mGPDH Deficiency leads to melanoma metastasis via induced NRF2

Abstract Oxidative stress critically influences carcinogenesis and the progression of melanoma, and aggressive malignant melanoma activity is due to its high metastatic ability. Some findings in several cancer cell lines have indicated that mGPDH, a component of the mitochondrial respiratory chain, also modulates oxidative stress. However, the role of mGPDH in melanoma remains elusive. Here, we report that the mGPDH protein level is decreased in human skin melanoma compared to normal skin and decreased in metastatic melanoma compared to primary melanoma. Our in vivo and in vitro experiments indicated that mGPDH depletion accelerated melanoma migration and invasion without affecting proliferation or apoptosis. Mechanistically, we found elevated NRF2 protein levels in human skin melanoma and mGPDH‐knockout (ko) metastatic xenografts in the lungs of nude mice. Moreover, in A375 melanoma cells, the loss of mGPDH‐induced NRF2 expression but did not affect NRF2 protein degradation. Additionally, melanoma metastasis induced by the loss of mGPDH was rescued by the further down‐regulation of NRF2 in vivo and in vitro. Consistently, mGPDH overexpression (oe) depressed NRF2 expression and attenuated the malignant properties of melanoma cells. In conclusion, our findings suggest that mGPDH suppresses melanoma metastasis by inhibiting NRF2 and downstream oxidative signals, highlighting the therapeutic potential of mGPDH for melanoma treatment.

patients with cancer that has spread to the lymph nodes and 25% among those with cancer with distant metastasis. 2 In recent decades, cumulative evidence, including that our data, 4 has indicated that oxidative stress critically influences carcinogenesis and the progression of cancer, especially tumour metastasis. Furthermore, redox imbalance plays a central role in the genesis and development of melanoma. 5 A promising molecule for regulating the cellular antioxidant response is mitochondrial glycerol-3-phosphate dehydrogenase (mGPDH), 6 which was previously considered mainly an integral component of the respiratory chain. In addition to stimulating glycerol production to promote obesity, 7 mGPDH regulates the oxidative phosphorylation (OXPHOS) rate and is effectively targeted by metformin. 8,9 Notably, recent research has shown that mGPDH can modulate cell growth in thyroid cancer 8 and reactive oxygen species (ROS) generation in the oxidative stress-induced progression of prostate cancer and that mGPDH functions as a crucial regulator of mitochondrial oxidative stress. [10][11][12] Oxidative stress and consequent oxidative damage are important contributors to tumour metastasis. 13 Our previous studies further demonstrated that nuclear factor (erythroid-derived-2)-like 2 (NRF2, also known as NFE2L2) modulates cell migration in both normal skin cells 14 and cancer cells. 4 Accumulating evidence has established that the NRF2 pathway plays a major role in the cellular antioxidant response. 15 Currently, NRF2 is considered a hallmark of cancer with both tumour-suppressive and tumour-promoting effects. 4,16 Thus, does mGPDH contribute to tumour metastasis, especially in malignant tumours with highly invasive characteristics such as melanoma? During pathogenesis and development, does mGPDH modulate oxidative stress, potentially via NRF2?
So far, studies investigating the role of mGPDH in melanoma, especially in the progression of metastasis of this cancer, are lacking.
In this study, we determined mGPDH expression in normal skin and melanoma tissues from patients. A tissue array containing numerous human tissues from primary and metastatic melanoma at various stages revealed a correlation between mGPDH and melanoma progression. In vivo and in vitro experiments using stable short hairpin RNA (shRNA)-transfected cells, nude mouse xenograft experiments and living cell fluorescence signal analysis demonstrated that mGPDH deficiency leads to melanoma cell migration and invasion as a result of activation of the endogenous antioxidant stressNRF2. We further observed that the inhibition of NRF2 had an antitumour effect on low mGPDH-induced melanoma metastasis.
These results have uncovered the role of mGPDH in melanoma metastasis and suggest the potential use of targeted therapies to treat melanoma with low mGPDH expression levels.  Figure S1A.

| Patients and samples
The final IHC staining score was determined by the following equation: IHC score = intensity score × percentage score. Mouse tissue sections were fixed in 10% buffered formalin and embedded in paraffin. The tissue sections (5 μm) were stained with haematoxylin and eosin (H&E) and then subjected to standard deparaffinization. Sample information and the mGPDH IHC scores are listed in Table S1.

| siRNA and plasmid transfection
Transfection with small interfering RNA (siRNA) and plasmids was performed with RNAiMAX or Lipofectamine 3000 (Invitrogen) according to the manufacturer's instructions. mGPDH siRNA, control siRNA, mGPDH plasmid and the vector were obtained from Ruibo Genetech. The siRNA and plasmid sequences are listed in Table S2.

| Construction of a stable shRNA-expressing melanoma cell line
Human mGPDH shRNA, control shRNA (ubi-MCS-firefly-Luciferase-IRES-Puromycin), sh-ko-control, sh-ko-mGPDH, sh-oe-control and sh-oe-mGPDH were purchased from the GeneChem Company. The sequences are listed in Table S2. Treated 96-well cell culture plates and DMEM containing 10% foetal calf serum were used. The cells reached 40% confluence before shRNA transduction. Two microlitres of lentiviral particles were added per well, and the cells were incubated for 6 hours before the medium was changed. After one day of maintenance, the cells that were illuminated after the addition of luciferin (Beyotime) under a bioluminescence imaging system were chosen.

| Cell migration and invasion
Cellular Transwell assays were used to test cell migration and invasion. Cells (2 × 10 4 /well) were inoculated into serum-free medium in the inserts of Transwell ® cell culture chambers (8 mm pore size; Corning) to assess cell migration and invasion. The lower chamber was filled with DMEM containing 20% FBS. The mean numbers of cells in five randomly chosen fields per well were compared between groups. For Matrigel invasion assays, inserts with an 8 µm pore size were coated with 1 mg/mL Matrigel (BD Bioscience) for 24 hours.
Cells were incubated for 24 hours and allowed to migrate or invade.
Cells that migrated to or invaded the lower surface of the membrane were fixed with 4% PFA and then stained with crystal violet as previously described. Five random fields per chamber were counted using an inverted microscope.

| In vivo tumour growth and metastasis
To examine in vivo tumour growth, control or mGPDH-knockout (ko) shRNA-injected stable A375 cells were subcutaneously injected into nude mice. Every 2 days for 2 weeks, tumour sizes were measured with a calliper, and the tumour volumes were calculated using the the tumour volume, W is the tumour width, L is the tumour length and D is the tumour depth. 18 At the end of the experiment, the mice were sacrificed, and the tumours were excised and weighed. Bioluminescence images were acquired with an in vivo imaging system (IVIS; Caliper Life Sciences). The mice were sacrificed after imaging, and the lungs were isolated, fixed or preserved by freezing. All animal protocols had been approved by the Army Medical University Institutional Animal Care and Use Committee.

| Quantitative real-time PCR (qRT-PCR)
Total RNA was extracted from the harvested cells, and 1 µg of RNA was reverse transcribed into cDNA as described above. qRT-PCR was performed with SYBR Premix Ex Taq II (Takara, Terra Bella Ave) using an Applied Biosystem 7300 system (Thermo Fisher Scientific).
The primer sequences are listed in Table S3.

| Western blot (WB) analysis
Immediately following dissection, the melanoma and skin tissue samples were rinsed with cold PBS and placed in a tube on ice. Small tissue samples (200 mg) were chopped with clean dissecting instruments. From cells, cell lysates were prepared in ice-cold lysis buffer.
Proteins (40 μg/lane) were separated by gel electrophoresis on 10% SDS-PAGE gels. The separated proteins were transferred onto PVDF membranes (Bio-Rad, Hercules) by electroblotting at 100 V for 90 minutes. The membranes were probed with the respective primary antibody, followed by incubation with an HRP-conjugated secondary antibody and detection by chemiluminescence. The primary antibodies were used at 1:800-1:1000 dilution, and the secondary antibodies were used at 1:3000 dilution.

| Determination of the NRF2 half-life
Cells were treated with mGPDH siRNA or control siRNA for 48 hours.
To block protein synthesis, cycloheximide (CHX, 50 μmol/L, BioVision) was added, and cell lysates were collected every 15-min after CHX treatment and subjected to immunoblotting with anti-NRF2 antibody. 19 The relative band intensities were quantified using the Fusion FX5s system (Vilber Lourmat), and ImageJ software was used to analyse the half-life values.

| Apoptosis assay
Cell apoptosis was evaluated with an Annexin V/PI kit (BD Biosciences). Cells were seeded into 6-well plates at a concentration

| Detection of the ATP content
Cellular ATP levels were determined using an ATP assay kit

| Statistical analyses
The data are presented as the mean ± standard deviation (SD), frequency or percentage. Statistical significance was determined using Student's t test, one-tailed Fisher's exact test, or one-way ANOVA with Newman-Keuls multiple comparisons test. The statistical analysis software GraphPad Prism (6.01) was used for statistical analyses, and P < 0.05 indicated statistical significance.

| mGPDH expression was down-regulated in melanoma
To investigate the possible association between mGPDH and melanoma, we first observed mGPDH protein expression in melanoma tissues. Our results showed that the protein level of mGPDH was significantly decreased in human melanoma skin tissue compared to adjacent normal skin ( Figure 1A). With the melanoma tissue array, we found that most primary melanoma tissues showed more obvious IHC staining for mGPDH than the metastatic tissues, as shown in Figure 1B. Specifically, the percentage of primary melanoma patients with a high IHC score (++, +++) was 33.66%, which was more than that for patients with metastatic melanoma (18.58%) ( Figure 1C).
Consistently, after sorting the melanoma tissues into high and low mGPDH expression groups, the Fisher's exact test further confirmed the significant correlation between mGPDH expression and the status of the melanoma as primary or metastatic ( Figure 1D). In addition, even in the primary melanoma tissue, the percentage of samples from more severe tumours (stages III and IV) negative for mGPDH staining was higher than that among samples from less severe tumours (stages I and II) ( Figure S1B), and similar mGPDH expression profiles were found in primary melanoma tissues with lymph node metastasis ( Figure S1C). In summary, down-regulated mGPDH expression was observed in melanoma tissues from patients and metastatic and primary melanoma tissues at higher tumour stages in the human tissue array. These results indicate the correlation of mGPDH and melanoma and suggest the potential involvement of decreased mGPDH expression in melanoma progression.

| mGPDH silencing induced melanoma cell metastasis in vitro and in vivo
To assess the possible role of mGPDH in the progression of melanoma, loss-of-function studies were performed in vitro and in vivo.  and control siRNA (t 1/2 = 19.8 minutes) ( Figure 3C). Consistent with the results of the in vitro experiment, we observed that NRF2 and HO-1 protein levels were significantly higher in skin melanoma tissues from the patients compared to adjacent normal tissues ( Figure 3D). Similarly, in a nude mouse model of melanoma metastasis, lungs from mice with mGPDH-ko xenografts were consolidated and exhibited more metastases, and their surface appeared haemorrhagic and granulated, unlike the lung surface of mice with control A375 xenografts ( Figure 3E).
Moreover, the lung tissue samples were histologically processed and stained with anti-human anti-mGPDH, anti-NRF2 and anti-HO-1 antibodies ( Figure 3E). Consistent with the above results, the mGPDH-ko A375 xenografts expressed elevated protein levels of NRF2 and HO-1.
Therefore, our data suggest that mGPDH silencing activates the NRF2 pathway in melanoma.

| Down-regulation of NRF2 rescued mGPDH loss-induced melanoma metastasis
To confirm the role of NRF2 in melanoma metastasis induced by the loss of mGPDH, we constructed stable A375 cell lines with  3). D-G, Luciferase sh-control shRNA (Sh-ko-control) and mGPDH-ko shRNA (Sh-ko-mGPDH) were transfected into A375 cells to construct stable melanoma cell lines. The cells (2 million) were injected into nude mice via the tail vein, and melanoma cell metastasis was observed in vivo. Transfected A375 cell lines were continuously cultured for 8 wk to ensure consistency with the timing of the in vivo experiment. D. Relative mRNA expression (n = 3). E, Protein levels of mGPDH were detected by intensity quantification. F, G, In vivo metastasis assays. At 8 wk postinjection of the stable cells, melanoma cell metastasis in nude mice was assessed by bioluminescence imaging, and the photon flux ratio was quantified (n = 5-7 mice per group). In A, C, D and G, the P-values were derived from Student's t tests. *P < 0.05, **P < 0.01, ***P < 0.001

| Elevated mGPDH expression alleviated melanoma metastasis
Finally, the therapeutic effect of mGPDH overexpression on melanoma was evaluated. The mGPDH plasmid (Oe-mGPDH) was transfected into A375 cells, which led to increased mGPDH protein levels in the A375 cells, accompanied by down-regulated NRF2 ( Figure 5A).
As expected, an in vitro metastatic assay showed decreased migration and invasion abilities in the mGPDH plasmid-transduced cells compared to the vector-transduced control cells ( Figure 5B-D). Then, we constructed A375 cells stably expressing high levels of mGPDH by transducing the cells with mGPDH-oe shRNA (Sh-oe-mGPDH).
This stable cell line showed consistently higher mGPDH mRNA and protein levels, even after culture for 8 weeks, consistent with the in vivo duration ( Figure 5E,F). Consistently, further in vivo experiments confirmed that increased mGPDH levels alleviated distant metastasis ( Figure 5G,H) and improved the pathological phenotype of the xenografts ( Figure 5I). According to the above experiments, enhanced mGPDH expression alleviated melanoma metastasis and further progression.
F I G U R E 3 mGPDH silencing activated NRF2. A, B, The protein and the mRNA levels of mGPDH and NRF2 signalling pathway in A375 cells without or with mGPDH siRNA treatment were detected by WB analysis and RT-PCR (n = 3). C, NRF2 protein half-life (t 1/2 ) in A375 cells transduced with mGPDH siRNA was determined by pulse-chase assay and immunoblotting. A375 cells were transfected with control siRNA and mGPDH siRNA for 48 h, after which cycloheximide (50 μmol/L) was administered to block protein synthesis. D. Protein levels of NRF2 and HO-1 in skin melanoma tissue and adjacent skin tissue from patients were determined by intensity quantification (n = 3). E. Lung tissues of melanoma metastasis from nude mice injected with Luciferase-control shRNA (Sh-ko-control)/mGPDH-ko shRNA (Sh-ko-mGPDH) A375 cells were harvested. Representative images of lungs and their corresponding tissue sections stained with H&E are shown. Scale bars, 500 μm (gross morphology) and 200 μm (H&E staining). The expression of mGPDH, NRF2 and HO-1 was detected by IHC, as shown in the right three images. Scale bar, 200 μm. The P-Values were derived from Student's t tests. *P < 0.05, **P < 0.01, ***P < 0.001

| D ISCUSS I ON
Melanoma is the most dangerous type of skin cancer and exhibits higher morbidity and mortality than other skin cancers. In the presence of distant metastasis, this cancer is generally considered incurable. 23,24 Redox homeostasis is a vital mechanism underlying melanoma oncogenesis and progression. 5 However, further development is required for an efficient biomarker to guide gene therapy. In this study, we found that mGPDH was decreased in melanoma tissue, especially metastatic melanoma, which predicted a poor clinical outcome, and that the loss of mGPDH promoted metastasis by up-regulating the NRF2 signalling pathway. Genetic inhibition of NRF2 rescued mGPDH ablation-induced melanoma increases in migration and invasion.
Correspondingly, mGPDH overexpression had anti-metastatic effects on the melanoma cells, indicating its therapeutic potential.
Previous studies have demonstrated that mGPDH acts as a mitochondrial oxidative modulator and is usually expressed in low amounts in several cancers 25,26 ; however, the effect of decreased mGPDH expression in tumour progression has remained largely unknown. In this study, we report several findings regarding the role of mGPDH in melanoma. Specifically, mGPDH protein expression was decreased in primary melanoma tissue from patients, especially those with melanoma at a more severe stage or melanoma with lymph node metastasis, suggesting a negative correlation between mGPDH expression and melanoma progression. Moreover, in in vivo and in vitro xenograft experiments, the deletion of mGPDH by siRNA or shRNA aggravated melanoma cell migration and invasion, and the overexpression of mGPDH abrogated distant melanoma metastasis. These results confirm the effect of mGPDH in modulating melanoma metastasis. In addition, we did not find that mGPDH had obvious effects on melanoma cell proliferation, ATP production or apoptosis. However, Shilpa Thakur et al reported that mGPDH regulates thyroid cancer (follicular and papillary thyroid cancer) growth and metabolism. 8 The unique and specific biological characteristics of different kinds of cancers might partially explain the differences in these results. 27,28 NRF2 senses oxidants and regulates antioxidant defence in melanoma metabolism, 16,29 which has been well proven to contribute to cancer progression. 4 Our results confirm the vital role of NRF2 F I G U R E 4 Down-regulation of NRF2 rescued mGPDH loss-induced melanoma metastasis. Control shRNA, mGPDH-ko shRNA (Sh-ko-mGPDH) and/or NRF2 -ko shRNA (Sh-ko-NRF2) were transfected into A375 cells to construct the following 3 stable cell lines: Sh-control, Sh-ko-mGPDH and Sh-ko-mGPDH Sh-ko-NRF2 A375 cells. These cell lines were cultured for 8 wk. A, WB analysis was used to detect mGPDH and NRF2 expression, and the intensity was quantified in the three cell lines. B, C, Cell migration and invasion were evaluated and quantified by Transwell migration and Matrigel assays, respectively (n = 3). Scale bar, 200 μm. D, In in vivo metastasis assays, nude mice were injected with the three stable cell lines. D, E, After 8 wk, melanoma cell metastasis in the nude mice was assessed by bioluminescence imaging and quantified as the photon flux ratio (n = 5 mice per group). F. Lung images and H&E staining for the three groups. Scale bars, 500 μm and 200 μm. The P-Values were derived from one-way ANOVA with the Newman-Keuls multiple comparisons test. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 F I G U R E 5 mGPDH-overexpression alleviated melanoma metastasis. A-D, A375 cells were transfected with a vector (Oe-vector) or mGPDH-overexpression plasmid (Oe-mGPDH). A, Relative protein levels were detected by WB analysis, and the intensity was quantified. B, Transwell migration (upper) and Matrigel invasion assays (lower). Scale bar, 200 μm. C, D, Relative results of cell quantification by Transwell migration and Matrigel invasion assays (n = 3). E-I, Luciferase-control vector shRNA (Sh-oe-vector) or luciferase-overexpression mGPDH shRNA (Sh-oe-mGPDH) was transfected into the A375 cell line to construct stable cell lines. These cells were intravenously injected into nude mice, and melanoma cell metastasis was observed for 8 wk in vivo. E, The relative mRNA expression of mGPDH was detected (n = 3). F, Transfected A375 cell lines were consistently cultured in vitro for 8 wk, and then mGPDH and NRF2 protein levels were detected. G-I, In vivo metastasis assays. G, H, At 8 wk post-injection of the stable cell lines, melanoma cell metastasis in the nude mice was assessed by bioluminescence imaging and quantified (n = 4 mice per group). I, Metastatic lung tissue images and H&E staining of the two groups. Scale bars, 500 μm and 200 μm. The p-values were derived from Student's t tests. ***P < 0.001, **P < 0.01 in the effect of decreased mGPDH on melanoma metastasis. First, not only is NRF2 a traditional tumour suppressor before tumorigenesis but also its hyperactivation has cancer-promoting functions. 4,29 Numerous studies have revealed the mechanism underlying the role of excessive NRF2 activation in tumour metastasis. 30 Comparably, our results demonstrate that NRF2 was over-activated in skin tissues from patients with melanoma and that inhibiting NRF2 could rescue mGPDH loss-induced melanoma distant metastasis. Second, mGPDH silencing in melanoma cells led to the up-regulation of members of the NRF2 signalling pathway, including the transcript and protein levels of NRF2 and its downstream genes, but did not have a significant effect on NRF2 protein degradation. Consistently, recent studies have reported that NRF2 can be modulated at the protein and transcriptional levels. 21,22 Third, as in previous studies, 29,31 we also found that the inhibition of NRF2 improved the prognosis of melanoma. These In summary, these findings confirm that mGPDH negatively regulates melanoma metastasis by modulating the NRF2 signalling pathway and provide several insights into mGPDH-based targeted therapy to inhibit NRF2, which might be an attractive anti-melanoma treatment approach in the future.