Acetate drives ovarian cancer quiescence via ACSS2-mediated acetyl-CoA production

Abstract

Quiescence is a reversible cell cycle exit traditionally thought to be associated with a metabolically inactive state. Recent work in muscle cells indicates that metabolic reprogramming is associated with quiescence. Whether metabolic changes occur in cancer to drive quiescence is unclear. Using a multi-omics approach, we found that the metabolic enzyme ACSS2, which converts acetate into acetyl-CoA, is both highly upregulated in quiescent ovarian cancer cells and required for their survival. Indeed, quiescent ovarian cancer cells have increased levels of acetate-derived acetyl-CoA, confirming increased ACSS2 activity in these cells. Furthermore, either inducing ACSS2 expression or supplementing cells with acetate was sufficient to induce a reversible quiescent cell cycle exit. RNA-Seq of acetate treated cells confirmed negative enrichment in multiple cell cycle pathways as well as enrichment of genes in a published G0 gene signature. Finally, analysis of patient data showed that ACSS2 expression is upregulated in tumor cells from ascites, which are thought to be more quiescent, compared to matched primary tumors. Additionally, high ACSS2 expression is associated with platinum resistance and worse outcomes. Together, this study points to a previously unrecognized ACSS2-mediated metabolic reprogramming that drives quiescence in ovarian cancer. As chemotherapies to treat ovarian cancer, such as platinum, have increased efficacy in highly proliferative cells, our data give rise to the intriguing question that metabolically-driven quiescence may affect therapeutic response.

Keywords: ACSS2; G0 phase; cell cycle; metabolism.

Figure 1.. Multi-omics analysis identifies ACSS2 as a candidate regulator of quiescence.

(A-B) Schematic representation of RNA-Seq (A) and CRISPR KO screen (B) of serum starved (serum free) Ovcar8 cells. (C) Cross-comparison of RNA-seq and CRISPR KO screen results (cutoffs for RNA-seq: fold-change ≥ 2, FDR ≤ 0.1; Cutoff for CRISPR screen: pval ≤ 0.1). (D) Volcano plot of RNA-seq of serum starved Ovcar8 results showing differential expression of common genes from (C). (E) ACSS2 protein expression was determined by western blotting in the indicated cells. β-actin was used as a loading control. Western blots shown are representative data from at least 2 independent experiments in each cell line and condition.

Figure 2.. Acetate-derived acetyl-CoA abundance is increased in quiescent cells.

(A) Schematic of acetate-mediated acetyl-CoA generation via ACSS2. Red dots indicate labeled carbons for isotopologue tracing. (B) Steady state acetyl-CoA abundance in Ovcar8 cultured in complete or serum free media was assessed by LC-MS. Pooled data from 6 independent experiments is shown (n=8). (C) Acetyl-CoA M+2 percent isotopologue enrichment from labeled acetate in the indicated cells. Representative data from at least 2 independent experiments in each cell line and condition. (D) Schematic of glucose-derived acetyl-CoA via ACLY and PDH. Red dots indicate labeled carbons for isotopologue tracing. (E) Acetyl-CoA M+2 percent isotopologue enrichment from labeled glucose in the indicated cells. Representative data from at least 2 independent experiments in each cell line and condition. Graphs represent mean ± SD. ****p<0.001

Figure 3:. Acetate and ACSS2 drive quiescence in HGSC models.

(A) Ovcar8 cells were transduced with lentivirus expressing empty vector control (EV control) or HA-tagged ACSS2 (ACSS2 OE). ACSS2 and HA protein expression was determined by western blotting. b-actin was used as a loading control. Immunoblots shown are representative data from at least 2 independent experiments. (B) Percent proliferation relative to EV control was assessed after 4 days. Cell counts were assessed using the Incucyte Live Cell Imaging system at endpoint. Pooled data from 3 independent biological replicates (n=4). Graphs represent mean ± SD. (C) Real time imaging of the proliferation of Ovcar8 and Kuramochi cells treated with the indicated doses of acetate. Cell numbers were normalized to the 0 h timepoint. Data represent 1 biological replicate (n=3) of at least 3 independent experiments. (D) Summary of endpoint proliferation data from (C). Pooled data from at least 3 independent experiments (n = 3–4). (E) Proliferation as assessed by real time imaging. Cell numbers were normalized to the 0 h timepoint. Data represent 1 biological replicate (n=3) of at least 3 independent experiments. Horizontal dotted line indicates when media was replaced with fresh media without acetate. (F) Ovcar8 and Kuramochi cells were transduced with a modified FUCCI reporter construct and treated with acetate. Quiescence (cells in G₀) was assessed by flow cytometry. Shown is the CDT1/p27 double positive G₀ population normalized to control. Pooled data from 3 independent experiments is shown. Graphs represent mean ± SD. ****p<0.001

Figure 4.. Acetate affects cell cycle transcriptional signatures in HGSC cells.

(A-B) Ovcar8 cells were treated with 10mM acetate for 72h, and RNA-Seq was performed. (A) Negatively enriched Wikipathways (WP), Hallmark, and KEGG gene signatures in acetate vs. control. (B) Heatmap of genes in a published G0 signature [29].

Figure 5.. ACSS2 expression corresponds to worse outcomes and is increased in ascites compared to matched primary HGSC tumors.

(A) Progression-free and overall survival in patients with ovarian serous carcinoma treated with platin stratified by ACSS2 expression. Data from KMplot.com. (B) ACSS2 expression in HGSC patients from TCGA stratified by their G0 signature [29]. (C) Platinum status in HGSC patients from TCGA with low and high ACSS2 expression. (D) ACSS2 expression in matched primary tumor cells and ascites associated cells was determined by immunofluorescence. n = 3. Graphs represent mean ± SD. ****p<0.01