POU1F1 transcription factor induces metabolic reprogramming and breast cancer progression via LDHA regulation

Oncogene (2021) 40:2725–2740 https://doi.org/10.1038/s41388-021-01740-6 ARTICLE POU1F1 transcription factor induces metabolic reprogramming and breast cancer progression via LDHA regulation Anxo Martínez-Ordoñez1,7 ●Samuel Seoane1●Leandro Avila1●Noemi Eiro2●Manuel Macía3●Efigenia Arias3● Fabio Pereira4●Tomas García-Caballero5●Noemi Gómez-Lado6●Pablo Aguiar6●Francisco Vizoso2● Román Perez-Fernandez 1 Received: 12 August 2020 / Revised: 19 February 2021 / Accepted: 25 February 2021 / Published online: 13 March 2021 © The Author(s) 2021. This article is published with open access Abstract Metabolic reprogramming is considered hallmarks of cancer. Aerobic glycolysis in tumors cells has been well-known for almost a century, but specific factors that regulate lactate generation and the effects of lactate in both cancer cells and stroma are not yet well understood. In the present study using breast cancer cell lines, human primary cultures of breast tumors, and immune deficient murine models, we demonstrate that the POU1F1 transcription factor is functionally and clinically related to both metabolic reprogramming in breast cancer cells and fibroblasts activation. Mechanistically, we demonstrate that POU1F1 transcriptionally regulates the lactate dehydrogenase A (LDHA) gene. LDHA catalyzes pyruvate into lactate instead of leading into the tricarboxylic acid cycle. Lactate increases breast cancer cell proliferation, migration, and invasion. In addition, it activates normal-associated fibroblasts (NAFs) into cancer-associated fibroblasts (CAFs). Conversely, LDHA knockdown in breast cancer cells that overexpress POU1F1 decreases tumor volume and [18F]FDG uptake in tumor xenografts of mice. Clinically, POU1F1 and LDHA expression correlate with relapseand metastasis-free survival. Our data indicate that POU1F1 induces a metabolic reprogramming through LDHA regulation in human breast tumor cells, modifying the phenotype of both cancer cells and fibroblasts to promote cancer progression. Introduction Cellular metabolism reprogramming is a hallmark of cancer cells [1,2]. Metabolic plasticity includes aerobic glycolysis, also called the Warburg effect, a well-known feature of tumors early recognized by Otto Warburg [3]. The glycolysis occurring in cancer cells gives rise to a high-rate glucose uptake and the conversion of pyruvate into lactate, instead of converting into Acetyl CoA and leading to the tricarboxylic acid (TCA) cycle and mitochondrial oxidative phosphorylation (OXPHOS). Moreover, the lactate output to the extracellular space results in acidification of tumor microenvironment (TME) [4–6]. In addition to the key role of lactate in tumor cell maintenance [7,8], lactate is known to affect stromal cells in the TME. For example, in immune cells, lactate induces M2-like polarization of tumorassociated macrophages (TAM) and TAM-induced *Román Perez-Fernandez [email protected] 1Department of Physiology-Center for Research in Molecular Medicine and Chronic Diseases (CIMUS), University of Santiago de Compostela, Santiago de Compostela, Spain 2Research Unit, Hospital Fundación de Jove, Gijón, Spain 3Department of Obstetrics and Gynecology, Health Research Institute of Santiago de Compostela (IDIS)-University of Santiago de Compostela, Santiago de Compostela, Spain 4Department of Radiation Oncology, Health Research Institute of Santiago de Compostela (IDIS)-University of Santiago de Compostela, Santiago de Compostela, Spain 5Department of Morphological Sciences, Health Research Institute of Santiago de Compostela (IDIS)-University of Santiago de Compostela, Santiago de Compostela, Spain 6Molecular Imaging Group. Department of Psychiatry, Radiology, Public Health, Nursing and Medicine, and Health Research Institute of Santiago de Compostela (IDIS). University of Santiago de Compostela, Santiago de Compostela, Spain 7Present address: Department of Pathology and Laboratory Medicine, Weill Cornell Medicine, New York, NY, USA Supplementary information The online version contains supplementary material available at https://doi.org/10.1038/s41388021-01740-6. 1234567890();,: 1234567890();,: angiogenesis [9]. It also impairs T and natural killer cell activation and compromises dendritic cell differentiation and maturation that suppresses anticancer immune responses [10,11]. In endothelial cells, lactate has been demonstrated to increase angiogenesis [12]. However, the role of lactate on other stromal cell populations has not yet been fully elucidated. For example, administration of exogenous lactate in breast cancer cells has been shown to induce the cancer stem cell (CSC) phenotype [13], but it is not clear which source of lactate could contribute to inducing stemness: the cancer cell-derived lactate, the CSC-produced lactate, or both. Regarding fibroblasts, cancer cells reprogram the surrounding fibroblasts to induce aerobic glycolysis. In turn, fibroblasts produce lactate that can be taken up by cancer cells and oxidized in the mitochondria for energy production and tumor progression. This metabolic coupling is referred to as the “reverse Warburg effect”[14]. In addition, a recent study indicated that lactate secreted by pancreatic ductal adenocarcinoma cells reprogram mesenchymal stem cells epigenetically to differentiate into cancer-associated fibroblasts (CAFs), thus leading to tumor progression [15]. Lactate dehydrogenase A (LDHA) is the key enzyme generating lactate from pyruvate. HIF-1, c-Myc, and p53 are likely the most widely studied lactate regulators. Upregulation of HIF-1 and c-Myc and suppression of p53 are responsible for the metabolic switch to glycolysis in cancer cells [16]. LDHA overexpression promotes cell proliferation and invasion in pituitary adenomas [17], whereas reduction in LDHA activity compromises the ability of tumor cells to proliferate under hypoxia and severely diminishes tumorigenicity and tumor cell maintenance [7]. Transcriptional regulation of prolactin (PRL) and growth hormone (GH) in the pituitary gland by the POU1F1 transcription factor is well known [18,19]. However, POU1F1 is also expressed in non-pituitary tissues, such as human breast, where it also regulates GH and PRL [20,21]. High POU1F1 levels in breast cancer cells induce cell proliferation, reduce apoptosis, and increase migration and invasion [22]. In fact, high expression of POU1F1 in breast cancer correlates with poor clinical outcome [23]. Hypomethylation profiles in clusters of circulating breast cancer cells have recently been identified in binding sites for several transcription factors related to stemness and proliferation, including POU1F1. This epigenetic mechanism has been related to metastasis seeding [24]. The current study analyzes the role of POU1F1 in breast cancer cell metabolism. Human breast cancer cell lines and primary human breast tumors were used to evaluate the effect of POU1F1 overexpression and POU1F1 knockdown in the glycolysis pathway. Using immunodeficient mice, we studied how cancer cells with POU1F1 overexpression and LDHA blockade could affect tumor growth and glucose uptake. In patients, POU1F1 and LDHA mRNA expression was correlated with breast cancer clinical outcome. Finally, in primary cultures of human breast tumors, we studied the effect of both POU1F1 and LDHA on fibroblast activation. Results POU1F1 in breast cancer cells induces metabolic reprogramming Bioinformatic analyses of human breast cancer datasets were carried out for glycolytic activity and other processes related with proliferation and metabolism (Fig. 1A–C). Glycolytic activity was found to be higher in breast tumors than in normal breast (GSE109169), in triple-negative breast cancer (TNBC) than in luminal A tumors (GSE45827), and in bone metastasis than in primary tumors (GSE103357). Interestingly, glycolytic activity was also found to be higher in the breast cancer subtypes Luminal B and HER2 as compared to luminal A (Supplementary Fig. S1A–B). Given that POU1F1 is related to pituitary gland development and cancer progression, we analyzed these biological processes in the context of breast cancer. We carried out an unbiased transcriptomic analysis by performing a microarray (GSE64101) in the control luminal A subtype of MCF7 cells with low endogenous POU1F1 expression (MCF7) and after transient transfection by a POU1F1 overexpression vector (MCF7-POU1F1). To analyze these data, a gene-set enrichment analyses was performed. We found that POU1F1 overexpression induced a clear enrichment of glycolysis signature (NES =2.01, P< 0.001) (Fig. 1D). Next, we selected well-known enzymes and carriers involved in the glycolytic pathway (Fig. 1E) and a real-time PCR was carried out before and after POU1F1 overexpression. Our data confirmed significant upregulation in mRNA expression of eight glycolysisrelated genes: HK2 (hexokinase 2), GAPDH (glyceraldehyde 3-phosphate dehydrogenase), PGK1 (phosphoglycerate kinase 1), PGAM1 (phosphoglycerate mutase), ENO1 (enolase 1), ENO2,PKM2 (pyruvate kinase M2), and LDHA (lactate dehydrogenase A). Downregulation of PFK1 (6-phosphofructokinase) and no changes of GPI (glucose-6phosphate isomerase), or ALDOA (aldolase A) glycolysisrelated genes (Fig. 1F) were observed. In addition, two key transporters in glucose metabolism, SLC2A1 (glucose transporter 1, GLUT1), and SLC2A4 (GLUT4), significantly decreased and increased, respectively, after POU1F1 overexpression. The mRNA expression of SLC16A3 (monocarboxylate transporter 4, MCT4), which is responsible for transporting lactate out of cells, also significantly increased after POU1F1 overexpression (Fig. 1F). Furthermore, the C12orf5 gene which codes for the TP53-inducible glycolysis and apoptosis regulator (TIGAR) protein, a well2726 A. Martínez-Ordoñez et al. known glycolysis p53-mediator [25,26], decreases after POU1F1 overexpression, and this could suggest a possible role for TIGAR in POU1F1-induced glycolysis. To analyze possible changes in secreted metabolites, conditioned medium (CM) from MCF7 and MCF7-POU1F1 cells after 24 h of culture was analyzed by 1H-NMR assay (Supplementary Fig. S2, A-B, and Supplementary Table S1). Glycolyticand TCA-derived metabolites showed dramatic changes after POU1F1 overexpression, with a significant decrease in glucose and a significant increase in lactate (both P< 0.001) (Fig. 1G). Several TCA-derived metabolites also presented significantly reduced levels in CMMCF7-POU1F1 cells (Fig. 1G). Altogether, our data indicate a clear Warburg effect induced by POU1F1., i.e., HALLMARK_P53_PATHWAY HALLMARK_CHOLESTEROL_HOMEOSTASIS HALLMARK_EPITHELIAL_MESENCHYMAL_TRANSITION HALLMARK_GLYCOLYSIS HALLMARK_MITOTIC_SPINDLE HALLMARK_ANGIOGENESIS HALLMARK_XENOBIOTIC_METABOLISM HALLMARK_ADIPOGENESIS HALLMARK_HEME_METABOLISM HALLMARK_MYOGENESIS HALLMARK_MYC_TARGETS_V2 HALLMARK_G2M_CHECKPOINT 0123 0.00 0.05 0.100.15 0.20 FDR NES G CM-MCF7 CM-MCF7-POU1F1 0 5000 10000 15000 Glucose (M) *** 0 200 400 600 800 Pyruvate ( M) *** 0 50 100 150 200 250 Succinate ( M ) *** 0 50 100 150 200 a-Ketoglutarate (M) *** 0 50 100 150 200 Isocitrate ( M) ** 0 5000 10000 15000 *** Lactate ( M) D +POU1F1 Control 0 0.1 0.2 0.3 0.4 HALLMARK_GLYCOLYSIS ES NES=2.01 q<0.001 MCF7-POU1F1 vs MCF7 FDR NES GSE64101 1,3-Bisphosphoglycerate TCA CYCLE GAPDH GLUCOSE LACTATE Glucose-6P Fructose-6P Fructose-1,6bisP Glyceraldehyde-3P 3-Phosphoglycerate 2-Phosphoglycerate Phosphoenolpyruvate Acetyl CoA HK2 GPI PFK1 ALDOA PGK1 PGAM1 ENO PKM LDHA Pyruvate E Pentose Phosphate pathway 0 1 2 3 0 1 2 3 4 5 ENO2 *** 0 1 2 3PKM2 ** 0 1 2 3 4LDHA *** 0 5 10 15 20 25 SLC16A3 *** 0 1 2 3SLC2A1 *** 0.6 0.8 1.0 1.2 SLC2A4 *** Relative mRNA 0.0 0.5 1.0 1.5 ns ** ns 0 1 2 3 4POU1F1 0 1 2 3 4 5 HK2 0.0 0.5 1.0 1.5 GPI 0.0 0.5 1.0 1.5 2.0 PFK1 0.0 0.5 1.0 1.5 2.0 2.5 ALDOA **** MCF7 MCF7-POU1F1 0 1 2 3 4 0 1 2 3 4*** *** *** ENO1 *** 0.0 0.5 1.0 1.5 2.0 2.5 GAPDH PGK1 PGAM1 F evita l eR AN Rm evitaleRAN Rm TIGAR *** AB C HALLMARK_E2F_TARGETS HALLMARK_MYC_TARGETS_V1 HALLMARK_G2M_CHECKPOINT HALLMARK_MYC_TARGETS_V2 HALLMARK_MITOTIC_SPINDLE HALLMARK_OXIDATIVE_PHOSPHORYLATION HALLMARK_GLYCOLYSIS 0123 0.00 0.02 0.04 0.06 0.08 FDR NES HALLMARK_E2F_TARGETS HALLMARK_G2M_CHECKPOINT HALLMARK_MYC_TARGETS_V1 HALLMARK_MITOTIC_SPINDLE HALLMARK_MYC_TARGETS_V2 HALLMARK_GLYCOLYSIS HALLMARK_OXIDATIVE_PHOSPHORYLATION 0123 0.0000 .002 0.0040 .0060.008 FDR HALLMARK_GLYCOLYSIS Breast tumor vs Normal breast Bone metastasis vs Primary tumor NES GSE109169 GSE103357 FDR NES Breast tumor Normal Breast NES=1.65 q=0.003 0 0.1 0.2 0.3 0.4 ES HALLMARK_GLYCOLYSIS NES=1.29 q=0.06 Bone Metastasis Primary Tumor 0.1 0.3 0.2 0 HALLMARK_E2F_TARGETS HALLMARK_G2M_CHECKPOINT HALLMARK_MITOTIC_SPINDLE HALLMARK_MYC_TARGETS_V1 HALLMARK_MYC_TARGETS_V2 HALLMARK_GLYCOLYSIS HALLMARK_CHOLESTEROL_HOMEOSTASIS 0123 0.00 0.01 0.02 0.03 0.04 0.05 FDR GSE45827 TNBC vs Luminal A HALLMARK_GLYCOLYSIS Luminal A TNBC NES=1.56 q=0.011 0 0.1 0.2 0.3 0.4 ES POU1F1 transcription factor induces metabolic reprogramming and breast cancer progression via LDHA. . . 2727 pyruvate to lactate production increased, while pyruvate to TCA-cycle products decreased at least at the extracellular level. To confirm the functional relevance of our findings, glycolytic activity of breast cancer cells was assayed by measuring extracellular acidification rate (ECAR), quantification of basal glycolysis, compensatory glycolysis, and mitochondrial oxygen consumption rate (OCR)/proton extrusion rate (glycoPER) basal ratio in both MCF7 and MCF7-POU1F1 cells. The data obtained show that POU1F1 significantly increased the glycolytic profile as compared to control MCF7 cells (Fig. 2A, B). Conversely, knockout of POU1F1 (MDAsgPOU1F1) using the CRISPR methodology in the MDA-MB-231 cell line (triple-negative subtype, with high endogenous levels of POU1F1) presented a significant decrease in the glycolytic profile compared to control cells (MDAsgC) (Fig. 2C, D and Supplementary Fig. S3). Primary cultures of two human breast tumors with low POU1F1 protein expression (PDT 1) and high expression (PDT 6) (see below) were selected for glycolytic activity analyses. As shown in Fig. 2E, F, similar data were obtained for the human breast tumors and the breast cancer cell lines, which strongly suggests that POU1F1 regulates the glycolysis pathway in breast tumors. POU1F1 regulates LDHA expression in breast cancer LDHA is a key enzyme in the glycolysis pathway, transforming pyruvate into lactate. Aberrantly high expression of LDHA has been demonstrated in multiple cancers, including breast cancer, and it has been associated with malignant tumor progression. In fact, bioinformatic search of human breast cancer tumors and normal breast samples (GSE22820) revealed a significant (P< 0.001) increase of LDHA mRNA expression in human breast tumors as compared with normal mammary tissue (Fig. 3A). Using the same database, we also classified breast tumors and normal samples according to POU1F1 mRNA expression (Fig. 3B). Further analyses also showed a significant (P< 0.05) correlation between POU1F1 and LDHA mRNA expression (Fig. 3C, D). In order to evaluate the hypothesis that POU1F1 could regulate the LDHA gene at transcriptional level, we searched for potential binding sites of POU1F1 in the LDHA promoter using the JASPAR database (http://jaspar.genereg. net/), and found two putative motifs for POU1F1 in the LDHA gene promoter region (Fig. 3E). These two consensus elements for POU1F1 were located at −76 to −62 bp upstream and +276 to +290 bp downstream from the start transcription site in the LDHA promoter (Fig. 3F). To evaluate the possible transcriptional regulation of LDHA by POU1F1, a luciferase reporter assay in the absence or presence of POU1F1 was carried out using three plasmids: (a) the wild LDHA promoter (LDHApromoter-WT, −101 to +315 bp), (b) a plasmid with deletion of the first POU1F1 binding site (LDHApromoter-DEL1, deletion of −76 to −62 bp), and (c) a plasmid with deletion of the second POU1F1 binding site (LDHApromoter-DEL2, deletion of +276 to +290 bp). HEK 293 cells co-transfected with the LDHApromoter-WT and the POU1F1 overexpression vector significantly increased luciferase activity as compared with control (P< 0.01), whereas deletion of both the first and the second POU1F1 binding sites significantly reduced luciferase activity (P< 0.01 and P< 0.001) (Fig. 3G). In fact, interrogation of the epigenetic status of LDHA gene promoter in breast cancer cell lines showed a peak enrichment in H3K27ac, an epigenetic mark associated with higher activation of transcription (Fig. 3H). Furthermore, transient POU1F1 overexpression in MCF7 cells significantly (P< 0.001) increased LDHA mRNA expression (Fig. 3I). Western blots also indicated a visible increase in LDHA protein expression after POU1F1 overexpression in both transient (Fig. 3J) and stable (Supplementary Fig. S4A) MCF7 cells. Conversely, transient knockdown of POU1F1 using a pLKO-shPOU1F1 vector in MDA-MB-231 cells significantly decreased LDHA mRNA (P< 0.001, Fig. 3K) and protein (Fig. 3L) levels. Confocal microscopy of MCF7 control cells transfected stably with the pTRE2 control vector (MCF7), MCF7 cells transfected stably with the pTRE2-POU1F1 vector (MCF7-POU1F1), MDA-MB-231 control cells transfected with the pLentiCRISPRv2 control (MDAsgC), and after POU1F1 knockout (MDAsgPOU1F1), indicated a correlation between POU1F1 (in red) Fig. 1 POU1F1 induces metabolic reprogramming in human breast adenocarcinoma cells. A–CDataset enrichment graphs related with metabolic and proliferation process categories from Hallmark collection (up) and GSEA plot of enrichment in Glycolysis geneset (down). Gene expression data obtained from GSE109169, GSE45827, and GSE103357 as indicated: ATumor breast cancer patient samples (n=25) vs normal breast (n=25); BTNBC breast tumors (n=41) vs Luminal A breast tumors (n=29); CBreast cell line derived from bone metastasis patient (n=3) vs primary tumor cells (n=2). FDR false discovery rate, NES normalized enrichment score. DDataset enrichment graphs related with metabolic, proliferation, and development process categories from Hallmark collection (left) and GSEA plot of enrichment in Glycolysis geneset (right) of microarray data from POU1F1-overexpressing MCF7 cells vs MCF7 cells (GSE64101). E Glycolysis pathway and catalytic enzymes (in red). FqPCR of glycolytic enzymes, carriers, and factors involved in the glycolytic pathway in control and POU1F1-overexpressing MCF7 cells. POU1F1,HK2 (hexokinase 2), GPI (glucose-6-phosphate isomerase), PFK1 (6-phosphofructokinase), ALDOA (aldolase A), GAPDH (glyceraldehyde 3-phosphate dehydrogenase), PGK1 (phosphoglycerate kinase 1), PGAM1 (phosphoglycerate mutase), ENO1 (enolase 1), ENO2,PKM2 (pyruvate kinase M2), LDHA (lactate dehydrogenase A), SLC2A1 (glucose transporter 1, GLUT1), SLC2A4 (GLUT4), SLC16A3 (monocarboxylate transporter 4, MCT4), and C12orf5 (TIGAR). G1H-NMR assay of glycolyticand TCA-derived metabolites in conditioned medium (CM) of MCF7 and MCF7-POU1F1 cells after 24 h culture. *FDR-adjusted pvalue (q-value) < 0.05; **< 0.01; ***0.001; ns not significant. 2728 A. Martínez-Ordoñez et al. and LDHA (in green) immunodetection (Fig. 3M). Supplementary Fig. S4B also shows a decrease in LDHA protein expression after knockout of POU1F1 in MDA-MB231 cells using three different CRISPR clones. POU1F1-induced lactate facilitates cancer progression in breast cancer cells To evaluate the effect of lactate on breast cancer cells, glycolytic activity (ECAR and basal glycolysis) was assayed in stable MCF7 overexpressing POU1F1 cells before and after treatment with either a pharmacological LDHA inhibitor (GSK2837808A, referred as LDHAi, 10 μM for 24 h) or after genetic LDHA knockdown (shLDHA, Supplementary Fig. S4C). Figure 4A, B indicate a significant decrease in ECAR and basal glycolysis in MCF7-POU1F1 cells after both LDHA procedures. Previous studies have shown an LDHA-dependence on cancer cell proliferation under hypoxic environment conditions [27]. Our data are in line with these reports. Proliferation of MCF7 cells significantly increased after POU1F1 overexpression in both normoxic and hypoxic cultures, as we previously demonstrated [28], but only under hypoxia does the treatment of MCF7-POU1F1 cells with the LDHAi and knockdown of LDHA significantly (P< 0.05) reduce cell proliferation (Fig. 4C, D). Next, migration of MCF7POU1F1 cells was evaluated using a trans-well assay. A significant increase in cell migration was observed in MCF7 020406080 0 1000 2000 3000 4000 Time (minutes) ECAR( mpH/min/1000cs) *** *** *** MCF7 MCF7-POU1F1 pmol/min/1000 cs mitoOCR/GlycoPER Basal glycolysis Compensatory glycolysis pmol/min/1000 cs A 0.0 0.2 0.4 0.6 0.8 1.0 0 5000 10000 15000 20000 0 5000 10000 15000 20000 25000 MCF7 MCF7-POU1F1 B 0 20000 40000 60000 80000 100000 0 50000 100000 150000 200000 0.0 0.1 0.2 0.3 0.4 0.5 ** ** * 0 20406080 0 10000 20000 30000 (pmol/min/1000 cs) (pmol/min/1000 cs) Compensatory glycolysis mitoOCR/GlycoPER MDAsgC MDAsgPOU1F1 ( RACEmpH )sc 0001/nim/ C MDAsgC MDAsgPOU1F1 Basal glycolysis D PDT 1 (low POU1F1) 020406080 0 20000 40000 60000 Time (minutes) ECAR (mpH/min/1000 cs) 0 100000 200000 300000 400000 0.00 0.05 0.10 0.15 0.20 0.25 0 100000 200000 300000 400000 Time (minutes) Compensatory glycolysis mitoOCR/GlycoPER Basal glycolysis *** ** ** pmol/min/1000 cs pmol/min/1000 cs PDT 1 (low POU1F1) PDT 6 (high POU1F1) EF PDT 6 (high POU1F1) Fig. 2 POU1F1 regulates the glycolytic profile of breast cancer cell lines and primary tumors in vitro. A Representative ECAR Glycolytic Rate Assay profile in MCF7 and MCF7-POU1F1 cells. cs cells. BQuantification of basal glycolysis, compensatory glycolysis and mitoOCR/glycoPER basal ratio in MCF7 and MCF7-POU1F1 cells. C Representative ECAR Glycolytic Rate Assay profile in control (MDAsgC) and POU1F1-knocked-out MDA-MB-231 (MDAsgPOU1F1) cells. DQuantification of basal glycolysis, compensatory glycolysis and mitoOCR/glycoPER basal ratio in MDAsgC and MDAsgPOU1F1 cells. ERepresentative ECAR Glycolytic Rate Assay profile in human primary breast tumor-derived cultures (see Fig. 6D) with low POU1F1 expression (PDT 1) and high POU1F1 expression (PDT 6). FQuantification of basal glycolysis, compensatory glycolysis and mitoOCR/glycoPER basal ratio in PDT 1 and PDT 6. Data are expressed as mean ± SEM. *P< 0.05, **P< 0.01, and ***P< 0.001. POU1F1 transcription factor induces metabolic reprogramming and breast cancer progression via LDHA. . . 2729 cells after POU1F1 overexpression as compared with control cells (Fig. 4E). However, treatment either with LDHAi (P< 0.001) or shLDHA (P< 0.01) significantly reduced cell migration (Fig. 4E). Finally, to analyze cancer cell invasion, we carried out organotypic cultures using control MCF7POU1F1 cells and MCF7-POU1F1 cells either treated with 0 1 2 3 4 5 0 1 2 3 4 R=0.16 P=0.02 N=186 D POU1F1 mRNA High POU1F1 Low POU1F1 N=47 N=40 pLKO pLKO-shPOU1F1 POU1F1 LDHA -actin +-- - +++ - 0 1 2 3 4 5 0 2 4 6 POU1F1/ -actin protein levels Ctrl 72h 48h LDHA/ -actin protein levels Ctrl 72h 48h 24h Bits 2 1 12345678910 1112 1314 Position TCTTGGAATATGATAATACAAGAA JASPAR score=9.20711 JASPAR score=8.41102 TTAATAAACATTAAAGAA B E F POU1F1 LDHA -actin pDream pDream-POU1F1 +--- +++ - L DAPI POU1F1 LDHA MCF7 MCF7-POU1F1 MDAsgC MDAsgPOU1F1 Merge M *** AHDL ANRm A N=176 N=10 C LDHA mRNA High POU1F1 Low POU1F1 N=47 N=40 * 0.0 0.5 1.0 1.5 POU1F1/ -actin protein levels Ctrl 72h 48h 24h 0.0 0.5 1.0 1.5 LDHA/ -actin protein levels Ctrl 72h 48h 24h LDHApromoter-DEL1 ++ LDHApromoter-DEL2 G 0 5000 10000 15000 20000 ** pDream pDream-POU1F1 LDHApromoter-WT ++ ++ ---- - - -- ---- LDHA promoter activity (RLU) ** *** *** 0 20 40 60 80 0 5000 10000 15000 20000 LDHAPOU1F1 ANRm evitaleR Relative mRNA pDream pDream-POU1F1 I *** 0.0 0.5 1.0 1.5 2.0 0.0 0.5 1.0 1.5 pLKO-shPOU1F1 A NRm evit aleR Relative mRNA pLKO POU1F1 K LDHA ** *** DEL1 -101 bp +315 bp +1 bp DEL2 LDHA gene promoter Putative POU1F1 binding sites POU1F1 LDHA 5 5 5 5 5 5 MCF7 ZR75 T47D MDAMB231 CAL51 SUM159 GSE69112 18,416 kb 18,416 kb chr11 (H3K27ac-ChIPseq) LDHA H J GSE22820 GSE22820 GSE22820 GSE22820 Tumor Normal 0 1 2 3 4 012345 0 1 2 3 4 5 2730 A. Martínez-Ordoñez et al. LDHAi or transfected with shLDHA. Our data indicate a significant decrease (P< 0.05) in cell invasion after both experimental manipulations (Fig. 4F). In summary, our data suggest that either blocking LDHA enzymatic activity or decreasing LDHA expression in MCF7-POU1F1 cells reduces proliferation, migration, and invasion of breast cancer cells. POU1F1-regulated LDHA influences tumor growth and tumor glucose uptake in vivo, and both POU1F1 and LDHA expression are related to clinical outcome To evaluate in vivo the effect of LDHA knockdown in POU1F1-overexpressing cells, immunodeficient BALB/cnu mice were injected with control MCF7 cells (stably transfected with the pTRE2 control vector, n=7, MCF7), MCF7 cells with POU1F1 overexpression (pTRE2POU1F1-overexpressing vector plus the pLKO control vector, n=7, MCF7-POU1F1), and with MCF7 cells with POU1F1 overexpression and LDHA knockdown (pTRE2POU1F1-overexpressing vector plus the pLKO-shLDHA plasmid, n=7, MCF7-POU1F1-shLDHA). Xenografted tumors were monitored every 3 days, and mice were sacrificed at day 15 post injection. Body weight remained stable among all three groups (Fig. 5A). Tumor volume in MCF7-injected mice was negligible at day 15, but a continuous growth of tumor volume in both the MCF7POU1F1 and the MCF7-POU1F1-shLDHA groups was observed during the study, being significatively higher at 9, 12, and 15 days in MCF7-POU1F1 mice comparing to LDHA knocked-down mice (Fig. 5B). POU1F1 and LDHA protein levels in three tumors from each group were assayed by western blot showing, as expected, reduced levels of LDHA in MCF7-POU1F1-shLDHA-injected mice with respect to MCF7-POU1F1 group (Fig. 5C). Immunohistochemistry analyses of tumor xenografts showed high cell proliferation (ki67 marker) in both MCF7-POU1F1 and MCF7-POU1F1-shLDHA groups with respect to MCF7injected group, but ki67 staining was reduced in MCF7POU1F1-shLDHA as compared to the MCF7-POU1F1 injected mice (Fig. 5D). In addition, four mice per group were assayed for glucose uptake just before being sacrificed, using [18F]FDG PET/CT scans. SUVmax indicates a significant (P< 0.01) increase in glucose uptake in tumors from MCF7-POU1F1 injected mice as compared to control group (Fig. 5E). However, tumors in mice injected with MCF7-POU1F1shLDHA cells had significantly reduced SUVmax (P< 0.05) as compared with MCF7-POU1F1 injected mice, but similar values to those found in control mice (Fig. 5E). Finally, to study the possible relationship between POU1F1/LDHA expression and clinical outcome, POU1F1 and LDHA mRNA was analyzed in a dataset of human breast cancer patients. We found a significant correlation between POU1F1/LDHA expression and both relapse-free survival (RFS) (P< 0.001) and overall survival (OS) (P< 0.001) (Fig. 5F, G, and Supplementary Fig. S5). POU1F1 expression in breast cancer tumor samples correlates with CAF activation Once we demonstrated that POU1F1 regulates LDHA and that pharmacological or genetic manipulation of LDHA induces phenotypic changes in breast cancer cells that modify cancer progression, we studied whether POU1F1 could have an impact on TME, specifically on fibroblasts. Numerous studies have shown increased NAF to CAF activation in breast cancer [29]. To study the role of POU1F1 in CAF activation, we first evaluated POU1F1 mRNA expression in 21 human breast tumors (Supplementary Table S2). Tumor samples were classified as high POU1F1 (POU1F1 higher than 75th percentile, n=6) and low POU1F1 (with levels below 25th percentile, n=5) mRNA expression (Fig. 6A). Based on this classification, actin alpha 2 smooth muscle mRNA (ACTA2) expression was evaluated as a key marker of CAFs. We found a significant (R=0.597, P=0.0043) correlation between POU1F1 and ACTA2 mRNA (Fig. 6B, C). POU1F1 and Fig. 3 POU1F1 regulates LDHA expression. A Dispersion plot of LDHA mRNA levels in human breast tumors (n=176) and normal tissues (n=10) (GSE22820). B,CPOU1F1 and LDHA mRNA levels according to POU1F1 mRNA levels (high POU1F1: higher than 75th percentile, and lower POU1F1: with levels below 25th percentile). D Spearman correlation analysis of POU1F1 and LDHA mRNA expression (log2) (n=186). E,FJASPAR analysis indicates two POU1F1 binding sites in the LDHA gene at positions −76/−62 bp and +276/+290 bp with respect to transcription start site. DEL1 and DEL2 indicate deletion of POU1F1 binding sites. GThe wild LDHA promoter (−101 to +315 bp from the transcription start site; LDHApromoter-WT) and the LDHA promoter with specific deletions at the POU1F1 binding sites (LDHApromoter-DEL1 and LDHApromoter-DEL2) were subcloned into the pRP vector, transfected in HEK 293 cells, and co-transfected with the control (pDream) or the pDream-POU1F1 overexpression vector for 48 h, and luciferase activity was measured. Normalized relative luciferase units (RLU) were calculated as the ratio of luciferase activity in control and POU1F1-transfected cells. H H3K27Ac occupancy of LDHA promoter in human breast cancer cell lines (GSE69112), analyzed by ChIPseq (MCF7, ZR75 and T47D: Luminal A; CAL51, SUM159, and MDA-MB-231: Triple Negative Breast Cancer). IMCF7 cells were transiently transfected with the pDream-POU1F1 overexpression vector and 72 h later a qPCR was carried out to evaluate LDHA and POU1F1 mRNA expression. J Western blot of LDHA, POU1F1, and β-actin in MCF7 cells after transient POU1F1 overexpression (n=2). KLDHA mRNA in MDAMB-231 cells after POU1F1 knockdown for 72 h. LWB of POU1F1, LDHA, and β-actin in MDA-MB-231 cells after POU1F1 knockdown (n=2). MRepresentative images of POU1F1 and LDHA after confocal microscopy in MCF7 and MDA-MB-231 cells after stable POU1F1 overexpression (MCF7-POU1F1) and POU1F1 knockout (MDAsgPOU1F1). DAPI was used as nuclei marker. Scale bar: 15 μm. Data are expressed as mean ± SEM. **P< 0.01, and ***P< 0.001. POU1F1 transcription factor induces metabolic reprogramming and breast cancer progression via LDHA. . . 2731 ACTA2 protein (α-SMA) expression were also evaluated in seven primary cultures of human breast tumors (Supplementary Table S3). Both western blot and immunohistochemistry (IHC) showed a positive relationship 0 1 2 3 Migration fold MCF7 MCF7-POU1F1 MCF7-POU1F1-shLDHA MCF7-POU1F1-LDHAi Migration fold 0.0 0.5 1.0 1.5 2.0 2.5 *** *** ** ** MCF7 MCF7POU1F1 MCF7-POU1F1 + LDHAi (upper) or shLDHA (bottom) 0 20000 40000 60000 80000 ** pmol/min/1000 cs Basal Glycolysis 020406080 0 2000 4000 6000 8000 Time (minutes) ECAR (mpH/min/1000 cs) E 0.0 0.5 1.0 1.5 0.0 0.5 1.0 1.5 CM-MCF7-POU1F1 CM-MCF7-POU1F1-LDHAi CM-MCF7-POU1F1-shLDHA Invasion fold Invasion fold ** AB MCF7-POU1F1 MCF7-POU1F1 Control LDHAi Control shLDHA F MCF7-POU1F1 MCF7-POU1F1-LDHAi 0 20406080 0 5000 10000 15000 Time (minutes) ECAR (mpH/min/1000 cs) MCF7-POU1F1 MCF7-POU1F1-shLDHA 0 5000 10000 15000 20000 25000 ** pmol/min/1000 cs Basal Glycolysis Proliferation fold 0 1 2 3 4 5 Proliferation fold 0 1 2 3 4 Normoxia Hypoxia MCF7 MCF7-POU1F1 MCF7-POU1F1-shLDHA MCF7-POU1F1-LDHAi C D 0 2 4 6 Proliferation fold 0 2 4 6 8 Proliferation fold *** ns ns ** ** * *** Fig. 4 POU1F1-induced lactate acts on cancer cells to induce cancer progression. A,BAdministration of a LDHA inhibitor (10 μM, GSK2837808A, referred as LDHAi) or LDHA knockdown (shLDHA) in MCF7-POU1F1 cells for 24 h significantly decreases extracellular acidification rate (ECAR) and basal glycolysis. Data were obtained using an XFp seahorse cell analyzer. C,DCell proliferation in control and POU1F1-overexpressing MCF7 cells treated with 10 μM LDHAi or transfected with the pKLO-shLDHA vector for 72 h under normoxic and hypoxic (1% O2) atmosphere. ERepresentative figure and quantitative analysis of trans-well migration assays in control cells (MCF7), POU1F1-overexpressing cells (MCF7POU1F1), and MCF7-POU1F1 cells either treated with LDHAi (upper panel) or after LDHA knockdown (shLDHA, bottom panel). Scale bar: 100 μm. FRepresentative images of H&E-stained sections of MCF7POU1F1 cells cultured in an organotypic system in the presence of a pharmacological LDHA inhibitor (LDHAi) or LDHA knockdown (shLDHA), and quantification of cell invasion. Data are expressed as mean ± SEM, *P< 0.05, **P< 0.01, ***P< 0.001. Scale bar: 50 μm. 2732 A. Martínez-Ordoñez et al. between POU1F1 and α-SMA protein expression (Fig. 6D–F). Immunohistochemistry shows a representative example of POU1F1 and α-SMA immunodetection in two of the above-mentioned human breast tumors with high (Fig. 6E) and low (Fig. 6F) POU1F1 protein levels. As noted, POU1F1 is expressed at nuclear level in cancer cells, whereas α-SMA immunoreactivity in CAFs is present at cytoplasmic level. In line with the results at the mRNA level, a significant (R=0.764, P=0.045) correlation between POU1F1 and α-SMA protein expression was found (Fig. 6G). Indeed, a significant (P=0.048, HR =1.13 (1–1.28)) relationship between POU1F1/ACTA2 mRNA levels and RFS was found in a human breast cancer dataset (n=3951 samples) (Fig. 6H), which suggests a clinical prognostic value for both POU1F1 and α-SMA in breast tumors. MCF7 MCF7-POU1F1 MCF7-POU1F1-shLDH A 16 18 20 22 24 Body weight (g) MCF7 MCF7-POU1F1 MCF7-POU1F1-shLDHA A 1.0 0.8 0.6 0.4 0.2 0.0 050100 150 200 250 HR=1.63 (1.44-1.81) P<0.001 N=3951 Time (months) RFS probability High POU1F1-LDHA Low POU1F1-LDHA MCF7 MCF7POU1F1 MCF7-POU1F1 -shLDHA POU1F1 LDHA -actin C 12312 3123 FG ns ki67 MCF7 MCF7-POU1F1 MCF7-POU1F1-shLDHA 0.0 0.2 0.4 0.6 0.8 SUVmax E ** * MCF7 MCF7-POU1F1 MCF7-POU1F1-shLDHA SUV 4 0 MCF7 MCF7-POU1F1 MCF7-POU1F1-shLDHA D 0 3 6 9 12 15 0 50 100 150 200 250 Tumor volume (mm) Days post-injection B * ** ** ** *** *** 1.0 0.8 0.6 0.4 0.2 0.0 01234 OS probability High POU1F1-LDHA Low POU1F1-LDHA HR=1.4 (1.17-1.68) P<0.001 Days (x103) N=42 5 POU1F1 transcription factor induces metabolic reprogramming and breast cancer progression via LDHA. . . 2733 43. Roland CL, Arumugam T, Deng D, Liu SH, Philip B, Gomez S, et al. Cell surface lactate receptor GPR81 is crucial for cancer cell survival. Cancer Res. 2014;74:5301–10. 44. Lee YJ, Shin KJ, Park SA, Park KS, Park S, Heo K, et al. Gprotein-coupled receptor 81 promotes a malignant phenotype in breast cancer through angiogenic factor secretion. Oncotarget. 2016;7:70898–911. 45. Brown TP, Bhattacharjee P, Ramachandran S, Sivaprakasam S, Ristic B, Sikder MOF, et al. The lactate receptor GPR81 promotes breast cancer growth via a paracrine mechanism involving antigen-presenting cells in the tumor microenvironment. Oncogene. 2020;39:3292–304. 46. Sherman MH, Yu RT, Engle DD, Ding N, Atkins AR, Tiriac H, et al. Vitamin D receptor-mediated stromal reprogramming suppresses pancreatitis and enhances pancreatic cancer therapy. Cell. 2014;159:80–93. 47. Seoane S, Perez-Fernandez R. The vitamin D receptor represses transcription of the pituitary transcription factor Pit-1 gene without involvement of the retinoid X receptor. Mol Endocrinol. 2006;20:735–48. 48. Seoane S, Arias E, Sigueiro R, Sendon-Lago J, Martinez-Ordoñez A, Castelao E, et al. Pit-1 inhibits BRCA1 and sensitizes human breast tumors to cisplatin and vitamin D treatment. Oncotarget. 2015;6:14456–71. 49. Xie H, Hanai J, Ren JG, Kats L, Burgess K, Bhargava P, et al. Targeting lactate dehydrogenase−a inhibits tumorigenesis and tumor progression in mouse models of lung cancer and impacts tumor-initiating cells. Cell Metab. 2014;19:795–809. 2740 A. Martínez-Ordoñez et al.