Loss of STARD7 Triggers Metabolic Reprogramming and Cell Cycle Arrest in Breast Cancer

Abstract

Cancer cells adapt their metabolism to support aberrant cell proliferation. However, the functional link between metabolic reprogramming and cell cycle progression remains largely unexplored. Mitochondria rely on the transfer of multiple lipids from the endoplasmic reticulum (ER) to their membranes to be functional. Several mitochondrial-derived metabolites influence cancer cell proliferation by modulating the epigenome. Here, the loss of STARD7, a lipid transfer protein whose expression is enhanced in breast cancer, is shown to lead a metabolic reprogramming characterized by the accumulation of carnitine derivatives and S-Adenosyl-L-methionine (SAM). Elevated SAM levels cause the increase of H3K27 trimethylation on many gene promoters coding for candidates involved in cell cycle progression. Likewise, STARD7 deficiency triggers cell cycle arrest and impairs ERα-dependent cell proliferation. Moreover, EGFR signaling is also impaired in triple negative breast cancer cells lacking STARD7, at least due to deregulated EGFR trafficking to lysosomes. Therefore, mitochondria rely on STARD7 to support cell cycle progression in breast cancer.

Keywords: EGFR signaling; breast cancer; cell cycle arrest; lipid transfer protein; metabolic reprogramming.

Figure 1

STARD7 overexpression in breast cancer. A) STARD7 expression in multiple human cancer types. TCGA data with mRNA expression levels of STARD7 in human cancer types are illustrated. B) STARD7 expression at the single cell level. Processed data were collected from.^{[ 53 ]} The UMAP clustering from breast tumor BC‐P1 (CID4471) is illustrated. C,D) Enrichment of STARD7 mRNAs in breast cancer versus normal adjacent tissues (TCGA data). E) STARD7 overexpression in human breast cancer. STARD7 mRNA levels were quantified in the following cases: 1) Breast (144), 2) Benign Breast Neoplasm (3), 3) Breast Carcinoma (14), 4) Breast Phyllodes Tumor (5), 5) Ductal Breast Carcinoma in Situ (10), 6) Invasive Breast Carcinoma (21), 7) Invasive Ductal and Invasive Lobular Breast Carcinoma (90), 8) Invasive Ductal Breast Carcinoma (1556), 9) Invasive Lobular Breast Carcinoma (148), 10) Medullary Breast Carcinoma (32), 11) Mucinous Breast Carcinoma (46), and 12) Tubular Breast Carcinoma (67), as described.^{[ 54 ]} F) Enrichment of STARD7 expression in the basal subtype of breast cancer. STARD7 expression in all indicated breast cancer subtypes was established using TISIDB (Tumor and Immune System Interaction Database) (http://cis.hku.hk/TISIDB/index.php).^{[ 55 ]} G) Enhanced STARD7 protein levels in human breast cancer. Extracts from ER⁺ and TNBCs as well as from corresponding adjacent normal tissues were subjected to Western blot (WB) analyses using the indicated antibodies.

Figure 2

STARD7 deficiency leads to changes in mitochondrial morphology in breast cancer cells. A) Enhanced number and size of mitochondria upon STARD7 deficiency. A quantitative analysis of mitochondria observed by Transmission Electron Microscopy (TEM) in control and STARD7‐depleted breast cancer cells is illustrated. The average number of mitochondria by cytoplasmic area and the average area of mitochondria are shown (top and bottom histograms, respectively) (***p < 0.001). WB analyses using extracts from control and STARD7‐depleted MCF7 cells are also illustrated on the top. B,C) Enhanced mitochondria‐ER contact sites in breast cancer cells lacking STARD7. TEM analysis of mitochondria/ER contacts in control and STARD7‐depleted MCF7 cells (B). On the left, illustration of the ultrastructure of mitochondria/ER contacts in both control and STARD7‐depleted MCF7 cells. On the right, number of mitochondria/ER contact sites and average percentage of mitochondrial perimeter in contact with the ER are plotted (top and bottom histograms, respectively) (see methods for details on statistical analyses). Control and STARD7‐depleted MCF7 cells were co‐stained with the marker of active mitochondria TMRE and with ER‐Tracker green (Endoplasmatic Reticulum marker) in order to show contact sites between both mitochondria and ER (C). Hoechst stainings were done to show the nucleus. Illustrated graphs represent the degree of colocalization between ER and mitochondria, using the Pearson's coefficient. A total of 51 control and 43‐depleted MCF7 cells were analyzed. The Prism10 program was used for statistical analyses (Welch's t‐test, **p < 0.01). D) Enrichment of some but not all MAMs markers in MAMs upon STARD7 deficiency. Extracts from MCF7 cells were biochemically fractionated to generate organelle‐enriched extracts. ER = endoplasmic reticulum, Cyto = cytoplasm, Crude mito = crude mitochondria, MAM = mitochondria‐associated membrane, Mito = mitochondria. The organelle‐enriched extracts were subjected to Western blots (WB) using the indicated antibodies. E) Elevated levels of ER stress markers in breast cancer cell lacking STARD7. Extracts from control and STARD7‐depleted MCF7 cells were subjected to WB analyses using the indicated antibodies. Experiments were repeated at least three times. Representative blots are shown.

Figure 3

STARD7 deficiency leads to a metabolic and an epigenetic‐mediated transcriptional reprogramming. A) The metabolic signature of both control and STARD7‐depleted MCF7 cells was established on cells from three independent depletions (see methods for details). An Enrichment overview was conducted and the top 20 upregulated pathways in cells lacking STARD7 are illustrated. B) Enrichment of carnitine derivatives and S‐Adenosyl‐L‐methionine (SAM) in breast cancer cells lacking STARD7. A HeatMap was generated using the metabolomic data from control and STARD7‐depleted MCF7 cells. Metabolites whose level dramatically change upon STARD7 deficiency are illustrated. C) Defective ACC expression upon STARD7 deficiency in breast cancer cells. Extracts from MCF7, T47D or MDA‐MB231 cells were subjected to WB analyses using the indicated antibodies. Experiments were repeated at least three times. Representative blots are shown. D) Enhanced SAM levels and methylation potential in breast cancer cells lacking STARD7. The plotted data were extracted from the metabolomic data. *p < 0.05, **p < 0.01, ***p < 0.001). E,F) Enrichment of H3K27 trimethylation on gene candidates involved in cell proliferation and in the response to oxidative stress. H3K27me3 ChIP‐seq analysis of control and STARD7‐depleted MCF7 cells (Sh Control, « CTL » and sh STARD7#2, « DEP », respectively) were conducted on cells from three independent depletions. Peak profiles and heatmaps depicting genome‐wide H3K27me3 peaks aligned by intensity of reads surrounding the TSS (n = 3 merged experiments) are illustrated (E). Color key indicates relative intensity of peaks. Annotated feature distribution of H3K27me3 peaks in control and STARD7‐depleted MCF7 cells (« CTL » and « DEP », respectively) within respective genomic regions are shown (F). G,H) Gene tracks of H3K27me3 enrichment on NUF2 (G) and TXNIP (H) genes, respectively (n = 3 merged experiments). TSS‐transcription start site. Extracts from control and STARD7‐depleted MCF7 (G and H) and MDA‐MB231 (H) cells were subjected to WB analyses using the indicated antibodies for validation purposes. Experiments were repeated at least three times. Representative blots are shown.

Figure 4

STARD7 deficiency in breast cancer cells leads to defects in cell cycle progression. A) Deregulated transcriptional signature in breast cancer cells lacking STARD7. A volcano plot illustrating the up and downregulated candidates (blue and red dots, respectively) upon STARD7 deficiency in MCF7 cells (n = 3). B) STARD7 deficiency causes cell cycle defects in breast cancer cells. A Gene Ontology for biological processes was carried out with transcriptomic data obtained with control versus STARD7‐depleted MCF7 cells (n = 3). C) Top 30 candidates downregulated upon STARD7 deficiency in breast cancer cells. A HeatMap showing the most downregulated candidates in MCF7 cells lacking STARD7 is shown (n = 3). D) STARD7 deficiency impairs the expression of candidates involved in G2M checkpoint as well as c‐Myc targets. GSEA were carried out with transcriptomic data obtained with control versus STARD7‐depleted MCF7 cells (n = 3). E) Impaired expression of candidates involved in cell cycle progression in breast cancer cells lacking STARD7. Extracts from control and STARD7‐depleted MCF7 or MDA‐MB231 cells were subjected to WB analyses using the indicated antibodies. Experiments were repeated at least three times. Representative blots are shown. F) Cell cycle progression defects upon STARD7 deficiency. The percentage of cells in G1, S, or G2 phases was quantified in both control and STARD7‐depleted MCF7 cells by FACS. Two experiments carried out in duplicates are shown and 10⁴ cells were analyzed for each experimental condition (Student's T‐test, **p < 0.01, ***p < 0.001, ****p < 0.0001).

Figure 5

STARD7 promotes the expression of candidates involved in DNA replication in breast cancer cells. A) HeatMap of differentially expressed candidates upon STARD7 deficiency in MCF7 cells from three independent experiments (proteomic analysis). B) STARD7 controls the expression of candidates involved in DNA replication. On the left, data sets of downregulated candidates upon STARD7 deficiency from both transcriptomic and proteomic analyses were compared. On the right, the top 19 significantly downregulated gene sets found in both transcriptomic and proteomic analyses upon STARD7 deficiency in MCF7 cells are shown (HALLMARK analysis) (Mann–Whitney U test) (n = 3). C) STARD7 promotes the expression of candidates involved in kinetochore formation. Protein extracts from control and STARD7‐depleted MCF7, T47D and BT549 cells were subjected to WB analyses using the indicated antibodies. Representative blots from three independent experiments are shown (MCF7 and T47D cells) and from one additional confirmatory experiment done in BT549 cells.

Figure 6

STARD7 deficiency in breast cancer cells impairs ERα signaling. A) Impaired ERα expression upon STARD7 deficiency in breast cancer cells. Extracts from control and STARD7‐depleted MCF7 cells from at least three independent depletions were subjected to WB analyses using the indicated antibodies. Representative blots are shown. B) Impaired estrogens‐dependent signaling upon STARD7 deficiency in breast cancer cells. Control and STARD7‐depleted T47D or MCF7 cells incubated in 5% charcoal‐treated serum were treated or not with estrogens (E2, 10 nM) for the indicated periods of time and the resulting extracts were subjected to WB analyses using the indicated antibodies. Experiments were repeated three times. Representative blots are shown. C) Impaired induction of GREB1 by estrogens in breast cancer cells lacking STARD7. Serum‐starved control and STARD7‐depleted T47D cells were treated or not with E2 (10 nM) for the indicated hours and the resulting extracts were subjected to Real‐Time PCR analyses. mRNA levels of GREB1 in untreated control cells were set to 1 and levels in other experimental conditions were relative to that after normalization with GAPDH levels. Data are from three independent experiments (mean ± SD, two‐way ANOVA, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). D) STARD7 deficiency in breast cancer cells impairs cell proliferation induced by estrogens. Control and STARD7‐depleted T47D cells incubated in 5% charcoal‐treated serum were treated or not with E2 (10 nM) for the indicated hours and the resulting cells were subjected to FACS analyses to quantify the percentage of proliferative cells. Experiments were repeated three times (mean ± SEM, two‐way ANOVA with Dunnettes post‐test **p < 0.01, ***p < 0.001).

Figure 7

STARD7 deficiency impairs EGFR signaling. A) STARD7 deficiency impairs EGFR expression in TNBC‐ but not in ER⁺‐derived breast cancer cells. Control and STARD7‐depleted MCF7 or MDA‐MB231 cells were subjected to WB analyses using the indicated antibodies (left and right panels, respectively). Experiments were repeated at least three times. Representative blots are shown. B) EGFR signaling relies on STARD7 in TNBC‐derived breast cancer cells. Serum‐starved control and STARD7‐depleted MDA‐MB231 or BT549 cells were treated or not with EGF (50 ng ml⁻¹) for the indicated periods of time and the resulting extracts were subjected to WB analyses using the indicated antibodies (top and bottom panels, respectively). Cell extracts were collected 12 days post‐infection. Experiments were repeated at least twice on each cell line. Representative blots are shown. C) STARD7 deficiency enhances EGFR degradation through lysosomes. Control or STARD7‐depleted MDA‐MB231 cells were pretreated with Cycloheximide (CHX, 50 µg ml⁻¹) and with DMSO (vehicle), MG132 (25 µM) or Bafilomycin A1 (BafA1) (0.1 µM) for 2 hours and subsequently subjected or not to EGF for the indicated periods of time (top, middle, and bottom panels, respectively). The resulting cell extracts from two independent experiments were subjected to WB analyses. Representative blots are shown. D) Enlarged early endosomes in breast cancer cells lacking STARD7. Control or STARD7‐deficient MDA‐MB231 cells were serum starved and subsequently untreated or stimulated with EGF for 30 minutes and subjected to immunofluorescence analyses. EGF‐bound to EGFR (green) as well as EEA1⁺ endosomes (red) were detected. Nuclei were visualized by DAPI stainings. On the right, a quantification of enlarged (> 1 µm) EEA1⁺ endosomes per cell in EGF‐stimulated control and STARD7‐depleted cells is illustrated. 6 fields for each experimental condition were analyzed (314, 325, and 232 cells for shRNAs Control, STARD7#2 and STARD7#4, respectively) (one‐way ANOVA, ***p < 0.001; *p < 0.05). E) STARD7 deficiency decreases the number of endosomes with tubular structures. Electron micrographs showing endosomes (E) with tubular structures (arrows) in control versus STARD7‐depleted MDA‐MB231 cells are shown. Around thirty photos were randomly taken in both control and STARD7‐depleted cells at the level of endosomes at a magnification of 10 000.

Figure 8

S‐Adenosylmethionine triggers molecular changes seen in breast cancer cells lacking STARD7. A. STARD7 overexpression does not change levels of candidates involved in cell proliferation. Extracts from MCF10A cells overexpressing an empty vector (negative control) or STARD7 were subjected to WB analyses using the indicated antibodies (n = 2). B,C) SAM stimulation decreased protein levels of proteins involved in cell proliferation. MCF7 (B), T47D (B), MDA‐MB231, or BT549 (C) breast cancer cells were treated or not with SAM for 72 hours at the indicated concentrations and the resulting extracts were subjected to WB analyses using the indicated antibodies. Experiments were repeated twice for each cell line (MCF7, T47D, and MDA‐MB231 cells). Results were confirmed once in BT549 cells. Representative blots are shown.

Figure 9

STARD7 deficiency potentiates ciliogenesis. A) A ciliogenesis signature is found in breast cancer cells lacking STARD7. A Gene Ontology for biological processes was carried out with RNA Sequencing data from control and STARD7‐depleted MCF7 cells (three independent depletions). B) Identification of a ciliogenesis signature upon STARD7 deficiency in breast cancer cells. A HeatMap generated with data from the RNA Sequencing experiments is shown. The most differentially expressed candidates between control and STARD7‐depleted MCF7 cells are illustrated (n = 3). C) STARD7 deficiency is linked to ciliogenesis. Gene Set Enrichment analyses (GSEA) were carried out with data from RNA sequencing experiments (n = 3). D) Enhanced mRNA levels of ciliogenesis genes in breast cancer cells lacking STARD7. Real‐Time PCR analyses were carried out with total RNAs from control and STARD7‐depleted T47D cells. mRNA levels of the indicated candidates were set to 1 in control cells and levels in other experimental conditions were relative to that after normalization with GAPDH levels (mean ± SEM, one‐way ANOVA with Dennett's post‐test, *p < 0.05, **p < 0.01, ***p <0.001, ****p < 0.0001, n = 8 distinct experiments). E) Enhanced ciliogenesis upon STARD7 deficiency. Immunofluorescence analyses were carried out with control and STARD7‐depleted hTERT‐RPE1 cells using the anti‐Arl13b to visualize the primary cilium. The length of the primary cilium in control and STARD7‐depleted cells was quantified (histogram on the right) (mean ± SEM, unpaired T‐test, **p < 0.01) (see methods for details). WB analyses with extracts from all experimental conditions are also illustrated.

Figure 10

STARD7 deficiency leads to autophagy. A) Enrichment of lysosomal proteins in breast cancer cells lacking STARD7. A Heatmap of proteins involved in GO: Lysosome 0005764 was generated with data from control and STARD7‐depleted MCF7 cells. Data was obtained from proteomics analysis, hierarchical clustering on raw was performed using one minus Pearson's correlation with average linkage method. Relative coloring scheme was applied. B) Enhanced levels of the conjugated form of LC3B in breast cancer cells lacking STARD7. On the top, extracts from control and STARD7‐depleted MCF7 or MDA‐MB231 cells treated or not with Chloroquine (25 µM, 24 hours) were subjected to Western blot analyses using the indicated antibodies. Experiments were repeated at least three times. Representative blots are shown. At the bottom, histograms showing the LCB3II/LCB3I (unconjugated and conjugated forms of LC3B, respectively) ratio after Chloroquine (25 µM) overnight treatment in MCF7 cells. Ratios were calculated for at least three independent experiments (mean ± SEM, paired T‐test, *p < 0.05, **p < 0.01). C) Enhanced autophagy in breast cancer cells lacking STARD7. On the left, extracts from control and STARD7‐depleted MCF7, T47D, and MDA‐MB231 cells were subjected to WB analyses using the indicated antibodies. Experiments were repeated at least three times. Representative blots are shown. In the middle, FACS analyses using the lysotracker dye to monitor autophagy in control and STARD7‐depleted breast cancer cells. On the right, histograms illustrating the quantification of FACS data using the lysotracker. The median fluorescence intensity was set to 1 in control cells and values in other experimental conditions were relative to that. Data are from at least three independent experiments per cell line (mean ± SEM, one‐way ANOVA with Dennett's post‐test, *p < 0.05, **p < 0.01, ***p < 0.001). D) STARD7 deficiency causes autophagy in breast cancer cells. Representative immunofluorescence analyses of control and STARD7‐depleted MCF7 cells (shRNA Control and shRNA STARD7#2) after an overnight treatment with Chloroquine (15 µM). Anti‐p62 (red) and LC3B (green) stainings were carried out. DAPI was used for nuclei stainings. The bar graph shows a quantitative analysis of cells undergoing autophagy. The quantification shows the percentage of cells with more than 10 co‐localized points to the total number of analyzed cells (Student's T‐test, **p < 0.01). E) The ciliogenesis signature of breast cancer cells lacking STARD7 results from autophagy. MCF7 cells were treated or not with DMSO (vehicle), Rapamycin (20 nM), Chloroquine (25 µM) or were serum‐starved and mRNA levels of the indicated candidates were quantified by Real‐Time PCR analyses. Data are from 5 independent experiments (mean ± EM, one‐way ANOVA with Dennett's post‐test, *p < 0.05, **p < 0.01, ***p < 0.001).