BRCA1 through Its E3 Ligase Activity Regulates the Transcription Factor Oct1 and Carbohydrate Metabolism

Abstract

The tumor suppressor BRCA1 regulates the DNA damage response (DDR) and other processes that remain incompletely defined. Among these, BRCA1 heterodimerizes with BARD1 to ubiquitylate targets via its N-terminal E3 ligase activity. Here, it is demonstrated that BRCA1 promotes oxidative metabolism by degrading Oct1 (POU2F1), a transcription factor with proglycolytic and tumorigenic effects. BRCA1 E3 ubiquitin ligase mutation skews cells toward a glycolytic metabolic profile while elevating Oct1 protein. CRISPR-mediated Oct1 deletion reverts the glycolytic phenotype. RNA sequencing (RNAseq) confirms deregulation of metabolic genes downstream of Oct1. BRCA1 mediates Oct1 ubiquitylation and degradation, and mutation of two ubiquitylated Oct1 lysines insulates the protein against BRCA1-mediated destabilization. Oct1 deletion in MCF-7 breast cancer cells does not perturb growth in standard culture, but inhibits growth in soft agar and xenograft assays. In primary breast cancer clinical specimens, Oct1 protein levels correlate positively with tumor aggressiveness and inversely with BRCA1. These results identify BRCA1 as an Oct1 ubiquitin ligase that catalyzes Oct1 degradation to promote oxidative metabolism and restrict tumorigenicity. Mol Cancer Res; 16(3); 439-52. ©2018 AACR.

Figure 1

Disruption of BRCA1 E3 ligase activity promotes a glycolytic phenotype. A, O₂ consumption rate (OCR), a measure of mitochondrial respiration, was assessed in WT and Brca1^I26A MEFs using a metabolic extracellular flux analyzer. Oligomycin inhibits ATP synthase, revealing mitochondria-mediated O₂ consumption. FCCP is a mitochondrial inner membrane uncoupler that reveals maximum respiratory capacity. Antimycin A and Rotenone are respiratory inhibitors. 4×10⁴ cells were plated in each well of a 24-well plate. B, Extracellular acidification rate (ECAR), a measure of glycolytic function, in WT and Brca1^I26A MEFs. C, Plot of individual metabolites identified in WT and Brca1^I26A MEFs using GC-MS. Log₂ of averaged normalized metabolite levels vs. –10×log₁₀ P-value is shown. Significantly altered metabolites (P<0.001, 2-fold change) are shown in blue (down-regulated in Brca1^I26A cells) or red (up-regulated in Brca1^I26A cells). Four biological replicates of each condition were used. Three glycolytic intermediates, glucose-6-phosphate, pyruvate and lactate, are highlighted. D, Bar graph showing absolute levels (GC-MS ion current area under curve) of glucose-6-phosphate, pyruvate and lactate. Error bars indicate ±SD. P-values (from Table S1) were <<0.001 in all cases. E, Similar to (A) except using primary MEFs. F, Similar to (B) except using primary MEFs. G, WT and Brca1^I26A MEFs were stained with JC-1 dye. Epifluorescence images of live cells are shown. H, Cell stained is in (G) were subjected to flow cytometry. Mean fluorescence intensity (MFI) was calculated for JC-1 red (aggregate) and green (monomeric), and ratios were taken. Three biological replicates of each condition were used. Error bars depict ±SD.

Figure 2

Disruption of BRCA1 E3 ligase activity stabilizes Oct1 protein. A, Oct1 protein levels are higher in Brca1^I26A MEFs compared to WT, three different clones shown. B, Oct1 protein levels in WT and Brca1^I26A MEFs, P-value = 0.029. Protein was quantified by with ImageLab software and normalized to β-actin internal standards. Four replicates were used. Error bars denote ±SD. C, Oct1 immunofluorescence in I26A and WT control MEFs. Alexa488 secondary antibodies were used. Slides were counterstained with DAPI to visualize nuclei. D, Oct1 (Pou2f1) mRNA levels in WT and Brca1^I26A MEFs, P-value = 0.064. Message levels are shown relative to Tbp control mRNA. Three replicates were used. Error bars denote ±SD. E, Similar to (A) except Western blots used lysates from primary early-passage WT MEFs and littermate Brca1^I26A MEFs derived in parallel. F, Brca1^I26A MEFs were infected with lentiviruses encoding Cas9, a mouse Oct1-directed gRNA, and GFP. Infected cells were sorted into GFP⁻ and GFP⁺ pools, and immunoblotted with antibodies against Oct1. G, Brca1^I26A MEFs infected with lentiviruses and sorted as in panel (F) were assessed using a metabolic extracellular flux analyzer. 1.5×10⁴ cells were plated in each well of a 96-well plate. All datapoints were derived from four biological replicates of each condition used. Error bars indicate ±SD. H, Plot of average mRNA expression differences in Brca1^I26A MEFs infected with control of Oct1-CRISPR lentiviruses. Infected cells were isolated on the basis of GFP positivity. Log₂ average gene expression fold change vs. –10×log₁₀ P-value is shown. Significantly altered genes (P<0.05, 2-fold change) are shown in blue (down-regulated with Oct1-CRISPR) or red (up-regulated). Four biological replicates of each condition were used. I, Genome tracks of four significantly altered genes: Gsta3, Mt2, Spp1 and Sod3.

Figure 3

Mutation of ubiquitylated Oct1 residues K9 and K403 stabilizes the protein only in cells with WT BRCA1. A, Top: Schematic of Oct1 protein showing known sites of Oct1 post-translational modification. K9 and K403 are functional ubiquitylation sites, determined by MS. Bottom: anti-Ub immunoblot showing Oct1 immunoprecipitates from HeLa cells treated with 50 µM MG132, or DMSO vehicle control, for 4 hr. the immunoprecipitates were immunoblotted with either pan-Ub antibodies (lanes 1-2) or antibodies specific for K48-linked chains (–4). B, WT and K9/403R Oct1 were expressed in HeLa, A549 and HCC1937 cells using retroviruses. Resulting lysates were probed for Oct1 levels by immunoblot. β-actin is shown as a loading control. C, Stabilization of Oct1 K9/403R in HCC1937 cells ectopically expressing BRCA1. Immunoblots are shown. D, Oct1 protein levels in HCC1937 and HCC1937-BRCA1 cells, as well as multiple other tumor cell lines: HeLa (cervical), A549 (lung), and MCF-7 (breast). Immunoblots for BRCA1, Oct1 and β-actin are shown. E, Averaged Oct1 levels from four experiments. Error bars denote ±SD. P-value = 0.004. F, Immunofluorescence images of cells treated similarly to (D). Cells were co-stained with DAPI to reveal nuclei. G, Pou2f1 (Oct1) mRNA levels in HCC1937 and cells complemented with BRCA1. Message levels are shown relative to Tbp control mRNA. Three replicates were used. Error bars denote ±SD.

Figure 4

BRCA1 targets Oct1 for Ub-mediated degradation via the proteasome. A, Parent HCC1937 or HCC1937-BRCA1 cells were treated with CHX (50µg/ml) and collected at the indicated time points. Prepared lysates were subjected to immunoblotting with antibodies against Oct1, CyclinB1 as a positive control, and β-actin as a loading standard. B, Decay curves generated from averages of three independent experiments ±SD. Oct1 protein levels are shown in HCC1937 cells in the absence (blue) or presence (red) of BRCA1. Black lines show single-exponential curve fit and calculated half-lives in the presence or absence of BRCA1. C, Oct1 deficient MEFs complemented with WT human Oct1 (hOct1) or K9/403R Oct1 were treated with CHX as in (A) for 12 hr and immunoblotted as in (A). For each cell type, the Oct1/β-actin at t=0 was arbitrarily set to 1. Three replicates were conducted and error bars depict ±SD. D, Effect of proteasome inhibition on the accumulation of Ub-modified Oct1 in WT and Brca1^I26A MEFs. Cells were treated with 50 µM MG132, or DMSO vehicle control, for 4 hr. Lysates were collected and used to immunoprecipitate Oct1 protein. An anti-Ub immunoblot is shown. Oct1 and β-actin are shown as loading controls. E, In vitro Ub transfer reactions contained lysates from Oct1 deficient MEFs, or cells complemented with WT or K9/403R hOct1. BARD1 was supplied in the lysate. Supplemental recombinant BRCA1 (ActiveMotif), E2 (UbcH5c, BostonBiochem) and E1 (Abcam) were provided where indicated. Following incubation, the reactions were immunoprecipitated with Oct1 antibodies and immunoblotted with anti-Ub antibodies. NS=nonspecific band. Input controls are shown below.

Figure 5

Oct1 CRISPR in MCF-7 cells blocks anchorage-independent growth and tumorigenicity in xenograft assays. A, MCF-7 cells were infected with lentiviruses expressing Cas9 alone (EV), or additionally expressing an Oct1-directed gRNA. The vectors additionally express GFP. Infected cells were sorted into GFP⁻ and GFP⁺ pools, and immunoblotted with antibodies against Oct1. B, 3×10⁴ sorted cells from (A) were placed into 6-well plates, incubated for 3 d, and counted. 3×10⁴ cells were re-plated for five total passages. Population doublings were calculated, accumulated and plotted. C, The same cells in (B) were plated in soft-agar for 3 weeks, and imaged using brightfield and GFP microscopy. D and E, Soft agar colony number and size were counted after staining with crystal violet, averaged and plotted. Error bars indicate ±SEM. F, MCF-7 cells expressing constitutive luciferase (MCF-7/luc) were infected with lentiviruses expressing Cas9 and the Oct1-specific gRNA from panel A above. Cells were injected into recipient NOD/SCID female mice to assess tumor formation. Images are shown at two weeks. G, Images of dissected fat pads using mice engrafted with either 5×10³ or 2.5×10⁴ MCF-7/luc cells. Tumors were collected at four weeks. H, Averaged weights of dissected tumors. Error bars indicate ±SEM. I, Example fixed and sectioned tumors were stained using either H&E or an anti-Oct1 antibody (ab178869) using a peroxidase-conjugated secondary antibody.

Figure 6

Levels of BRCA1 and Oct1 inversely correlate in breast cancer cell lines and primary tissues. A, H&E-stained sections of tumor specimens were assessed for grade, and Oct1 and BRCA1 images from adjacent sections were assessed for low or high expression of either BRCA1 or Oct1 in a blinded fashion. The percentage of cells in a given sample was then compared with its grade and plotted. Averages across multiple samples are shown, using 14 prophylactic normal, 40 non-invasive (NI), grade 1 (G1) or G2, and 26 G3. Error bars indicate ±SEM. Student T-test P-values: NS=non-significant, **<0.01, ***<0.001. B, Representative H&E and IHC images of tumors of different grade are shown.

Figure 7

Model for BRCA1/BARD1 activity on Oct1 and downstream metabolism. NES: nuclear exclusion sequence. NLS: nuclear localization sequence. BRCA1/BARD1 heterodimerization is mediated by the N-termini of both proteins, where the E3 ligase activity (within the RING domain) is also localized.