Akt-dependent metabolic reprogramming regulates tumor cell histone acetylation

Abstract

Histone acetylation plays important roles in gene regulation, DNA replication, and the response to DNA damage, and it is frequently deregulated in tumors. We postulated that tumor cell histone acetylation levels are determined in part by changes in acetyl coenzyme A (acetyl-CoA) availability mediated by oncogenic metabolic reprogramming. Here, we demonstrate that acetyl-CoA is dynamically regulated by glucose availability in cancer cells and that the ratio of acetyl-CoA:coenzyme A within the nucleus modulates global histone acetylation levels. In vivo, expression of oncogenic Kras or Akt stimulates histone acetylation changes that precede tumor development. Furthermore, we show that Akt's effects on histone acetylation are mediated through the metabolic enzyme ATP-citrate lyase and that pAkt(Ser473) levels correlate significantly with histone acetylation marks in human gliomas and prostate tumors. The data implicate acetyl-CoA metabolism as a key determinant of histone acetylation levels in cancer cells.

Figure 1. Glucose availability regulates histone acetylation in several cancer cell lines

(A) Acetylation of acid-extracted histones from cells cultured under indicated glucose conditions for 24 hours. Total histones were stained by Coomassie or Ponceau. (B) Acetylation of acid extracted histones from LN229 cells treated with 1 or 10 mM glucose, +/− 5 mM NaOAc for 24 hours (C) Relative levels of acetyl-CoA in LN229 cells after 24 hours of indicated treatment, mean +/− SEM of triplicates (*, p<0.05; **, p<0.005) (D) Venn diagram of genes regulated by glucose and/or acetate in LN229 cells, with the overlap representing genes designated as acetyl-CoA-regulated (E) Heatmap of 881 genes represented in the overlap from the Venn diagram. See Table S3 for associated gene list and cluster ID. (F) DAVID functional annotation of pathways regulated by acetyl-CoA, using the gene list identified from the indicated clusters on the heatmap. (G) Doubling time for LN229 cells treated as indicated for 24 hours, mean +/− SEM of triplicates (*, p<0.05; **, p<0.005).

Figure 2. Acetyl-CoA: CoASH ratio is a determinant of histone acetylation in cancer cells

(A) Molar concentrations of acetyl-CoA, CoASH, and ratio of acetyl-CoA: CoASH over 24 hours in LN229 cells, mean +/− SEM of triplicates. Result is representative of 3 independent experiments. (B) IL-3-dependent bax−/−bak−/− cells were cultured for 48 hours +/− glucose, +/− IL-3 as indicated. Acetyl-CoA and CoASH were measured and normalized to cell volume, mean +/− SD of triplicates. Significance as compared to Glc+IL-3+ samples (**, p<0.005, ***, p<0.0005) (C) Representative Western blot of acetylated histones upon incubation of isolated nuclei with varying concentrations of acetyl-CoA and CoASH. Total histones were stained by Ponceau. Data was quantified from four independent experiments with 25:0 value set to 1, mean +/− SEM. Repeated measures one-way ANOVA with post test for linear trend was performed for 25:10–25:100 values. Post test for linear trend significance for AcH4: H4, p<0.0001 (***); for AcH3: H3, p=0.0034 (**).

Figure 3. Oncogenic Kras increases histone acetylation in vivo

Pancreata from mice expressing Kras-G12D (KPCY, previously described(Rhim et al., 2012)) were harvested at either 6 weeks (KPCY-Normal Area), 8 weeks (KPCY-PanIN) or 10 weeks (KPCY-PDA) of age, along with pancreata from control (Kras WT) mice (n≥3, each group). Immunohistochemistry against AcH4 (A), AcH3 (B) or Ki67 (C) was performed on paraffin-embedded tissue sections and nuclei were counterstained with hematoxylin. Representative images are shown, with magnification of areas of interest. Scale bar: 50 μm. See Fig. S3 for quantification of AcH4 in acinar cells in Kras WT and KPCY mice.

Figure 4. Oncogenic Kras enhances histone acetylation in vitro in an Akt- and ACL-dependent manner

(A) Mouse pancreatic primary cells derived from KPCY mouse PanIN lesions were treated with indicated inhibitors for 24 hours. Acetylation of histones and phosphorylation of signaling proteins were assessed by Western Blot in acid extracts and RIPA lysates, respectively. Ponceau staining is shown as loading control for histones. (B) Glucose consumption and Lactate production were measured in PanIN-derived primary mouse cells treated as in (A), mean +/− SD of triplicates (*, p<0.05). (C) Acetyl-CoA and CoASH levels were measured in PanIN cells, +/− Akt inhibitor, mean +/− SD of triplicates (***, p<0.001). (D) PanIN-derived primary cells were transduced with control (shCtrl) or ACL-targeting (shACL) short hairpin RNA. Cells were cultivated under indicated glucose concentrations, +/− Akt inhibitor, +/− 5 mM acetate for 24 hours. Histones were acid-extracted and analyzed by Western Blot. Ponceau staining is shown as loading control for histones.

Figure 5. Akt activation allows sustained histone acetylation in glucose-limited conditions

(A) Western blot analysis of proteins and histones from LN229 and LN229-myrAkt cells. Quantitation represents 5 independent experiments with LN229 4mM samples set to 1, mean +/− SEM (*, p ≤0.05). (B) Western blot analysis of histone acid extracts from LN229 cells stably expressing vector (EV), wt ACL, ACL-S455A, or ACL-S455D. Cells were treated with 1 or 10mM glucose for 24 hours before lysis. Quantitation represents the ratio of acetylated to total histones in four independent experiments with EV 10 mM set to 1, mean +/− SEM (*, p≤0.05) (C) Relative acetyl-CoA concentrations and percent enrichment after treatment in 1 mM or 10 mM [U-¹³C₆]-glucose for 20 hours, mean +/− SEM of triplicates. Total acetyl-CoA levels as well as M+2 acetyl-CoA (enriched from glucose) were significantly suppressed (p≤0.05) in 1 mM as compared to 10 mM glucose in EV and WT-ACL cells, but not ACL-S455D cells. (D) Control (MTB) and transgenic (MTB-tAkt) mice were administered doxycyclin for 96 hours (n≥5, each group). Immunohistochemistry was performed on paraffin-embedded mammary tissue sections. Representative images are shown, along with magnification of areas of interest. Scale bar: 50 μm.

Figure 6. pAkt correlates with histone acetylation in human glioma

(A) Representative images (20X) of pAkt and AcH4 in a grade IV GBM and a grade II oligodendroglioma. (B) pAkt expression showed a significant correlation with AcH4 levels in 56 human glioma samples (r=0.4721, p=0.0002). (C) pAkt and AcH4 in astrocytic tumors (grade IV astrocytomas = GBM) and oligodendrogliomas, mean +/−SD. (D) pAkt and AcH4 levels in 10 IDH mutant and 21 IDH wt gliomas, mean +/− SD.

Figure 7. pAkt correlates with histone acetylation levels in human prostate cancer

(A) Representative images of H3K18ac, H3K9ac, H4K12ac, and pAkt expression detected by immunohistochemistry in Gleason grade 3–5, metastatic (n=25) and non-metastatic (n=24) prostate cancer tumors. (B) H-scores were determined and correlations between marks determined. pAkt expression showed a significant correlation with: H3K18ac levels (r=0.7452, p≤0.0001); H3K9Ac (r=0.6283, p≤0.0001); and H4K12ac (r=0.5276, p≤0.0001). (C) Box plot showing H-scores for H4K12ac, H3K18ac, H3K9ac, and pAkt in prostate cancer tumors in patients who developed PSA failure (Yes) or not (No) (**, p<0.005, *, p<0.05).