TRPM7 silencing modulates glucose metabolic reprogramming to inhibit the growth of ovarian cancer by enhancing AMPK activation to promote HIF-1α degradation

Background Tumor cell metabolic reprogramming is crucial for the malignant behavior of cancer cells by promoting their proliferation. However, little is known on how transient receptor potential 7 (TRPM7) modulates metabolic reprogramming in ovarian cancer. Methods The effects of TRPM7 silencing on transcriptome profile, glucose uptake, lactic acid production, extracellular acidification rate (ECAR), oxygen consumption rate (OCR), intracellular ROS and ATP levels, and NAD+/NADH ratios in ovarian cancer cells were examined. The impacts of TRPM7 silencing on the levels of glycolysis-related HK2, PDK1 and oxidative phosphorylation (OXPHOS)-related IDH3B and UQCRC1, HIF-1α expression and AMPK phosphorylation were determined in ovarian cancer. The effect of AMPK activity on HIF-1α ubiquitination degradation was investigated in ovarian cancer cells. Results Compared with the control, TRPM7 silencing suppressed the proliferation of ovarian cancer cells by shifting preferable glycolysis to OXPHOS. In parallel, TRPM7 silencing decreased the glucose uptake of tumor-bearing mice and TRPM7 levels were negatively correlated with IDH3B and UQCRC1, but positively with HK2 and PDK1 expression in ovarian cancer tissues. Mechanistically, TRPM7 silencing significantly increased AMPK phosphorylation and decreased HIF-1α protein levels in ovarian cancer, particularly in HIF-1α silencing cells. The shifting from glycolysis to OXPHOS by TRPM7 silencing was abrogated by HIF-1α over-expression and impaired by inhibiting AMPK activity in ovarian cancer cells. Moreover, enhanced AMPK activation inhibited glycolysis, which was abrogated by HIF-1α over-expression in ovarian cancer cells. Moreover, the enhanced AMPK activation promoted HIF-1α ubiquitination degradation. Conclusions TRPM7 silencing enhanced AMPK activation to shift glycolysis to oxidative phosphorylation by promoting HIF-1α ubiquitination degradation in ovarian cancer. Hence, TRPM7 may be a therapeutic target for intervention of ovarian cancer. Supplementary Information The online version contains supplementary material available at 10.1186/s13046-022-02252-1.


Introduction
Ovarian cancer is a common gynecological malignancy in women. Its prevalence and incidence remain high, affecting many women in the world. Currently, the standard therapies for ovarian cancer include surgical resection and chemotherapy, but ovarian cancer patients still have a 5-year survival rate of < 50% due to the tumor recurrence and distant metastasis [1][2][3]. Hence, discovering the molecular pathogenesis, and novel biomarkers and targets is of high signi cance in prognosis and treatment of ovarian cancer.
It is well known that the majority of healthy cells prefer glucose metabolism by oxidative phosphorylation (OXPHOS) to produce high levels of ATP and reactive oxygen species (ROS), which can cause cell injury and promote apoptosis. In contrast, malignant tumor cells can reprogram their energy metabolism and escape from immune destruction [4,5]. Tumor cells can alter their glucose metabolism by preferring glycolysis even in a normoxic environment to support their malignant behaviors, proliferation, invasion and metastasis, called the "Warburg effect" [6][7][8]. However, it is still unclear how what factors regulate the Warburg effect and OXPHOS to support the growth of ovarian cancer.
The hypoxia inducible factor (HIF)-1α and adenosine 5'-monophosphate (AMP)-activated protein kinase (AMPK) signal pathways are the major regulators of glycolysis and OXPHOS, and they are crucial for the metabolic reprogramming in tumor cells [9,10]. AMPK is an evolutionarily conserved serine / threonine kinase. AMPK can be activated by many factors, such as AMP / ATP and ADP / ATP ratios [11,12] and intracellular calcium through CaMKK2-mediated phosphorylation [13][14][15]. HIF-1α is commonly expressed by mammalian cells, but it is rapidly degraded by the intracellular oxygen-dependent ubiquitin protease degradation pathway in a normoxygenic condition. HIF-1α expression is signi cantly up-regulated and HIF-1α protein stabilizes in a hypoxic condition. The AMPK and HIF pathways can also modulate the expression and function of several transcription factors, such as FoxO, NF-κB, NRF2, and p53, as well as protein kinases and other factors, including mTOR, ULK1, HDAC5, and SIRT1 [16]. However, how little is known on the AMPK and HIF pathways modulate the metabolic reprogramming in ovarian cancer.
Transient receptor potential melastatin 7 (TRPM7) is a unique cation channel protein, and functionally responsible for transportation of cations (such as calcium, magnesium, and others), thereby increasing [Ca2 +]i levels. In addition, TRPM7 has a kinase activity and can activate itself and other substrates, dependent on ATP. TRPM7 expression is up-regulated in a variety of tumors and can enhance malignant behaviors of cancer cells [17]. Our previous study has revealed that up-regulated TRPM7 expression is related to pelvic lymph node metastasis and poor prognosis of human ovarian cancer and inhibition of TRPM7 can suppress the invasion and metastasis of ovarian cancer cells through attenuating the Ca2 + -PI3K/AKT signaling [18]. Interestingly, inhibition of TRPM7 can promote HIF-1a degradation in prostate cancer cells [19]. Given that glycolysis inhibition is associated with HIF-1α degradation we hypothesize that TRPM7 activation to increase [Ca2 +]i may modulate the HIF-1α/AMPK signaling to regulate glucose metabolism and promote ovarian cancer cell proliferation.
This study investigated the impacts of TRPM7 silencing on ovarian cancer cell proliferation, apoptosis, glucose metabolism and the AMPK / HIF-1a signaling.
We found that TRPM7 silencing enhanced OXPHOX, inhibited glucose uptake, glycolysis, lactic acid production and suppressed the proliferation of ovarian cancer cells by enhancing AMPK activation and HIF-1a degradation. Together, our ndings indicate that TRPM7 is a novel therapeutic target and may shed new lights on the regulation of metabolic reprogramming and pathogenesis of ovarian cancer. Results 1. TRPM7 silencing attenuates the growth of ovarian cancer by shifting glucose metabolic reprogramming.
Given that up-regulated TRPM7 expression is related to poor prognosis of ovarian cancer [17], we determined the impact of TRPM7 expression on cell proliferation. First, we generated TRPM7 stably silencing SKOV3 and HO8910 cells ( Figure S1A). To understand how TRPM7 silencing regulated gene expression in ovarian cancer cells, we performed RNA-seq analyses between SKOV3 sh-Control and SKOV3 sh-TRPM7 cells. We found that with a fold change of ≥ 2.0 and P < 0.05, there were 2873 differentially expressed genes (DEGs) between SKOV3 sh-Control and SKOV3 sh-TRPM7 cells, and among the DEGs, 1222 were up-regulated and 1651 were down-regulated in SKOV3 sh-TRPM7 cells (Fig. 1A, Figure S1B and C), which were supported by the Gene Set Enrichment Analysis (GSEA). The GO and KEGG analyses indicated that the DEGs were involved in the process of cell cycling, positive regulation of mitotic cell cycling and acute in ammatory response in ovarian cancer (Fig. 1B, Table 1). To test the functional outcomes of TRPM7 silencing, we quanti ed the proliferation of different groups of ovarian cancer cells by the EdU assay. As shown in Fig. 1C, there were less numbers of EdU + cells in the TRPM7-silenced SKOV3 or HO8910 cells than their sh-control cells. Flow cytometry analysis exhibited that TRPM7 silencing signi cantly increased the frequency of apoptotic SKOV3 and HO8910 cells (P < 0.001, Fig. 1D). Furthermore, TRPM7 silencing signi cantly decreased the tumor sizes and weights in a mouse xenograft of SKOV3 tumors (P < 0.01, Fig. 1E). Given that rapid proliferation and metabolic reprogramming are hallmarks of cancer cells, we further performed the GO analyses and found that the DEGs were enriched in metabolic processes, such as cellular processes, DNA metabolism, macromolecule metabolism, primary metabolism, cellular metabolism, regulation of metabolism, and others (Fig. 1F). It is notable that many DEGs regulated glycolysis and OXPHOS, and they included the glycolysis-related PDK1, HK2, LDHB and OXPHOS-related IDH3B, UQCRC1, UQCRC2. TRPM7 silencing also down-regulated HIF-1a expression in SKOV3 cells (Fig. 1G). Further GSEA analyses revealed the potential signaling pathways, such as the AMPK signal pathway, were enriched in SKOV3 cells Table 2). Accordingly, we speculate that TRPM7 silencing may suppress the proliferation of ovarian cancer cells by shifting metabolic reprogramming.  Tumor cells usually prefer glycolysis to provide necessary energy and metabolites for their proliferation and switching to OXPHOS can promote high levels of ROS production and cell apoptosis [20]. To test whether TRPM7 silencing could shift the glucose metabolic reprogramming, we measured the levels of glucose consumption and lactic acid production in the sh-control and TRPM7 silencing ovarian cancer cells. In comparison with the sh-control cells, TRPM7 silencing signi cantly reduced the levels of glucose consumption by > 50% and lactic acid production by 35-40% in both SKOV3 and HO8910 cells (P < 0.05, P < 0.01, Figs. 2A and B). Seahorse assays exhibited that TRPM7 silencing also decreased ECRA, but increased OCR ( Fig. 2C and D). TRPM7 silencing also increased the ratios of NAD + /NADPH and the levels of ATP and ROS in both SKOV3 and HO8910 cells (P < 0.05, P < 0.01, Fig. 2E-G). IHC displayed that carvacrol treatment reduced the expression of TRPM7 in xenograft ovarian tumor tissues in mice (Fig. 2H). More importantly, PET-CT imaging revealed that carvacrol treatment not only reduced the tumor size, but also attenuated 18 F-FDG uptake in tumor-bearing mice by signi cantly decreasing mean standard uptake value and total volume (P < 0.01, Fig. 2I). Collectively, these results indicate that TRPM7 silencing shifts glycolysis to OXPHOS in ovarian cancer, contributing to its inhibition of ovarian cancer growth.
3. TRPM7 silencing attenuates the glycolysis process and enhances the OXPHOS process of ovarian cancer cells To understand how TRPM7 modulated metabolic reprogramming in ovarian cancer, we analyzed the expression of TRPM7, glycolysis-related HK2, PDK1, and OXPHOS-related IDH3B and UQCRC1 in 60 ovarian cancer tissues by IHC (Fig. 3A). The information of those patients is shown in Table S1. Correlation analysis indicated that TRPM7 expression levels in ovarian cancer tissues were positively correlated with HK2 (P < 0.01) and PDK1 (P < 0.01), but negatively correlated with IDH3B (p < 0.01) and UQCRC1 in this population (P < 0.05, Fig. 3B), supporting the notion that TRPM7 promotes glycolysis in ovarian cancer.
Consequently, TRPM7 silencing decreased the relative HK2 and PDK1 mRNA transcripts, but increased IDH3B and UQCRC1 mRNA transcripts in both SKOV3 and HO8910 cells (P < 0.05, P < 0.01, Fig. 3C). Similar patterns of the relative levels of HK2, PDK1, IDH3B and UQCRC1 proteins were detected in these cells by Western blot and immuno uorescence ( Fig. 3D and E, Figure S2D). PKM2 is a key enzyme for glycolysis and its nuclear translocation is associated with malignant behaviors of tumor cells [21]. To test the importance of PKM2, we characterized the intracellular distribution of PKM2 and found that TRPM7 silencing obviously decreased the levels of nuclear PKM2, but increased its cytoplasmic PKM2 in both SKOV3 and HO8910 cells (Fig. 3F, Figure S2F). Finally, decreased levels of HK2 and PDK1 proteins, but increased levels of IDH3B and UQCRC1 expression were observed in TRPM7-silenced SKOV3 tumors (Fig. 3G). Hence, TRPM7 silencing modulated the levels of glycolysis-and OXPHOS-related enzyme expression to shift glycolysis to OXPHOS in ovarian cancer.

Activation of the AMPK signal pathway inhibits glycolysis that is dependent on high levels of HIF-1α in ovarian cancer cells
It is well known that AMPK activation can promote OXPHOS and inhibit glycolysis by down-regulating the function of HIF-1α that can promote glycolysis in tumor cells under hypoxic conditions [22][23][24][25]. To understand how TRPM7 silencing modulates the HIF-1α and AMPK signaling, we characterized the levels of HIF-1α and AMPK expression and activation in the indicated ovarian cancer cells by Western blot and IHC. Clearly, TRPM7 silencing elevated the relative levels of AMPK phosphorylation, but minimized the HIF-1α expression in SKOV3 and HO8910 cells and SKOV3 tumors (Fig. 4A). Hence, TRPM7 silencing enhanced AMPK activation and reduced HIF-1α protein to switch preferable glycolysis to OXPHOS in ovarian cancer cells. To further understand the importance of AMPK activation in shifting glycolysis to OXPHOS, we tested the effect of altered AMPK activation on the HIF-1α-regulated glycolysis in ovarian cancer cells. We rst established stable HIF-1α silencing SKOV3 and HO8910 cells. Compared with the vehicle-treated control cells, treatment with Compound C (CC, an inhibitor of AMPK) signi cantly inhibited AMPK activation ( Figure S3A-B), but increased glucose uptake, lactate acid production and ECAR in both SKOV3 and HO8910 cells (Fig. 4B, C and F), which were signi cantly mitigated or abrogated in the HIF-1α silencing SKOV3 and HO8910 cells. In contrast, treatment with metformin (an activator of AMPK) not only increased AMPK phosphorylation ( Figure S3C-D), but also signi cantly decreased glucose uptake, lactic acid production and ECAR in both SKOV3 and HO8910 cells (Fig. 4D, E and G). However, the decreased glycolysis by metformin was abrogated by HIF-1α over-expression in ovarian cancer cells. These two lines of evidence demonstrated that AMPK activation inhibited glycolysis that was dependent on high levels of HIF-1α presence in ovarian cancer. 5. AMPK activation or HIF-1α over-expression mitigates the TRPM7 silencing-inhibited glycolysis in ovarian cancer cells.
To understand how HIF-1α and AMPK activation regulate the TRPM7 silencing-induced metabolic reprogramming in ovarian cancer cells, we rst generated HIF-1α over-expression in TRPM7 silencing ovarian cancer cells. Second, we tested the impact of TRPM7 silencing on the HIF-1α-mediated glycolysis in SKOV3 and HO8910 cells. Comparison with the control cells, TRPM7 silencing signi cantly decreased glucose uptake, lactic acid production and ECAR in SKOV3 and HO8910 cells, but did not signi cantly change glucose uptake, lactic acid production and ECAR in the HIF-1α over-expressing SKOV3 and HO8910 cells ( Fig. 5A-C). These data indicated that TRPM7 de ciency suppressed glycolysis in ovarian cancer cells, which was abrogated by HIF-1α over-expression.
Next, we tested how inhibition of AMPK modulated the TRPM7 silencing-promoted OXPHOS in ovarian cancer cells. While TRPM7 silencing signi cantly increased the production of ATP, ROS and the ratios of NAD + /NADPH as well as OCR in ovarian cancer cells, which were signi cantly mitigated or abrogated by treatment with CC ( Fig. 5D-G). Thus, HIF-1α over-expression signi cantly mitigated or abrogated the TRMP7 silencing-decreased glycolysis and inhibition of AMPK abrogated the TRPM7 silencing-enhanced OXPHOS in ovarian cancer cells.
6. TRPM7 silencing enhances the AMPK activation and decreases HIF-1α protein levels to shift glycolysis to OXPHOS in ovarian cancer cells.
Next, we explored how TRPM7 silencing modulates AMPK activation, HIF-1α expression, OXPHOS and glycolysis in ovarian cancer cells. We initially quanti ed the IDH3B and UQCRC1 expression in the different groups of cells. Relative to the untreated control cells, treatment with metformin to activate AMPK obviously increased IDH3B and UQCRC1 expression, which were mitigated by HIF-1α silencing in SKOV3 and HO8910 cells (Fig. 6A, Figure S4A). In contrast, treatment with CC to inhibit AMPK activation remarkably decreased the relative levels of IDH3B and UQCRC1 expression, which were further reduced in the HIF-1α silencing SKOV3 and HO8910 cells (Fig. 6B, Figure S4B). Similarly, TRPM7 silencing increased AMPK phosphorylation, IDH3B and UQCRC1 expression in those cells, which were limited by CC (Fig. 6C, Figure S4C). In contrast, TRPM7 silencing reduced the HIF-1α, HK2 and PDK1 expression in SKOV3 and HO8910 cells, which were rescued by induction of HIF-1α over-expression (Fig. 6D, Figure S4D). Together, TRPM7 silencing enhances the AMPK activation and decreases HIF-1α protein levels to shift glycolysis to OXPHOS in ovarian cancer cells.
7. TRPM7 silencing enhances the AMPK activation to promote HIF-1α ubiquitination and degradation in ovarian cancer cells.
Finally, we investigated how TRPM7 silencing decreased HIF-1α protein in ovarian cancer cells. Hypoxia induced HIF-1α protein expression in ovarian cancer cells, but not in the TRPM7 silencing ovarian cancer cells (Fig. 7A). Treatment with MG132, a proteasome inhibitor, remarkably restored HIF-1α protein levels in ovarian cancer cells regardless of hypoxia condition and TRPM7 silencing (Fig. 7B), indicating that TRPM7 silencing promoted the ubiquitination and proteasomal degradation of HIF-1α. Furthermore, treatment with CC to inhibit AMPK activation restored HIF-1α protein levels in both control and TRPM7 silencing ovarian cancer cells under hypoxic and normoxic conditions ( Fig. 7C and D). Moreover, treatment with CC also decreased the ubiquitination of HIF-1α in both control and TRPM7 silencing ovarian cancer cells under hypoxic and normoxic conditions ( Fig. 7E and F). Therefore, TRPM7 silencing enhanced AMPK activation to promote the ubiquitination and proteasomal degradation of HIF-1α, shifting glycolysis to OXPHOX in ovarian cancer cells.

Discussion
Exploring the potential mechanisms for ovarian cancer growth and nding new therapeutic targets have become a current research hotspot [26]. TRPM7 expression is up-regulated in various tumors [19,27,28]. Furthermore, up-regulated TRPM7 expression is positively related to pelvic lymph node metastasis and poor prognosis of ovarian cancer [29]. Inhibition of TRPM7 can minimize the malignant behaviors of ovarian cancer [17,18]. Here, we reported that TRPM7 silencing changed the transcriptomes involved in modulating cell cycle and acute in ammation, which are linked to tumor cell proliferation and apoptosis [30][31][32] and modulate the tumor environment [33][34][35][36]. Actually, we found that TRMP7 silencing suppressed ovarian cancer cell proliferation and enhanced their apoptosis in vitro and xenograft tumor growth in mice. Hence, TRPM7 acts as an oncogenic factor to promote ovarian cancer growth and is a potential target for development of therapies for ovarian cancer.
It is well known that glucose metabolism by the Warburg effect and OXPHOS can provide nutrients and energy to support the proliferation and biosynthesis in tumor cells [8, 37,38]. Our results indicated that TRPM7 silencing shifted glycolysis to OXPHOS in ovarian cancer cells. Evidently, TRPM7 silencing signi cantly minimized glucose uptake, lactic acid production, ECAR, but elevated ATP and ROS levels, NAD+/NADH ratios and OCR in ovarian cancer cells.
Consistently, PET-CT imaging revealed that treatment with carvacrol to inhibit TRPM7 activity also decreased 18F-FDG uptake by xenograft ovarian cancer in mice. Furthermore, TRPM7 silencing also decreased the relative levels of glycolysis-related HK2 and PDK1 expression as well as PKM2 nuclear translocation, but increased OXPHOS-related IDH3B and UQCRC1 expression in ovarian cancer cells and tissues. The PKM2 is a speci c isoform of pyruvate kinase and a rate-limiting enzyme controlling the glycolysis process. The PKM2 is highly expressed in many types of tumors, including ovarian cancer [39][40][41][42][43]. Studies have shown that PKM2 can translocate to the nucleus, where it forms a complex with HIF-1α and acts as a transcription coactivator of HIF-1α to promote the transcription of HIF-1α targeted genes (such as glycolytic rate-limiting enzymes). Hence, TRPM7 silencing inhibited the proliferation of ovarian cancer cells by shifting from glycolysis to OXPHOS [44][45][46]. OXPHOS can produce high levels of ATP and ROS, which can induce oxidative in ammation and excessive ROS levels can damage DNA and trigger cell apoptosis [33,36,47]. These may explain why TRPM7 silencing signi cantly increased the frequency of apoptotic ovarian cancer cells.
Glucose metabolic reprogramming by converting glycolysis to OXPHOS or tumor cell metabolic plasticity has been noticed in several types of tumor cells and is regulated by the HIF-1α and AMPK signaling [9,10,24,25]. Actually, the enhanced AMPK activation inhibits the expression of HIF-1α and glycolysis-related regulators of HK2, PKM, glucose transporter (GLUT1), fructose-6-phosphate-2-kinase / fructose-2,6-bisphosphatase 4 (PFKFB4) and lactate dehydrogenase A (LDHA) [10,[48][49][50]. Our data revealed that TRPM7 silencing down-regulated HIF-1α expression, but enhanced AMPK activation in ovarian cancer cells, extending previous observations [51][52][53][54]. Furthermore, treatment with metformin to enhance AMPK activation promoted the shifting from glycolysis to OXPHOS in ovarian cancer cells, which was abrogated by HIF-1α over-expression. This shifting may supply fewer nucleotides and lipid precursors for macromolecular synthesis in ovarian cancer cells. These, together high levels of ROS to promote the apoptosis of ovarian cancer cells, inhibit ovarian cancer growth. In contrast, treatment with CC to inhibit AMPK activation minimized the shifting from glycolysis to OXPHOS, dependent on high HIF-1α levels in ovarian cancer cells [51,52,55,56]. Apparently, the HIF-1α and AMPK signaling has opposite roles in regulating glycolysis and OXPHOS in ovarian cancer.
More importantly, we found that TRPM7 silencing promoted HIF-1α ubiquitination and degradation in ovarian cancer cells. It is possible that TRPM7 silencing to reduce [Ca ++ ]i may activate the AMPK to promote HIF-1α ubiquitination proteasomal degradation that attenuates the HIF-1α-enhanced glycolysis to shift glycolysis to OXPHOS, inhibiting the proliferation of ovarian cancer cells and promoting their apoptosis. Conceivably, these ndings may uncover the regulation of the TRPM7/AMPK/HIF-1α axis on the glucose metabolic reprogramming in ovarian cancer.

Conclusion
Our data indicated that TRPM7 silencing shifted glycolysis to OXPHOS in ovarian cancer cells and attenuated the proliferation of ovarian cancer cells and promoted their apoptosis. Mechanistically, TRPM7 silencing enhanced AMPK activation and promoted HIF-1α ubiquitination and degradation in ovarian cancer cells. Hence, our ndings may uncover the new function of TRPM7 in the growth of ovarian cancer and imply that TRPM7 may be a new target for design of therapies for ovarian cancer.

Cells and culture
Human ovarian cancer SKOV3 and HO8910 cells were provided by the Cancer Institute, Central South University and they were identi ed by STR. SKOV3 and HO8910 cells were cultured in RPMI1640 medium containing 10% fetal bowel serum (FBS, Life Technologies, Carlsbad, USA) at 37°C in 5% CO 2 . Some cells were cultured in a hypoxic incubator of 1% O 2 , 5% CO 2 , and 94% N 2 at 37°C.

Transduction and transfection
The lentiviruses for TRPM7 knockdown, and HIF-1α overexpression or knockdown were purchased from Genepharma (shanghai, China). SKOV3 and HO8910 cells (5×10 5 /well) were cultured overnight and transduced with lentivirus at a MOI of 5 for expression of Con-sh, TRPM7-sh1, TRPM7-sh2, TRPM7-sh3, TRPM7-sh4, HIF-1α-sh or HIF-1a. The target sequences are shown in the Supplementary Tables S2. After 48 hours of incubation, their total RNAs and proteins were extracted for veri cation of gene silencing or over-expression. Some cells were cultured in the presence of G418 (500 µg/ml, Biofrox, Germany) for 14 days to generate stable gene silencing or over-expressing cells. Subsequently, we characterized the cell clones by quantitative real-time PCR (qRT-PCR) and Western blot.

qRT-PCR
We extracted total RNAs from each specimen or group of cells using Trizol reagent, and reversely transcribed them into cDNA using Revert Aid First Strand cDNA Synthesis Kit (K1622, Thermoscienti c, USA). We quanti ed the relative levels of target gene to Tubulin mRNA transcripts by qRT-PCR using the FastStart Essential DNA Green Master kit (06924204001, Lifescience, Roche, Mannheim, Germany) and speci c primers in the RocheLightCycler® 96 instrument and software (05815916001, Lifescience). The sequences of primers are shown in the Supplementary Tables S3. We performed the PCR reactions in triplicate and analyzed the data using the 2 −ΔΔCt method.

Western blot
We performed Western blot assays to quantify the relative levels of target protein expression in different groups of cells [58]. Brie y, we performed sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) to separate the cell lysate proteins (30-50 µg/lane) using 10% gels and transferred them onto polyvinylidene di uoride (PVDF) membranes. We treated the membranes with 5% BSA in TBST and probed the membranes overnight at 4 ° C with antibodies against HK2, PDK1, IDH3B, UQCRC1, AMPK, phosphor-AMPK, TRPM7 and α-Tubulin. Subsequently, we incubated the membranes with HRP-labeled second antibodies and developed the blots with the ECL substrate (32109, ThermoScienti c). Finally, we quanti ed the relative levels of individual proteins to α-Tubulin using the ImageJ software (Madison, USA).

Animal experiments and PET/CT study
We obtained female BALB/c nude mice (6-week-old, ~ 20 g) from SLA Laboratory Animal (Changsha, China) and maintained them in a speci c pathogen-free room. In the rst set of experiments, we injected each mouse subcutaneously with ve millions of SKOV3-Con-sh or SKOV3-TRPM7-sh cells in 0.15 mL of saline. Four weeks later, we dissected, weighed, photoimaged the tumors. In another set of experiments, we injected each mouse subcutaneously with 5x10 6 SKOV3 cells. Three weeks later, the experimental group of mice were intraperitoneally injected 50 mg/kg carvacrol (an inhibitor of TRPM7, Sigma-Aldrich, 282197) in DMSO vehicle daily for 7 consecutive days while the control mice received vehicle alone. At the end of treatment, we used PET/CT to measure the tumors in mice and euthanized them. We dissected, weighed and frozen tumors. Subsequently, we xed some tumors from each group of mice in 10% formalin and para n-embedded them as well as froze some tumors in liquid nitrogen for subsequent experiments ( Figure S5). To perform the PET/CT imaging, we fasted the mice overnight and injected individual mice intravenously with approximately 200 ± 10 µCi 18-uoro-6-deoxy-glucose (FDG, Wuhan Union Hospital PET Center, Wuhan, China). One hour later, we anesthetized the mice with 2% iso urane and imaged them using PET with a static mode of 10 min and using CT scan of normal mode in the TransPET Discoverist 180 system (Raycan Technology, Suzhou, China). We reconstructed the PET images using the three-dimensional (3D) OSEM method with a voxel size of 0.5×0.5×0.5 mm3. We reconstructed the CT images using FDK algorithm with 256×256×256 matrix. We displayed the images with software Carimas (Turku PET Center, Turku, Finland). We calculated the mean standardized uptake value (SUV) using the following formula: mean pixel value with the decay-corrected region-of-interest activity (µCi /kg)/(injected dose [µCi]/weight [kg]). The experimental protocols were approved by the Animal Research and Care Committee of our hospital (approval number: 2019-014).

Cell proliferation
The cell proliferation was tested by EdU assay, using the BeyoClick ™ EdU-594 Cell Proliferation Assay Kit (Byotime), per the manufacturer's protocol. Brie y, the different groups of cells (3×10 5 cells/well) were cultured in 6-well plates for 48 h and labeled in triplicate with EdU and DAPI. The uorescent signals were captured under a uorescence microscope (Zeiss, Germany).

Flow Cytometry
The different groups of cells were stained in duplicate using the Annexin V-FITC/PI Apoptosis Detection Kit (Vazyme, Nanjing, China). The frequency of apoptotic cells was mwasured by ow cytometry in the BD Flow Cytometry System.

Intracellular ROS levels
The contents of intracellular ROS in each group of cells were quanti ed using Reactive Oxygen Species Assay Kit (Byotime), per the manufacturer's protocol. The frequency of uorescence positive cells was quanti ed by ow cytometry using CellQuest software.

Measurement of glucose uptake and lactate production
The impact of TRPM7 silencing on glucose uptake and lactate production in ieach group of ovarian cancer cells was tested using the glucose uptake colorimetric assay kit and lactate colorimetric assay kit (Biovision, USA), following the manufacturer's protocols.

Seahorse assay
We performed seahorse assays to quantitatively measure ECAR and OCR of individual groups of cells using Seahorse XF Glycolysis Stress Test Kit and Seahorse XF Cell Mito Stress Test Kit in the Seahorse XFe 96 Extracellular Flux Analyzer (Seahorse Bioscience, USA), following the manufacturer's protocols.
Brie y, individual groups of cells (1 × 10 4 cells/well) were cultured into a Seahorse XF 96-well microplate and their baseline measures were determined.
Subsequently, the cells were treated sequentially with glucose, oligomycin (the oxidative phosphorylation inhibitor), and 2-DG (the glycolytic inhibitor) at indicated dose and time points for measurement of ECAR. Similarly, the cells were treated sequentially with oligomycin, oxidative phosphorylation FCCP (ptri uoromethoxy carbonyl cyanide phenylhydrazone, the reversible inhibitor), and rotenone/antimycin A (Rote/AA, the mitochondrial complex I inhibitor and the mitochondrial complex III inhibitor, respectively). Data were analyzed by Seahorse XF-96 Wave software and expressed as pmols/min for OCR and mpH/min for ECAR, respectively.

ATP assay
The levels of ATP production in each group of cells were determined by ATP assays using ATP Assay Kit (Byotime). Brie y, individual groups of cells were suspended in an ice-cold ATP-releasing buffer, and centrifuged. Individual supernatants or standard (100 µL each) were mixed with 100 µL ATP detection solution and analyzed using the Dual-Luciferase Reporter Assay System (Promega). After normalized with protein concentrations, the luminescent signals were used to quantify the levels of ATP, according to the standard curve prepared with known concentrations (1 nM − 1 mM) of ATP.
In vivo ubiquitination assays The impact of TRPM7 silencing on the stability of HIF-1α in ovarian cancer cells was examined by in vivo ubiquitination assays. Brie y, the different groups of cells were treated in triplicate with CC (20 µM) under a normoxic or hypoxic condition for one day. The cells were treated with MG132 (10 µM) for 8 h, and the levels of HIF-1α ubiquitination were determined by IP with anti-HIF-1α antibody, followed by Western blot using anti-Ub antibody (10201-1-AP, Porteintech, 1:1000).

Statistical analysis
We expressed the data as representative images or the mean ± SD of each group from three independent experiments. We compared the data from multiple groups using ANOVA and post hoc Bonferroni analysis and the data from two groups by Student's T-test using the SPSS version 18.0 (SPSS, Chicago, IL, USA). We analysed the potential correlation of two variables by Spearman's rank test. We de ned statistical signi cant difference when a P-value of < 0.05 and labelled the signi cant levels as *P < 0.05, **P < 0.01, ***P < 0.001 in the gures.  (H) GSEA analysis predicted that TRPM7 was involved in the AMPK pathway.   AMPK activation enhanced glycolysis, which was mitigated by HIF-1α silencing, by measuring glucose uptake (B), lactic acid production (C) and ECAR (F) in SKOV3-sh-control, SKOV3-sh-TRPM7, HO8910-sh-control and HO8910-sh-TRPM7 cells. (D, E, G) Treatment with metformin to enhance the AMPK activation attenuated glycolysis, which was abrogated by HIF-1α over-expression, by measuring glucose uptake (D), lactic acid production (E) and ECAR (G) in the indicated cells.

Figure 5
The shifting glycolysis to OXPHOS by TRPM7 silencing is abrogated or mitigated by HIF-1α over-expression and AMPK inhibition in ovarian cancer cells. HIF-1α over-expression abrogated the TRPM7 silencing-promoted glycolysis in ovarian cancer cells, determined by the glucose uptake (A), lactic acid production (B) and ECAR (C). Treatment with compound C (20 µM) to inhibit AMPK activation mitigated the TRPM7 silencing-induced OXPHOS in ovarian cancer cells, determined by the ATP levels (D), ROS levels (E), NAD+/NAPH rations (F) and OCR (G).

Figure 6
Altered AMPK activation or HIF-1α expression modulates the TRPM7 silencing-decreased expression of glycolysis-relative regulators and increased expression of OXPHOS-relative regulators in ovarian cancer cells. (A-D) Western blot analysis of IDH3B, UQCRC1, HIF-1α, HK2, PDK1 expression and AMPK phosphorylation in the indicated groups of cells.