Loss of FBP1 by Snail-mediated repression provides metabolic advantages in basal-like breast cancer

Abstract

The epithelial-mesenchymal transition (EMT) enhances cancer invasiveness and confers tumor cells with cancer stem cell (CSC)-like characteristics. We show that the Snail-G9a-Dnmt1 complex, which is critical for E-cadherin promoter silencing, is also required for the promoter methylation of fructose-1,6-biphosphatase (FBP1) in basal-like breast cancer (BLBC). Loss of FBP1 induces glycolysis and results in increased glucose uptake, macromolecule biosynthesis, formation of tetrameric PKM2, and maintenance of ATP production under hypoxia. Loss of FBP1 also inhibits oxygen consumption and reactive oxygen species production by suppressing mitochondrial complex I activity; this metabolic reprogramming results in an increased CSC-like property and tumorigenicity by enhancing the interaction of β-catenin with T-cell factor. Our study indicates that the loss of FBP1 is a critical oncogenic event in EMT and BLBC.

Figure 1. FBP1 expression inversely correlates with Snail in breast cancer

(A) Box-plots indicate FBP1 expression in different subtypes of breast cancer. (B) Statistical analysis of 150 cases of breast tumor samples immuno-stained using antibodies against FBP1, ERα, and a control serum. (C) Expression of FBP1, ERα and Snail were analyzed on fresh frozen tumor samples from six cases of luminal and six cases of triple-negative breast cancer. (D) Expression of FBP1, ERα, E-cadherin, Snail and other EMT markers was determined by Western blotting on five luminal and six BLBC cell lines (MDA-MB231, MDA-MB435, MDA-MB157 and MDA-MB361 are abbreviated to MDA231, MDA435, MDA157 and MDA361 in all Figures). See also Figure S1.

Figure 2. Snail represses FBP1 expression

(A) Schematic diagram showing that Snail was co-expressed with vector (GFP) or FBP1 (CMV promoter) in luminal-subtype breast tumor cells. (B) Snail was co-expressed with vector (GFP) or FBP1 in MCF7 and T47D cells for four days. Morphologic changes indicative of EMT are shown in the phase contrast images; expression of FBP1 and E-cadherin were analyzed by immunofluorescent staining. Nuclei were visualized with DAPI (blue). Scale bar = 20 µm. (C) The mRNA levels of E-cadherin and FBP1 were quantitated by real-time PCR (mean ± SD in three separate experiments). (D) Expression of E-cadherin, FBP1, Snail, luminal markers (ERα and FoxA1) and basal markers (EGFR, SPARC and Caveolin-1) for cells in (B) was analyzed by Western blotting. (E) Schematic diagram showing positions of nine potential Snail-binding E-boxes on the FBP1 promoter and FBP1 promoter luciferase construct used. (F) FBP1 promoter luciferase construct (FL1) was co-expressed with Snail or vector in HEK293, HeLa and MCF7 cells, respectively. After 48 h, luciferase activities were determined and normalized (mean ± SD in three separate experiments). (G) FBP1 promoter luciferase constructs (FL1 and FL2) were co-expressed with Snail or vector in HEK293 cells. Luciferase activities were determined as in (F). (H) FBP1 promoter luciferase constructs (FL2, FL3 and FL4 as well as their E-box mutants) were co-expressed with Snail or vector in HEK293 cells. Luciferase activities were determined as in (F). See also Figure S2.

Figure 3. Snail and G9a are required for H3K9me2 and DNA methylation at the FBP1 promoter

(A) Three sets of primers used for FBP1 promoter ChIP are shown. (B–C) The association of Snail and G9a, and the level of H3K9me2 and H3K9ac at the FBP1 promoter in cells that have undergone Snail-mediated EMT (B) and cell lines from Figure 1D (C) were analyzed by ChIP. DNA methylation at the FBP1 promoter was analyzed by MSP. (D) The association of G9a, and the level of H3K9me2 and DNA methylation at the FBP1 promoter in luminal (25 cases) and triple-negative (16 cases) breast cancer tissues were analyzed by ChIP and MSP, respectively. Horizontal lines represent mean values. Statistical analyses (mean ± SD in three separate experiments) are shown below. (E) Expression of FBP1 and E-cadherin was examined in MDA-MB231 cells with knockdown of G9a expression (top panel). Their mRNA levels were also quantified by real-time PCR (bottom panel). Data are presented as a percentage of non-target control (NTC) values (mean ± SD in three separate experiments in duplicates). See also Figure S3.

Figure 4. FBP1 inhibits glucose uptake and sensitivity and suppresses cell growth under hypoxia

(A) Stable clones with FBP1 expression or knockdown were established in six BLBC and two luminal cell lines, respectively. (B) Glucose uptake was measured. (C) Cells were deprived for glucose for 12 hr followed by glucose stimulation for additional 3 hr. TXNIP expression was examined by Western blotting. (D) Lactate excretion was measured. (E) Cell growth under hypoxic condition was measured by cell-count assay for 2 days. Data are presented as a percentage of vector control values for BLBC cells, whereas data are presented as a percentage of FBP1-knockdown groups for luminal cells (mean ± SD in three separate experiments in triplicates). (F) Oxygen consumption was measured (mean ± SD in three separate experiments in triplicate). For B and D, data are presented as a percentage of vector control values (mean ± SD in three separate experiments in triplicates). For B, D and E, * and #p< 0.01 for vector control cells compared with their FBP1-expressing or FBP1-knockdown clones, respectively. See also Figure S4.

Figure 5. FBP1 inhibits glycolysis and increases OXPHOS

(A) ¹³C₆-Glucose uptake, ¹³C₃-lactate production and the conversion of ¹³C₆-Glucose to ¹³C₃-lactate were measured by 1D ¹H NMR analysis of the media of vector- and FBP1-expressing MDA-MB231 cells grown in ¹³C₆-Glucose (mean ± SEM in duplicate). ¹H NMR spectra from the media is showed in Figure S5D. (B) A pair of representative 1D ¹H{¹³C} HSQC NMR spectra show the changes in ¹³C abundance (represented by the intensity of ¹³C-attached ¹H peaks) of various assigned metabolites elicited by FBP1 expression in MDA-MB231 cells (black: control vector; red: FBP1, bottom panel). The relative ¹³C abundance of indicated metabolites from cell extracts was quantified from their HSQC peak intensity (Fan and Lane, 2008) (mean ± SEM in duplicate). Lac: lactate; AXP: adenine nucleotides; UXP: uracil nucleotides; UDPG: UDP-glucose. (C) Top left panel shows the expected ¹³C (●) labeling patterns of glycolytic and TCA cycle metabolites with ¹³C₆-Glc as tracer. The doubly ¹³C labeled TCA cycle metabolites are derived from the 1^st turn of the TCA cycle while the quadruply ¹³C labeled citrate is produced from the 2^nd turn of the cycle. The levels of several indicated ¹³C isotopologues of glycolytic and TCA cycle metabolites were obtained from the GC-MS analysis of the same cell extracts as in (B) (mean ± SEM in duplicate). See also Figure S5.

Figure 6. FBP1 suppresses PKM2 activation and increases complex I activity and ROS production

(A) F-1,6-BP was measured in FBP1-expressing and FBP1-knockdown clones. (B) Cells were either treated with (+) or without (−) 1% formaldehyde (crosslinker; CL) for 20 min immediately after cell lysis. The states of monomer, dimer and tetramer of PKM2 were analyzed by Western blotting. (C) Schematic diagram showing the electron transfer from mitochondrial complex I to IV. Cells were treated with Rotenone and TTFA, respectively. Oxygen consumption was measured (mean ± SD in three separate experiments in triplicate). (D) Complex I activity was measured from mitochondria isolated from cells in (C). (E) and (F) ROS generation was analyzed by flow cytometry (mean ± SD in three separate experiments in duplicates). Representative images are shown (F). ⁺ and ^‡p< 0.05 for vector control cells compared with their FBP1-expressing or FBP1-knockdown clones, respectively. For A and D, Data are presented as a percentage of vector control values (mean ± SD in three separate experiments in triplicates). * and #p< 0.01 for vector control cells compared with their FBP1-expressing or FBP1-knockdown clones, respectively. See also Figure S6.

Figure 7. FBP1 inhibits tumorsphere-formation and reduces CSC population

(A) Tumorsphere-formation was assessed under normoxic condition. (B–C) The CSC population (CD44^high/CD24^low/EpCAM⁺) was analyzed by flow cytometry. Representative images for MDA-MB231 and MCF7 cells are shown in (C). (D) Schematic diagram of the interaction of β-catenin with TCF4 and FOXO3a. (E) FBP1-expressing BLBC cells (F) as well as in FBP1-knockdown luminal cells (S) were treated with or without NAC for overnight; the interactions of β-catenin with TCF4 and FOXO3a were examined by immunoprecipitating β-catenin following immunoblot of TCF4 and FOXO3a. For A and B, Data are presented as a percentage of vector control values (mean ± SD in three separate experiments in duplicates). * and #p< 0.01 for vector control cells compared with their FBP1-expressing or FBP1-knockdown clones, respectively. See also Figure S7.

Figure 8. FBP1 suppresses tumorigenicity in vitro and in vivo

(A) Data of soft-agar assay are presented as a percentage of vector control cell lines (mean ± SD in three separate experiments with duplicates). ND indicates no colonies detected. (B) Tumor growth was monitored every three days; tumor size and weight were recorded. Data are represented as a mean ± SEM from 6 mice. * and #p< 0.01 for vector control cells compared with their FBP1-expressing or FBP1-knockdown clones, respectively. (C) Kaplan–Meier overall survival curve separates the tumors into two groups based on FBP1 expression. (D) A proposed model to illustrate the transcription repression of FBP1 by Snail in EMT and BLBC, which results in the switch to aerobic glycolysis and increased β-catenin activity. See also Figure S8.