Preventing BRCA1/ZBRK1 repressor complex binding to the GOT2 promoter results in accelerated aspartate biosynthesis and promotion of cell proliferation

Breast cancer susceptibility gene 1 (BRCA1) has been implicated in modulating metabolism via transcriptional regulation. However, direct metabolic targets of BRCA1 and the underlying regulatory mechanisms are still unknown. Here, we identified several metabolic genes, including the gene which encodes glutamate‐oxaloacetate transaminase 2 (GOT2), a key enzyme for aspartate biosynthesis, which are repressed by BRCA1. We report that BRCA1 forms a co‐repressor complex with ZBRK1 that coordinately represses GOT 2 expression via a ZBRK1 recognition element in the promoter of GOT2. Impairment of this complex results in upregulation of GOT2, which in turn increases aspartate and alpha ketoglutarate production, leading to rapid cell proliferation of breast cancer cells. Importantly, we found that GOT2 can serve as an independent prognostic factor for overall survival and disease‐free survival of patients with breast cancer, especially triple‐negative breast cancer. Interestingly, we also demonstrated that GOT2 overexpression sensitized breast cancer cells to methotrexate, suggesting a promising precision therapeutic strategy for breast cancer treatment. In summary, our findings reveal that BRCA1 modulates aspartate biosynthesis through transcriptional repression of GOT2, and provides a biological basis for treatment choices in breast cancer.

Breast cancer susceptibility gene 1 (BRCA1) has been implicated in modulating metabolism via transcriptional regulation. However, direct metabolic targets of BRCA1 and the underlying regulatory mechanisms are still unknown. Here, we identified several metabolic genes, including the gene which encodes glutamateoxaloacetate transaminase 2 (GOT2), a key enzyme for aspartate biosynthesis, which are repressed by BRCA1. We report that BRCA1 forms a co-repressor complex with ZBRK1 that coordinately represses GOT2 expression via a ZBRK1 recognition element in the promoter of GOT2. Impairment of this complex results in upregulation of GOT2, which in turn increases aspartate and alpha ketoglutarate production, leading to rapid cell proliferation of breast cancer cells. Importantly, we found that GOT2 can serve as an independent prognostic factor for overall survival and disease-free survival of patients with breast cancer, especially triple-negative breast cancer. Interestingly, we also demonstrated that GOT2 overexpression sensitized breast cancer cells to methotrexate, suggesting a promising precision therapeutic strategy for breast cancer treatment. In summary, our findings reveal that BRCA1 modulates aspartate biosynthesis through transcriptional repression of GOT2, and provides a biological basis for treatment choices in breast cancer.

Introduction
Metabolic reprogramming is a well-recognized hallmark of cancer (Hanahan and Weinberg, 2011). To sustain rapid proliferation and survive in a nutrientpoor environment, tumor cells have evolved a sophisticated adaptive metabolic signaling to generate ATP as energy source, maintain redox balance and commit resources to biosynthesis, including increased glycolysis, glutaminolysis and lipid synthesis (Hirschey et al., 2015;Martinez-Outschoorn et al., 2017). Moreover, tumor cells arising in different tissues or even within the same tissue have different metabolic profiles, even when they have identical genetic lesions, suggesting that tumor cells rewire their metabolic pathways to adapt to different tissue environments or tumor microenvironments (TME) (Anastasiou, 2017;Danhier et al., 2017;Munoz-Pinedo et al., 2012;Yuneva et al., 2012). Although metabolic variability and flexibility support tumor progression, they also render tumor cells more susceptible to perturbations within the metabolic network. However, the dynamic regulatory processes and heterogeneity of tumor cell metabolism are far from being understood. Thus, investigating the metabolic dependency of tumor cells may open a promising new avenue for therapeutic intervention.
Breast cancer susceptibility gene 1 (BRCA1) is a major breast cancer suppressor gene, which is most frequently mutated in hereditary breast cancer and is often downregulated in sporadic breast cancer (Miki et al., 1994;Yarden and Papa, 2006). The BRCA1 protein is a pluripotent regulator of cellular functions in breast cancer, including DNA double-strand break repair, cell cycle control, transcriptional regulation, ubiquitination, apoptosis and resistance to anticancer agents (Silver and Livingston, 2012). There is growing evidence linking BRCA1 to metabolic reprogramming. BRCA1 has been shown to interact with many key metabolic regulators, including p53, Myc, Akt (Xiang et al., 2008), HIF-1a (van der Groep et al., 2008;Kang et al., 2006), Oct1 (Vazquez-Arreguin et al., 2018) and acetyl-CoA carboxylase (Moreau et al., 2006). Maud Privat and colleagues have revealed that BRCA1 inhibited glycolysis while activating TCA cycle and oxidative phosphorylation through global transcriptional and metabolite profiling (Privat et al., 2014). Except for these correlations, the casual roles and underlying molecular mechanisms of BRCA1 in metabolic adaptation are still unclear, especially for the transcriptional regulation functions of BRCA1 in metabolism. Because BRCA1 lacks the ability to recognize these regulatory sequences, it must associate with sequence-specific binding transcription factors to execute its transcriptional function (Yun and Lee, 2003). Zinc finger and BRCA1-interacting protein with KRAB domain-1 (ZBRK1, also known as ZNF350), which was first identified in a yeast two-hybrid screening for proteins associated with BRCA1, is a typical KRABcontaining zinc finger protein that contains a highly conserved KRAB domain at the NH 2 terminus, and a CTRD domain for BRCA1 interactions at the COOH terminus (Chen et al., 1996). Its zinc finger repeats were suggested to recognize a consensus DNA-binding element, GGGxxxCAGxxxTTT, or to be involved in protein interactions (Bellefroid et al., 1991). BRCA1 transcriptionally repressed the expression of several genes by forming a complex with ZBRK1, which including GADD45A (Zheng et al., 2000), HMGA2 (Ahmed et al., 2010), MMP9 (Lin et al., 2010) and ANG1 (Furuta et al., 2006). However, whether ZBRK1 plays a direct role in BRCA1-associated metabolism reprogramming has yet to be elucidated.
In the present study, we sought to determine the metabolic target genes of BRCA1/ZBRK1 transcriptional repressor complex. To this end, by integrating mRNA microarray profiles and bioinformatics analysis, we identified glutamate-oxaloacetate transaminase 2 (GOT2), a key enzyme for aspartate production, as a direct transcriptionally repressed target of BRCA1/ ZBRK1. Our findings revealed that BRCA1/ZBRK1 complex could have a function in aspartate metabolism, which was partially dependent on transcriptional regulation of GOT2. Importantly, GOT2 was overexpressed in breast cancer (BC), especially in triple-negative breast cancer (TNBC), and was significantly correlated with poor survival of patients with these diseases. Our study also demonstrated that GOT2 was crucial for breast cancer progression and could serve as a predictor of methotrexate (MTX) sensitivity.

Materials and methods
2.1. Analyzing the mRNA microarray data and screening consensus ZBRK1 DNA binding element Previous mRNA microarray profiles of BRCA1-deleted mouse embryonic fibroblast MEF-BRCA1 M/M and the wild-type counterparts (MEF-BRCA1 +/+ ) were used for analysis, specifically focusing on metabolic enzymes and transporters. The differential expression genes were defined by the relative expression fold change (MEF-BRCA1 M/M vs MEF-BRCA1 +/+ ) > 2 or < 0.5.
For consensus ZBRK1 DNA binding element screening, 2000-bp upstream sequences of the transcriptional start site of each metabolic gene were downloaded from UCSC Genome Browser (https://ge nome.ucsc.edu/cgi-bin/hgTables). Consensus ZBRK1 DNA binding element GGxxxCAGxxxT was searched for in these sequences by R software.

Cell lines and cell culture
Human breast cancer cell lines MDA-MB-231, MCF-7, CAL-51, MDA-MB-468, human embryonic kidney cell line HEK293T and mouse embryonic fibroblast MEF-BRCA1 M/M and the wild-type counterparts MEF-BRCA1 +/+ (kindly provided by Professor Chuxia Deng of National Institute of Diabetes, Digestive, and Kidney Diseases) were maintained in Dulbecco's modified Eagle's medium (DMEM, Thermo Fisher Scientific, New York, NY, USA) supplemented with 10% FBS. For aspartate or alpha ketoglutarate (a-KG) rescue assay, cells were treated with 10 lM aspartate or 2 mM a-KG for the indicated times. To inhibit transaminases, cells were treated with 1 mM aminooxyacetic acid hemihydrochloride (AOA) from Med-ChemExpress (HY-107994) in complete medium for 48 h. All of these cells were maintained at 37°C with 5% CO 2 .

RNA isolation and quantitative real-time PCR
Total RNA of breast cancer cells was extracted with TRIzol reagent (Invitrogen, Carlsbad, CA, USA). The cDNA was synthesized with the PrimeScript RT reagent Kit (Promega, Madison, WI, USA). Real-time PCR was carried out using an ABI 7500 real-time PCR system (Applied Biosystems, Foster City, CA, USA). The mRNA expression level of GOT2 and ASS1 were normalized to the endogenous expression of GAPDH. Primers were provided by Invitrogen as described below: GAPDH forward, 5 0 -TCTCTGCTCCTCCTGTTC-3 0 , GAPDH reverse, 5 0 -GTTGACTCCGACCTTCAC-

Plasmid construction, transfection and lentivirus infection
Plasmid pcDNA3.1-BRCA1 and pcDNA3.1-ZBRK1 (also known as ZNF350) were constructed by inserting the corresponding human cDNA into the HindIII/XhoI sites of pcDNA3.1 vector. The region À929 to À130 of GOT2 promoter, which containing the consensus ZBRK1 DNA binding element (ZBE) GGGaggCA-GaggTTG, was inserted in the NheI/XhoI sites of pGL3-basic plasmid, hereafter termed GOT2 promoter Wt. The mutant of GOT2 promoter (GOT2 promoter Mut) was constructed by deleting the ZBE of GOT2 promoter Wt. After constructing pLVX-GOT2 by inserting the corresponding human cDNA into the EcoRI/BamHI sites of pLVX-puro vector, pLVX-GOT2 was packaged into the lentivirus system with pMD2.G and pSPAX2 vectors using HEK293T cells. Lentiviral shRNA of GOT2 were purchased from GeneChem Co. Ltd. Shanghai, China. The GOT2 overexpression or shRNA lentiviruses were infected into cells using polybrene (8 lgÁmL À1 ) to enhance infection efficiency. Puromycin (1 lgÁmL À1 ) was added to the cells for 72h to screen for stably infected cells.

ChIP and ChIP/Re-ChIP assay
ChIP and ChIP/Re-ChIP were performed in MCF-7 cells with a Pierce TM Agarose ChIP Kit (Thermo Fisher Scientific, New York, NY, USA) according to the manufacturer's instructions and using the protocol of Zhang et al. (2004,2006,2007) as reference. ChIP experiments were conducted in MCF-7 cells with antibodies against BRCA1 or ZBRK1, and antibody against RNA Polymerase II as positive control. ChIP/Re-ChIP experiments were conducted in MCF-7 cells with antibody against BRCA1 in the first round ChIP assay; the eluted DNA was then applied to Re-ChIP assay with antibody against ZBRK1, and antibody against RNA Polymerase II as positive control. The enrichment of the DNA template was analyzed by real time PCR using primers: forward: 5 0 -GGAGGCAGAGGTTGTAGT-3 0 and reverse: 5 0 -CACGTCGTTCCTGAATGT-3 0 specific for GOT2 gene promoter, and primers: forward: 5 0 -CATG GGTGTGAACCATGAGA-3 0 and reverse: 5 0 -GTCTT CTGGGTGGCAGTGAT-3 0 specific for GAPDH gene promoter.

Aspartate, a-KG and mitochondrial NADH/NAD + redox measurement
Transfected cells (after 48 h) or MEF-BRCA1 M/M and MEF-BRCA1 +/+ cells were dissociated and counted. The intracellular levels of aspartate and a-KG were examined by Aspartate Assay Kit (Abcam) and Alpha Ketoglutarate Assay Kit (Abcam). Briefly, 2 9 10 6 cells were homogenized in PBS and the supernatant collected. The flow-through containing the metabolites was used for the measurement of aspartate and a-KG, following the manufacturer's instructions. The abundance of aspartate and a-KG were normalized to the cell number. For mitochondrial aspartate, a-KG and NADH/NAD + redox detection, mitochondria were separated from 5 9 10 6 transfected cells or MEF-BRCA1 M/M and MEF-BRCA1 +/+ cells via commercial kits (Beyotime, Shanghai, China), which has been reported to separate mitochondria from cytosol (Wang et al., 2016a). The mitochondria were homogenized in PBS and the supernatant was collected. The flowthrough containing the metabolites was used for the measurement of aspartate, a-KG and NADH/NAD + redox (Beyotime), following the manufacturer's instructions.

Cell proliferation and colony formation assay
Cell proliferation was measured by MTS assay kit. Briefly, 3000 cells were seeded in each well of a 96-well plate and at least four duplicate wells were set for each cell line. From five duplicate 96-well plates, the relative cell viability by MTS assay kit was measured according to the manufacturer's instructions every day. The relative cell viability (A 450nm ) of Day 0 was measured after cell adherence, and the A 450nm of subsequent day points was normalized to the Day 0.
For colony formation assay, 1000 cells were seeded onto 6-well culture plates and incubated for about 10 days. Culture plates were performed in duplicate. After washing with pre-cooled PBS, wells were fixed with methanol for 5 min and stained with crystal violet for 10 min. Colonies were examined and automatically calculated by G:BOX (Syngene).

Ethics statement, tissue specimens and clinicopathological characteristics
Tissue microarrays (TMA) of breast cancer (BC) specimens were obtained from Shanghai Outdo Biotech Co., Ltd. (SOBC; Shanghai, China), with the approval of the Institutional Review Board. The clinicopathological characteristics of all specimens are summarized in Table S5. The samples were collected by surgery in 2001-2004 and were followed up until 2013. An additional 48 paraffin-embedded TNBC tissue samples of surgical biopsies were obtained, and histopathologically and clinically diagnosed at, the Cancer Institute and Hospital, Chinese Academic of Medical Sciences & Peking Union Medical College (Table S6). Written informed consent was obtained from all patients prior to the study. The use of the clinical specimens for research purposes was approved by the Institutional Research Ethics Committee.

Immunohistochemistry
Immunohistochemistry (IHC) analysis was performed and diagnosed by two pathologists blindly on the 130 paraffin-embedded BC tissue sections (Tissue Microarray, BC TMA cohort), 90 matched adjacent normal tissues (Tissue Microarray) and the 48 TNBC tissue samples. In brief, the sections were deparaffinized with xylene and rehydrated in graded ethanol. Sections were submerged into EDTA antigenic retrieval buffer (pH 8.0) and microwaved for antigenic retrieval. The sections were then treated with 3% hydrogen peroxide in methanol to quench the endogenous peroxidase activity, followed by incubation with 1% goat serum albumin to block nonspecific binding. The tissue sections were incubated with rabbit anti-BRCA1 (1 : 200; Cell Signaling), ZBRK1 (1 : 200, Abcam) and GOT2 (1 : 200, Abcam) overnight at 4°C. After washing, the tissue sections were treated with goat anti-mouse/rit IgG HRP-polymer (Dako) for 60 min. 3,3 0 -Diaminobenzidine was used as the chromogen.

MTX treatment assay
GOT2 stable knockdown cell line CAL-51 shGOT2-1/ CAL-51 shGOT2-2 and their counterparts, GOT2 stable overexpression cell line MDA-MB-468 GOT2 and its counterpart, were used for MTX treatment assay. Briefly, 3000 cells were seeded in each well of a 96-well plate and at least three duplicate wells were set for each cell line. After 12 h at 37°C with 5% CO 2 , cells were treated with different concentrations (10, 5, 1, 0.1, 0.01 and 0 lM) of MTX and returned for further incubation for 72 h. MTS assay was used to measure the relative cell viability.

Statistical analysis
Statistical analysis was carried out using IBM SPSS Statistics 19 (New York, NY, USA) or GRAPHPAD PRISM 5 (San Diego, CA, USA) for Windows. Twotailed Student's t-test was used to analyze those results, expressed as mean AE SEM. The two-tailed Pearson chi-square test was used to analyze the association of different molecular expression and clinicopathological parameters. Correlation was determined by Pearson's correlation coefficient test or Spearman's rank correlation coefficient analysis. The survival curves were plotted using Kaplan-Meier analysis and compared by log-rank test. Survival data were also evaluated by multivariate Cox regression analysis. COMPUSYN software was used to calculate the IC 50 . Differences were considered significant when the Pvalue was less than 0.05.

BRCA1/ZBRK1 potentially regulates cell metabolism
To determine the metabolic pathways regulated by BRCA1, we reanalyzed our previous mRNA microarray profiles of BRCA1-deleted mouse embryonic fibroblast MEF-BRCA1 M/M and the wild-type counterparts (MEF-BRCA1 +/+ ), with a specific focus on metabolic enzymes and transporters. As expected, a subset of metabolic genes was perturbed by BRCA1 deletion (Fig. 1A, Table S1). Pathway enrichment analysis by GENECODIS3 (Tabas-Madrid et al., 2012) according to the differential expressed genes revealed that 51   Table S3). These findings further confirmed that BRCA1 deficiency could induce metabolism reprogramming.
Given that BRCA1 often serves as a transcriptional co-repressor through physical interaction with ZBRK1, we sought to determine whether those genes upregulated in MEF-BRCA1 M/M cells harbored consensus ZBRK1 DNA-binding element. Because a previous study has verified that the strict consistency of ZBRK1 DNA-binding element was not required (Zheng et al., 2000), we searched the promoter sequences of BRCA1repressed genes with a comparatively strict ZBRK1 DNA-binding element (GGxxxCAGxxxT); 21 of 51  promoters contained this element (Fig. 1D, Tables S1 and S4). These data indicated that ZBRK1 may be a major mediator in BRCA1-induced metabolic rewiring. Since the genome is different between human and mouse, we next attempted to determine whether ZBRK1 was also harbored in human genes. Among these candidate BRCA1/ZBRK1-regulated genes, 19 promoters of human genes contained the GGxxxCAGxxxT elements (Fig. 1D, Table S4), suggesting that these metabolic genes were also regulated by BRCA1/ZBRK1 in human cells.
As BRCA1 and ZBRK1 deficiency contributed to the progression of breast cancer, we supposed that this effect might be partially dependent on its functions in metabolism regulations. Thus, we evaluated the prognostic association of these 19 BRCA1/ ZBRK1 candidate repressed genes in breast cancer through the BreastMark database (http://glados.ucd. ie/BreastMark/mRNA_custom.html) (Madden et al., 2013). Interestingly, the mRNA levels of six genes (PGK1, PAICS, GOT2, MGST1, ALDOC and PECR) were significant positively associated with both poor overall survival (OS) and disease-free survival (DFS) ( Table S4). These genes were involved in glycolysis (PGK1 and ALDOC), glutathione metabolism (MGST1), aspartate metabolism (GOT2), purine metabolism (PAICS) and peroxisomal lipid metabolism (PECR), indicating that BRCA1/ZBRK1 might play a role in remodeling these metabolic processes through transcriptional regulation of the expression of these genes. We were particularly intrigued by GOT2, a key enzyme of malate-aspartate shuttle, which was critical for cell redox balance and aspartate production. Aspartate is required for rapid cell growth, which is the most important intermediate for protein, purine nucleotide and pyrimidine nucleotide synthesis (Chen et al., 1996;Sullivan et al., 2015). Dysfunction of BRCA1 is correlated with accelerated growth and progression of breast tumors (Stoppa-Lyonnet et al., 2000;Xu et al., 1999). Maintaining redox balance and committing resources to biosynthesis should be essential for the proliferation of BRCA1-deficient breast cancer cells. Therefore, we focused our interest on BRCA1/ZBRK1-mediated aspartate metabolism reprogramming through GOT2 in the present study. The other metabolic pathways are currently being studied by our group.

BRCA1/ZBRK1 suppresses GOT2 expression
Aspartate is a critical metabolic intermediate for protein, purine nucleotide and pyrimidine nucleotide synthesis and is required for cell proliferation (Sullivan et al., 2015). We queried whether BRCA1 deficiency could perturb aspartate metabolism. Indeed, the key aspartate biosynthesis enzyme GOT2 and the argininosuccinate synthase 1 (ASS1) were significantly upregulated in BRCA1-deficient MEF-BRCA1 M/M cells, indicating that the aspartate metabolism was tightly regulated by BRCA1 ( Fig. 2A,B). Interestingly, the promoter of GOT2, the key enzyme in this pathway, harbored a consensus ZBRK1-DNA binding element (Fig. 2C), whereas there was no consensus ZBRK1-DNA binding element found in the promoter of ASS1, suggesting GOT2 might be coordinately regulated by BRCA1/ZBRK1 repressor complex. To verify this, we first confirmed that BRCA1 indeed physically interacted with ZBRK1 in mouse MEF-BRCA1 +/+ cells and human breast cancer cell lines MCF-7 (Fig. 2D). GOT2 and ASS1 were shown to be significantly upregulated in MEF-BRCA1 M/M cells compared with their counterparts in mRNA microarray assay, and this observation was also confirmed by real-time PCR (Figs 2B,E and S1A). Knockdown of BRCA1 increased, whereas overexpression of BRCA1 decreased the expression of GOT2 and ASS1 at the mRNA and protein level (Figs 2F and S1B). Concordantly, ZBRK1 exerted a similar effect on GOT2; however, knockdown of ZBRK1 did not affect the mRNA or protein levels of ASS1 (Fig. S1C), suggesting that BRCA1 might transcriptionally regulate the expression of ASS1 via an indirect mechanism or other transcription factor other than ZBRK1. Collectively, these findings suggest that BRCA1/ZBRK1 plays a role in repressing GOT2 expression.
3.3. BRCA1 and ZBRK1 form a repressor complex occupied on GOT2 promoter BRCA1 and ZBRK1 have been shown to form a repressor complex to suppress the expression of target genes. To determine a potential transcriptional regulation of GOT2 by BRCA1/ZBRK1, we first performed chromatin immunoprecipitation (ChIP) to estimate the physical occupancy of BRCA1 and ZBRK1 within the promoter region of GOT2. ChIP experiments were conducted in MCF-7 cells with antibodies against BRCA1 or ZBRK1, and antibody against RNA polymerase II as positive control (Fig. S2). As expected, BRCA1 and ZBRK1 were both co-precipitated with the promoter of GOT2 (Fig. 3A). In addition, ChIP/ Re-ChIP assay (Zhang et al., 2004) confirmed that BRCA1 and ZBRK1 exist in the same protein complex on the GOT2 promoter (Fig. 3B). These experiments not only support the idea that GOT2 is targeted by the BRCA1/ZBRK1 complex but also confirm that  BRCA1 is physically associated with and is an integral component of ZBRK1.
To examine further the transcriptional activity of BRCA1/ZBRK1 on GOT2, we measured the luciferase reporter activity of GOT2 promoter constructs with wild-type (contained the consensus ZBRK1-DNA binding element) or mutant (with deletion of the consensus ZBRK1-DNA binding element, Fig. 3C) after ectopic overexpression of BRCA1 or ZBRK1 (Fig. 3D,E) in HEK293T cells. Only the wild-type luciferase reporter construct showed a significant decrease in the activity as the expression of BRCA1 or ZBRK1 gradually increased (Fig. 3D); the mutant construct did not show any significant changes (Fig. 3E). These results imply that this region (À548 to À562) is essential for transcriptional repression by BRCA1/ZBRK1.

BRCA1/ZBRK1 modulates aspartate metabolism through repression of GOT2 expression
Given that GOT2 converts oxaloacetate (OAA) and L-glutamate (Glu) to generate L-aspartate (Asp) and 2-oxoglutarate (also known as a-ketoglutarate), we next investigated whether BRCA1/ZBRK1 could modulate aspartate metabolism through transcriptionally repressing GOT2. It has been reported that GOT2, GPT2 and GDH1 could catalyze glutamate to generate a-KG (Jin promoter Wt, À929 to À130 region with wild-type consensus ZBRK1 DNA binding element (ZBE). Bottom, GOT2 promoter Mut, À929 to À130 region with deletion of ZBE. (D) Relative luciferase activity of reporter constructs GOT2 promoter Wt in HEK293T cells transfected with indicated concentrations of BRCA1 or ZBRK1 plasmids for 48 h. Top, Quantification of relative luciferase activity. Bottom, western blot analysis of ectopic expression of BRCA1 and ZBRK1. (E) Relative luciferase activity of reporter constructs GOT2 promoter Wt and Mut in HEK293T cells transfected with BRCA1 or ZBRK1 plasmids for 48 h. Top, Quantitative of the relative luciferase activity. Bottom, western blot analysis of ectopic expression of BRCA1 and ZBRK1. Two-tailed Student's t-test was used to evaluated the difference; error bars represent the mean AE SEM from three independent experiments. ***P < 0.001; **P < 0.01; *P < 0.05; ns, not significant. ; we therefore sought to determine which was predominantly responsible for BRCA1 deficiency. Interestingly, BRCA1 dysregulation did not affect the protein levels of GDH1 and GPT2 in MEF-BRCA1 M/M cells compared with the wild-type counterparts (Fig. S3A). Consistently, knockdown BRCA1 or ZBRK1 in MDA-MB-231 cells only increased the protein level of GOT2, not GDH1 or GPT2 (Fig. S3B). We then examined the intracellular and mitochondrial aspartate and a-KG levels in MEF-BRCA1 M/M and MEF-BRCA1 +/+ cells. As expected, intracellular and mitochondrial aspartate and a-KG levels were significantly increased in MEF-BRCA1 M/M cells compared with MEF-BRCA1 +/+ cells (Fig. 4A,B). Additionally, the mitochondrial NADH/ NAD + redox was significantly elevated in MEF-BRCA1 M/M cells compared with MEF-BRCA1 +/+ cells (Fig. 4C).To uncover further the underlying mechanisms of BRCA1/ZBRK1 modulation of aspartate biosynthesis, we simultaneously knocked down BRCA1 with GOT2 or ZBRK1 with GOT2 in MCF-7 and MDA-MB-231 cells, and evaluated the intracellular and mitochondrial Asp and a-KG levels and NADH/NAD + redox 48 h after RNAi transfection. As expected, knockdown of GOT2 in BRCA1 or ZBRK1 downregulated cells reduced the intracellular and mitochondrial aspartate and a-KG levels compared with BRCA1 or ZBRK1 single knockdown cells, although they did not reach to the level of the control cells (Figs 4D-G and S4A-C). Concordantly, the mitochondrial NADH/NAD + redox was decreased as well (Fig. 4H,I). This result suggested that the BRCA1/ZBRK1-GOT2 axis accounted for the Asp and a-KG production partially and that there another metabolic signaling may be regulated by BRCA1/ ZBRK1.

GOT2 mediates metabolic adaptation to promote malignant phenotypes of breast cancer cells
Given that aspartate is a critical metabolite for cell proliferation and biomass production, we further asked whether GOT2 could promote the malignant phenotypes of breast cancer cells through mediating aspartate biosynthesis. Interestingly, analysis using CELLMINER (http://discover. nci.nih.gov/cellminer) revealed that GOT2 mRNA expression level was negatively associated with cell doubling time of NCI-60 cell lines (Pearson r = À0.521, P < 0.001) or breast cancer cell lines (Pearson r = À0.770, P = 0.128; Fig. S5A). For further validation, GOT2 was knocked down or overexpressed via lentiviral infection. We observed that downregulation of GOT2 significantly attenuated, and upregulation of GOT2 promoted, cell proliferation and colony formation of breast cancer cells ( Fig. 5A-E). We treated the GOT2 ectopically overexpressed cell line MDA-MB-468 and the control cells with aminooxyacetate (AOA), an inhibitor of transaminases, to inhibit the malate-aspartate shuttle and measured the cell proliferation by MTS assay after 48 h. The results showed that AOA abolished the cell proliferation promotion capacity of GOT2 (Fig. 5F). Given that GOT2 contributes to a portion of aspartate and a-KG generation, we asked whether adding aspartate could restore the capacities of cell proliferation and colony formation upon GOT2 elimination. Indeed, aspartate addition partially restored the cellular behaviors mentioned above in the context of GOT2 knockdown (Fig. 5G,H). Additionally, we treated GOT2 stable knockdown cell lines CAL-51 and MCF-7 and their control cells with a-KG and found that a-KG could partially rescue the growth suppression by GOT2 knockdown as well (Fig. 5B,C). Taken together, our results revealed that GOT2 could partially evoke malignant phenotypes of breast cancer cells via its catalytic function.

BRCA1/ZBRK1-GOT2 axis is associated with clinicopathological characteristics of breast cancer
We then validated the association between BRCA1/ ZBRK1 and GOT2 in BC TMA and another cohort of TNBC tissue samples using immunohistochemistry (IHC) assay. The results revealed that the protein levels of both ZBRK1 and BRCA1 were significantly negative compared with GOT2 in the BC TMA cohort (Spearman's coefficient test, rho = À0.283, P = 0.001 and rho = À0.190, P = 0.038, respectively) (Fig. 6A,B) and TNBC cohort (Spearman's coefficient test, rho = À0.253, P = 0.029 and rho = À0.384, P = 0.001, respectively) (Fig. S6A). The protein level of GOT2 was significantly higher in the BC tissue samples than in the adjacent normal tissue samples (Student's test, P < 0.001; Fig. 6C). Additionally, the there was a substantially higher frequency of high protein level of GOT2 in the TNBC cohort (73.33%, 55/75) than in the BC cohort (59.69%, 77/129). We next explored the association between GOT2 protein level and different clinicopathological features of patients with BC. In 130 paraffin-embedded, archived BC TMA cohort, statistical analysis revealed that GOT2 protein level was significantly positively correlated with lymph node metastasis (P = 0.039), pathological grade (P = 0.019) and American Joint Committee on Cancer (AJCC) stage (P = 0.036; Table 1).
Notably, the mRNA level of GOT2 was strongly associated with poor overall survival and disease-free survival in BC patients in the BreastMark Database cohort (Madden et al., 2013) (P < 0.001, Kaplan-Meier survival analysis and log-rank test, Fig. 6D) and KMplot database cohort (Gyorffy et al., 2010) (P < 0.001 for OS and P = 0.034 for DFS, Kaplan-Meier survival analysis and log-rank test, Fig. S6B), and was also correlated with overall survival of BC patients in the TCGA cohort analyzed by the Human Protein Atlas (https://www.proteina tlas.org/, P = 0.015, Fig. S6C). Kaplan-Meier survival RNAi with or without GOT2 RNAi simultaneously for 48 h. Two-tailed Student's t-test was used to evaluated the difference; error bars represent the mean AE SEM from three independent experiments. ***P < 0.001; **P < 0.01; *P < 0.05; ns, not significant. analysis of our BC TMA cohort also demonstrated that the protein level of GOT2 was significantly correlated with the poor OS of BC patients (Fig. 6E). Multivariate Cox regression survival analysis adjusting for age, T stage, LNM, pathological grade and GOT protein level consistently reported strong correlation between GOT2 overexpression and shorter survival (P = 0.027, Hazard ratio (HR) = 2.383, 95% CI 1.104-5.141) (Fig. 6F, Additional file 1: Table S7), indicating that GOT2 expression was an independent prognostic factor for outcome in BC. In fact, the stratification by GOT2 level displayed even higher prognostic significance than the widely employed LNM (P = 0.014, HR = 2.249, 95% CI 1.180-4.285) (Fig. 6F, Table S7). Consistently, the protein level of GOT2 was also strongly associated with poor OS and DFS of TNBC patients (P = 0.021 and 0.005, respectively, Kaplan-Meier survival analysis and log-rank test ) (Fig. 6G). Multivariate Cox regression survival analysis also revealed a strong correlation between GOT2 overexpression and shorter OS and DFS (G) Cell proliferation was evaluated in GOT2 knockdown CAL-51 or MCF-7 cells and their counterparts with or without aspartate. (H) Colony formation was examined in GOT2 knockdown CAL-51 or MCF-7 cells and their counterparts with or without aspartate. Two-tailed Student's t-test was used to evaluated the difference; error bars represent the mean AE SEM from three independent experiments. ***P < 0.001; **P < 0.01; *P < 0.05.
Collectively, these findings demonstrated the correlation between BRCA1/ZBRK1 and GOT2 in breast cancer samples and suggested the promising prognostic value of GOT2 in clinical practice.

GOT2 is a potential molecular biomarker indicating the sensitivity of MTX treatment
To determine whether GOT2 could serve as a potential molecular biomarker for cancer treatment, we used Cell-Miner, a web-based suite of bioinformatics tools designed to explore the drug activity in the NCI-60 cell lines (Wang et al., 2016b) to mine the significantly associated drugs related to the transcription level of GOT2. CellMiner tools allow rapid data retrieval of transcripts for 22 379 genes, 92 proteins and 360 microRNA along with activity reports for more than 20 503 chemical compounds, which include 102 drugs approved by the U.S. Food and Drug Administration (FDA). Interestingly, MTX, a classic chemotherapeutic drug for breast cancer treatment, was the top ranked drug positively correlated with GOT2 mRNA level among FDA approved drugs or those under clinical trials (Fig. 7A, Table S9). However, the expression level of ZBRK1 was weakly negatively correlated with the drug activity of MTX (r = À0.226, P = 0.0832; Table S10), and BRCA1 showed no significant correlation with the drug activity of MTX (r = 0.214, P = 0.100) (Table S11). For further validation, we treated breast cancer cells that were stable ectopic overexpression or knockdown of GOT2 with different concentrations of MTX. The results showed that knockdown of GOT2 substantially increased the IC 50 of MTX, whereas overexpression of GOT2 reduced the IC 50 in breast cancer cells (Fig. 7B,  C). These findings suggested that GOT2 expression might act as a predictor of the MTX activity.

Discussion
BRCA1 has been implicated with cell metabolism in previous studies (Privat et al., 2014); however, the underlying mechanisms are still unclear. As a transcriptional co-repressor, there is still no direct evidence to support the notion that BRCA1 mediates tumor metabolic adaptation through its transcriptional function. In the present study, we demonstrated for the first time that BRCA1 formed a co-repressor complex with ZBRK1 on the promoter of GOT2 via the ZBRK1 recognition element, and that BRCA1 deficiency promoted aspartate and a-KG production and accelerated tumor cell growth along with GOT2 upregulation (Fig. 8). Importantly, we found that GOT2 was frequently overexpressed in BC, especially in TNBC, which was significantly associated with poor survival of patients with these diseases and could serve as an independent prognostic biomarker. These findings indicated that GOT2 was a malignant metabolic driver in the absence of BRCA1 or ZBRK1 in BC. Interestingly, our preliminary results revealed that GOT2-overexpressed BC cells were more sensitive to MTX treatment, suggesting a promising therapeutic strategy for precision treatment of BC patients.
Depletion of BRCA1 in mouse embryonic fibroblast induced differentiated expression of a dozen metabolic genes, and a majority of these genes contained the consensus ZBRK1 DNA-binding element, indicating that BRCA1 and ZBRK1 share common regulation pathways in metabolism modulation. BRCA1 has been reported to form a co-repressor complex with ZBRK1 to suppress the expression of many genes, including GADD45A (Yun and Lee, 2003), HMGA2 (Ahmed et al., 2010), MMP9 (Lin et al., 2010) and ANG1 (Furuta et al., 2006). However, whether this co-repressor complex participates in metabolism regulation is still unclear. Strikingly, our results provided the first evidence that aspartate transaminase GOT2 was a direct target of BRCA1/ZBRK1. A series of observations supported this conclusion. First, GOT2 was upregulated both in BRCA1 depletion or knockdown cells and in ZBRK1 knockdown cells. Secondly, ChIP assay using antibodies against BRCA1 or ZBRK1 revealed that BRCA1 and ZBRK1 co-occupied the GOT2 promoter. Moreover, by ChIP/re-ChIP assay analysis, BRCA1 and ZRBK1 were further confirmed to form a complex on the GOT2 promoter. Thirdly, luciferase reporter assay verified that BRCA1 and ZBRK1 exerted transcriptional repression activity on GOT2 expression. Deletion of the ZBRK1 recognizing motif GGGxxxCAGxxxTTG of GOT2 abolished the transcriptional repression effect of BRCA1 ectopic expression, further suggesting that BRCA1 regulated the expression of GOT2 through ZBRK1. A previous study has demonstrated that CtIP formed a complex with BRCA1 and ZBRK1 to repress the expression of ANG1 (Furuta et al., 2006). It seems that BRCA1 and CtIP facilitate ZBRK1 binding to the recognition site in vivo (Tan et al., 2004); however, whether CtIP is essential for the regulation of GOT2 by BRCA1/ ZBRK1 remains to be elucidated. Interestingly, the expression of BRCA1 was previously reported to be controlled by the NAD + /NADH ratio (Di et al., 2010). An increased NAD + /NADH ratio led to disassociation of HDAC1 from BRCA1 promoter, thus promoting BRCA1 expression. Here, we showed that BRCA1 deficiency resulted in transcriptional activation of GOT2, concomitantly leading to increased NADH/NAD + redox. Accordingly, it may be speculated that GOT2 overexpression can cause BRCA1 downregulation and subsequently 972 influence the canonical functions of BRCA1. The potential feed-forward loop of BRCA1/ZBRK1/GOT2 may be a crucial metabolic driver signaling BC progression. Recently, GOT2 was shown to activate transcriptionally STAT3/NF-jB in IL-10 responsiveness (Feist et al., 2018). One may propose that GOT2 is fine-tuned with different cell metabolic status and different stimuli in the tumor microenvironment. The specific context and the mechanisms of how the repressor complex is removed from the ZBRK1 site, and how these two pathways are coordinated in regulating GOT2 expression, warrant further investigation.
Aspartate is required for rapid cell growth, which is the most important intermediate for protein, purine nucleotide and pyrimidine nucleotide synthesis (Chen et al., 1996;Sullivan et al., 2015). Given that aspartate has one of the lowest levels of all circulating amino acids (Mayers and Vander Heiden, 2015), obtaining aspartate from the blood for tumor cells is a great challenge. Thus aspartate synthesis is essential for cancer progression. Actually, aspartate metabolism reprogramming exists in many cancers, including hepatocellular carcinoma (Darpolor et al., 2014), pancreatic cancer (Yang et al., 2015(Yang et al., , 2018, leukemia (Sullivan et al., 2015), lung cancer, cervical cancer, osteosarcoma and glioblastoma (Sullivan et al., 2015). GOT2 converts glutamate and oxaloacetate to generate aspartate and a-KG, which is the critical enzyme for maintaining the aspartate pool (Feist et al., 2018). Previous studies have revealed that a crucial role of cellular respiration is to maintain redox homeostasis to support aspartate and therefore biomass production and rapid proliferation (Birsoy et al., 2015;Sullivan et al., 2015). Coloff et al. (2016) demonstrated that mammary epithelial cells can switch from oxidative deamination of Glu to a-KG catalyzed by GDH1 to the conversion of OAA into aspartate by transamination of Gln catalyzed by GOT. Interestingly, Hui Yang et al. elucidated a novel mechanism whereby SIRT3dependent GOT2 acetylation affects the malate-aspartate NADH shuttle activity and promotes pancreatic tumor growth (Yang et al., 2015). Additionally, GOT2 overexpression could inhibit senescence of pancreatic tumor cells and therefore contribute to tumor progression (Yang et al., 2018). Thus, it is reasonable to propose that GOT2 is a metabolic hub, providing cells with aspartate and/or a-KG for energy demands and biosynthetic processes. In our present study, we showed that BRCA1 or ZBRK1 deficiency resulted in GOT2 overexpression, and enhanced the aspartate and a-KG production, subsequently promoting cell proliferation. These observations provided a novel perspective for understanding how BRCA1 and ZBRK1 contribute to BC progression. However, simultaneous knockdown of GOT2 and BRCA1 or ZBRK1 did not completely restore aspartate and a-KG to basal level. This discrepancy indicates that BRCA1/ZBRK1 might also The bold P-value indicated the statistical significance. coordinate aspartate and a-KG metabolism by other metabolic pathways, and is worthy of further investigation. GOT2 has not been associated with the sensitivity of MTX in previous studies. Here, we found that GOT2 overexpression sensitized breast cancer cells to MTX. GOT2 is a key enzyme providing aspartate for nucleotide synthesis, which sustains rapid cell proliferation (Feist et al., 2018). It is considered that MTX treatment causes cancer cells death mainly by reducing the cellular pool of tetrahydrofolate (THF), which inhibits nucleotide synthesis (Hitchings and Burchall, 1965;Kanarek et al., 2018). It is probable that cancer cells with high GOT2 expression are addictive in nucleotide synthesis, supporting rapid cell proliferation and thus leading to THF decline. Recently, Naama Kanarek and colleagues reported that histidine ammonia lyase (HAL), the rate-limiting enzyme of the histidine degradation pathway, contributed to the sensitivity of cancer cells to MTX by competitive consumption of THF (Kanarek et al., 2018). Interestingly, data mined by CellMiner revealed that histidine degradation-related gene HAL and nucleotide synthesis-related genes IMPDH2, MTHFD1, PFAS, PPAT and ATIC were significantly coexpressed with GOT2 (data not show). Therefore, we proposed that GOT2 overexpression might be coupled with activation of histidine degradation pathway and nucleotide synthesis pathways to Fig. 8. Schematic representation of the BRCA1/ZBRK1-GOT2 proposed mode of action in modulating breast tumor responses. In normal mammary cells, BRCA1 and ZBRK1 forms a co-repressor complex on the promoter of GOT2 and suppresses its transcription, whereas in BRCA1-or ZBRK1-deficient breast tumor cells, impairment of BRCA1/ZBRK1 co-repressor complex fails to inhibit the transcription of GOT2 and thus leads to GOT2 upregulation and subsequently accelerated aspartate and a-KG production.  decrease the cellular level of THF, thus sensitized breast cancer cells to MTX.

Conclusions
Our study provides evidence that BRCA1 and ZBRK1 possess an additional role in tumor suppression, regulating the aspartate metabolism through transcriptional repression of GOT2. Importantly, our data reveal that GOT2 can serve as an independent prognostic factor for survival outcome of patients with BC and a potential predictor for sensitivity of MTX treatment. These findings extend our knowledge of tumor suppressive functions of BRCA1, providing a new perspective on metabolism regulation and raising the possibility that synthetically targeting aspartate metabolism may be a promising strategy for treatment of BRCA1-deficient breast tumors. supports pancreatic cancer growth through senescence inhibition. Cell Death Dis 9, 55. Yarden RI and Papa MZ (2006)

Supporting information
Additional supporting information may be found online in the Supporting Information section at the end of the article. Fig. S1. ASS1 was regulated by BRCA1 but not ZBRK1.  Fig. S3. GDH1 and GPT2 were not regulated by BRCA1 or ZBRK1. Fig. S4. BRCA1 and ZBRK1 affected the protein level of GOT2 and the mitochondrial a-KG level. Fig. S5. GOT2 promoted cell proliferation. Fig. S6. GOT2 associated with clinicopathological characteristics. Table S1. Significant differentially expression of metabolic genes in MEF BRCA1 M/M and MEF BRCA1 +/+ cell lines. Table S2. BRCA1 repressed metabolic pathways. Table S3. BRCA1 upregulated metabolic pathways. Table S4. Survival analysis of the potential ZBRK1 transcriptional repressed metabolic genes by Breast-Mark database.  Table S9. Correlations between the GOT2 mRNA level and drug activities. Table S10. Correlations between the ZBRK1 mRNA level and drug activities. Table S11. Correlations between the BRCA1 mRNA level and drug activities.