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

This article is featured in Highlights of This Issue, p. 359

BRCA1 is a well-known tumor suppressor. Mutation of BRCA1 is associated with familial, aggressive basal, and serous breast and ovarian tumors that have expanded stem cell signatures and are associated with poor prognosis (1–3). BRCA1 is also epigenetically silenced in some sporadic high-grade tumors (4, 5). Growing evidence implicates BRCA1 in epithelial tumor types apart from breast and ovarian cancers, e.g., colorectal and pancreatic ductal adenocarcinoma (6–9).

BRCA1 encodes a large 220-kDa protein that promotes repair of double-strand DNA breaks via homologous recombination, the most accurate form of double-strand DNA break repair. The C-terminal BRCT phosphopeptide-binding domains largely mediate DNA damage repair activity. BRCA1 forms an obligate heterodimer with BARD1 (BRCA1-associated RING domain-1), and their N-terminal RING domains function as a ubiquitin (Ub) E3 ligase that transfers Ub moieties to targets such as BRCA1 itself, CtIP, and Claspin (10–14). The resulting mono- or poly-Ub can alter protein function (15, 16). In some contexts, BRCA1 targets proteins for proteasomal degradation by catalyzing canonical K48 linkages (13, 17).

Oct1/POU2F1 is a POU-domain transcription factor related to the pluripotency master regulator Oct4 (18). Like BRCA1, it is widely expressed. Oct1 insulates critical target genes against inhibition by oxidative stress, dampens ROS, and promotes glycolytic metabolism (19–22). Oct1 also promotes normal and cancer stem cell phenotypes such as dye efflux and ALDH activity, through target genes such as Abcg2 and Aldh1a1 (23). In addition, elevated Oct1 protein expression is a negative prognostic factor in human gastric, colon, prostate, lung, cervical, and breast cancers (24). Although it does not promote proliferation and is dispensable for immortalization of MEFs by serial passage, Oct1 is required for growth in soft agar following oncogenic transformation and promotes thymic lymphoma generation in Tp53^−/− mouse models (20). Oct1 ablation is associated with oxidative metabolism, elevated ROS, increased sensitivity to oxidative and genotoxic stress, decreased soft agar colony formation, and decreased tumor initiation in xenograft models (20, 21, 23).

Oct1 protein levels are elevated in gut stem cells without changes in mRNA (23, 25). Further work extended these findings to malignant human stem-like (CD44^hiCD24^lo) breast cancer cells: samples with high stem cell content had high levels of Oct1 protein without a concomitant increase in mRNA (23). These results raise the question of how Oct1 protein levels are regulated. Here, we show that BRCA1 E3 Ub ligase activity promotes oxidative glucose metabolism by ubiquitylating and degrading Oct1. Loss of BRCA1 catalytic activity decreases oxygen consumption and mitochondrial membrane potential, increases extracellular acidification rates, and increases levels of glycolytic metabolites. CRISPR/Cas9-mediated Oct1 inactivation reverses this metabolic phenotype, consistent with a model in which Oct1 operates downstream of BRCA1. RNAseq using BRCA1 E3 Ub ligase–deficient cells with and without Oct1 CRISPR identifies multiple deregulated metabolic genes, including genes controlling glutathione metabolism, associated with reversion of the cells to an oxidative phenotype. Mutation of two conserved ubiquitylated Oct1 lysine residues stabilizes the protein and insulates it from BRCA1 destabilization. MG132 treatment leads to accumulation of polyubiquitylated Oct1 species specifically in the presence of BRCA1. Oct1 CRISPR in MCF-7 cells, which have low but detectable levels of genetically normal BRCA1, leaves growth in standard conditions unperturbed but severely impairs growth in soft agar and xenograft assays. Using a panel of primary human breast cancer specimens, we show that elevated Oct1 protein levels correlate with invasive tumors and with low levels of BRCA1. Cumulatively, the results delineate a novel pathway acting downstream of BRCA1 to promote oxidative metabolism and suppress tumor growth.

HCC1937 cells ± BRCA1 were obtained from the laboratory of Dr. Vicente Planelles and were validated using the University of Utah Health Sciences Center DNA Sequencing Core and short tandem repeat (STR) analysis. Cells were cultured in RPMI-1640 and maintained according to ATCC instructions. Cells were screened monthly for mycoplasma by PCR and confirmed to be mycoplasma free. The presence of functional BRCA1 diminishes the growth rate of these cells and can be used as a quality control. Cells were used for only 2 to 5 passages before discarding. For half-life studies, HCC1937 ± BRCA1 cells were treated with 50 μg/mL CHX and harvested at the indicated time points. Immortalized and primary BRCA1 I26A homozygous, S1598F homozygous, and WT control MEFs were cultured as described (20). For proteasome inhibition assays, cells were incubated in MG132 or DMSO vehicle control for 4 hours, followed by immunoprecipitation and Western blotting. JC-1 dye was purchased from ThermoFisher Scientific. JC-1 staining and ratiometric detection used vendor protocols.

Metabolic measurements used an XF24 and XF96 metabolic flux analyzers (Seahorse Bioscience) in 24-well or 96-well microplates, respectively, provided by the manufacturer. Glycolysis was measured per the glycolysis stress test kit and displayed as extracellular acidification rate (ECAR). Pyruvate oxidation was measured using oxygen consumption rate (OCR) when cells were incubated in unbuffered Seahorse media containing 10 mmol/L sodium pyruvate as the only respiratory substrate. A total of 2 × 10⁴ cells/well were plated 16-18 hours prior to analysis.

Analysis of steady-state cellular metabolites was performed as previously described (20), with modifications. Cold methanol precipitation was used to remove protein from metabolites. The supernatant containing the extracted metabolites was transferred to fresh disposable tubes and completely dried en vacuo. GC-MS analysis was performed with a Waters GCT Premier mass spectrometer fitted with an Agilent 6890 gas chromatograph and a Gerstel MPS2 autosampler. Dried samples were suspended in 40 μL of a 40 mg/mL O-methoxylamine hydrochloride (MOX) in pyridine and incubated for 1 hour at 30°C. To autosampler vials was added 25 μL of this solution. N-methyl-N-trimethylsilyltrifluoracetamide (MSTFA, 40 μL) was added via the autosampler and incubated for 1 hour at 37°C with shaking. After incubation, 3 μL of a fatty acid methyl ester standard solution was added via the autosampler, then 1 μL of the prepared sample was injected to the gas chromatograph inlet in the split mode with the inlet temperature held at 250°C. A 10:1 split ratio was used for analysis. The gas chromatograph had an initial temperature of 95°C for 1 minute followed by a 40°C/minute ramp to 110°C and a hold time of 2 minutes. This was followed by a second 5°C/minute ramp to 250°C, a third ramp to 350°C, then a final hold time of 3 minutes. A 30-m Phenomex ZB5-5 MSi column with a 5-m long guard column was used for chromatographic separation. Helium was used as the carrier gas at 1 mL/minute. Due to high amounts of some metabolites, samples were analyzed a second time at a 10-fold dilution. Data were collected using MassLynx 4.1 software (Waters). Metabolites were identified and their peak area was recorded using QuanLynx (Waters). Data analysis was performed using Metaboanalyst 3.0 (http://www.metaboanalyst.ca) and custom R scripts.

Antibodies used for immunoblots were as follows: Oct1, Bethyl #A310-610; β-actin, Santa Cruz Biotechnology #sc-47778 (C4); GAPDH, Millipore Sigma #MAB374; ubiquitin, Abcam #ab7254; BRCA1, Bethyl #A300-000A; Cyclin-B1, Santa Cruz Biotechnology #sc-245 (GNS1); PGC-1α, Cell Signaling Technology #2178S (3G6); HIF-1α, Santa Cruz Biotechnology #sc-10790 (H-206); c-Myc, Millipore Sigma #05-724 (4A6).

Oct1 in pBabePuro (22) was mutated at K9 and K403 (to arginine) using QuickChange (Agilent). Primers were designed using vendor software: K9R top, CGTCAGAAACCAGTAGACCATCTATGGAGAG; K9R bot, CTCTCCATAGATGGTCTACTGGTTTCTGACG; K403R top, TCTTGGAGAATCAAAGGCCTACCTCGGAAGA; K403R bot, TCTTCCGAGGTAGGCCTTTGATTCTCCAAGA.

RNA was isolated using TRIzol (Invitrogen). cDNA was synthesized using SuperScript VILO (Invitrogen). Sequences for primers used were: hPou2f1 forward: 5′-GGACCAGCAGCTCACCTATTA-3′; reverse 5′-AAAGATTCGTCACAGCAGCA-3′; hTbp forward: 5′-AGCCTGCCACCTTACGCTCAG-3′; reverse: 5′-TGCTGCCTTTGTTGCTCTTCC-3′; mPou2f1 forward 5′-GACTTTCAGAAACAGCCCGTGC-3′, reverse 5′-CTGAGCAGCAGCCTGTAAACTTG-3′; mTbp forward: 5′-ACATCTCAGCAACCCACACA-3′; reverse: 5′-CTGGTGTGGCAGGAGTGATA-3′.

HCC1937 cells and MEFs were fixed in methanol at −20°C for 3 minutes and permeabilized in PBS-T (PBS with 0.1% Triton X-100) for 5 minutes at room temperature. Fixed cells were blocked with antibody dilution buffer (2% BSA, 0.1% Triton X-100 in TBS) for 1 hour at room temperature, and incubated in primary antibody at 4°C overnight. A rabbit anti-Oct1 antibody was used (1:500; Santa Cruz Biotechnology, sc-232X). Cells were probed with a secondary goat anti-rabbit antibody conjugated to Alexa488 (Invitrogen #A11008). Images were collected using a Zeiss Axio Observer Z1 imaging system.

Unique gRNA sequences were cloned into CRISPR-Cas9 lenti-CRISPR-v2-GFP (Addgene #52961). This vector was used as previously described (26) and modified to coexpress GFP (27). The gRNA sequences used were: mPou2f1, 5′-CTACCTGGTGAAGATGTCCG-3′; hPou2f1, 5′-ATCATCTCACAGACGCCCCA-3′. The control vector contained a nontargeting gRNA, MGLibA_66407, that was taken from a prior study (26), with a sequence of 5′-GCTTTCACGGAGGTTCGACG-3′. Briefly, to produce virus, 293T cells were transfected with lenti-CRISPR-v2-GFP plasmids containing control or specific gRNA sequences against mouse Pou2f1, alongside the packaging plasmids pVSVg, pRSV-Rev, and pMDLg/RRE. MEFs were transduced with virus at low MOI (0.4) using 10 μg/mL polybrene. Cells were allowed to grow for 2 days before sorting GFP⁺ and GFP⁻ cells, then allowed to expand for 1 to 2 weeks before testing. Cells were used to confirm Oct1 status, assess metabolic phenotype, and perform RNAseq. Cells were discarded immediately so as to prevent the emergence of dominant clones.

RNA concentration was determined using a Quant-iT RNA assay kit and a Qubit fluorometer (Invitrogen). Intact poly(A) RNA was purified from total RNA samples (100–500 ng) with oligo(dT) magnetic beads, and stranded mRNA sequencing libraries were prepared as described using the Illumina TruSeq mRNA library preparation kit. Purified libraries were qualified on an Agilent Technologies 2200 TapeStation using a D1000 ScreenTape assay. Molarity of adapter-modified molecules was defined by qPCR using the Kapa Biosystems Library Quant Kit. Individual libraries were normalized to 10 nmol/L and equal volumes were pooled in preparation for Illumina sequencing. Sequencing libraries (25 pmol/L) were chemically denatured and applied to an Illumina HiSeq v4 paired-end flow cell using an Illumina cBot. Hybridized molecules were clonally amplified and annealed to sequencing primers with reagents from an Illumina HiSeq PE Cluster Kit v4-cBot. Following transfer of the flowcell to an Illumina HiSeq 2500 instrument (HCS v2.2.38 and RTA v1.18.61), a 50-cycle single-end sequence run was performed using HiSeq SBS Kit v4 sequencing reagents. Fastq data quality were checked using Fastqc version 0.10.1 (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/). The resulting 50-base reads were aligned to the mouse mm10 genome (GRCm38, December 2011) plus splice junctions using novoalign version 2.08.01 (http://www.novocraft.com) following the Extended Splice Junction protocol (http://useq.sourceforge.net/usageRNASeq.html). Alignments to splice junctions were translated back to genome coordinates using the SamTranscriptomeParser application in the USeq package (https://sourceforge.net/projects/useq/). Aligned reads were quality checked using the Picard tools' CollectRnaSeqMetrics command (https://broadinstitute.github.io/picard/). On average, 99.0% of the reads aligned to the mouse genome, with 78% of reads providing unique alignments, and 86% of reads providing alignments to protein coding and UTR regions of the genome. Tests for differential gene expression were performed with DESeq2, version 1.10.0 (https://bioconductor.org/packages/release/bioc/html/DESeq2.html). Genes with a count of at least 10 across all samples were tested. Genes showing at least 2-fold change of expression and an adjusted P value of <0.05 were selected as differentially expressed. Pathway analysis was performed using Ingenuity Pathway Analysis (Qiagen).

In vitro ubiquitylation was performed as described (http://www.abcam.com) with modifications. Purified recombinant proteins BRCA1 (Active Motif, 31113), E1 (Abcam, ab188686), UbcH5c (Boston Biochem, E2-627), and biotin-Ub (Enzo, BML-UW8705-0100) were used. Lysates from three different MEF lines were used as the source for Oct1 substrate (Oct1^−/−, Oct1^−/−::hOct1 and Oct1^−/−::hOct1-K9R/403R). The reaction mixture contained 20 μg/mL biotinylated Ub, 5 nmol/L E1, 100 nmol/L UbcH5c, 20 nmol/L BRCA1, 2 mmol/L ATP and ubiquitylation buffer (5 mmol/L Tris-Cl pH 8.0, 0.5 mmol/L MgCl₂, 0.01% Tween-20, 0.1 mmol/L DTT). The final reaction mixture (50 μL) was incubated at room temperature for 1 hour followed by Oct1 immunoprecipitation (Bethyl #A310-610A) and Western blot analysis.

Soft agar assays were performed as described previously (20). Images were taken at 10× magnification. Images were quantified using ImageScope software (Aperio) using the vendor instructions.

Animal studies were conducted in accordance with an Institutional Animal Care and Use Committee (IACUC). MCF-7 cells expressing constitutive luciferase (MCF-7/luc) were obtained from Cell Biolabs. Seven-week female NOD/SCID mice (The Jackson Laboratory) were used as recipients (6 per group). Surgical sites were prepared by hair removal followed by alcohol and betadine scrubs. A 1.5-cm incision was made to expose the inguinal mammary fat pad. Different numbers of cells suspended in 25 μL of matrigel (Corning) were injected into the fat pad with a 27-gauge insulin syringe. Mice received subcutaneous estrogen pellets at the time of cell implantation [∼18 mg beeswax pellet, 25 mg β-estradiol (Sigma) per mg beeswax]. Incisions were closed with wound clips and mice were allowed to recover on a 37° circulating warm water pad before being returned to their cage. Wound clips were removed 10 days after surgery. Prior to in vivo imaging, the mice were anesthetized with isoflurane. d-Luciferin solution (200 μL; 0.0167 mg/μL in PBS) was injected i.p. After 10 minutes, mice were imaged using an IVIS Spectrum. Exposure time was 60 seconds. Bioluminescent signals were quantified using Living Image 4.3.1 (Caliper Life Sciences).

Slides were heated at 60°C for 10 minutes, followed by deparaffinization in xylene for 10 minutes, twice, and rehydration for 3 minutes each in 100% ethanol thrice, followed by 95%, 85%, and 70% ethanol. The slides were washed in IHC Buffer I (50 mmol/L Tris-Cl, pH 7.6, 150 mmol/L NaCl) prior to inactivation of endogenous peroxidase by incubation in methanol + 3% H₂O₂ for 45 minutes with shaking. After a second wash in IHC Buffer I, antigen retrieval was performed in 10 mmol/L Tris-EDTA buffer (10 mmol/L Tris base, 1 mmol/L EDTA, 0.05% Tween-20, pH 9.0) for 30 minutes using a steamer. The slides were rinsed in IHC Buffer I before blocking for 30 minutes at 37°C in PBS + 3% BSA. Anti-Oct1 (Abcam ab178869, rabbit monoclonal, 1:500) and anti-BRCA1 (Calbiochem MS110, mouse monoclonal, 1:100) Abs were used at 4°C overnight. After two washes with IHC Buffer II (50 mmol/L Tris-HCl, pH 7.6, 150 mmol/L NaCl, 0.05% Tween-20), slides were incubated in HRP-conjugated donkey anti-rabbit or HRP-conjugated goat anti-mouse secondary antibodies (GE Healthcare #NA9340 and #NA9310) for 1 hour at room temperature. The slides were developed with DAB peroxidase substrate (Vector Laboratories, SK-4100) as per manufacturer's instructions and were counterstained with hematoxylin. After dehydration (3 minutes washes each of 70%, 85%, 95%, and 100% ethanol), the slides were incubated in xylene for 3 minutes twice and mounted using Permount (Fisher Scientific, SP15-100). IHC breast cancer specimens were evaluated for histology and immunohistochemistry (IHC) by a single pathologist to maintain precision. BRCA1 and Oct1 are not standard immunohistochemical stains, so their evaluation was against controls. Analysis was conducted in a blinded fashion. Some sections had no tumor, and these were noted and excluded from the analysis. For the sections containing tumor, grade was assessed using the well-documented Nottingham grading system (28), which gives a score based on tubule formation, nuclear pleomorphism, and mitotic index. In the Nottingham scoring system, mitoses are based on counting 10 fields at 40 ×. In the case of tissue microarrays, the punch is too small to evaluate 10 fields, so the grade of the larger tumor may not be entirely represented. Intensity of the stain and percentage of tumor cells staining were noted. If excessive background staining of normal tissue was present, the tumor was considered negative.

We determined oxygen consumption and extracellular acidification rates with SV-40 large T antigen–transformed MEFs derived from BRCA1 E3 ligase homozygous-deficient (Brca1^I26A) animals (29, 30) and littermate WT controls. I26A MEFs are morphologically normal and proliferate normally (30) but consume less oxygen and acidify extracellular media rapidly compared with controls (Fig. 1A and B), consistent with higher rates of glycolysis and lactate production. GC-MS analysis of steady-state metabolic intermediates showed that I26A MEFs were metabolically more similar to each other than to controls (Supplementary Fig. S1). We identified 86 metabolites, 44 of which were unchanged, 4 were significantly decreased, and 38 were significantly increased (>2-fold change and P < 0.001; Fig. 1C). Glucose-6-phosphate, pyruvate, and lactate were elevated in I26A MEFs, consistent with a shift toward glycolytic metabolism (Fig. 1D). Decreased metabolites included α-ketoglutarate and inositol (Supplementary Table S1). We also used primary early-passage WT and Brca1^I26A MEFs derived from littermate embryos. As with immortalized cells, primary Brca1^I26A MEFs showed decreased oxygen consumption and increased extracellular acidification rates compared with controls (Fig. 1E and F).

To determine if mitochondria are the source of the increased oxygen consumption capacity, we used the mitochondrial membrane potential (ΔΨ_m) marker JC-1 (5,5′,6,6′-tetrachloro-1,1′,3,3′-tetraethylbenzimidazolylcarbocyanine iodide) with WT and I26A MEFs. JC-1 efficiently formed mitochondrial aggregates (Fig. 1G). Quantification of mean fluorescence of aggregated (red) versus monomeric (green) provides a measure of ΔΨ_m. We identified a significant decrease in ΔΨ_m in I26A MEFs (Fig. 1H). Cumulatively, these findings show that ablation of the BRCA1 Ub E3 ligase activity dampens mitochondrial function and promotes a glycolytic metabolic phenotype.

Oct1 promotes a glycolytic metabolic profile and its loss increases mitochondrial function (20). Steady-state metabolites altered by Oct1 loss include glucose, pyruvate, and lactate (down), as well as α-ketoglutarate and inositol (up). Because BRCA1 promotes a metabolic program similar to that induced by Oct1 loss, and because Oct1 and BRCA1 have been reported to interact (31–34), we tested whether BRCA1 reduces Oct1 levels. We analyzed Oct1 protein and mRNA levels in three independent clones of immortalized WT and I26A MEFs. Oct1 protein was markedly elevated in E3 ligase–deficient I26A MEFs compared with WT (Fig. 2A). Quantification of multiple experiments confirmed that the differences were statistically significant (Fig. 2B). Immunofluorescence analysis of the same cells confirmed these findings (Fig. 2C). As a control, we used Oct1-deficient MEFs (35), which showed no antibody reactivity (Supplementary Fig. S2). The elevated Oct1 associated more strongly with nuclear periphery, consistent with prior reports (36, 37). In contrast to the elevated protein levels, mRNA levels were not significantly altered in I26A MEFs (Fig. 2D). Oct1 protein levels were also consistently elevated in primary I26A MEFs compared with WT (Fig. 2E).

We also tested MEFs with a C-terminal BRCA1 point mutation (S1598F, the equivalent of human S1655F), which disrupts a BRCT repeat and compromises the ability of BRCA1 to interact with phosphorylated substrates (30). BRCA1 S1598F mutation resulted in a slight elevation of Oct1 protein levels compared with WT, though not to the same degree as in I26A MEFs (Supplementary Fig. S3A). The same cells were also slightly more glycolytic than WT, but consumed more oxygen than I26A MEFs (Supplementary Fig. S3B). C-terminal BRCA1 mutations can affect N-terminal RING ligase activity and substrate availability, for example through changes in subcellular localization (16, 38, 39), offering a possible explanation for this finding. Because the effects of this mutant were weak compared with the I26A mutant, we did not study BRCA1 C-terminal mutations further.

A model in which BRCA1 controls metabolism through Oct1 predicts that Oct1 loss in I26A MEFs would shift cells back toward oxidative metabolism. We transduced I26A MEFs with lentiviruses encoding Cas9 and gRNA sequences targeting murine Oct1. Control vectors expressed Cas9 and a nonspecific gRNA. The vectors also encode GFP, allowing isolation of transduced cells. Transduction frequencies for the different constructs were equivalent (Supplementary Fig. S4). Oct1 protein was efficiently eliminated in sorted cells infected with the Oct1-specific gRNA (Fig. 2F), which also partially restored oxygen consumption (Fig. 2G) and extracellular acidification (not shown). These findings are consistent with a model in which BRCA1 promotes oxidative metabolism in part through Oct1 protein destabilization.

To identify gene expression changes associated with Oct1 loss in I26A MEFs, we performed RNAseq. Four replicates were performed using GFP⁺ MEFs infected with either Oct1-specific gRNA or nonspecific control vectors. For all samples, between 22 and 27 million reads mapped uniquely to the mouse mm10 reference genome assembly. Gene expression in replicates with Oct1 CRISPR was distinct from controls (Supplementary Fig. S5A). Using a 2-fold change and P < 0.05 cutoff, we identified 704 genes upregulated and 233 downregulated with Oct1 loss (Fig. 2H; Supplementary Table S2). Among the top upregulated and downregulated genes were Ppargc1a (Pgc1a, up), Spp1/Opp (down), Gsta3 (up), Dhrs3 (up), Mt2 (down), and Sod3 (down). Genome tracks for Gsta3, Mt2, Spp1, and Sod3 are shown in Fig. 2I. We confirmed the upregulation of PGC-1α protein in Oct1 CRISPR cells (Supplementary Fig. S5B). Pathway analysis identified “Metabolic Disease” as significantly altered by Oct1-specific CRISPR (Supplementary Table S3). Other significantly altered pathways include “Inflammatory Response,” “Tissue Development,” “Cancer,” and “Lipid Metabolism” (Supplementary Table S3).

We also studied the effect of Oct1 CRISPR on HIF-1α and c-Myc in I26A MEFs. Both c-Myc and HIF-1 are potent metabolic regulators. Loss of Oct1 in the context of BRCA1 I26A mutation had little effect on the observed low levels of either HIF-1α or c-Myc (Supplementary Fig. S6A). Similarly, deletion of Oct1 in human MCF-7 cells, which have low but detectable levels of genetically WT BRCA1 (see below), resulted in unaltered HIF-1α and c-Myc (Supplementary Fig. S6B). Again, infection rates were comparable with the different (human) Cas9/gRNA constructs (Supplementary Fig. S7). These results suggest that Oct1 acts either in parallel to or downstream of HIF-1α and c-Myc to promote glycolytic metabolism.

A model in which Oct1 protein levels are controlled by BRCA1 Ub ligase activity predicts that BRCA1 regulates Oct1 protein stability by ubiquitylation. We previously identified multiple Oct1 posttranslational modifications by mass spectroscopy (22, 36, 40). In the most recent of these studies (36), we identified two ubiquitylated lysines, K9 and K403 (Fig. 3A, top). We verified that Oct1 was ubiquitylated by treating HeLa cells with MG132 to inhibit proteasome-mediated degradation, immunoprecipitating lysates with anti-Oct1 antibodies, and blotting for Ub (Fig. 3A, bottom, lanes 1–2). Antibodies specific to K48-linked Ub chains identified the same set of bands (lanes 3–4), indicating that K48-linked chains are being catalyzed. The same ubiquitylated bands were identified with urea, indicating that they were not due to Oct1-associated proteins (Supplementary Fig. S8). To determine the function of the two identified ubiquitylated lysines, we mutated them to arginine, which prevents Ub ligation, and retrovirally transduced three cancer cell lines, HeLa, A549, and HCC1937, with either an empty vector (EV), WT Oct1, or the lysine double mutant (K9/403R). HeLa and A549 cells accumulated Oct1 K9/403R to higher levels compared with EV and WT Oct1 (Fig. 3B). These cells express genetically normal BRCA1. However, in the breast cancer cell line HCC1937 (ER⁻/PR⁻/Her2⁻;Brca1^5382insC/−;Pten^−/−), Oct1 protein levels were similar following transduction with WT or K9/403R Oct1. HCC1937 cells synthesize a truncated, cytoplasmic-localized BRCA1^5382insC protein (41).

To determine if the lack of BRCA1 is responsible for the equivalence between WT and K9/403R Oct1 levels, we used HCC1937 cells in which WT BRCA1 had been reintroduced (41). WT BRCA1 expression in HCC1937 cells was sufficient to reduce steady-state levels of WT Oct1, generating a similar pattern to HeLa, A549, and MCF-7 cells (Fig. 3C). This higher level of Oct1 was sufficient to further promote glycolytic metabolism in MCF-7 cells, as measured by oxygen consumption (Supplementary Fig. S9) and extracellular acidification rate (not shown). Direct comparison of HCC1937 ± BRCA1 cells confirmed elevated Oct1 protein levels in the absence of BRCA1 (Fig. 3D–F). In contrast to the low protein levels, Pou2f1 (Oct1) mRNA levels were slightly elevated in HCC1937 cells expressing WT BRCA1 compared with parent cells (Fig. 3G). These results indicate that K9 and K403 regulate Oct1 posttranscriptionally in a BRCA1-dependent manner.

To distinguish between increased Oct1 protein degradation and decreased protein synthesis in the presence of functional BRCA1, we treated HCC1937 ± BRCA1 cells with cycloheximide (CHX), an inhibitor of protein synthesis, and analyzed Oct1 protein levels during a 24-hour time course. Cyclin B1, a known BRCA1 target (13), was used as a positive control for this analysis. β-Actin by contrast should remain stable as a loading standard. The half-life of Oct1 in cells expressing BRCA1 was approximately 10-fold lower than in cells lacking functional BRCA1, indicating that BRCA1 facilitates Oct1 protein degradation (Fig. 4A). Replicate experiments confirmed the decreased Oct1 stability in cells containing BRCA1 (T_1/2 = 13.5 hours) compared with those lacking BRCA1 (T_1/2 = 162.5 hours, Fig. 4B). Oct1 stability was also decreased in WT compared with I26A MEFs (T_1/2 = 54.1 vs. 7.0 hours; Supplementary Fig. S10).

To determine if the accelerated Oct1 decay in the presence of functional BRCA1 is dependent on lysines 9 and 403, we also used Oct1 deficient MEFs transduced with WT or K9/403R Oct1. Cells were treated with CHX or vehicle control as above for 12 hours, and immunoblotted for Oct1. Quantification of protein levels from three separate experiments indicated WT Oct1 decayed to about 70% of t = 0 levels, while K9/403R Oct1 was stable (Fig. 4C). These results indicate that Oct1 destabilization depends not only on BRCA1 but also on Oct1 lysines 9 and 403.

Because BRCA1 is a Ub E3 ligase, we tested if Oct1 ubiquitylation is dependent on BRCA1 E3 ligase activity. We treated WT and I26A MEFs with the proteasome inhibitor MG132 and subjected lysates to Oct1 immunoprecipitation, followed by immunoblotting with anti-Ub antibodies. With WT BRCA1 and in the presence of MG132, multiple high-molecular weight forms of Oct1 were visualized (Fig. 4D, lane 2). In contrast, loss of BRCA1 E3 ligase function severely diminished detection of the high-molecular-weight forms of Oct1 (lane 4).

Oct1 and BRCA1 are known to physically interact (31–34), though BRCA1 has never been tested as a possible Oct1 E3 Ub ligase. To establish Oct1 as a BRCA1 ubiquitylation target, we performed an in vitro ubiquitylation assay, in which all components necessary for transfer of Ub molecules to target proteins are present in a reconstituted system. We used a reaction system containing overexpressed human Oct1 in lysates from Oct1-deficient murine fibroblasts, added purified E3 ligase (BRCA1/BARD1), and the E1. We used several E2 activities known to collaborate with BRCA1/BARD1 (e.g., UbcH1, UbcH5a, and UbcH5b), identifying UbcH5c as an active E2 (data not shown). Empty vector–transduced Oct1-deficient fibroblast lysates were used as a control (Fig. 4E, lanes 1–5). We have found that Oct1 mutation in MEFs augments BRCA1 levels (unpublished), accounting for the higher BRCA1 input protein in the mutant MEFs (Fig. 4E, lanes 1–5). Only when all components in the reaction system were present was robust Oct1 ubiquitylation observed (Fig. 4E, lane 10). Omitting BRCA1 or the E1 and E2 components eliminated the signal (lanes 6–9). Similarly, immunoprecipitation using control rabbit IgG resulted in no Oct1 signal (Supplementary Fig. S11). These results strongly suggest that BRCA1 mediates Oct1 Ub transfer, though it is formally possible that BRCA1 modifies other components of the reaction to indirectly catalyze Oct1 ubiquitylation. To determine if Oct1 ubiquitylation is dependent on lysines 9 and 403, we also used cells transduced with K9/403R Oct1. Even in the presence of all assay components ubiquitylation was not observed (lane 15). These results indicate that BRCA1 ubiquitylates Oct1 largely through either or both K9 and K403 in vitro. This experiment also identifies UbcH5c as the E2.

To identify functional effects of Oct1 deletion in a breast cancer line, we used the same constructs with MCF-7 cells (ER⁺PR⁺Her2⁻;Brca1^+/+). MCF-7 cells have low but detectable levels of BRCA1 (Fig. 3D). Infected and uninfected cells were sorted on the basis of GFP, and the sorted populations were tested for growth in normal culture conditions and soft agar. In fibroblasts, Oct1 loss does not diminish growth rate or immortalization by serial passage in 2D culture, but blocks growth in soft agar following oncogenic transformation (20). Transduction with Cas9 and either of the two gRNAs significantly reduced measurable Oct1 (Fig. 5A, lanes 4 and 6). CRISPR/Cas9-mediated Oct1 deletion resulted in no change in growth rate in standard conditions (Fig. 5B), but blocked soft agar colony formation (Fig. 5C). Microscopic inspection of the agars indicated that equal numbers of GFP⁺ cells were plated, and that the cells remain viable but do not proliferate into colonies in anchorage-free conditions. Quantification of colony number (Fig. 5D) and size (Fig. 5E) indicated that the results were robust and reproducible. These results indicate that Oct1 loss specifically compromises growth in anchorage-independent conditions.

We tested the effect of Oct1 knockdown in MCF-7 cells expressing luciferase in xenograft assays using orthotopic injection into female NOD/SCID mice. CRISPR-mediated Oct1 knockout in MEF-7/luc cells was as efficacious as in the parent MCF-7 cell line (data not shown). As with RNAi in other cell lines (20, 23), Oct1 CRISPR in MCF-7 cells significantly reduced tumor burden in recipient animals, in particular at low numbers of transferred cells (Fig. 5F). Dissected fat pad/tumor tissue (Fig. 5G) consistently showed decreased tumor mass (Fig. 5H), though the tumors looked histologically similar (Fig. 5I).

We previously showed that ablation of Oct1 in the breast cancer line MB-MDA-231 decreased tumor initiation in xenograft models (23). To determine if Oct1 protein levels vary in human primary breast cancer specimens of different grade, we studied 80 human breast tumor and normal prophylactic samples combined from the Huntsman Cancer Institute Biorepository and Molecular Pathology Core and from a commercial breast cancer tissue microarray (TMA). Tumor grade was assessed by analysis of H&E-stained sections, and Oct1/BRCA1 expression was assessed by IHC using adjacent sections. The IHC signal was grouped into no staining, weak staining or positive staining, and percentages of cells expressing Oct1 or BRCA1 at a particular level in the field was determined. Prophylactic normal specimens showed positive BRCA1 and weak Oct1 staining. In malignant samples, Oct1 protein trended higher in tumors of intermediate grade (Fig. 6A, noninvasive, grades 1 and 2) while BRCA1 declined. Data and other clinicopathological features for the individual samples are shown in Supplementary Table S4, and the individual data points superimposed on the averages are shown in Supplementary Figure S12. The effect was strongest in high-grade tumors. Representative images are shown in Fig. 6B. These results demonstrate an increase in Oct1 levels with tumor grade and an inverse correlation between Oct1 and BRCA1.

Here, we show that the tumor suppressor BRCA1 promotes oxidative, antitumorigenic metabolic profiles, via its E3 ligase activity and reduction of Oct1 protein levels (Fig. 7). BRCA1/BARD1 plays an important role in the DNA damage response and also functions as an E3 Ub ligase (17), catalyzing nonconventional K6 Ub linkages and mono-ubiquitylating certain targets such as the core histones, H2AX and gamma-tubulin (15, 16). BRCA1/BARD1 also targets proteins such as Cyclin B and Cdc25C for proteasomal degradation via canonical K48 linkages (13, 17). We show that BRCA1 E3 Ub ligase–deficient cells are more glycolytic and that BRCA1 targets Oct1 for Ub-mediated degradation. BRCA1/BARD1 controls metabolism through Oct1, as CRISPR/Cas9-mediated Oct1 deletion in cells lacking BRCA1 E3 ligase activity reverts the glycolytic phenotype. The effect of Oct1 loss partially reverses the phenotype, suggesting that BRCA1/BARD1 may have multiple targets that control metabolism. One study provides evidence that BRCA1 inhibits glycolysis and activates oxidative phosphorylation, possibly through negative regulation of pAKT (42).

Gene expression profiling in I26A MEFs identifies metabolic disease as the most affected pathway. Example deregulated genes include Sod3, Ppargc1a (Pgc1a), Mt2, Gsta3, and Cox6a2. The findings are consistent with prior results showing that Oct1 loss increases mitochondrial function with Ppargc1a, the gene encoding PGC-1α, a direct repressed Oct1 target (20). Oct1 is also subject to upstream regulation by activities that control metabolism including AMPK and ERK (43–46).

Oct1 protein is frequently elevated in epithelial cancers, including gastric, colorectal, and lung cancers, compared with normal tissue. In many of these cases, strong Oct1 protein expression is a negative prognostic factor (24). In gastric cancer, where the Oct1 gene Pou2f1 is frequently amplified (47), elevated message levels correlate with poor patient outcome (Supplementary Fig. S13A). Previous reports of mRNA expression in breast cancer do not reveal consistent mRNA elevation (reviewed in 24), and consistently patient outcome cannot be stratified based on Oct1 mRNA expression (e.g., Supplementary Fig. S13B, data from Ref. 54). We and others showed that Oct1 protein but not mRNA is elevated in gut stem cells and in primary breast cancer stem–like cells (23, 25). We also showed that Oct1 promotes stem cell potency (23). One other likely Oct1 Ub ligase, TRIM21, has been identified, though direct ubiquitylation was not demonstrated in this study (48).

Physical and functional interactions between Oct1 and BRCA1 have been described before in other contexts (31–34, 48). Our data show that Oct1 protein levels are lower and its decay rate increased in the presence of BRCA1. Oct1 is ubiquitylated at two lysine residues, K9 and K403. K403 is in the DNA binding domain, but this residue does not make DNA contacts (49). Mutation of these residues to arginine stabilizes Oct1 and insulates it against the destabilizing effects of BRCA1. Proteasome inhibition with MG132 results in the accumulation of high-molecular-weight forms of Oct1 specifically in the presence of BRCA1, consistent with a model in which BRCA1/BARD1 destabilizes Oct1 through Ub-mediated degradation to promote oxidative metabolism. This model is likely to be an oversimplification because the metabolic phenotypes BRCA1 and Oct1 promote are not precisely opposite. For example, both Oct1 and BRCA1 reduce ROS levels and promote oxidative stress resistance (20, 21, 50–52). The BRCA1 RING domain is essential to suppress ROS (50, 51). One way to explain these findings is to postulate that BRCA1/BARD1 also controls metabolism through additional targets. For example, the interaction of BRCA1 with Nrf2 prevents KEAP1-mediated degradation and promotes antioxidant transcriptional responses (52). ROS is a byproduct of increased oxidative metabolism, and therefore BRCA1 suppression of ROS may be a means of achieving oxidative homeostasis.

Using MCF-7 cells, we show that Oct1 knockout does not affect proliferation rates in 2D culture, but compromises anchorage-independent growth. Selective effects of metabolic changes on growth depending on specific conditions have been observed before in MEFs using Oct1 shRNAs (20). Changes in growth conditions are also known to alter cellular carbon metabolism (53). Using primary patient samples, we show that Oct1 protein levels increase with tumor grade and inversely correlate with BRCA1. IHC for BRCA1 is not routinely performed in clinical laboratories, while genetic tests lack complete coverage and do not take into account epigenetic silencing. Our findings suggest that Oct1 may be used as a readout of tumor aggressiveness and BRCA1 activity. Cumulatively, the findings pinpoint Oct1 as a central component of a new metabolic regulatory pathway operating downstream of BRCA1.

Conception and design: J. Maddox, D. Tantin

Development of methodology: K. Vázquez-Arreguín, J. Maddox

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): K. Vázquez-Arreguín, J. Maddox, R.R. Cano, T. Ludwig

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): K. Vázquez-Arreguín, J. Maddox, R.R. Cano, R.E. Factor, T. Ludwig

Writing, review, and/or revision of the manuscript: K. Vázquez-Arreguín, T. Ludwig, D. Tantin

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J. Kang, D. Park

Study supervision: D. Tantin

NIH R01AI100873 to D. Tantin. Ruth L. Kirschstein National Research Service Award 5T32DK091317 to K. Vázquez-Arreguín. Pelotonia Graduate Research Fellowship to D. Park. P30CA042014 awarded to Huntsman Cancer Institute and to the Nuclear Control program at Huntsman Cancer Institute, of which D. Tantin is a member.

The authors thank J. Neilson, S. Tavtigian, A. Welm, J. Rutter, and D. Stillman for critical reading of the manuscript. We thank T. Dahlem and the Health Science Center Mutation Generation and Detection Core for generation of Oct1 lentiviral CRISPR knockout constructs. We thank C. Villanueva and members of his laboratory for advice and reagents. We thank A. Shakya for help with soft agar assays and J. Jafek for help with human Oct1 CRISPR/Cas9 constructs. We thank D. Lum and the Huntsman Cancer Institute Preclinical Research Resource core facility for assistance with xenograft experiments. We thank James Marvin in the Flow Cytometry Core Facility for assistance with FACS. Primary human breast cancer samples were obtained from the Huntsman Cancer Institute Biomolecular Repository.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.