HMG-CoA synthase 1 is a synthetic lethal partner of BRAFV600E in human cancers

Contributions of metabolic changes to cancer development and maintenance have received increasing attention in recent years. Although many human cancers share similar metabolic alterations, it remains unclear whether oncogene-specific metabolic alterations are required for tumor development. Using an RNAi-based screen targeting the majority of the known metabolic proteins, we recently found that oncogenic BRAFV600E up-regulates HMG-CoA lyase (HMGCL), which converts HMG-CoA to acetyl-CoA and a ketone body, acetoacetate, that selectively enhances BRAFV600E-dependent MEK1 activation in human cancer. Here, we identified HMG-CoA synthase 1 (HMGCS1), the upstream ketogenic enzyme of HMGCL, as an additional “synthetic lethal” partner of BRAFV600E. Although HMGCS1 expression did not correlate with BRAFV600E mutation in human melanoma cells, HMGCS1 was selectively important for proliferation of BRAFV600E-positive melanoma and colon cancer cells but not control cells harboring active N/KRAS mutants, and stable knockdown of HMGCS1 only attenuated colony formation and tumor growth potential of BRAFV600E melanoma cells. Moreover, cytosolic HMGCS1 that co-localized with HMGCL and BRAFV600E was more important than the mitochondrial HMGCS2 isoform in BRAFV600E-expressing cancer cells in terms of acetoacetate production. Interestingly, HMGCL knockdown did not affect HMGCS1 expression levels, whereas HMGCS1 knockdown caused a compensating increase in HMGCL protein level because of attenuated protein degradation. However, this increase did not reverse the reduced ketogenesis in HMGCS1 knockdown cells. Mechanistically, HMGCS1 inhibition decreased intracellular acetoacetate levels, leading to reduced BRAFV600E-MEK1 binding and consequent MEK1 activation. We conclude that the ketogenic HMGCS1-HMGCL-acetoacetate axis may represent a promising therapeutic target for managing BRAFV600E-positive human cancers.

HMGCS1 inhibition decreased intracellular acetoacetate levels, leading to reduced BRAF V600E -MEK1 binding and consequent MEK1 activation. We conclude that the ketogenic HMGCS1-HMGCL-acetoacetate axis may represent a promising therapeutic target for managing BRAF V600E -positive human cancers.
The importance of metabolic alterations in cancer has been increasingly recognized over the past decade. Cancer cells appear to coordinate bioenergetics (aerobic glycolysis), anabolic biosynthesis, and appropriate redox status to promote cancer cell proliferation and tumor growth (1,2). Interestingly, although increasing evidence suggests that different human cancers may share common metabolic properties, such as the Warburg effect, it is unclear whether distinct oncogene mutations including oncogenes and tumor suppressor genes (TSGs) may regulate and require distinct metabolic alterations for tumor development.
We approached this question by identifying a unique metabolic vulnerability required by distinct oncogenic mutations (3). We chose oncogenic BRAF V600E -positive melanoma as a platform. Melanoma is one of the most common human cancers. BRAF V600E mutation is found in Ն 60% of malignant melanomas (4 -6), as well as in 10% of colorectal cancer (7,8), 100% of hairy cell leukemia (HCL) (9), and 5% of multiple myeloma (10). The pathogenic importance of BRAF V600E in diverse human cancers makes it a promising therapeutic target in cancer treatment. However, although BRAF-mutant and MEK inhibitors have been demonstrated to be clinically effective in treating BRAF V600E -positive melanoma patients, resistance eventually develops that leads to cancer relapse of treated patients (4 -6). Thus, identification of co-dependent targets in BRAF V600E -positive melanomas may inform more successful long-term treatment strategies.
We designed a novel shRNA library that targets the majority of metabolism-related proteins in the human genome (3). Using this library, we identified synthetic lethal (SL) metabolic partners of oncogenic BRAF V600E in human melanoma cells, attenuation of which only inhibited BRAF V600E -expressing melanoma cells but not cells expressing NRAS mutant or wild type (WT) BRAF and NRAS (3). Top candidates include two key enzymes in the ketogenic pathway, HMG-CoA lyase (HMGCL) 3 and HMG-CoA synthase 1 (HMGCS1) (3).
Ketogenesis mainly occurs in the mitochondria of liver cells, which normally produces ketone bodies as a result of fatty acid breakdown to generate energy when glucose levels in the blood are low (11,12). ␤-oxidation breaks down fatty acids to form acetyl-CoA, which, under normal conditions, is further oxidized in the TCA cycle. However, if TCA cycle activity is low, or the acetyl-CoA generation rate of ␤-oxidation exceeds the capacity of the TCA cycle, ketogenesis will be activated to convert acetyl-CoA to ketone bodies via HMG-CoA. HMGCS converts acetyl-coA to HMG-CoA, whereas HMGCL converts HMG-CoA to acetyl-CoA and a ketone body, acetoacetate (AA), which can be further converted to two other ketone bodies, including D-␤-hydroxybutyrate (3HB) and acetone. Ketone bodies can be transported from the liver to other tissues, where acetoacetate and 3HB but not acetone will be further oxidized via the TCA cycle to produce acetyl-CoA for energy production. Organs, including the heart and brain, can use acetoacetate and 3HB for energy. Acetoacetate, if not used for energy, will be decarboxylated to acetone that is removed as waste (13,14).
We found that BRAF V600E activates a transcription factor, Oct-1, leading to HMGCL up-regulation in BRAF V600E -expressing human melanoma and hairy cell leukemia cells (3). HMGCL increases intracellular levels of acetoacetate that selectively promotes BRAF V600E (but not WT) binding to MEK1 and subsequent MEK1 activation, which represents a new "wiring" that only occurs in cancer cells expressing BRAF V600E but not normal cells (3). Moreover, we recently reported that a high-fat ketogenic diet resulted in increased serum levels of acetoacetate, leading to enhanced tumor growth potential of BRAF V600E -expressing human melanoma cells in xenograft mice. Treatment with hypolipidemic agents to lower circulating acetoacetate levels or an inhibitory homologue of acetoacetate, dehydroacetic acid (DHAA), to antagonize acetoacetate-BRAF V600E binding attenuated BRAF V600E tumor growth. These findings together reveal a signaling basis underlying a pathogenic role of dietary fat-fueled ketogenesis in BRAF V600E -expressing human cancer (15).
Here we demonstrate that HMGCS1 as a key ketogenic enzyme also functions as a synthetic lethal partner of BRAF V600E in human melanoma cells. There are two isoforms of HMGCS, including cytosolic HMGCS1 and mitochondrial HMGCS2 (16). Although mitochondrial HMGCS2 has been suggested to be the main regulatory site in ketogenesis, our results suggest that cytosolic HMGCS1 is important to regulate intracellular acetoacetate levels that enhance BRAF V600E -dependent MEK1 activation and consequent cell proliferation and tumor growth potential in BRAF V600E -positive melanoma.

HMGCS1 is a synthetic lethal partner of BRAF V600E in human melanoma cells
We previously identified HMGCS1 as a top synthetic lethal partner candidate of BRAF V600E , a systematic RNAi screen using human melanoma cells expressing BRAF V600E mutant and control BRAF WT-expressing melanoma cells (3). Future cell-based analyses revealed that stable knockdown of HMGCS1 resulted in more attenuated cell proliferation rates in BRAF V600D/E -expressing melanoma cells including A2058, A375, and WM-266-4 cells compared with control PMWK cells expressing BRAF WT as well as HMCB and SK-ME-2 cells expressing active NRAS Q61K or Q61R mutant, respectively, suggesting selective importance of HMGCS1 in BRAF V600Einduced melanoma transformation (Fig. 1A). Consistent with this finding, we found that stable knockdown of HMGCS1 in BRAF V600E -expressing human colon cancer WiDr and HT-29 cells also resulted in reduced cell proliferation rates, but not in control colon cancer HCT-116 cells (Fig. 1B).
Surprisingly, unlike HMGCL that is up-regulated by BRAF V600E in human melanoma cells, we found that HMGCS1 protein expression levels did not correlate with BRAF V600E mutational status in diverse human melanoma cell lines (Fig.  1C). In addition, BRAF V600E mutation did not affect mRNA levels of HMGCS1 in primary human tumor tissue samples from a group of BRAF V600E -positive and -negative melanoma patients (Fig. 1D). Consistent with these findings, gene expression levels of HMGCS1 were not up-regulated in human melanoma patients harboring distinct mutations at V600 including V600E, V600K, and V600R compared with patients expressing BRAF WT as assessed using TCGA Provisional melanoma dataset (Fig. 1E). These data suggest that the physiological levels of HMGCS1 in human melanoma cells are sufficient to fulfill the elevated ketogenic flux and consequent acetoacetate production because of increased HMGCL expression that is upregulated by BRAF V600E .

HMGCS1 signals through acetoacetate to promote cell proliferation and BRAF V600E -dependent MEK1 activation in BRAF V600E -expressing cells
We next sought to explore the underlying molecular mechanism. Previously we demonstrated that HMGCL product acetoacetate selectively promotes BRAF V600E but not BRAF WT binding to MEK1 and consequent MEK1 phosphorylation and activation (3). Similar as observed in HMGCL knockdown cells (3), silencing HMGCS1 by shRNA resulted in significantly decreased intracellular concentration of acetoacetate in diverse human melanoma ( Fig. 2A) and colon cancer cells (Fig. 2B), despite BRAF V600E status. However, stable knockdown of HMGCS1 selectively reduced cell proliferation rates in BRAF V600E -expressing melanoma A2058 and A375 cells but not in control NRAS Q61K-expressing HMCB cells (Fig. 2C), as well as in BRAF V600E -expressing colon cancer WiDr and HT-29 cells but not in control KRAS G13D-expressing HCT-116 cells (Fig. 2D). Addition of AA in the culture media significantly reversed the decreased cell proliferation rates only in BRAF V600E -expressing cells with stable HMGCS1 knockdown ( Fig. 2, C and D).
In addition, stable knockdown of HMGCS1 selectively attenuated phosphorylation of MEK1 and ERK1/2 only in BRAF V600E -expressing A2058 and A375 melanoma cells but not control HMCB cells, whereas AA treatment reversed such decreased phosphorylation and activation of MEK-ERK pathway because of HMGCS1 knockdown only in BRAF V600E -expressing melanoma (Fig. 3A) and colon cancer cells (Fig. 3B). Moreover, AA treatment resulted in increased MEK1 binding to BRAF V600E in A2058 and A375 cells, but not to BRAF WT in control HMCB cells, which was reversed by treatment with AA ( Fig. 3C). Similar results were obtained using colon cancer cells with HMGCS1 knockdown (Fig. 3D). These data together sug-gest that HMGCS1 plays an important role in ketogenesis-fueled BRAF V600E transformation in human melanoma and colon cancer cells through regulation of downstream HMGCL product acetoacetate to promote a BRAF V600E -activated MEK-ERK signaling cascade.

Cytosolic HMGCS1 but not mitochondrial HMGCS2 is selectively important for BRAF V600E cancer cells
To distinguish the different roles of the two HMGCS isoforms, cytosolic HMGCS1 and mitochondrial HMGCS2, we examined the expression levels of HMGCS1 and HMGCS2 in a group of human cancer cells. We found that HMGCS1 protein expression levels are comparable in all the tested melanoma and colon cancer cells despite BRAF V600E mutational status   (Fig. 3A). In contrast, HMGCS2 was only detected in BRAF V600E -expressing colon cancer WiDr and HT-29 cells but not in KRAS G13D-expressing HCT-116 colon cancer cells or NRAS mutant-expressing HMCB and BRAF V600E -expressing A375 and A2058 melanoma cells (Fig. 4A). In addition, we found that the relative HMGCS2 mRNA levels were approximately Ͼ 450-fold higher in colon cancer WiDr and HT-29 cells compared with HCT-116 cells and melanoma HMCB, A375, and A2058 cells, whereas the relative HMGCS1 mRNA levels were detectable in all of the tested cell lines (Fig. 4B). Consistent with these results, mitochondrial HMGCS2 protein expression was detected in WiDr cells and a control colon cancer Caco 2 cell line (Abcam), whereas the protein expression levels of HMGCS2 were much lower and almost undetectable in West- ern blot experiments using cell lysates from diverse human melanoma cell lines despite BRAF and NRAS mutational states (Fig. 4C).

HMGCS1 enhances MEK1 activation by BRAF V600E
We thus next examined the effect of stable knockdown of HMGCS2 in the two BRAF V600E -expressing colon cancer cell lines including WiDr and HT-29, which have relatively high levels of HMGCS2 protein expression. In contrast to HMGCS1 knockdown (Figs. 2 and 3), we found that HMGCS2 knockdown did not affect the intracellular levels of acetoacetate (Fig.  4D) or phosphorylation levels of MEK1 and ERK1/2 (Fig. 4E) in BRAF V600E -expressing human colon cancer cells.

HMGCS1 co-localizes with HMGCL and BRAF V600E in cytosol and regulates HMGCL protein stability
Consistent with these results, we found that HMGCS1 co-localized with HMGCL and BRAF in cytosolic fractions in human melanoma cells expressing BRAF WT or V600E (Fig. 5A). To determine the potential interplay between HMGCS1 and HMGCL in BRAF V600E -expressing cancer cells, we examined the effect of stable knockdown of one protein on the other. We found that stable knockdown of HMGCL did not affect the protein expression and mRNA levels of HMGCS1 in HMCB, A375, and A2058 melanoma cells despite BRAF V600E mutational status (Fig. 5B, left and right, respectively). Interestingly, stable knockdown of HMGCS1 resulted in increased protein levels of HMGCL (Fig. 5C, left) but with decreased HMGCL mRNA levels (Fig. 5C, right). These results suggest that attenuation of HMGCS1 leads to a compensating increase in HMGCL protein levels, which is likely involving alteration in protein stability rather than gene expression. Indeed, we found that treatment with cycloheximide (CHX) induced protein degradation of HMGCL in HMCB and A375 cells but not in cells with stable knockdown of HMGCS1 (Fig. 5D). Together, these data suggest that HMGCS1 knockdown caused a compensating  Figure 5. HMGCS1 co-localizes with HMGCL and BRAF V600E in cytosol and regulates HMGCL protein stability. A, subcellular localization of HMGCS1, BRAF, and HMGCL in cytoplasm, nucleus, and mitochondria in BRAF WT cells HMCB and BRAF V600E -positive cells A2058 and A375. Actin, PARP, and TOM40 were included as markers for cytoplasm, nucleus, and mitochondria, respectively. B and C, effect of stable knockdown of HMGCL (B) or HMGCS1 (C) on HMGCS1 or HMGCL protein expression (left panels) and mRNA levels (right panels), respectively. D, results of Western blot showing effect of cycloheximide (CHX) treatment on HMCGL1 in the melanoma cells in the presence or absence of stable knockdown of HMGCS1. increase in downstream HMGCL protein level because of attenuated protein degradation, which, however, could not reverse the reduced ketogenesis in cells with HMGCS1 knockdown.

HMGCS1 is important for transformation potential of BRAF V600E in human melanoma cells
We next examined whether HMGCS1 is required for BRAF V600E -induced transformation. We found that stable knockdown of HMGCS1 resulted in attenuated colony-formation potential of BRAF V600E -expressing A2058 and A375 cells but not control HMCB cells expressing NRAS Q61K mutant (Fig. 6A). Similar results were obtained using KRAS G13D-expressing HCT-116 and BRAF V600E -expressing WiDr and HT-29 colon cancer cells (Fig. 6B).
Moreover, in a xenograft nude mouse model, silencing HMGCS1 did not affect tumor growth rates, masses, or sizes of tumors harvested from xenograft nude mice injected with HMCB cells or cells with stable knockdown of HMGCS1 (Fig.  7A). In contrast, tumors derived from A375 cells with stable knockdown of HMGCS1 demonstrated decreased growth rates and masses with reduced tumor sizes compared with those derived from control A375 cells in xenograft nude mice (Fig.  7B). Mechanistic studies revealed that HMGCS1 knockdown resulted in decreased phosphorylation levels of MEK1 and ERK1/2 only in tumors derived from A375 cells but not HMCB cells in xenograft nude mice (Fig. 7C), despite the fact that reduced intracellular acetoacetate levels were observed in both A375 and HMCB xenograft tumors with stable knockdown of HMGCS1 compared with xenograft tumors derived from corresponding parental cells (Fig. 7D). These data together suggest a synthetic lethal interaction between BRAF V600E and HMGCS1 in human cancer cells.

Discussion
Our results suggest that a key ketogenic enzyme, HMGCS1, is specifically important for BRAF V600E -dependent transformation signaling in human melanoma cells. HMGCS1 regulates the intracellular levels of acetoacetate, the product of its downstream enzyme HMGCL in the ketogenesis pathway, to selectively promote BRAF V600E -MEK1 binding and consequent activation of MEK-ERK pathway. This is consistent with our previous findings that support a novel "feed-forward" model, where BRAF V600E activates a ketogenesis pathway by up-regulating HMGCL and subsequent intracellular levels of acetoacetate, which specifically binds to BRAF V600E mutant but not BRAF WT, leading to enhanced MEK1 and ERK1/2 activation in human melanoma cells (3). Interestingly, although mitochondrial HMGCS2 has been suggested to be the main regulator of ketone body production, it was not identified as one of the top candidates for BRAF V600E synthetic lethal partner (3), and consistently our results herein suggest that HMGCS2 is not selectively important for BRAF V600E cancer cell proliferation. Our results suggest that both HMGCS1 and HMGCL are localized in cytosol with BRAF proteins in human melanoma cells expressing either BRAF WT or V600E mutant, suggesting that ketogenesis and acetoacetate production in cytosol are more critical than those in mitochondria for BRAF V600E -dependent transformation in human melanoma cells.
Our results demonstrate that although oncogenic BRAF V600E mutant "rewires" metabolic and cell signaling networks in human cancer cells by up-regulating HMGCL subsequent intracellular acetoacetate levels through activation of transcription factor Oct-1, the expression levels of HMGCS1 do not correlate with BRAF V600E mutational status in human melanoma cells, suggesting that HMGCS1 is not directly involved in the process by which oncogenic BRAF V600E rewires the ketogenesis pathway to MEK-ERK signaling. However, physiological activity of HMGCS1 plays an important role in such a rewiring and is sufficient to cooperate with up-regulated HMGCL by BRAF V600E to achieve elevated production of acetoacetate that enhances BRAF V600E transformation potential in human melanoma cells. Interestingly, knockdown of HMGCS1 appears to cause a compensating increase in downstream HMGCL protein level because of attenuated protein degradation through an unknown mechanism, which, however, could not reverse the reduced ketogenesis in cells with HMGCS1 knockdown. Future studies are warranted to explore the molecular and signaling basis underlying HMGCS1-dependent regulation of protein stability in cells.
Our findings also suggest that the ketogenic HMGCS1-HMGCL-acetoacetate axis represents a promising therapeutic target in treatment of BRAF V600E -positive human cancers. Indeed, in our recent report, DHAA functions as an inhibitory derivative of acetoacetate by binding to BRAF V600E with a higher affinity that enables DHAA to compete with acetoacetate for BRAF V600E binding, which selectively inhibits cell proliferation and tumor growth potential of melanoma cells expressing BRAF V600E but not melanoma cells expressing wild type BRAF (15). Thus, future anti-ketogenic drugs such as small molecule inhibitors of HMGCS and HMGCL, as well as DHAA and its next generation derivatives, may represent alternative clinical treatments for patients with BRAF V600E -positive cancer. Furthermore, these findings provide insights into the development of a precision diet (15), which should be designed based on individual genetic background to lower cancer risk and provide cancer prevention.

Cell culture
PMWK, A2058, and A375 cells were cultured in Dulbecco's modified Eagle's medium (DMEM) with 10% fetal bovine serum (FBS). HMCB, WM-266-4, and SK-MEL-2 cells were cultured in Eagle's minimal essential medium (EMEM) with 10% FBS. AA treatment was performed by incubating 10 mM AA for about 12 h. For the cell proliferation assay, 2 mM AA were added in the cell culture medium.

Acetoacetate extraction and measurement
To determine the cellular concentration of AA, 100 l of packed cell pellets were homogenized in 500 l of lysis buffer (10 mM HEPES, pH 7.0, 150 mM NaCl, 1% Nonidet P-40, 5 mM NaF, 5 mM sodium pyrophosphate and protease inhibitor mixture). The cell lysates were filtered with 10-kDa Amicon® Ultra-4 Centrifugal Filters (Millipore). The flow through containing the metabolites was used for AA measurement according to the instructions for the Acetoacetate Assay Kit (BioVision).

Cell proliferation
Cell proliferation assays were performed by seeding 5 ϫ 10 4 cells in a 6-well plate and culturing the cells at 37°C in 5% CO 2 . Cell proliferation was determined by cell numbers recorded at 1 to 6 days after being seeded and normalized to that of each of the cell lines at the starting time (T ϭ 0 h) by trypan blue exclusion using a TC10 Automated Cell Counter (Bio-Rad).

Colony-forming assay
Melanoma cell lines transduced with lentivirus harboring pLKO.1 or shHMGCS1 (300 cells) were seeded in 6-well plates for monolayer colony formation. After 2 to 3 weeks, colonies were stained with 0.5% crystal violet, counted microscopically, and photographed.

Co-immunoprecipitation assays
For co-immunoprecipitation, cells were harvested in 1 ml Nonidet P-40 buffer as described previously. 1-mg cell lysates were incubated with anti-BRAF antibody overnight at 4°C. After incubation, protein G-Sepharose beads were used for precipitation. The beads were washed and eluted with SDS sample buffer for immunoblotting analysis.

Cell fractionation
Nuclear and cytoplasmic extraction was performed according to NE-PER Nuclear and Cytoplasmic Extraction Kit instructions (Pierce). Mitochondrial subfractionation was performed according to Mitochondria Isolation Kit instructions (Thermo Scientific).

Xenograft studies
Approval of use of mice and designed experiments was given by the Institutional Animal Care and Use Committee of Emory University. Nude mice (nu/nu, female 4 -6 weeks old) (Harlan Laboratories) were subcutaneously injected with 2 ϫ 10 6 melanoma A375 or HMCB cells harboring empty vector on the left flank, and cells with stable knockdown of endogenous HMGCS1 on the right flank. Tumor sizes were measured every 2 days by using Vernier calipers. Tumor growth was recorded by measurement of two perpendicular diameters using the formula 4/3 ϫ (width/2) 2 ϫ (length/2). The tumors were harvested and weighed at the experimental end point, and the masses of tumors (mg) derived from cells with or without stable knockdown HMGCS1 were compared. Statistical analyses were optioned by two-way analysis of variance (ANOVA). No blinding was done for these animal studies.

Study approval
Approval of use of human specimens was given by the Institutional Review Board of Emory University School of Medicine. All clinical samples were obtained with informed consent with approval by the Emory University Institutional Review Board. Clinical information for the patients was obtained from the pathology files at Emory University Hospital under the guidelines and with approval from the Institutional Review Board of Emory University School of Medicine and according to the Health Insurance Portability and Accountability Act. Tissues were harvested for total RNA extraction.

Statistics
In studies in which statistical analyses were performed, a 2-tailed Student's t test was used to generate p values. p values less than or equal to 0.05 were considered significant. A twoway ANOVA was used for tumor volume and tumor weight.

Reproducibility of experiments
The results of one representative experiment from at least two independent experiments are shown except data obtained from primary patient samples and animal experiments shown in Figs. 1D and 7, respectively. There is no estimate of variation in each group of data and the variance is similar between the groups. No statistical method was used to predetermine sample size. The experiments were not randomized. The investigators were not blinded to allocation during experiments and outcome assessment. All data are expected to have normal distribution.