ACSL4-mediated H3K9 and H3K27 hyperacetylation upregulates SNAIL to drive TNBC metastasis

Abstract

Triple-negative breast cancer (TNBC) has profound unmet medical need globally for its devastating clinical outcome associated with rapid metastasis and lack of targeted therapies. Recently, lipid metabolic reprogramming especially fatty acid oxidation (FAO) has emerged as a major driver of breast cancer metastasis. Analyzing the expression of major FAO regulatory genes in breast cancer, we found selective overexpression of acyl-CoA synthetase 4 (ACSL4) in TNBC, which is primarily attributed to the absence of progesterone receptor. Loss of ACSL4 function, by genetic ablation or pharmacological inhibition significantly reduces metastatic potential of TNBC. Global transcriptome analysis reveals that ACSL4 activity positively influences the gene expression related to TNBC migration and invasion. Mechanistically, ACSL4 modulates FAO and intracellular acetyl-CoA levels, leading to hyperacetylation of particularly H3K9ac and H3K27ac marks resulting in overexpression of SNAIL during the course of TNBC metastatic spread to lymph node and lung. Further, human TNBC metastasis exhibits positive correlation among ACSL4, H3K9ac, H3K27ac, and SNAIL expression. Altogether, our findings provide molecular insights regarding the intricate interplay between metabolic alterations and epigenetic modifications, intertwined to orchestrate TNBC metastasis, and posit a rational understanding for the development of ACSL4 inhibitors as a targeted therapy against TNBC.

Keywords: ACSL4; SNAIL; TNBC; histone acetylation; metastasis.

Fig. 1.

ACSL4 is selectively upregulated in TNBC subtype and PR negatively regulates ACSL4 expression. (A) Heat map depicting mRNA expression analysis of the key components of the FAO pathway in human Non-TNBC and TNBC tumors from the TCGA BRCA dataset, which was publicly accessible via UCSC Xena Browser (Left). The violin plot represents the relative ACSL4 expression between Non-TNBC and TNBC tumors according to the TCGA dataset (Right). (B) mRNA expression analysis of different ACSL isoforms in a panel of human breast cancer cell lines. The color key at the top indicates high (in blue-reduced expression) and low (in red-increased expression) ∆Ct values in which ∆Ct is the difference between Ct value of the target gene minus Ct value of the reference gene. (C) Western blot analysis of ACSL4 protein levels in a panel of human breast cancer cell lines. β-actin was used as a loading control. (D) Analysis of ACSL4 expression in TNBC compared to Non-TNBC patient tumor tissue samples. The dot plot represents the fold change values of ACSL4 mRNA expression, data represent mean ± SEM, n = 15 (TNBC), and n = 18 (Non-TNBC) patient tumor tissue samples. (E) Immunohistochemical staining of formalin-fixed paraffin-embedded sections of human Non-TNBC and TNBC primary tumors using anti-ACSL4 antibody (Left). Representative photomicrographs were shown at 40× magnification. [Scale bar, 10 μm (40×).] Quantitative H-scores for Non-TNBC (n = 10) and TNBC (n = 11) primary tumors were calculated for ACSL4 expression and represented as dot plot (Right); error bar, mean ± SEM. (F) Immunoblot analysis of ACSL4 protein levels in control, ESR1, PGR, and ERBB2 KD MCF-7 cells. GAPDH was used as a loading control. (G) mRNA expression analysis of ACSL4 in ESR1, PGR, and ERBB2 KD compared to control MCF-7 cells by RT-PCR. Data points represents mean ± SD for three independent experiments. (H) The MCF-7 cells were treated with either vehicle control or 20 μM of RU486 (PR antagonist) for 36 h followed by the analysis of ACSL4 expression by qPCR. Data points represent mean ± SD for three independent experiments. (I) Diagrammatic scheme showing genomic location for the PREs on the ACSL4 promoter (Top). Bottom panel showing PR binding motif obtained from JASPAR database. (J) ChIP-qPCR data in control MCF-7 cells depicting recruitment of PR on the ACSL4 promoter using primer pairs targeting −0.9, −0.6, −0.5, and −0.3 kb of ACSL4 promoter region. (K) ChIP-qPCR data depicting the enrichment of PR on the ACSL4 promoter at −0.3 kb and −0.6 kb regions upstream of TSS in control or RU486 (20 μM) treated MCF-7 cells for 36 h. (J and K) Each bar represents a mean of triplicate readings ± SD of fold enrichment normalized to IgG. (L) Luciferase reporter activity of the ACSL4 promoter −1 kb region from TSS in control and PGR KD MCF-7 cells. (M) Same as in (L) except either vehicle control or RU486 (20 μM) treated MCF-7 cells were used. (L and M) Data points represent a mean of triplicate readings of samples ± SD. (N) Western blot analysis of PR B/A and ACSL4 in PGR B/A over expression (OE) MDA-MB-231 cells with their respective control. β-actin and/or GAPDH was used as a loading control for the respective immunoblots. For panels, (A) (Right), (D) unpaired two-tailed Welch’s t test, (E) (Right), (H, L, and M) Student’s t test (two-tailed), (G) one-way ANOVA Dunnett’s multiple comparisons test, (J and K) two-way ANOVA Sidak’s multiple comparisons test were applied.

Fig. 2.

Loss of ACSL4 function impairs TNBC growth and metastasis. (A) In vivo bioluminescence images of 4T-1 tumor-bearing control (Top), and ACSL4 KD (Bottom) mice (n = 4 per group). The color scale indicates the bioluminescence intensity (counts/sec) emitted from each group. Quantitative bar graph representation of the metastatic burden in terms of bioluminescence intensity, calculated from the region of interest (ROI) is shown, (n = 4) each group (Middle). Each bar represents mean ± SEM. Representative bioluminescence images of the lung, liver, and spleen harvested from control and ACSL4 KD tumor-bearing mice (Right). (B) The schematic representation of the in vivo experimental plan. (C) Representative flow cytometry-derived scatter plots (Left) showing control (Td-Tomato⁺) and ACSL4 KD (Td-Tomato⁻) cells in primary tumors and metastatic organs [CTCs (n = 6), lymph node (n = 4), lung (n = 3), liver (n = 2), and peritoneal fluid (n = 6)]. The quantitative bar graph represents the flow cytometry analysis of the percentage of control (Td-Tomato⁺) and ACSL4 KD (Td-Tomato⁻) cells (Right). Data are represented as mean ± SD. (D) Representative bioluminescence images of 4T-1 tumor-bearing nude mice (Left), treated with either vehicle (Top) or PRGL493, Bottom panel (n = 4 per group). The color scale indicates the bioluminescence intensity (counts/sec) emitted from each group. Quantitative bar graph representation (Top Right) of the metastatic burden, (n = 4) each group. Each bar represents mean ± SEM. Representative bioluminescence images of the lung, liver, and spleen harvested from control and PRGL493-treated tumor-bearing mice (Bottom Right). 4T1 allograft experiments described in (A and D), same tumor-bearing mice (n = 4) were used as control for both the cases. (E) Representative bioluminescence images of MDA-MB-231 tumor-bearing NSG mice (n = 4 per group), treated with either vehicle (Top) or PRGL493 (500 μg/kg/every other day), Bottom. The color scale indicates the bioluminescence intensity (counts/sec) emitted from each group. Quantitative bar graph representation (Top Right) of the metastatic burden, (n = 4) each group. Each bar represents mean ± SEM. Bioluminescence analysis of the metastatic signal in the organs harvested from control and PRGL493-treated tumor-bearing mice (Bottom Right). For panels (A, C, D, and E), two-way ANOVA Sidak’s multiple comparisons tests were applied.

Fig. 3.

Transcriptome analysis revealed that loss of ACSL4 function markedly inhibits pathways central to TNBC migration and invasion. (A) Schematic representing the key steps involved in assessing the impact of loss of ACSL4 function on the global transcription network through high-throughput transcriptome analysis with RNA isolated from control, ACSL4 KD, and PRGL493 (20 µM, for 48 h)-treated MDA-MB-231 cells (n = 3 per group). (B) Volcano plots depicting differential gene expression pattern in ACSL4 KD vs. control (Top) and PRGL493 vs. control (Bottom). The downregulated and upregulated genes are shown in blue and red, [Fold change (FC) >1.5 for upregulated and FC < 1.5 for downregulated genes, P < 0.05], respectively. (C) Heat maps showing the differential gene expression pattern in ACSL4 KD and PRGL493-treated cells compared to control [Color scale: green (downregulated), black (no change), and red (upregulated)]. (D) Venn diagram depicting the number of DE genes upon ACSL4 KD and PRGL493 treatment with respect to control. (E) Gene Ontology Enrichment Analysis was performed using the 312 common DEGs that were altered in both ACSL4 KD and PRGL493-treated conditions compared to control to investigate the different biological processes that were affected, as predicted by Panther software. (F) Heat map shows differential expression of top 23 downregulated genes that are related to the regulation of cell migration, cell motility, and EMT. Expression is represented as a plot of normalized read counts for control, ACSL4 KD, and PRGL493 treatment (n = 3 per group). (G) qPCR validation of top 25% of downregulated cell migration-related genes identified by RNA-seq analysis in ACSL4 KD compared to control. Each bar represents mean ± SD, n = 3. (H) Same as (G), except PRGL493 treatment was compared to control. For panels, (G and H) two-way ANOVA Sidak’s multiple comparisons test was applied.

Fig. 4.

ACSL4 promotes histone H3 acetylation and positively regulates SNAIL expression in TNBC. (A and B) ACSL activity was assessed in ACSL4 KD (A) and PRGL493-treated (20 µM for 48 h). (B) MDA-MB-231 cells compared to control using acyl-CoA synthetase assay kit following the manufacturer’s instruction. (C) Estimation of AC levels in ACSL4 KD and PRGL493-treated MDA-MB-231 cells compared to control (n = 5; mean ± SD). (D and E) The acetyl-CoA production was detected using the acetyl-CoA assay kit following the manufacturer’s protocol. (D) The quantitative bar graph representation of acetyl-CoA levels in ACSL4 KD MDA-MB-231 cells compared to control. (E) Same as (D), except MDA-MB-231 cells treated either with vehicle control or PRGL493 (20 µM) for 48 h. (F–I) Immunoblot analysis of different histone H3 acetylation marks in MDA-MB-231 cells. β-actin and total H3 were used as loading control. H3K9, H3K27, H3K14, H3K23, H3K18, and H3K56 acetylation levels in control and ACSL4 KD cells (F). H3K9, H3K27, H3K14, H3K23, and H3K18 acetylation levels in control and ACSL4 OE cells (G), and control and PRGL493 (20 µM) for 48 h treated cells (H). (I) The histone acetylation levels were assessed in the presence or absence of synthetic acetyl-CoA (acetyl-CoA trisodium salt) (20 µM) in control and ACSL4 KD cells. (J) Representative images of the wound healing assay in control and ACSL4 KD cells after exogenous supplementation of synthetic acetyl-CoA (20 µM), magnification 10×. [Scale bar, 50 µm (Left).] The quantitative bar graph of the wound healing assay is shown (Right). (K) Representative images of the transwell chamber invasion assay in control and ACSL4 KD MDA-MB-231 cells after supplementation of synthetic acetyl-CoA (20 µM), magnification 10×. [Scale bar, 50 µm (Left).] The quantitative bar graph for invasion assay is shown (Right). (L) mRNA expression of genes that exhibited significant alterations in the RNA-seq analysis were examined by qPCR after supplementation of acetyl-CoA in control and ACSL4 KD MDA-MB-231 cells, represented in bar graph. Data points represent a mean of triplicate readings of samples ± SD. (M) Scatter plot showing the positive correlation between ACSL4 and SNAI1 mRNA expression in TNBC samples as analyzed by the bc-GenExMiner v5.0 web tool. Pearson pairwise correlation coefficient on 293 TNBC samples shows a positive correlation (r = 0.22) between ACSL4 and SNAI1 and is significantly associated with each other (P = 0.0002). (N) Scatter plot depicting insignificant correlation between ACSL4 and EGR3 mRNA expression in TNBC samples (r = 0.04, P = 0.5388). (O) Kaplan–Meier survival analysis for ACSL4 and SNAI1 expression in TNBC patients (n = 116). Patients were categorized into four groups based on the median expression levels of ACSL4 and SNAI1: High ACSL4 & High SNAI1, High ACSL4 & Low SNAI1, Low ACSL4 & High SNAI1, and Low ACSL4 & Low SNAI1 respectively. Data were retrieved from TCGA-BRCA dataset. (P–R) Immunoblot analysis of SNAIL expression in control, ACSL4 KD, ACSL4 KD supplemented with acetyl-CoA (P), control and ACSL4 OE (Q), and after treatment with vehicle control or PRGL493 (R) in MDA-MB-231 cells. β-actin was used as a loading control. For panels (A, B, D, E, J, K, and L) each bar represents mean of triplicate readings of samples ± SD. For panels (A, B, D, and E) Student’s t test (two-tailed), (C and J) two-way ANOVA Dunnett’s multiple comparisons test, (K) two-way ANOVA Sidak’s multiple comparisons test, (L) two-way ANOVA Tukey’s multiple comparisons test and (O) log rank test were applied.

Fig. 5.

ACSL4 upregulates SNAIL expression by enriching acetylated H3K9 and H3K27 in the SNAI1 promoter. (A) Diagram illustrating the different sites examined for the enrichment of histone acetylation marks at SNAI1 promoter. (B) ChIP-qPCR data depicting the enrichment of H3K9ac, H3K27ac, H3K23ac, and H3K14ac using walking primer pairs targeting −1.5 kb to TSS of the SNAI1 promoter. (C–F) ChIP-qPCR data showing the enrichment of H3K9ac and H3K27ac at the SNAI1 promoter, respectively. (C and D) In control, ACSL4 KD, and after supplementation of synthetic acetyl-CoA in ACSL4 KD MDA-MB-231 cells. (E and F) In ACSL4 OE MDA-MB-231 cells compared to control. Each bar represents a mean of triplicate readings ± SD of fold enrichment normalized to IgG. (G) Representative confocal microscopy images of control and ACSL4 KD cells MDA-MB-231 cells costained with ACSL4 (red) and H3K9ac (green), or ACSL4 (red) and H3K27ac (green), or ACSL4 (red) and SNAIL (green) antibodies, magnification 60×. (Scale bar, 50 μm.) For panels, (B) two-way ANOVA Dunnett’s multiple comparisons test, (C and D) two-way ANOVA Tukey’s multiple comparisons test, and (E and F) two-way ANOVA Sidak’s multiple comparisons test were applied.

Fig. 6.

Loss of H3K9ac, H3K27ac, or SNAIL expression attenuates migration and invasion in ACSL4 overexpressing TNBC cells. (A) Immunoblot analysis of ACSL4, H3K9ac, H3K27ac, and SNAIL in control and ACSL4 OE MDA-MB-231 cells, either treated with vehicle control or C646 (P300 inhibitor) (10 µM). (B) Same as (A) except either vehicle control or PU139 (pan-HAT inhibitor) (10 µM) treated control and ACSL4 OE MDA-MB-231 cells were used. β-actin and total H3 were used as loading control. (C) Representative images of the wound healing assay in control and ACSL4 OE cells, either treated with vehicle, C646, or PU139, respectively at 0 h and 24 h. Magnification 10×. (Scale bar, 50 µm.) (D) The quantitative bar graph of the wound healing assay is shown. (E) Representative images of the transwell chamber invasion assay in control and ACSL4 OE cells treated with either vehicle, C646, or PU139, respectively at 24 h. Magnification 10×. (Scale bar, 50 µm.) (F) The quantitative bar graph for invasion assay is shown. (G) Immunoblot analysis to confirm SNAI1 knockdown by using two different shRNAs in ACSL4 OE MDA-MB231 cells. β-actin was used as a loading control. (H) Representative images of the wound healing assay upon SNAI1 knockdown in ACSL4 OE MDA-MB-231 cells at 0 h and 24 h, magnification 10×. (Scale bar, 50 µm.) (I) The quantitative bar graphs of the wound healing assay are shown. (J) Representative images of the transwell chamber invasion assay upon SNAI1 knockdown in ACSL4 OE MDA-MB-231 cells at 24 h, magnification 10×. (Scale bar, 50 µm.) (K) The quantitative bar graphs of the invasion assay are shown. For panels (D, F, I, and K) each bar represents mean of triplicate readings of samples ± SD, For panels, (D and F) two-way ANOVA Tukey’s multiple comparisons test, (I) two-way ANOVA Sidak’s multiple comparisons test, and (K) one-way ANOVA Dunnett’s multiple comparisons test were applied.

Fig. 7.

ACSL4-H3K9/H3K27ac-SNAIL expression is significantly elevated in human TNBC metastasis. (A) MDA-MB-231 cells isolated from primary tumor, metastatic lung, and lymph node and subjected to confocal microscopy analysis after having either costaining ACSL4 (red) and H3K9ac (green), or ACSL4 (red) and H3K27ac (green), or ACSL4 (red) and SNAIL (green) antibodies. In all cases, DAPI was used to stain the nuclei. Magnification 60×. (Scale bar, 50 and 10 μm.) (B) Immunohistochemical staining of formalin-fixed paraffin-embedded sections of human TNBC primary tumors and TNBC metastatic tissues using anti-ACSL4, anti-H3K9ac, anti-H3K27ac, and anti-SNAIL antibodies. Representative photomicrographs are shown at 10× and 40× magnifications (Inset). [Scale bar, 50 μm (10×) or 10 μm (40×).] (C–F) Quantitative H-scores for TNBC primary tumors (n = 11) and TNBC metastatic tissues (n = 7) were calculated for ACSL4 (C), H3K9ac (D), H3K27ac (E), and SNAIL (F) expression and represented as dot plots; error bar, mean ± SEM. For panels (C–F) Student’s t test (two-tailed) were applied.

Fig. 8.

ACSL4-driven histone H3K9 and H3K27 hyperacetylation and SNAIL upregulation promote TNBC metastasis Illustration depicting heightened ACSL4 activity plays a pivotal role in fostering elevated FAO and cellular acetyl-CoA levels. This, in turn, leads to enhanced enrichment of histone H3K9ac and H3K27ac on the SNAI1 promoter, resulting in SNAIL upregulation. Consequently, this molecular cascade drives increased migration, invasion, and metastasis in TNBC.