Overcoming Chemoresistance to Vemurafenib in Melanoma via Targeted Inhibition of PCK1 Using 3-mercaptopropionic Acid

Background:BRAF inhibitors are the mainstay treatment for melanoma with the V600E mutation, but its resistance to BRAFi remains a clinical challenge. Therefore, it is necessary to explore the mechanism of BRAFi resistance and develop new therapeutic targets. Methods:We established an A2058 melanoma cell line with acquired resistance to vemurafenib in vitro. RNA sequencing was used to identify the target gene and signaling pathway which were related to the resistance.A series of in vitro assays were applied to conrm the function of PCK1, including test of scratch,transwell,cell viability,ow cytometry,sphere formation,western blot and ROS detection.Western blot was taken to identify the activation of PI3K/Akt pathway.We constructed the subcutaneous xenograft model of melanoma,and the mice were randomly injected with DMSO,vemurafenib,and combination of vemurafenib and 3-mercaptopropionic acid.Finally, tumor size,ROS in tissue,and immunohistochemistry were analyzed to validate the ndings. Results:We identied that the activation of PI3K/Akt pathway led to the overexpression of phosphoenolpyruvate carboxykinase 1, a key enzyme of gluconeogenesis. An elevated PCK1 level induced intracellular metabolic reprogramming, thereby lowering oxidative stress contributed to the chemoresistance to vemurafenib. 3-mercaptopropionic acid , an antihyperglycemic agent, could inhibit the viability of PCK1 then bring oxidative damage to drug-resistant cells. As a result, 3-mercaptopropionic acid sensitized the cells to the killing effect of vemurafenib, exerting a synergistic anti-tumor effect in combination with vemurafenib. Conclusions:Our study demonstrates that the PI3K/Akt-PCK1-ROS axis plays an important role in BRAFi-resistant melanoma and that using the antihyperglycemic agent 3-MPA is a feasible strategy to restore its therapeutic sensitivity.


Background
The incidence of melanoma, a malignant tumor of melanocytes that arises in the skin, has been increasing in recent years. 1 BRAF mutation is a prominent feature of melanoma, with studies pointing to the presence of serine/threonine protein kinase B-Raf activating mutation in 50 ~ 60% of patients with advanced melanoma, and with the V600E mutation being the most common, comprising 74-86% of all BRAF mutations. 2,3 This oncogenic mutation activates the downstream kinase MEK/ERK within the MAPK pathway, promoting melanoma progression. 4 Inhibitors designed to target the BRAF mutation, such as vemurafenib and dabrafenib, have achieved pioneering results. The median PFS of vemurafenib was found to be higher than that of dabrafenib (5.8 months versus 7.3 months). 5,6 However, patients can only bene t from the use of these inhibitors in the early stages of treatment, since inevitable acquired resistance is the major factor contributing to treatment failure. Studies show that 'paradoxical activation' includes mutations in NRAS or MEK1/MEK2, and a high expression level of COT, EGFR, PDGFRβ, IGF1R, or MET, which has encouraged clinicians to investigate a combination of inhibitors that effectively block the MAPK signaling pathway. 7,8 Based on the 'paradoxical activation' of the MAPK pathway via BRAF inhibition, the BRAF-MEK inhibitor combination was further investigated in phase III clinical trial COMBI-d.
The results showed that patients could bene t more from combination therapy than from vemurafenib alone (median PFS, 11.4 months; OS, 25.1 months versus 7.3 months, 18.7 months). 9 Therefore, "Darabfenib + Trametinib" combination therapy is recommended as the main chemotherapy for advanced melanoma patients. Innovative strategies are still urgently needed to improve the precision and persistence of melanoma therapies.
The process of metabolic remodeling is crucial for melanoma cells to adapt to the tumor microenvironment (TME) and maintain growth and metastasis. 10 Phosphoenolpyruvate carboxykinase 1 (PCK1, PEPCK-C), a cytoplasmic form of PEPCK, is a rate limiting enzyme, which decarboxylate in the second step of gluconeogenesis after the carboxylation of pyruvate, and then phosphorylates oxaloacetic acid to form phosphoenolpyruvate (PEP). 11,12 As a central molecule regulating glycolysis, the tricarboxylic acid cycle (TCA), and gluconeogenesis, PCK1 exerts diametrically opposite biological effects in different tissues. 13 In gluconeogenic tissues, such as the liver and kidneys, the overexpression of PCK1 acts as a suppressor 14 because it exacerbates gluconeogenesis and hinders both glycolysis and the TCA cycle, leading to nutrient stress and energy homeostasis. 15 Interestingly, in the colon or lungs, where glycogen is not produced, the overexpression of PCK1 abundantly generates NADPH to reduce the levels of reactive oxygen species (ROS) through the pentose phosphate pathway (PPP). 16 Thus, a high level of PCK1 promotes cancer exacerbation when carcinogenesis occurs in intestinal and lung cancers. However, the mechanism of PCK1 during melanoma progression and chemoresistance remains unknown and regulation of metabolic reprogramming may be an Achilles' heel of cancer, with important implications.
In this study, A2058, a cell line harboring the BRAF V600E mutation, was used to construct a vemurafenibresistant cell line in vitro. Using next-generation sequencing, we found that hyperactivation of the Akt pathway and its downstream molecule PCK1 was associated with resistance to BRAF inhibitors (BRAFi) in melanoma cells. PCK1, a key enzyme of PPP, promotes resistance in non-gluconeogenic tissues to oxidative stress by synthesizing a large amount of NADPH to maintain oxidation homeostasis, which keeps melanoma cells alive and proliferating under drug pressure. Based on this, we revealed the role of PCK1 inhibitor 3-mercaptopropionic acid (3-MPA) in reversing the metabolic reprogramming of melanoma cells using in vivo and in vitro experiments. We found that 3-MPA can signi cantly enhance the therapeutic effect of BRAFi and is therefore a novel strategy for the clinical treatment of melanoma.

Methods
Establishment of BRAFi-resistant melanoma cells A2058 cells were inoculated in a culture ask, and a high dose of vemurafenib(PLX4032, Cat#HY-L032) (MedChemExpress, USA) was used for treatment. When most of the cells had died, the dead cells were washed away and the medium was changed to a drug-free medium. Post reaching con uence, the cells were digested and re-inoculated into a new culture ask, and some cells were cryopreserved for RNA-seq.
The cells were treated with high density vemurafenib until the cells reached 30% con uence. This process was repeated 3 times, and we obtained R0, R1, R2, and R3 cells, which were then used for RNA-seq analysis. Melanoma cells in the exponential growth phase were inoculated into culture bottles. After 48 h, the cells were transferred to a drug-free medium and expanded until the next mitotic phase. Then, the above steps were repeated using vemurafenib that was two times more concentrated. At the same time, cell death was observed every day, the fresh complete medium containing vemurafenib was replaced, and CCK8 analysis was performed regularly until melanoma cells grew stably in the medium. This process lasted for six months, and transcriptome sequencing was performed once every two months; thus, 4 samples (R0, R1, R2, and R3) were obtained.
Cell proliferation and drug treatments A cell proliferation experiment was carried out on a 96 well plate. The cells in the exponential growth phase were analyzed using the CCK8(Beyotime,China) method. The inoculation density used was 4 × 10 3 cells/well. The treatment started at 24 h post inoculation and lasted for 72 h. Then, 10 μl of a CCK8 solution was added to each well, and the cells were cultured at 37℃ under 5% CO 2 until the solution turned signi cantly yellow. The optical density (OD) was measured at 450 nm on a microplate. The cell viability was calculated using the following formula: (OD of dosing well -OD of culture medium)/(OD of control well -OD of culture medium) × 100%. The 50% inhibitory concentration (IC50) calculator (https://www.aatbio.com) was used to prepare the survival rate and drug concentration curves, and the best tting IC50 was obtained. The statistical chart of the IC50 value was prepared using GraphPad prism 6 software. The following formula was used: Resistance index (RI) = IC50 of resistant cells/IC50 of parental cells.

Western blot analysis and antibodies
Western blot (WB) was performed as reported previously, 17 and all the antibodies used are listed in supplementary table 1.

Cellapoptosis analysis
Flow cytometry was performed as reported previously, 17 and the Annexin V Apoptosis Detection Kit APC was obtained from eBioscience.

ROSdetection in cells and tissues
CellROX Regent orange solution was used to culture or cover the tissue and avoid light during the whole process. After incubation for 30 min at 37℃, we used a High Content Imaging System to observe ROS production in the cells. The dye solution was removed from the tissues and they were xed. Then, photos were taken under a microscope, and the tissues were stored.

Immunohistochemistry
The tumor tissues were dehydrated and xed in a 10% formalin buffer solution for 24 h. The frozen sections were embedded in OCT, frozen, and continuously sectioned to 15 μm sections at 20℃.
Immunohistochemistry was performed as previously described. The primary antibodies used are listed in Table S1.

Animal models
Parental A2058 cells or drug-resistant melanoma A2058R cells (1 × 10 6 ) in 200 μl of PBS were subcutaneously transplanted into SPF BALB/c nude mice (purchased from Zhongshan Hospital Laboratory). After 7-10 days of transplantation, the palpable tumor size (40-100 mm) was reached, and mice with a similar tumor size were randomly assigned to different four treatment groups. The parallel group design was adopted, and the mice were treated with 30 mg/kg vemurafenib ,and/or 3-MPA(SKF-34288 hydrochloride, Cat#320386-54-7) (MedChemExpress, USA),and equal DMSO,once a day for 21 days. All animal protocols were approved by the animal protection and the use committee of Zhongshan Hospital, Fudan University.

Calculation of the combination index
CompuSyn is available for download from http://www.combosyn.com. According to the Chou-Talalay mathematical model of drug interaction, the combination index (CI) of cells receiving combination therapy was calculated. Chou-Talalay's Ci theorem provides a quantitative de nition of the additive effect (CI = 1), synergistic effect (CI < 1), and antagonistic effect (CI > 1) of drug combination 17 .

Statistical analysis
Each experiment was repeated three times. The results are shown as the mean ± standard deviation. SPSS software 19.0 was used for the statistical analyses, including Student's t-test (two-tailed), Pearson's correlation analysis, and the Log-Rank test. The signi cance threshold was set at 0.05 for each test.

Results
Establishment and phenotype of the drug resistance model.
To investigate the molecular mechanisms underlying BRAFi resistance, we exposed the melanoma cell line A2058 harboring the BRAF V600E mutation to high concentrations of vemurafenib. After 6 months, we obtained the resistant cell line A2058R. We found that the cells became slender under the action of vemurafenib at a high concentration ( Figure 1A, 1B). Results of the CCK8 experiment con rmed that the IC 50 of drug-resistant A2058R cells was higher than that of primary A2058 cells, (A2058 IC50 = 0.71, A2058R IC 50 = 21.95, RI = 30.9, P < 0.005) ( Figure 1C, 1D). Scratch and Transwell assays con rmed that A2058R cells have greater invasion and migration abilities than A2058 cells ( Figure 1E -1I). Meanwhile, WB assays showed that the levels of CD271, SOX10, TWIST, and SLUG were elevated in A2058R cells, indicating that they possess higher stemness and EMT abilities. PCK1 protein and AKT/PI3K signaling were key factors in the transcriptome sequencing results.
After normalizing the transcriptome sequencing results (Figure 2A), we found that there were differences in the gene expression pro les among the four samples ( Figure 2B), which was further corroborated by PCA results ( Figure 2C). Differential expression analysis was performed, and we focused on PCK1, a gene whose expression level is high and constant in R1, R2, and R3, and we considered it as the key target gene for conferring of drug resistance ( Figure 2D). In KEGG enrichment analysis ( Figure 2E), we observed that the PI3K/AKT signaling pathway molecules were continuously highly expressed in the three drugresistant samples. Our results in combination with the literature on this topic indicate that the PCK1 and PI3K/Akt signaling pathways are involved in the process of acquired drug resistance induced by BRAFi in melanoma cells. Furthermore, western blot analysis showed that the expression of the PCK1 and PI3K/Akt signaling pathways was up-regulated and activated in drug-resistant cells. In addition, when we inhibited the PI3K/Akt pathway, the expression of PCK1 decreased, suggesting that PCK1 is a crucial downstream molecule for the PI3K/Akt signaling pathway for achievement of drug resistance ( Figure 2F). PCK1 activity determined melanoma cell resistance to vemurafenib.
Next, we constructed low expression and overexpression PCK1 cell lines using siRNA and overexpression viruses, respectively. Western blot results indicated that the transfection was successful ( Figure 3A). After PCK1 knockdown, cell viability signi cantly reduced; thus, the IC50 of siPCK1 and the resistance index (RI) both decreased (RI = 4.46), P < 0.005 ( Figure 3B, 3C).
However, the cell viability and resistance index (RI) were higher (RI = 9.27), P < 0.005, in the cell line overexpressing PCK1 ( Figure 3B, 3D). Therefore, the results indicated that PCK1 is a crucial molecule that can enhance the tolerance of melanoma cells to vemurafenib. PCK1 promoted the proliferation, migration, and stemness of melanoma cells with V600E mutation.
Based on the hypothesis that the development of drug resistance may be caused by related genes that induce cancer stemness, EMT, and apoptosis inhibition, we rst designed scratch and invasion experiments and found that the PCK1 gene status was related to the proliferation and invasion of melanoma cells, while the activation of PCK1 could promote drug resistance in melanoma and increase the severity of the disease ( Figure 4A, 4B). Next, ow cytometry was conducted, and the results showed that the knockdown and overexpression of PCK1 did not have a direct effect on apoptosis, although statistical differences were noted. This suggested that PCK1 might indirectly induce BRAFi resistance through PPP ( Figure 4C-4G). In addition, the results of the tumor spheroidization test on a low adhesion plate showed that the spheroidization ability of A2058R cells was signi cantly higher than that of A2058 cells ( Figure 4H), that of the SiPCK1 group was decreased, and that of the OE group was restored after the overexpression of PCK1. Combined with the western blot results, these results suggested that CD271 and SOX10, which have been recognized as indicators of melanoma stemness, were highly expressed in A2058R and OE cells ( Figure 4I), indicating that PCK1 could modify cancer stemness in the development of drug resistance. Meanwhile, in this experiment, PCK1 could not alter the EMT of melanoma for achievement of drug resistance ( Figure 4J). These lines of evidence illustrated that the activation of PCK1 potentiated melanoma proliferation, migration, and tumor stemness, all the phenotypes associated with drug resistance. The results revealed that PCK1 did not contribute to drug resistance directly by inhibiting apoptosis but, most likely, contributed indirectly to acquired drug resistance by triggering some of the cascades of the biochemical metabolic chain.
Synergistic effect of PCK1 inhibitor 3-MPA combined with vemurafenib on drug-resistant cells.
A CCK8 assay was used to detect the activity of A2058R cells under different drug treatment regimens to evaluate the advantages of a combined treatment strategy. The IC50 of A2058R cells treated with 3-MPA or vemurafenib alone did not change signi cantly ( Figure 5A, 5C). However, the IC50 of 3-MPA combined with vemurafenib was signi cantly lower than that of vemurafenib alone, which indicated that 3-MPA increased the resistance inhibition of vemurafenib ( Figure 5B, 5C). In order to evaluate the e cacy of the combination therapy, we used the Chou-Talalay method to compare the best tting IC50 of each treatment group again. In the experiments in which drug-resistant A2058R cells were treated, the IC50 of the 3-MPA + vemurafenib group was 6.37, that of vemurafenib monotherapy was 14.47, that of 3-MPA monotherapy was 27.47, and the combination index (CI) = 0.3362 < 1, which showed that the combination of 3-MPA and vemurafenib had a synergistic effect on pharmacology ( Figure 5D, 5E). Moreover, only when the reaction part of the system reaches an 0.8 < FA < 0.9, the synergistic effect will be reversed to the antagonistic effect, which indicates that change in the doses of the two drugs has little effect on the synergistic effect and that this effect is stable ( Figure 5F). Inhibition of PCK1 by 3-MPA led to ROS accumulation and oxidative damage in drug-resistant cells.
Since PCK1 is a key enzyme in the PPP process and PPP controls the intracellular ROS levels, the supplement of reduction equivalent is undoubtedly of great value to melanoma cells. Therefore, we speculated that the downstream mechanism of PCK1 might be related to the regulation of ROS levels. We used the High Content Live Cell Imaging System to observe the ROS levels. As shown in Figure 6A, we found that the ROS level in primary A2058 cells was signi cantly higher than that in A2058R cells ( Figure  6B), which demonstrated that drug-resistant cells have a better ROS reduction ability. When A2058R cells were exposed to vemurafenib, the ROS level was still lower than that in A2058 cells. In addition, the ROS levels of primary A2058 cells and drug-resistant A2058R cells were higher than those in A2058R cells after adding 3-MPA. When A2058 and A2058R cells were treated with a combination of 3-MPA and vemurafenib, the ROS level was increased signi cantly, which suggested that the ROS level of drugresistant cells was determined by the activity of PCK1.
We generated a subcutaneous xenograft model from nude mice, forming cryosections that were analyzed using a ROS probe ( Figure 6C). Combined with the tumor volume/wt ( Figure 7D, 7E), we found that the ROS levels were higher in the A2058 + vemurafenib group than in the A2058 + DMSO group, whereas the number of tumor cells was reduced relative to the A2058 + DMSO group, indicating that vemurafenib inhibited V600E mutated melanoma progression by generating strong oxidative damage. Compared with the A2058R + vemurafenib group, the A2058 + vemurafenib group exhibited higher intracellular ROS levels and decreased tumor cell numbers, indicating that the drug-resistant cells hedge the killing effect of vemurafenib via metabolic reprogramming. Compared with the A2058R + vemurafenib group, the A2058R + 3-MPA + vemurafenib group showed a signi cant increase in ROS levels and a decrease in the number of tumor cells, demonstrating that, in drug-resistant cells, the inhibition of PCK1 sensitizes melanoma to BRAFi inhibition. The increase in ROS levels was mainly caused by the obstruction of the PPP pathway. However, when we added 3-MPA (20 μmol/l) to A2058 and A2058R cells, the WB results showed that the expression of KEAP1 was increased after PCK1 silencing, which indicated that PCK1 inhibited the expression of KEAP1 ( Figure 6D). These pieces of evidence all pointed the spearhead of resistance toward ROS reduction, which was the result of PCK1 activity. 3-MPA blocked PCK1, thereby enabling ROS accumulation and suppressing chemoresistance in melanoma cells.

3-MPA improved the inhibitory effect of BRAFi in vivo
Next, we evaluated the effect of combination therapy in melanoma using a subcutaneous xenograft mouse model ( Figure 7A, 7B). According to the wt and volume data ( Figure 7C, 7D), we found that the combination of vemurafenib and 3-MPA in the drug-resistant group could reverse the drug-resistant situation and signi cantly reduce the wet and volume of the tumor tissue. Compared with A2058 primary melanoma cells, the inhibition ability of the combination group was even more prominent, which suggested that the combination of BRAFi and PCK1 could achieve an excellent inhibition effect in BRAFi-resistant cells with PCK1 mutation. Combined with the immunohistochemical results, as mentioned above, we further con rmed that the main effect of 3-MPA was to increase the ROS levels of melanoma cells, disrupt the oxidative balance in drug-resistant melanoma cells, and expose them to an oxidative attack again.
The immunohistochemistry results showed that in the subcutaneous tumor tissues of nude mice inoculated with primary A2058 cells, the expression of PCK1 in the vemurafenib group was up-regulated compared with the DMSO group, while KEAP1 was inhibited ( Figure 6D). In the tissues of A2058R cells, the up-regulated expression of PCK1 in the group treated with vemurafenib in combination with 3-MPA was limited compared with that in the vemurafenib group, but the expression of KEAP1 was up-regulated. Therefore, both the in vitro and in vivo experiments con rmed that vemurafenib could induce high expression of PCK1, and then inhibit the synthesis of KEAP1, thus inducing drug resistance. However, when 3-MPA was used to inhibit PCK1, KEAP1 was also activated and ROS accumulated in the cells, which made the drug-resistant melanoma cells suffer from oxidative attack and inhibited tumor progression.

Discussion
Acquired resistance is the main clinical obstacle to improving the prognosis of melanoma patients, and the known mechanisms involve various components. 18, 19 The most common pathological mechanisms are the reactivation of the BRAF/MEK pathway or other proliferation promoting signal transduction pathways, the increase in the BRAF V600E protein copy number, RAS mutations, and the activation of ARAF and CRAF, which promote the formation of a dimer that BRAFi cannot inhibit. 20,21,22 Moreover, the activation and mutation of MEK1/MEK2 and the high expression of COT lead to the activation of the MAPK/ERK pathway downstream of BRAF. 23 Overexpressed RTKs and the compensatory activation of the Akt/PI3K/mTOR pathway contribute to drug resistance. 24,25 Our study also demonstrated that the adaptive hyperactivation of AKT in response to BRAFi compensated for BRAF inactivation, and the mutually compensatory relationship between the Akt/PI3K/mTOR signal transduction pathway and the ERK pathway supported tumor adaptation and resistance to speci c interventions. Therefore, we concluded that the AKT/PI3K pathway was an extremely important signaling axis, promoting cell proliferation, migration, and stemness, which contributed to BRAFi resistance in V600E mutated melanoma.
Although there are many downstream effectors of PI3K and Akt that can change the function of melanoma cells, PCK1 synthesis and activation are important processes mediated after AKT phosphorylation. The Akt/PI3K/mTOR pathway was mainly related to the mechanism of metabolic reprogramming, 26 including the activation of mTOR complex 1 (mTORC1), glycogen synthase kinase 3 (GSK3), and forkhead box O (FoxO) transcription factor family members. 27,28 In our study, the activity of the Akt/PI3K/mTOR pathway and the tPCK1 levels always paralleled each other in vitro, probably because the activation of mTORC1 molecules induced cytosolic SREBP to promote PCK1 synthesis. 29 Additionally, FoxO1 deacetylation synergizes PGC-1α action to promote PCK1 transcription. It has recently been reported that activated Akt can directly phosphorylate PCK1 at Ser-90. 27,30 Ultimately, the biological processes mediated by PCK1 under the agonism of AKT are critical steps for melanoma cells to survive and proliferate.
A redox homeostasis imbalance elevates ROS levels and is pivotal in driving the processes of carcinogenesis, metastasis, and drug resistance in cutaneous melanoma. 31,32,33 Lim pointed out that was the process of redox metabolism rewiring in melanoma cells. 34 During the course of long-term treatment with BRAFi, the upregulated MITF-PGC1α axis and the long non-coding RNA (lncRNA) SAMMSON-p32 led to an inevitable accumulation of ROS. 35 High levels of ROS should have induced apoptosis, but genetic changes led cells to survive at high levels of ROS by increasing the NADPH level through the pentose phosphate pathway (PPP) or by activating the KEAP1/Nrf2 antioxidant signaling pathway. 36,37 One of the crucial switches driving redox remodeling was PCK1, which is a key enzyme involved in the PPP process, 38 playing a role in the antioxidant function of tumors. 39 This requires the presence of PCK1 to synthesize abundant reductive NADPH via the PPP, and its inhibition impedes melanoma growth. 40,41 Moreover, PCK1 repressed KEAP1 and the Nrf2 restriction was removed, further activating the antioxidant repair ability of cells. Thus, a large amount of reductive NADPH was produced, which counteracted the cell damage induced by high levels of ROS and enabled cells no close to achieving redox equilibrium to survive.
Studies on the role of redox fragility in melanoma progression have been reported, and BRAFi-resistant melanoma cells with elevated ROS levels are more sensitive to the death induced by oxidants. 42,48 This is highly consistent with our results, which showed that 3-MPA sensitized the biological effect of vemurafenib to restore the sensitivity of tumors. When 3-MPA was used, drug-resistant cells lost the ability of supplementing the reduction equivalent, the redox level was unbalanced again, and the cells experienced oxidative damage. The speci c mechanism underlying this effect might be that 3-MPA blocks the PPP pathway to synthesize NADPH and nucleic acids, inhibiting cell proliferation and leading to the accumulation of intracellular ROS. In addition, interestingly, the increase in ROS level should play a protective role through the KEAP1/Nrf2 axis. However, when 3-MPA was used, the expression of KEAP1, a repressor of Nrf2, increased. 49 As the KEAP1/Nrf2 axis had cellular protective effects on the antioxidant and detoxi cation activities, 50 inhibiting Nrf2 undoubtedly aggravated the oxidative burden of tumor cells. These results suggested that 3-MPA not only down-regulates the PPP pathway, but also indirectly up-regulates KEAP1, leading to a loss of the self-antioxidant repair ability of drug-resistant cells; as a result, the combination of 3-MPA and vemurafenib resulted in outstanding therapeutic effects.
One limitation of our study is that only the melanoma A2058 cell line was used for modeling. Moreover, the drug resistance was a relatively long-term change, and the observation time used in our in vitro and in vivo experiments was relatively short. Therefore, the role of PCK1 in the drug resistance process of other tumor cells needed to be further revealed, and the observation cycle also needs to be optimized to explore the speci c functions of PCK1 from a long-term and objective perspective.

Conclusions
Oue study found that the activation of the Akt/PI3K/mTOR signaling pathway compensated for the tumor promoting effect in the BRAFi-resistant melanoma cells. Akt/PI3K upregulated PCK1, which turned on the PPP switch and inhibited KEAP1, and then synthesized enough reductive equivalent to decrease the ROS levels (Figure 8),leading to BRAFi resistance. PCK1 also promoted cell proliferation, migration, and tumor stemness. As AKT/PI3K/mTOR pathway widely distributed, patients with direct suppression of it may have more side effects. Based on the strategy of precision treatment,we used an antihyperglycemic agent 3-MPA combined with vemurafenib and found it enhanced the sensitivity of drugs obviously.Finally, we clari ed the mechanism of resistance to BRAFi and provided new options for clinically targeted combination therapy in patients with advanced melanoma.

Abbreviations
Kelch-like ECH-associated protein 1 KEAP1 Availability of data and materials The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests    Functions of PCK1 in melanoma proliferation, migration, apoptosis, stemness, and EMT. A. Scratch assay. B. Results of the scratch assay: the y-axis is the area of change before and after scratching, and the proliferation area of SIPCK1 cells was smaller than that of Ctrl cells, and the proliferation area of PCK1 cells was higher than that of EV (P < 0.05). C.Flow cytometry of apoptosis: the y-axis represents the PIY585 staining and the x-axis represents the Annexin V staining. the Q2 + Q3 is the proportion of apoptotic cells. The percentage of apoptotic cells was 0.9% in the control group . D.Flow cytometry of apoptosis: the percentage of apoptotic cells was 0.53% in the EV group. E. Flow cytometry to assess apoptosis: there was no signi cant difference in the distribution of apoptotic cells.F. Flow cytometry to assess apoptosis: the percentage of apoptotic cells in the siPCK1 group was 1.05%. G. Flow cytometry to assess apoptosis: the percentage of overexpression of the PCK1 group was 1.11%. H. Results of the sphere formation assay: the y-axis indicated the number of cell spheres with a diameter larger than 50 µm under a eld of view, which was signi cantly lower in the siPCK1 group than in the Ctrl group, and higher in the overexpression PCK1 group than in the EV group, P < 0.05. I. Stemness test. J. EMT test.  PCK1 led to drug resistance by reducing ROS accumulation. A. ROS imaging: The red area is the result of the post development of the ROS probe combined with intracytoplasmic ROS, and the white area represents the strongly refractile nuclear components. B. Result of ROS imaging: the y-axis represents the average light intensity of the probe, and the ROS content of A2058 cells was higher than that of A2058R cells. A2058 and A2058R cells showed increased ROS levels when vemurafenib was used alone, and A2058 cells showed higher ROS levels than A2058R cells (P < 0.05). After the addition of 3-MPA alone, the ROS content of A2058R cells was signi cantly higher than that of A2058 cells (P < 0.05). A2058 and A2058R cells showed increased ROS levels after the combination treatment with 3-MPA and vemurafenib, compared with the treatment with vemurafenib alone (P < 0.05). C. Detection of ROS in frozen section: the sliced tissues were derived from the subcutaneous tumorigenesis nude mice model, P represents the subcutaneous tumor formed after inoculation with melanoma A2058 cells, R represents the subcutaneous tumor formed after inoculation with drug-resistant A2058R cells, red represents the ROS, blue represents the nuclei (DAPI). ROS activity was higher within the tissues from P which received vemurafenib than from the P which received DMSO and was lower within the tissues from R which received vemurafenib; however, when 3-MPA and vemurafenib were combined, ROS activity was signi cantly higher within the tissues from R which received vemurafenib alone. D. Expression of KEAP1: in A2058 and A2058R cells, KEAP1 expression increased after the addition of 3-MPA. In non-drug resistant tissues P, the expression of PCK1 increased and that of KEAP1 decreased after using vemurafenib. In R, the level of PCK1 was higher than that in P; the combination of 3-MPA and vemurafenib decreased the levels of PCK1 and increased the levels of KEAP1. Supplementary Files