia, hyperthermia and autophagy, we tested proteins related to autophagy signaling pathway by Western blot and qRT-PCR in Cal-27 and SCC-15 cells, and found that HIF-1α and BNIP3 were hardly expressed in the untreated control group and "HT" group (hyperthermia treatment alone), and there were no statistical difference in the expression of HIF-1α and BNIP3 in HT group compared the control group (P>0.05) (Fig.  4A, B and C). This indicated that HIF-1α and BNIP3 were not expressed in abundance under normal oxygen condition, and hyperthermia could not induce the expression of HIF-1α and BNIP3 in normoxia condition.
The expression of autophagy related genes such as Beclin1 and LC3II were higher in "HT" group compared the control group (P<0.05) (Fig. 4A, B and C), showing that Beclin1 and LC3IIwere expressed under constant oxygen condition and hyperthermia promoted Beclin1 and LC3IIexpression. In addition, the expression of P62 protein sharply decreased in "HT" group compared the control group (P<0.05) (Fig. 4A, B and C). Taken together, the results showed that hyperthermia could induce autophagy under normal oxygen condition.
However, the expression level of HIF-1α, BNIP3, Beclin1, LC3II signi cantly increased in the "Hy" group (hypoxia treatment alone) and "HT+Hy" group (combined hypoxia and hyperthermia treatment) compared the control group and "HT" group (P<0.05). In contrary, the expression of P62 sharply decreased in the "Hy" group and "HT+Hy" group compared to the control group and "HT" group (P<0.05) (Fig. 4A, B and C), What's more, we found that compared with the "Hy" group, the expression of HIF-1α, BNIP3, Beclin1, LC3II were signi cantly higher, whereas that of p62 proteins were remarkably lower in the "HT+Hy" group (P<0.05) (Fig. 4A, B and C).
These data indicated hyperthermia induced autophagy through activating HIF-1α/BNIP3/Beclin1 signaling pathway. 3-MA had a stronger effect in inhibiting autophagy than YC-1, and combined use of 3-MA and YC-1 provided further inhibition of autophagy in OSCC cells. Therefore, we concluded that hyperthermia might not only induce autophagy through activating HIF-1α/BNIP3/Beclin1 signaling pathway, but also involve other pathways in hypoxia and starvation microenvironment. Moreover, hyperthermia combined with chemotherapy inhibits autophagy.

Hyperthermia combined chemotherapy inhibited the secretion of HMGB1
HMGB1 was translocated from the nucleus to the cytoplasm and secreted or passively released through the permeabilized plasma membrane of succumbing/dead cells. We measured the extracellular HMGB1 protein by ELISA , and found that compared with the control group, the secretion of HMGB1 signi cantly increased with hypoxia or hyperthermia treatment alone and combined in Cal-27 and SCC-15 cells, with protein level in HT+Hy group > HT group > Hy group > control group (P<0.05, Fig. 6), suggesting that hyperthermia and hypoxia promoted HMGB1 secretion under normal oxygen conditions, and that hyperthermia was stronger than hypoxia in promoting HMGB1 secretion. On the contrary, addition of 3-MA and YC-1 signi cantly reduced the secretion of HMGB1, with protein level in HT+Hy+YC-1+3-MA group < HT+Hy+3-MA group < HT+Hy+YC-1 group < HT+Hy group (P<0.05, Fig. 6). Based on the above experimental data, we reached a conclusion that hyperthermia and hypoxia might facilitate HMGB1 secretion in starvation tumor microenvironment, and the use of chemotherapy drugs to could also inhibit the secretion of HMGB1 in addition to autophagy inhibition.

Inhibition of autophagy reduced tumor cell migration in hypoxia microenvironment
In order to determine the effect of autophagy and hypoxia on Cal-27 and SCC-15 cell migration, we performed a wound healing assay in Cal-27 and SCC-15 cells and the results showed compared with untreated control cells, cell mobility increased in "Hy" group and decreased in "HT" group (P<0.05, Fig. 7A and B), which demonstrated hypoxia might promote cell migration while hyperthermia inhibit cell migration. In addition, "HT+Hy" group had worse migratory ability than the "Hy" group had (P<0.05) (P<0.05, Fig. 7A and B), which demonstrated that hyperthermia could inhibit cell migration whether in normoxia or hypoxia conditions. Compared with "HT+Hy" group, the cell migration of "HT+Hy+3-MA" group, "HT+Hy+YC-1" group and "HT+Hy+YC-1+3-MA" group were all signi cantly reduced (P<0.05, Fig. 7A and B) and "HT+Hy+YC-1+3-MA" group had the strongest inhibition of cell migration, which indicated that inhibiting autophagy might reduce tumor cell migration in hypoxia microenvironment. However, there was no signi cant difference in migration between "HT+Hy+3-MA" group and "HT+Hy+YC-1" group (P<0.05, Fig. 7A and B)

Inhibition of autophagy enhanced hyperthermia-induced apoptosis in OSCC cells
Apoptosis was investigated using ow cytometry analysis. In terms of early cell apoptosis, compared "Con" group, the cell apoptosis rate of Cal-27 and SCC-15 were increased in "HT" group and "HT+Hy" group while the group of "HT" was signi cantly higher than the group of "HT+Hy" (P<0.05, Fig. 8A, B and Ca-b), and take "HT+Hy" group as a contrast, early cell apoptosis were signi cantly enhanced in group of "HT+Hy+YC-1", "HT+Hy+3-MA" and "HT+Hy+YC-1+3-MA" (P<0.05, Fig. 8A, B and Ca-b), with there was no signi cant change in "Hy" group compared "Con" group and in "HT+Hy+3-MA" group vs "HT+Hy+YC-1" group (P>0.05, Fig. 8A, B and Ca-b). Moreover, when talking about late cell apoptosis, compared "Con" group, the cell apoptosis rate of Cal-27 and SCC-15 were decreased in "HT" group and "Hy" group (P<0.05, Fig. 8A, B and Cc-d) as well there was no statistical change in "HT+Hy" group (P>0.05, Fig. 8A, B and Cc-d); However, the apoptosis rate of "HT+Hy" group was enhanced that compared with "Hy" group in two cell lines, and compared with "HT" group, which promote cell late apoptosis in Cal-27 cells. besides, compared with "HT+Hy" group, "HT+Hy+YC-1" group and "HT+Hy+3-MA" , "HT+Hy+YC-1+3-MA" group obviously promoted late cell apoptosis (P>0.05, Fig. 8A, B and Cc-d). In conclusion, We found that compared with the control group, the cell apoptosis rate increased in "HT" group and decreased in "Hy" group (P<0.05, Fig. 8A, B and Ce-f). Meanwhile, the cell apoptosis rate of "HT+Hy" group was higher than "Hy" group (P<0.05, Fig. 8A,B and Ce-f), which supported the conclusion that hyperthermia signi cantly promoted tumor cell apoptosis at normal oxygen concentration and at hypoxia condition, while hypoxia inhibited cell apoptosis in nutrition de ciency microenvironment. In addition, YC-1 and 3-MA both facilitated cell apoptosis (P<0.05, Fig. 8A,B and Ce-f) in approximately equivalent scale (P>0.05) and combined use of both drugs had the strongest effect on cell apoptosis (P<0.05, Fig. 8A,B and Ce-f). The results revealed that hyperthermia combined with chemotherapy promoted OSCC cell apoptosis under hypoxia and nutrition de ciency microenvironment in vitro. In general, the results from the two cell lines (Cal-27 and SCC-15) showed a similar tendency.

Discussion
OSCC is the most common solid tumor in head and neck, and a growing body of evidence indicates that the microenvironment of solid tumor with hypoxia, low PH, poor nutrition, poor perfusion and abnormally high interstitial uid pressure change the biological behavior of tumor, which signi cantly reduces the effect of radiotherapy and chemotherapy and promotes tumor cell proliferation, invasion, and migration [25,26].The adaptation and survival of tumor cells in heterogeneous microenvironment requires the coordination of complex pathways and mechanisms, such as hypoxia induction factor 1 (HIF-1), unfolded protein reaction (UPR), rapamycin (mTOR) and autophagy [27]. Hypoxia is a crucial microenvironment condition for solid tumor pathophysiology, including tumor proliferation, invasion, and metastasis, and HIF-1α is a key molecule that is highly expressed under hypoxia. In general, under the condition of normoxia, the biosynthesis of mitochondrial respiration and anabolism is facilitated by oncoprotein MYC, allowing cancer cells to proliferate under conditions of adequate oxygen and nutrition [28,29]. But under hypoxia, energy metabolism is regulated by HIF-1 rather than MYC, HIF-1 is a powerful mediator of carbohydrate reprogramming from oxidative phosphorylation to glycolytic metabolism in hypoxic reactions by regulating oxygen transport (angiogenesis) and oxygen consumption (glycolysis metabolism) [30]. And contrary to MYC, HIF-1 strongly inhibits mitochondrial respiration and biogenesis [31]. Metabolic pathway changes of tumor cells during hypoxia or malnutrition are considered to be a feature of cancer cells, namely metabolic reprogramming [32]. In addition, when tumor cells are exposed to stressful microenvironment, especially low PH, low oxygen and nutrient de ciency, autophagy is activated to cycle cellular metabolic substrate to meet their high metabolism and energy demand, and to suppress the body's in ammatory response induced by tumor to prevent the cytotoxicity accumulation and to promote tumor cell survival. Therefore, autophagy constitutes a way to prolong the survival of tumor cells [25,26,33,34].
Ribeiro et al. had investigated 93 OSCC samples for HIF-1α expression, demonstrating that metastatic lymph nodes and intratumoral regions of corresponding primary tumors expressed HIF-1α at a high frequency [35]. Therefore, HIF-1α is considered to be a potential prognostic marker of many cancers, including OSCC [36].The formation of HIF-1α is oxygen-dependent. At the tumor margin, blood vessels grow to provide su cient nutrients and oxygen, allowing the synthesis and rapid ubiquitin-mediated degradation within 10 min of HIF-1α under normoxia. In contrast, HIF-1α protein activity is prolonged under hypoxia [28]. Therefore, one important consequence of hypoxia is the induction of HIF-1α, which activates a series of downstream genes that facilitate tumor cell survive in hypoxia microenvironment, such as autophagy-related genes Beclin1 [37]. In this study, we detected the expression level of HIF-1α and Beclin1 in 80 pairs of OSCC tissues and adjacent normal tissue, and our results suggested that HIF-1α and Beclin1 were both highly expressed in OSCC tissues compared with normal tissues and were signi cantly associated with large tumor size, advanced TNM grade, high pathological grade and lymphocytic in ltration. This nding is consistent with previous research results.
Literatures have reported that autophagy is a self-degrading process and plays an indispensable role in sustaining cellular homeostasis under stress, which act as a housekeeper to clear damaged organelles such as fragmentized mitochondria, to remove misfolded proteins and to recycle cellular components [38,39]. The role of autophagy in cancer is most dramatic and dynamic. In normal cells, autophagy inhibits tumor occurrence, however, in hypoxia and nutrient de ciency tumor microenvironment, autophagy promotes tumor cell survival as an alternative energy supply pathway besides the "Warburg effect". Autophagy initiation occurs under stress conditions, such as nutrient and energy de ciency, hypoxia, reactive oxygen species (ROS), protein aggregation and production of damaged organelles [40,41]. As is well known, the conversion of cytosolic LC3-I to LC3-II, which binds autophagic vacuoles, is the primary hallmark of autophagy. In addition, the protein p62 (also known as SQSTM1) has been reported to interact with the autophagic effector protein LC3 and to be degraded through an autophagy-lysosome pathway [42]. Therefore, activation of autophagic ux leads to decrease in p62 and LC3-I levels and an increase in LC3-II level. Our research showed that OSCC cells Cal-27 and SCC-15 also underwent autophagy under hypoxia condition. In physiological conditions, Beclin-1 and Bcl-2 form a complex compound that results in inhibiting the activation of the autophagy pathway. It has been con rmed that BNIP3 is the target molecule of HIF-1α. Under hypoxia conditions, the expression of HIF-1α signi cantly augments, which upregulates BNIP3, and BNIP3 interacts with Bcl-2 or Bcl-XL and ultimately forms heterodimer, which will prevent the binding of Bcl-2 to Beclin-1, thereby the released Beclin-1 will activate the autophagy pathway [43]. As a result, HIF-1α/BNIP3/Beclin-1 signaling pathway is an important way of inducing autophagy under hypoxic conditions.
Previous researches have demonstrated that mild hyperthermia could directly damage proteins and organelles and thus trigger cytoprotective autophagy to tolerate the cellular stresses and prolong the survival of cancer cells [44][45][46]. Literatures reported that heat stress induced autophagy in several types of cancer cells, such as hepatocellular carcinoma cells (SMMC7721 and Huh7), cervical cancer cell (HeLa) and lung cancer cell (A549 cell) [28,42]. Our experimental data also indicated that hyperthermia could induce autophagy in both hypoxia and normoxic starvation microenvironments, and that autophagy was further enhanced in hypoxia condition. This might stem from the following reasons: HT-induced protein denaturation and aggregation results in the up regulation of HSPs, which are reported to up-regulate the autophagy mediator Beclin-1 [14]. Moreover, Hyperthermia can induce oxidative stress in cells and can further augment the generation of ROS [47]. ROS is a known inducer of autophagy and apoptosis. It has been reported that ROS acted on the complex formed by Beclin-1 and anti-apoptotic Bcl-2 homologs such as Bcl-2 and Bcl-xL. Moreover, this complex repressed the pro-autophagic activity of Beclin-1 and ROS could induce the dissociation of autophagy molecules Beclin 1 and Bcl-2, thus activating the Beclin1-induced autophagy pathway, increasing the expression of LC3-II, thereby initiating autophagy-associated pathways [48,49]. In addition, ROS is also reported to upregulate the activity of HIF1-α, and some scholars have also shown that the expression and activity of HIF-1a is not only induced in response to limited oxygen supply, but it is also regulated through related signaling pathways, including the extracellular signal-regulated kinase (ERK) [50] and the protein kinase B (AKT) [51] pathway, and hyperthermia promoted the expression of HIF-1α by activating phosphatidylinositol 3-kinase(PI3K) PI3K/AKT and mitogen-activated protein kinase (MAPK) MAPK/ERK signal pathway [52] while relative study have demonstrated that heat treatment can cause hypoxia in the local tissue and increase HIF-1a expression levels, which can induce the over-proliferation of any residual tumors [51]. In conclusion, hyperthermia and hypoxia induced the activation of HIF-1α, which furthermore facilitates the activation of HIF1-α/BNIP3/Beclin1 autophagy signaling pathway. In our study, the expressions of Beclin1 and LC3-II signi cantly increased when cells were treated with hypoxia alone and increased even futher when exposed to hyperthermia, which supported the above mechanism.
Namely, HIF1-α/BNIP3/Beclin1 autophagy signaling pathway was activated under hypoxia conditions and hyperthermia further enhanced the activation of the pathway by secreting ROS. The endothelial cells of the tumor microvessels proliferate and are more sensitive to heat than normal cells. Therefore, the temperature change caused by hyperthermia in the tumor area leads to apoptosis and necrosis of tumor cells. Hyperthermia can enhance the expression of apoptotic genes, such as p53, thereby impeding the cell cycle, inhibiting tumor cell proliferation, and leading to tumor cell apoptosis [53]. And our research results demonstrated that cell migration was inhibited and cell apoptosis rate was signi cantly augmented when untreated or hypoxia-treated cells were exposed to hyperthermia. Existing literatures show that HMGB1 release occurs passively as cell permeability breaks down upon necrosis [54] and late stage of apoptosis [55]. And in our experimental results, the secretion of HMGB1 was upregulated by hyperthermia. Nevertheless, during tumor development and cancer therapy, HMGB1 has been reported to play paradoxical roles in promoting both cell survival and death by regulating multiple signaling pathways. It has been demonstrated HMGB1 increases pro-survival autophagy in a Beclin1-dependent way during chemotherapy [56].
Moreover, it has been demonstrated ATG5-mediated autophagy pathway promoted the secretion of HMGB1 in starvation and lipopolysaccharide (LPS) treatment, and ROS signaling was required in this process [57]. What's more, several of the secondary messengers, such as cytosolic free calcium and ROS can regulate HMGB1 secretion [56].
We also demonstrated that both hyperthermia and hypoxia facilitated the secretion of HMGB1. Furthermore, a previous research has con rmed that secreted HMGB1 activated receptors for advanced glycation end products and Toll-like receptor-4 and induced autophagy in skeletal muscle [58].
Recently, more and more researchers have paid their attention to the manipulation of autophagy to enhance the e cacy of cancer therapy. YC-1 is a guanylate cyclase-activator and inhibitor of HIF-1α [59] while 3-MA is a key drug in studying autophagy, which can block autophagy [60]. Therefore we used YC-1 and 3-MA to examine the effect on autophagy. In the present study, we found the use of YC-1 signi cantly downregulated the expression of HIF-1α induced by CoCl 2 , although the exact mechanism is uncertain. We observed that inhibition of HIF-1α signi cantly suppressed BNIP3 expression and the administration of YC-1 and 3-MA alone or in combination signi cantly downregulated the expression of autophagy related genes LC3II, Beclin1and HMGB1, and increased the expression level of P62 when exposed to mild hyperthermia, hypoxia and nutrition de ciency microenvironment. Furthermore, we found that YC-1 and 3-MA treatment suppressed cell migration and increased cell apoptosis in Cal-27 and SCC-15 cell lines, which suggested that hyperthermia and hypoxia induced a protective effect of autophagy by activating HIF-1α/BNIP3/Beclin1 pathway and by stimulating the secretion of HMGB1, and autophagy act as a survival mechanism to alleviate hyperthermia and hypoxia injury. But even more importantly, this result could be reversed by the use of autophagy inhibitor and by blocking HIF-1α. In view of clinical application, when thermochemotherapy was used in combination, HT-related elevated blood perfusion also supported higher intra-and peritumoral drug concentrations, changed the tumor microenvironment to improve the e cacy of chemotherapy, while chemotherapy could effectively inhibit HT-induced autophagy to increase cell apoptosis. Autophagy and apoptosis are often inseparable and highly interactional. Research found that speci cally blocking autophagy enabled ROS to increase signi cantly in malignant tumor cells. Thus, autophagy can aggravate the apoptosis of tumor cells [61]. Previous study shows that both apoptosis and autophagy are activated in response to metabolic stress [62], and accumulating evidence reveals that autophagy and apoptosis can cooperate, antagonize or assist each other, thus in uencing the differential the fate of the cell [63].

Conclusion
Although HT is considered to be a promising cancer treatment regimen, cellular damage caused by heating could be repaired and reversed by the production of HSPs and autophagy process in hypoxia and starving environment, resulting in incomplete cell necrosis and attenuating the effects of HT therapy. In this study, we demonstrated that exposure to hypoxia and hyperthermia could induce autophagy in the OSCC cells Cal-27 and SCC-15. This process could be reversed by the use of autophagy inhibitor and by blocking HIF-1α. In summary, our ndings might bene t further understanding of the biological effects of thermo-chemo-therapy on cancer cells, and we believed that inhibition of autophagy might be a useful and promising therapeutic strategy to enhance the therapeutic effect of HT in hypoxia and nutrient de ciency tumor environment. In addition, further research on animal model is required in the future. Declarations