Mutant p53 Attenuates Oxidative Phosphorylation and Facilitates Cancer Stemness through Downregulating miR-200c–PCK2 Axis in Basal-Like Breast Cancer

Breast cancer is the most commonly diagnosed cancer type and the leading cause of cancer-realted deaths in females worldwide (1), accounting for 30% of all female cancers (2). On the basis of molecular profiling, breast cancer is classified into 5 molecular subtypes: luminal A, luminal B, HER2⁺, normal breast–like, and basal-like (3). Most basal-like breast cancers (BLBC) are triple-negative, lacking the expression of estrogen receptor (ER), progesterone receptor (PR), and HER2 (4). Although basal-like breast cancer/triple-negative breast cancer (BLBC/TNBC) comprises 15% to 20% of all breast cancers, it is the most aggressive subtype and harbors the worst clinical outcome (3). Patients with BLBCs had a higher chance of metastasis and tumor-recurrence rate, resulting in shorter disease-free survival and overall survival (OS; ref. 5). However, due to the high heterogeneity and lack of hormone receptors, current endocrine and HER2-targeted therapies do not apply to BLBCs, making treating BLBC a clinical challenge. Patients with BLBC receive surgical resection followed by traditional systemic chemotherapy or radiotherapy as the primary treatment option (6). Therefore, novel treatment strategies for BLBCs are urgently needed.

miRNAs are small noncoding RNAs with regulatory abilities on gene expression through inducing mRNA degradation or translational inhibition (7). Dysregulation of miRNAs is involved in human diseases, including developmental disorders and human cancers (8). Alterations in miRNA expression have been observed in a wide range of cancers, resulting in either oncogenic or tumor-suppressive effects (9). The miR-200 family plays an important role in regulating epithelial-mesenchymal transition (EMT) and stemness. In BLBCs, the expression of miR-200 family members is downregulated due to hypermethylation of miR-200 family gene promoters, which is in line with the EMT phenotypes and aggressive behaviors of BLBCs. Among the miR-200 family members, miR-200c is the most strongly associated with high-grade breast cancers (10). Downregulation of miR-200c in breast cancers has been reported to increase cell motility, promote anoikis resistance, decrease chemosensitivity, and enhance metastasis (11). Notably, miR-200c is a direct target gene of p53. The expression of miR-200c is downregulated by p53 deficiency or p53 mutations that interfere with the DNA binding of wild-type (WT) p53 onto miR-200c promoter. Restoring miR-200c expression interferes with the EMT and stemness phenotype caused by loss of p53 expression, suggesting a critical role of miR-200c in mutant p53–promoted tumorigenesis and tumor progression in TNBC/BLBC (12).

In breast cancers, the frequency of TP53 mutation ranges from 12% in luminal A to over 80% in BLBCs (13). The unusually high frequency of TP53 mutation in BLBCs reflects its indispensable oncogenic role. Beyond the guardian of the genome, p53 protein has emerged as an important regulator of cellular metabolism and is involved in tumor-associated metabolic reprogramming (14). Generally, WTp53 is thought to suppress glycolysis while increasing oxidative phosphorylation (OXPHOS) to inhibit tumor progression (14). However, recent studies also indicate that p53 regulates cancer-cell metabolism in a cell-type specific manner. Kim and colleagues demonstrated that hepatocellular carcinoma (HCC) cells retaining WTp53 can undergo a metabolic switch from OXPHOS to glycolysis by WTp53 activation to promote tumorigenesis (15). Conversely, in a mesenchymal stem-cell based model, tumor cell lines derived from murine mesenchymal stem cells bearing mutant p53 (MTp53) exhibit higher OXPHOS activity under the treatment of hypoxia-mimicking agent, CoCl₂ (16). Furthermore, even expression of various MTp53 impacts cancer-cell metabolism to varying extents (17), suggesting that the metabolic effect and mechanisms exerted by individual MTp53 require further clarification. Further, our understanding of how p53 regulates the EMT program and stemness by altering cancer-cell metabolism is limited. Recently, MTp53^R270H has been found to downregulate mitochondrial respiration to promote EMT and invasive ability in pancreatic ductal adenocarcinoma (PDAC; ref. 18). In lung cancer, the R72 variant of MTp53 increases metastasis along with elevated mitochondrial function through interacting with PGC-1α [19]. In breast cancers, although the loss of p53 by genetic mutations has been shown to drive gene expression alterations toward establishing the Warburg effect (20), and restoring LDHA rescues WTp53-induced tumor-suppressive effects in the luminal breast-cancer cell line MCF7 (21), how p53 regulates EMT and stemness, particularly, by modulating metabolism in BLBC remains to be elucidated. Since p53 regulates metabolism by modulating the expression or activity of proteins involved in metabolic pathways via transcriptional activation or repression (22), direct binding and inactivation (23), as well as indirect downregulation through signaling pathways (24) and miRNAs involving in regulating glycolysis (25), elucidating the complex functions and roles of p53 downstream targets in cellular metabolisms is expected to clarify the links between p53-mediated metabolic alterations with EMT or cancer stemness in BLBC.

Here, we demonstrated that p53 regulates stemness via modulating miR-200c–PCK2–OXPHOS axis in BLBCs. MTp53 decreases the expression of miR-200c–upregulated PCK2, the key enzyme in gluconeogenesis, which links glycolysis and tricarboxylic acid (TCA) cycle, subsequently reducing OXPHOS activity to promote cancer stem cell (CSC) properties, leading to poor prognosis in patients with BLBC. Our study unraveled the complicated molecular mechanism accounting for p53-mediated regulation of stemness via modulation of microRNA and the downstream metabolic reprogramming, not only providing a possible explanation for the unusually high mutation rate of TP53 in BLBC but also paving the way to developing novel treatments or prevention strategies for patients with BLBC.

Human normal mammary epithelial cell line, MCF12A (ATCC catalog no. # CRL-10782, RRID:CVCL_3744), human BLBC cell lines, BT549 (ATCC catalog no. # HTB-122, RRID:CVCL_1092), MDA-MB-231 (ATCC catalog no. # HTB-26, RRID:CVCL_0062), HS578T (ATCC catalog no. # HTB-126, RRID:CVCL_0332), and murine breast-cancer cell line, 4T1 (ATCC catalog no. # CRL-2539, RRID:CVCL_0125), were purchased from ATCC during 2016 to 2017. All of the cell lines are mycoplasma free according to the provider's (ATCC) declaration. Cryopreservation of cell lines was performed at passages 3 (grown for 7–8 days) after receipt. Cells have been cultured more than 10 to 15 passages (30 passages for CRISPR-mediated KO clones) were discarded to ensure the cell authentication and avoid mycoplasma contamination. Since after every 10 to 15 passages new cells were resuscitated from cryopreserved stocks for experimental usages, cell authentication and contamination of mycoplasma were not reexamined in our laboratory. All of the cell lines were cultured according to the guideline given by ATCC.

To obtain pCDH/miR-200c, PCR-amplified miR-200c precursor DNA fragment from Lenti–miR-200c precursor (System Biosciences) was subcloned into pCDH-CMV-MCS-EF1-Puro lentiviral plasmid. Genome-editing constructs, lentiviral genome-editing constructs, pAll-Cas9.Ppuro-control, pAll-Cas9.Ppuro-miR-200c-Sg1 and pAll-Cas9.Ppuro-miR-200c-Sg2, and lentiviral shRNA constructs for knockdown of PCK2, ZEB1, ZEB2, Twist, Slung, Foxc2, and Bmi1 were provided by the National RNAi Core Facility services at Academia Sinica in Taiwan. MTp53-expressing constructs (pLenti6-V5-p53R175H, RRID:Addgene_22936; pLenti6/V5-p53R249S, RRID:Addgene_22935; pLenti6-V5-p53R273H, RRID:Addgene_22934; and pLenti6-V5-p53R280K, RRID:Addgene_22933), short hairpin RNA (shRNA) against p53 (pLKO-p53-shRNA-427, RRID:Addgene_25636) and their control vectors were purchased from Addgene (12). Cloning strategies, sequences of cloning primer sets and plasmid maps will be provided upon request.

MCF12A, BT549, MDA-MB-231, HS578T, and 4T1 cells were infected with lentiviral constructs for gene overexpression or knockdown. Infected cells were selected with Puromycin (2 μg/mL) or Blasticidin (10 μg/mL) for 2 weeks to establish stable clones. To establish miR-200c–KO MCF12A cells (MCF12A-200c-Sg1 and MCF12A-200c-Sg2), cells were infected with genome-editing constructs and selected with Puromycin for 1.5 months then examined the expression of miR-200c by qRT-PCR.

Upon cells reaching confluence, the culture medium was replaced by XF Assay Medium Modified DMEM (Agilent) without glucose for 1 hour, then 10 mmol/L of glucose was added for 20 minutes. The extracellular L-lactate was determined by using L-lactate assay kit (Cayman Chemical) according to the manufacturer's instruction. The fluorescence intensity was measured by CLARIOstar microplate reader (BMG LABTECH) and normalized to cell numbers.

Cells (5 × 10⁴ cells/well) were treated with corresponding drugs and the survival rate was determined by the MTS colorimetric assay solution (Promega) at indicated time points. The absorbance was measured by CLARIOstar microplate reader (BMG LABTECH).

Total RNA was purified by RNeasy mini kit (Qiagen) and reverse-transcribed by using Transcriptor First Strand cDNA Synthesis kit (Roche). Expression of PCK2 mRNA was examined by qRT-PCR (KAPA SYBR FAST kit, Merck) using the primer set: PCK2-F, TTCTCAGATGTTTGTTGTGTGGATT; PCK2-R, AAACTTGCTCAAATGCTATTGCTG. Actin gene served as an internal control for quantitation using the primer set: ACTB-F, CACCATTGGCAATGAGCGGTTC; ACTB-R, AGGTCTTTGCGGATGTCCACGT. For quantification of microRNA expression, a stem-loop reverse transcription (RT) primer for mature miR-200c and snRNA (RNU6B) was designed: STRT-miR200c: GTCGT ATCCAGTGCAGGGTCCGAGGTATTCGCACTGGATCCATC. STRT-RNU6B: GTTGGCTGGTGCAGGGTCCGAGGTATTCGCACCAGAGCCAACAAAAATA. RNA isolated by using miRNeasy kit (Qiagen) was reverse-transcribed with stem-loop RT primer, and real time-PCR was performed with primer sets: miR-200c-F: GCATAGTAATACTGCCGGGTA; RNU6B-F: TTCCTCCGCAAGGATGACACGC; and Universal-R: GTGCAGGGTCCGAGGT. The emission intensity of SYBR Green was recorded and analyzed by CFX Connect Real-Time PCR Detection System (Bio-Rad) following manufacture's protocol. All experiments were performed in triplicate, and quantification of mRNA or miR-200c expressions were normalized to internal control genes, Actin or RNU6B.

The commercial available antibodies and dilutions used in the immunoblotting analysis were listed below: anti-ZEB1 (1:1,000, Cell Signaling Technology, #3369), anti-p53 (1:2,000, Santa Cruz Biotechnology, sc-126), anti–E-Cadherin (1:500, Santa Cruz Biotechnology, sc-8426), anti–N-Cadherin (1:250, Santa Cruz Biotechnology, Sc-59987), anti-vimentin (1:500, Santa Cruz Biotechnology, sc-6260), anti-PCK2 (1:2,000, Cell Signaling Technology, #6924), and anti-Actin (1:10,000, Sigma, A5441).

Cells (4–5 × 10⁴) were grown in 1.5 mL of MammoCult medium (STEMCELL Technologies) using ultra-low attachment 6-well plates (Corning) for 6 to 7 days, mammospheres (>40 μm in diameter) were counted manually under a microscope.

For migration assay, cells (1 × 10⁵ cells/well for MCF12A) were seeded into the top chamber of a transwell insert (24-well insert; pore size, 8-μm; BD Biosciences) in 100 μL of serum-free medium. For invasion Assay, cells (2 × 10⁵ cells/well for MCF12A) were seeded into the top chamber of a transwell insert precoated with 100 μL of 2% Matrigel (Corning). Complete medium was added to the bottom chamber. After 16 to 24 hours, cells were fixed and then stained with 0.05% crystal violet solution. Numbers of migrated or invaded cells were quantified by measuring staining intensity with ImageJ software (RRID:SCR_003070).

The oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) were examined with Seahorse XFp Extracellular Flux analyzer (Agilent) according to the instructions of manufacturer. Cells (3.5 × 10⁴/well for MCF12A, 4.5 × 10⁴/well for BT549) were plated into Seahorse assay plates and incubated overnight. For OCR measurement, the medium was replaced by XF Assay Medium Modified DMEM with 10 mmol/L glucose for 45 minutes. Next, assay medium including oligomycin, p-trifluoromethoxy carbonyl cyanide phenylhydrazone (FCCP), rotenone + antimycin A (Rot + AA) was serially injected into the analyzer. For ECAR measurement, the medium was replaced by XF Assay Medium Modified DMEM without glucose for 1 hour, then assay medium including glucose, oligomycin, 2-deoxyl-D-glucose (2-DG) was sequentially injected. The results were analyzed using software XFp Wave.

The library preparation and sequencing on an Illumina NovaSeq 6,000 platform with 150 bp paired-end reads were performed by Genomics, BioSci & Tech Co.

All the sequenced paired-end reads were trimmed with Trimmomatic (RRID:SCR_011848; ref. 26). After the adapter and low-quality ends (Q Score < 30) were removed, reads longer than 50 bps were used for expression quantification. The expected counts and transcripts per million (TPM) values were derived by RNA-Seq by Expectation-Maximization (RSEM) (RRID:SCR_013027; ref. 27) with the human annotation (gencode.v33) download from GENCODE database (RRID:SCR_014966; ref. 28). DESeq2 (RRID:SCR_000154; ref. 29) calculated the fold change (FC) and adjusted p value for each gene, and 1,059 genes with absolute FC larger than 4 and adjusted p value smaller than 0.05 were identified as differentially expressed genes (DEG). DEGs and their corresponding FCs were then uploaded to KEGG mapper (RRID:SCR_012773; ref. 30) to provide metabolic interpretation in KEGG pathways. Heatmaps were generated for the z-transformed TPM expression levels with the R package ComplexHeatmap (RRID:SCR_017270; ref. 31).

All the RNA-sequenced reads are available under BioProject ID: PRJNA683086. (https://www.ncbi.nlm.nih.gov/Traces/study/?acc=PRJNA683086).

The expression levels of miRNAs and mRNA of The Cancer Genome Atlas (TCGA) breast invasive carcinoma cohort (BRCA) were downloaded from Broad GDAC Firehose (https://gdac.broadinstitute.org). The mRNA expression values were calculated with RSEM (RRID:SCR_013027; ref. 27). To compare the expression of PCK2 in miR-200c–high expression and -low expression cohort, all the patients with breast cancer were divided into 2 equal sized groups based on the expression level of has-miR200c-3p in the tumor samples.

The quantification of 116 metabolites involved in glycolysis, pentose phosphate pathway, TCA cycle, urea cycle, and amino acid metabolisms was conducted by the C-SCOPE service of [Human Metabolome Technologies, Inc., (HMT)] using capillary electrophoresis time-of-flight mass spectrometry (CE-TOFMS, Agilent CE-TOFMS system Machine No. 3, Agilent Technologies) for cation analysis and CE-tandem mass spectrometry (CE-QqQMS; CE system with Agilent 6460 TripleQuad LC/MS Machine No, QqQ1, Agilent Technologies) for anion analysis. To prepare sample for metabolome analysis, 5–10 × 10⁶ cells were harvested according to the protocol suggested by HMT [Document ID:(E-170620)]. Briefly, cells were washed with 5% mannitol solution twice, and cellular metabolites were extracted by methanol on plates. Extracts were added with internal standards (H3304–1002, HMT) and then filtered through a Millipore 5-kDa cutoff filter (UltrafreeMC-PLHCC, HMT) to remove macromolecules. Filtrates were then subjected to metabolomic analysis at HMT. The metabolites were annotated by HMT, and the concentrations of all metabolites were calculated by normalizing with internal standards and cell numbers. Hierarchical cluster analysis (HCA) was performed by statistical analysis software developed by HMT. Detailed methods for CE-TOFMS and CE-QqQMS will be provided upon request.

Female NOD/SCID mice (NOD.CB17-Prkdc^scid/JNarl; purchased from National Laboratory Animal Center; 6–8 weeks old) were inoculated with 2 × 10⁶ of MDA-MB-231-Control, miR-200c overexpression or miR-200c-siPCK2 cells mixed with Matrigel (Matrigel: PBS = 1:1) into the third mammary fat pad. Tumor size was measured in every 3 days with a caliper, and tumor volume was determined with the formula: (d1x d2²)/2 (d1: larger diameter; d2: smaller diameter). Animal experiments and animal care were handled according to the protocol approved by the Institutional Animal Care and Use Committee (IACUC number: NCTU-IACUC-106030) at our university.

Recently, miR-200c was found to regulate cancer metabolism through targeting tryptophan-2,3-dioxygenase (TDO2; 32) and LDHA (33), raising the possibility that p53 might induce metabolic reprogramming via modulating miR-200c to regulate EMT and stemness in BLBC. To further investigate the role of miR-200c in regulating cell metabolism, we firstly established miR-200c knockout (KO) mammary epithelial cell lines by using CRISPR/Cas9-mediated genome editing (Fig. 1A) with 2 previously described small guiding-RNAs (34). The expression of miR-200c is depleted in miR-200c KO mammary epithelial cell lines (MCF12A-miR-200c-Sg1 and MCF12A-miR-200c-Sg2) as validated by quantitative-PCR (Fig. 1B). Consistent with previous findings, miR-200c KO cells exhibited elevated expression of ZEB1, accompanied by prominent EMT phenotype (Fig. 1C), and a substantial enhancement in cell migration (Supplementary Fig. S1A) and invasion abilities (Supplementary Fig. S1B). Additionally, the percentage of CD44⁺/CD24⁻ stem cell–enriched population (Supplementary Fig. S1C) and mammosphere-forming ability (Supplementary Fig. S1D) was significantly increased. Moreover, loss of miR-200c resulted in cell transformation in normal basal-type mammary epithelial cells as a significant increase in anchorage-independent growth was observed in miR-200c KO cells (Supplementary Fig. S1E), further supporting the notion that miR-200c is a tumor suppressor (11). miR-200c deficiency in MCF12A cells induced the expression of EMT inducers, ZEB2, Snail, Slug, FOXC2, and Twist, whereas the expression of these genes were differentially suppressed by miR-200c in human BLBC/TNBC cell lines, BT-549, or MDA-MB-231 (35, 36) (Supplementary Fig. S2A). To investigate the effects of miR-200c manipulation on the expression of cancer stemness genes, we directly compared the miR-200c-KO MCF12A, miR-200c-overexpressing BT-549, and MDA-MB-231 cells with their control counterparts, without needing to further sort the stem/CSC population, since the CD44⁺/CD24⁻ populations were approximately 96% in miR-200c KO cells (Supplementary Fig. S1C) and nearly 100% in MDA-MB-231 cells (37, 38). As shown in Supplementary Fig. S2B, the expression of BMI1, Nanog, and Sox2 were induced by miR-200c deficiency in mammary epithelial cells but was suppressed by miR-200c overexpression in BLBC cell lines, suggesting that miR-200c regulates stem/cancer stemness by downregulating stemness genes.

Next, 2 MCF12A-miR-200c KO cell lines were subjected to the analysis of mitochondrial OXPHOS by measuring cellular OCR (Fig. 1D; Supplementary Fig. S3A). miR-200c KO cells showed a 40% to 50% reduction in basal respiration activity, a 50% to 60% reduction in spare respiration activity and a 40% decrease in ATP production-associated respiration compared with control cells, indicating loss of miR-200c attenuates OXPHOS. Moreover, the reduced OXPHOS activity in miR-200c cells was accompanied by a concomitant enhancement in aerobic glycolysis by measuring the ECAR (Fig. 1E; Supplementary Fig. S3B). In accordance with this observation, a 2.5-fold higher extracellular lactate level is evidenced by direct measuring this metabolite in the culture medium of miR-200c-Sg2 cells (Fig. 1F). To further investigate the regulatory role of miR-200c in cancer metabolism, we overexpressed miR-200c in BT549 (Fig. 1G), MDA-MB-231 and the murine TNBC/BLBC breast cancer cell line, 4T1 (Supplementary Fig. S4A; refs. 39, 40). Although restoring miR-200c expression failed to suppress aerobic glycolysis in BT549 cells (Fig. 1H), the basal respiration, spare respiration activity, as well as ATP production–associated respiration were upregulated in BT549 (Fig. 1I), MDA-MB-231, and 4T1 cells (Supplementary Fig. S4A–S4C), indicating miR-200c serves as a positive regulator of OXPHOS activity, and enhancing OXPHOS might contribute to its tumor-suppressor function in BLBC cells.

To further address whether miR-200c-deficiency results in a dependency on its preferential utilization of glycolysis for cell survival, we inhibited glycolysis and evaluated cell viability upon treating 2-DG, a competitive inhibitor of hexokinase-2. As shown in Fig. 2A, our results revealed that miR-200c KO cells displayed higher susceptibility to 2-DG–mediated inhibition on cell survival as the IC₅₀ of 2-DG in control cells was more than 20 mmol/L, but significantly dropped to 2.1 mmol/L in KO cells, suggesting miR-200c deficient cells are highly dependent on aerobic glycolysis for cell growth.

To investigate the role of metabolic reprogramming in miR-200c deficiency–induced biological effects, we pretreated miR-200c KO cells with 2-DG (non-cytotoxic dosage, 1 mmol/L or 2 mmol/L) for 2 days, and pretreated cells were then subjected to serial analyses in the absence of 2-DG to examine whether interference with metabolic alteration could contribute to miR-200c deficiency–induced EMT, migration, invasion, stemness, and cell transformation. As shown in Fig. 2B, low dosage treatment of 2-DG failed to affect EMT, and only induced a slight to moderate effect on the migration (Fig. 2C) and invasion activity (Fig. 2D) of miR-200c KO cells. However, when 2-DG pretreated miR-200c KO cells were subjected to mammosphere forming assay and soft-agar foci formation assay without continuous treatment of 2-DG, significant decreases in sphere-forming ability (Fig. 2E) and anchorage-independent growth (Fig. 2F), were found, indicating subcytotoxic level of 2-DG treatment exerts more profound effects on stemness and cell transformation than EMT-related phenotypes. Notably, the suppression on sphere forming and anchorage-independent growth were not due to the growth-suppressive effect caused by 2-DG, as 2-DG pretreatment did not affect the growth of miR-200c KO cells without continuous treatment of 2-DG for 7 days (Supplementary Fig. S5). Taken together, these observations indicated that the elevated aerobic glycolysis contributes to miR-200c deficiency–induced biological effects, particularly stemness and cell transformation.

It has been shown that a comparatively low concentration of 2-DG is capable of reactivating OXPHOS while suppressing glycolysis, and this metabolic shift is associated with the reduction of in vitro aggressiveness of osteosarcoma cells (41). Since an enhanced OXPHOS activity was also found in MCF12A-miR-200c KO cells treated with a nontoxic concentration of 2-DG (Supplementary Fig. S6A), it is possible that reactivated-OXPHOS contributes to low-dosage 2-DG treatment-exerted inhibition on miR-200c deficiency–induced stemness, which also implies that decreased mitochondrial respiration contributes to miR-200c deficiency–induced stemness. To test this hypothesis, we fed miR-200c KO cells with NADH, the reductant fuels mitochondrial Complex I for 2 days to promote mitochondrial OXPHOS (Supplementary Fig. S6B), and repeated the series of EMT, migration, and mammosphere formation assay to examine whether restoration of OXPHOS activity is able to counteract the effects caused by loss of miR-200c. Similar to the effect exerted by low dosage 2-DG, restoration of OXPHOS by NADH treatment in miR-200c KO cells failed to reverse EMT phenotype (Supplementary Fig. S7A) and miR-200c deficiency–enhanced migration (Supplementary Fig. S7B). However, the enhanced sphere-forming ability was significantly compromised by NADH treatment (Fig. 2G), indicating the contribution of diminished OXPHOS to miR-200c deficiency enhanced stemness. Moreover, accumulated evidence indicates reactive oxygen species (ROS) generated by OXPHOS plays a critical role in regulating CSC properties in various cancer types (42); therefore, we further treated BT549-miR-200c cells with an antioxidant, N-Acetyl cysteine (NAC), to further address whether miR-200c inhibits cancer stemness via elevating ROS production by enhancing OXPHOS. As shown in Fig. 2H, NAC exerted a dose-dependent effect on rescuing the suppression of tumorosphere formation induced by miR-200c, indicating inhibition on OXPHOS-mediated ROS production is able to rescue miR-200c–mediated suppression on cancer stemness. This result strongly suggested that miR-200c may regulate cancer stemness through modulating mitochondrial OXPHOS in BLBCs.

Taken together, our results indicated metabolic rewiring plays a critical role in maintaining miR-200c deficiency–mediated biological effects, including stemness and cell transformation.

Previous study indicates that the hot-spot mutants of p53, R175H, R249S, R273H, and R280K, not only induce EMT but also stemness in MCF12A cells (12). To further confirm the metabolic effect of these mutants, we established MTp53-overexpressing cell lines in MCF12A cells (Fig. 3A), and found that p53 hot-spot mutants induce the Warburg effect in basal-type mammary epithelial cells (Fig. 3B and C; Supplementary Fig. S8). Additionally, our metabolomic analysis using CE-TOFMS and CE-QqQMS to measure the absolute concentration of 116 metabolites in miR-200c-KO and p53-R280K MCF12A cells revealed that both of miR-200c KO and overexpressing of p53-R280K lead to similar alterations in the steady-state concentration of metabolites involved in multiple metabolic pathways including glycolysis and TCA cycle (Fig. 3D and E; Supplementary Fig. S9). Consistent with the ECAR measurement in miR-200c KO and MTp53 overexpressing MCF12A cells, the increased lactate/pyruvate ratio also supported our observation that miR-200c deficiency and p53 mutations lead to increased aerobic glycolysis (Supplementary Fig. S9). Moreover, decreased TCA metabolites in miR-200c KO and p53-R280K–overexpressing cells also indicated attenuated TCA cycle might contribute to the declined OXPHOS caused by miR-200c deficiency and p53 mutations (Fig. 3E; Supplementary Fig. S9). The similar effects on EMT, stemness, Warburg effect, and metabolite profiles further imply that p53 might regulate these biological events through miR-200c.

To confirm this notion, we ectopically expressed miR-200c in MTp53-overexpressing MCF12A cell lines, and found that the EMT phenotype and enhanced stemness is reversed by restoring miR-200c expression (Fig. 4A and B). Moreover, miR-200c not only compromised the elevated aerobic glycolysis induced by p53-R280K (Fig. 4C), but also rescued the attenuated OXPHOS caused by p53-R249S, p53-R273H, and p53-R280K (Fig. 4D). All these results together indicated that MTp53 downregulates miR-200c, which in turn induces metabolic alteration and subsequently facilitates EMT and stemness.

To understand the underlying mechanisms by which MTp53 causes metabolic alterations via downregulating miR-200c, we firstly investigated the molecular mechanism accounting for miR-200c deficiency–induced metabolic reprogramming by a genome-wide RNA sequencing (RNA-seq) analysis (Fig. 5A). We identified several metabolic enzymes with differential gene expression caused by miR-200c KO in glycolysis/gluconeogenesis, TCA cycle, and mitochondrial OXPHOS pathways (Fig. 5B; Supplementary Fig. S10). Among those influenced genes, PCK2 (phosphoenolpyruvate carboxykinase mitochondrial isoform, PEPCK-M) drew our attention. PCK2 is a critical enzyme that connects the TCA cycle and gluconeogenesis (43), and declined expression of PCK2 has been linked to HIF-1α–promoted growth of breast tumorigenic cells (44). Hence, we further confirmed the downregulated mRNA and protein expression of PCK2 in miR-200c KO cells (Fig. 5C and D). Additionally, PCK2 expression is downregulated by overexpression of p53 hot-spot mutations (Fig. 5E) or p53 KD (Fig. 5F), and restoring the expression of miR-200c rescued the expression of PCK2 in p53 mutant-expressing MCF12A cells (Fig. 5G), BLBC cell line bearing an endogenous p53 mutation (BT549, MDA-MB-231, and HS578T; Fig. 5H) and p53-null murine breast cancer cell line, 4T1(Fig. 5H), suggesting PCK2 is positively regulated by miR-200c, and p53 deficiency leads to attenuated miR-200c–PCK2 axis.

PCK2 is regulated by miR-200c indirectly due to the lack of miR-200c seeding sequence in the 3′-UTR of PCK2 mRNA (data not shown). To underline the molecular mechanisms by which miR-200c upregulates PCK2 expression, the expression of miR-200c deficiency–induced EMT-inducing transcription factors were firstly examined in MCF12A-R273H-miR-200c cells. Restoring miR-200c expression in p53 mutant-expressing MCF12A cells not only upregulated PCK2 expression (Fig. 5G), but also suppressed the expression of ZEB1, ZEB2, BMI1, FOXC2, Twist, and Slug (SNAI2; Supplementary Fig. S11A). We further knocked down these transcription factors individually in miR-200c-KO MCF12A, p53-R273H MCF12A, BT549, or MDA-MB-231 cells, and found that knockdown of ZEB-1 and BMI1, but not ZEB2, Twist, Slug, and FOXC2, lead to an elevated expression of PCK2 in miR-200c–deficient MCF12A (MCF12A-miR-200c-Sg2-siZEB1-D4, MCF12A-miR-200c-Sg2-siBM1-A3, B3, and C3), p53-mutant MCF12A (MCF12A-p53-R273H-siZEB1-C3, D3, D4, E4 and F4, and MCF12A-p53-R273H-siBMI1-A3, and C3), BT-549 and MDA-MB-231 cells (Supplementary Fig. S11), suggesting miR-200c might upregulated PCK2 expression through downregulating its direct targets, ZEB1 and BMI1.

To elucidate whether downregulated PCK2 contributes to miR-200c deficiency–induced metabolic rewiring, we established PCK2-KD MCF12A cell lines (Fig. 6A). Although our results revealed that declined expression of PCK2 does not enhance aerobic glycolysis (Fig. 6B), a significant reduction in basal respiration, spare respiration capacity, and ATP production-associated respiration was observed in 2 independent PCK2-KD clones (Fig. 6C). These findings strongly implied that diminished PCK2 expression might account for the miR-200c deficiency–induced attenuation of OXPHOS.

To investigate the biological effects caused by declined expression of PCK2 and further link PCK2–OXPHOS axis to miR-200c deficiency- or MTp53-induced biological effects, particularly stemness and EMT-associated phenotypes, we examined EMT-associated phenotypes in PCK2-KD MCF12A cells. Our results revealed that declined PCK2 expression was not able to induce EMT (Fig. 6D) or affect migration ability significantly (Fig. 6E). Contrarily, we observed 5- to 8-fold increases in mammosphere-forming ability in PCK2-KD cell lines (Fig. 6F), indicating that PCK2 might serve as a negative regulator of stemness in mammary epithelial cells, and miR-200c-deficiency or MTp53 could lead to the declined expression of PCK2 which in turn attenuates OXPHOS and subsequently enhances stemness.

To further confirm whether p53 regulates OXPHOS and cancer stemness via the miR-200c–PCK2 axis, we knocked down PCK2 while restoring miR-200c expression in BT-549 cells, which contain an endogenous p53 mutation (R249S; Fig. 6G). As shown in (Fig. 6H), attenuation of miR200c-enhanced PCK2 expression could counteract the enhanced OXPHOS exerted by miR-200c in p53-mutated BT-549 cells. PCK2 KD reduced the basal- and ATP production-associated respiration enhanced by miR-200c overexpression, indicating miR-200c enhances OXPHOS activity via upregulating PCK2. Moreover, we found that interference of PCK2 expression was able to rescue the inhibition of cancer stemness caused by exogenous expression of miR-200c in BT549 cells (Fig. 6I), as the 50% reduction in sphere formation number in miR-200c overexpressed BT549 cells was diminished to 20% and 10% when knocking down PCK2 expression by 2 independent siRNAs. These two pieces of evidence strongly indicated that the miR-200c–PCK2 axis could regulate OXPHOS and stem/CSC properties, and loss of miR-200c–PCK2 axis might contribute to the attenuated OXPHOS and elevated cancer stemness prompted by MTp53.

Since declined expression of PCK2 was observed in miR-200c KO and p53 mutant–expressing cells which harbor EMT phenotype and expanded stem-cell properties, we further examined whether this in vitro observation could be recaptured in vivo by bioinformatic analysis using TCGA data sets on patients with breast cancer data sets. As shown in Supplementary Fig. S12A, our data revealed that the expression of PCK2 is downregulated in patients with breast cancer with lower miR-200c expression (P < 0.0001). We also found a slight positive correlation between PCK2 and miR-200c (R = 0.12, P < 0.0001, Supplementary Fig. S12B), while observing negative correlations with ZEB1 (R = −0.32, P < 0.0001; Supplementary Fig. S12C and S12D). The PCK2 expression is also downregulated in p53-mutated samples (Fig. 7A). Consistent with the negative association of PCK2 with ZEB-1 and lower expression of PCK2 in patients bearing p53 mutations, a reverse correlation between PCK2 and EMT-phenotype (PCK2 vs. CDH1, R = 0.24, P < 0.0001; PCK2 vs.CDH2, R = −0.16, P < 0.0001; PCK2 vs. VIM, R = −0.34, P < 0.0001; PCK2 vs. FN1, R = −0.19, P < 0.0001; Supplementary Fig. S13A and S13B), or several breast CSC markers (PCK2 vs.CD44, R = −0.1, P < 0.0001, PCK2 vs. ALDH1A1–3, R = −0.21∼−0.28, P < 0.0001; PCK2 vs. CD133, R = −0.33, P < 0.0001; Supplementary Fig. S13A and S13C) were observed, indicating PCK2 expression might be regulated by p53–miR-200c axis and associated with EMT and stemness phenotype in vivo. In viewing that BLBCs are featured with high p53 mutation rate, low miR-200c expression, and EMT phenotype with higher CSC properties, we further discovered a significant lower expression of PCK2 in BLBCs (P < 0.0001) by analyzing RNA-seq data from the TCGA database with (Fig. 7B) or using a DNA microarray-based web platform, GENT2 (Fig. 7C). To understand the clinical significance of PCK2 expression in patients with breast cancer, we examined the association of differential PCK2 expression with clinical outcomes by using three online genomics analysis platforms, named R2: Kaplan–Meier Scanner, Kaplan–Meier Plotter, and the GENT2 database. Patients with breast cancer could be divided into 2 prognostic groups according to the expression of PCK2, where lower PCK2 expression was predicted a shorter overall survival (OS; Fig. 7D). In the cohort of patients with p53-mutated breast cancer, lower PCK2 expression also shown a tendency of shorter disease-free survival (DFS), although only a borderline statistical significance was founded (P = 0.06; Fig. 7E). Furthermore, the interference of PCK2 expression impaired the tumor-suppressive ability of miR-200c in p53-mutated BLBC cells in vivo, as the miR-200c–mediated suppression on the tumor growth of MDA-MB-231 cells was rescued by PCK2 knockdown in the orthotopic xenograft mouse model (Supplementary Fig. S14). Together, the clinical validation and in vivo animal experiments support our finding that decreased PCK2 expression plays a critical role in metabolic reprogramming and MTp53- or miR-200c deficiency–induced biological effects, particularly cancer stemness, in breast cancers.

In the past 2 decades, accumulating evidence indicates that microRNAs play a critical role in regulating gene expression, modulating a wide spectrum of biological events such as cell proliferation, apoptosis, cell differentiation, and tumorigenesis (45, 46). Notably, microRNAs are also found to regulate metabolism in cancers via targeting metabolic enzymes or genes involved in metabolic pathways through direct or indirect mechanisms; therefore, understanding the complexity of microRNA-mediated metabolic rewiring and its connection with tumor progression will pave the way toward the development of novel therapies for cancers (47).

As a key microRNA that regulates EMT and stemness, expression of miR-200c is altered in various tumor types, and downregulation of miR-200c is often associated with metastasis, drug resistance, cancer stemness, or advanced tumor stages. Hence, miR-200c is generally considered a tumor-suppressor gene and plays an important role in controlling tumor progression (11, 12, 48). Although the molecular mechanisms accounting for miR-200c-regulated EMT, metastasis, and cancer stemness are well clarified, the role of miR-200c in regulating metabolic reprogramming in cancers, particularly in BLBCs, is still unclear. In this study, we knocked out the expression of miR-200c in an immortal normal basal-type mammary epithelial cell line, MCF12A, by Cas9/CRISPR-mediated genome editing, and a metabolic reprogramming that diminishes mitochondrial OXPHOS but enhances aerobic glycolysis was identified. Interference of this metabolic rewiring by inhibiting aerobic glycolysis or restoring OXPHOS leads to a profound inhibition on stemness in miR-200c KO cells, indicating a metabolic dependency for miR-200c deficiency–induced stemness/cancer stemness. This observation strongly suggests the metabolic alterations induced by deprivation of miR-200c are not secondary biochemical events triggered by the acquisition of stemness but drivers that govern necessary transcriptional or epigenetic reprogramming for acquiring or maintaining stem cell/cancer stem cell traits. This phenomenon is consistent with the “metabostemness” coined by Menendez and Alarcón for describing the fact that CSC properties are not only controlled by genetic alterations but also the cellular metabotypes which render normal cancer cells susceptible to epigenetic rewiring for escaping from epigenetic barriers to enter CSC-like status in response to genetical or environmental stimulations (49, 50). Indeed, metabolic pathways such as glycolysis, TCA cycle, and fatty-acid oxidation have been shown to provide metabolites as critical cofactors for epigenetic modifications (51). In this study, we observed a persistent effect from short term treatment of 2-DG on stemness and transformation, as the influences of 2 days of treatment can last for 7 days to inhibit mammosphere forming and 4 weeks to suppress anchorage-independent growth, respectively, which implies inhibition of glucose metabolism might interrupt the epigenetic reprogramming required for the acquisition of stemness directed by miR-200c deficiency–induced transcriptional network. In view of this, identifying which miR-200c–associated epigenetic changes are induced by metabolic rewiring and clarifying the linkage with miR-200c–mediated biological events will help us understand the underlying mechanisms accounting for miR-200c ablation–induced stemness and shed light on the development of novel therapeutic approaches targeting CSCs in BLBC.

In addition to glycolysis, multiple lines of evidence in this study suggest that OXPHOS plays a critical role in miR-200c–regulated stemness. First, we found 2-DG treatment can enhance OXPHOS in MCF12A KO cells (Supplementary Fig. S6A), which raises the speculation that restoring OXPHOS might also compromise miR-200c deficiency–induced stemness. Second, the fact that administration of NADH not only fuels mitochondrial respiratory chain reaction (Supplementary Fig. S6B) but also interferes with sphere-forming ability further links OXPHOS to miR-200c–regulated stemness (Fig. 2G). Third, restoration of miR-200c expression in BT549 cells is only able to recover OXPHOS activity (Fig. 1I) but not inhibit aerobic glycolysis (Fig. 1H), and the attenuation of CSC properties exerted by miR-200c can be rescued by NAC administration (Fig. 2H), which diminishes the ROS produced by OXPHOS. Our results further conclude that miR-200c inhibits cancer stemness through upregulating OXPHOS rather than inhibiting glycolysis in BLBC cells.

In addition to miR-200c deficiency, our results revealed that the expression of MTp53 leads to reduced OXPHOS in mammary epithelial cells. However, these hot-spot mutants of p53 impact mitochondrial respiration to varying extents (Fig. 3C; Supplementary Fig. S8C). Different p53 hot-spot mutations are reported to result in mutant proteins with distinct structural and biochemical properties (52), which might confer different binding affinities to noncanonical DNA structures or different preference to interacting partners, consequently resulting in differential gene expression signatures. Therefore, this differential suppression of OXPHOS activity by individual p53 mutants might result from their varying degrees of regulation on OXPHOS-related genes by differential binding to DNA through DNA structure-selective binding or interaction with different transcription factors (52).

The similarity of miR-200c deficiency- and MTp53-induced effects on EMT, stemness, and cellular metabolism strongly implies a functional correlation between miR-200c and p53. Indeed, miR-200c is a direct target of p53, and loss of p53 leads to EMT through downregulating miR-200c-ZEB1/ZEB2 axis in breast and liver cancer (12, 53). Moreover, the MTp53-induced EMT, stemness, and metabolic alteration could be reversed by restoring miR-200c in MTp53-overexpressing cells (Fig. 4), and abolishing miR-200c–induced PCK2 expression interferes with the biological effects of miR-200c restoration on OXPHOS and cancer stemness in BLBC cells containing an endogenous p53 mutation (Fig. 6G–I), further revealed that MTp53 attenuates OXPHOS through downregulating miR-200c-ZEB1/BMI1–PCK2 axis, which in turn contributes to the elevation of stemness/cancer stemness (Fig. 7F).

It is well known that p53 regulates multiple miRNAs which contributes to its biological function including apoptosis, cell cycle arrest, EMT, stemness, and metabolism (54). Among these p53 downstream miRNAs, miR-34a has been shown to regulate EMT and stemness in breast cancers, which is similar to the biological functions of miR-200c (55). Besides, miR-34a suppresses breast cancer growth through downregulating glycolysis by targeting LDHA (56). Although Kim and colleagues showed p53 downregulates glycolytic enzymes (hexokinase 1, hexokinase 2, glucose-6-phosphate isomerase) and pyruvate dehydrogenase kinase 1 through transactivating the expression of miR-34a, and anit-miR-34a inhibitor treatment partially rescues the DNA-damage agent (Adriamycin)–induced suppression on lactate production in colon cancer cell line HCT116 (57), whether restoring miR-34a expression/activity could counteract MTp53-induced metabolic reprogramming, and the links between miR-34a–mediated control on glucose metabolism and other biological effects including EMT and stemness are not fully addressed. Our study clearly demonstrated that metabolic reprogramming is a prerequisite for miR-200c deficiency–induced biological effects, particularly cancer stemness, and the miR-200c–PCK2 axis plays a crucial role in linking metabolic rewiring and MTp53-induced cancer stemness, which offers a novel therapeutic approach toward BLBC or breast cancers with p53 mutations.

R.T. Mai reports grants from Ministry of Science and Technology, Taiwan during the conduct of the study. No disclosures were reported by the other authors.

C.H. Chao: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. C.Y. Wang: Data curation, validation, investigation, visualization, writing–original draft, writing–review and editing. C.H. Wang: Data curation, investigation, visualization, methodology. T.W. Chen: Data curation, funding acquisition, investigation, visualization. H.Y. Hsu: Data curation, investigation, visualization, methodology. H.W. Huang: Data curation, investigation, visualization, methodology. C.W. Li: Funding acquisition, writing–review and editing. R.T. Mai: Funding acquisition, visualization.

We would like to thank the National Core Facility for Biopharmaceuticals (NCFB, MOST 106–2319-B-492–002) and the National Center for High-performance Computing (NCHC) of National Applied Research Laboratories (NARLabs) of Taiwan for providing computational resources and storage resources. We also thank the National RNAi Core Facility at Academia Sinica in Taiwan for providing shRNA reagents and related services.

This work was financially supported in part by the following: Ministry of Science and Technology (105-2320-B-009-004, 106-2320-B-009-002, 107-2628-B-009-002, 108-2628-B-009-002, and 109-2628-B-009-004 to C.H. Chao; 109-2311-B-009-002 to T.W. Chen; 109-2314-B-001-002 and 109-2314-B-001-008 to C.W. Li; 107-2320-B-009-006-MY2 and 109-2320-B-009-002 to R.T. Mai); the "Smart Platform of Dynamic Systems Biology for Therapeutic Development" and "Center for Intelligent Drug Systems and Smart Bio-devices (IDS²B)" from The Featured Areas Research Center Program within the framework of the Higher Education Sprout Project of the Ministry of Education (MOE) in Taiwan; Kaohsiung Medical University Research Center Grant (KMU-TC108A04-0).

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.