ACSS2 drives senescence-associated secretory phenotype by limiting purine biosynthesis through PAICS acetylation

Abstract

Senescence-associated secretory phenotype (SASP) mediates the biological effects of senescent cells on the tissue microenvironment and contributes to ageing-associated disease progression. ACSS2 produces acetyl-CoA from acetate and epigenetically controls gene expression through histone acetylation under various circumstances. However, whether and how ACSS2 regulates cellular senescence remains unclear. Here, we show that pharmacological inhibition and deletion of Acss2 in mice blunts SASP and abrogates the pro-tumorigenic and immune surveillance functions of senescent cells. Mechanistically, ACSS2 directly interacts with and promotes the acetylation of PAICS, a key enzyme for purine biosynthesis. The acetylation of PAICS promotes autophagy-mediated degradation of PAICS to limit purine metabolism and reduces dNTP pools for DNA repair, exacerbating cytoplasmic chromatin fragment accumulation and SASP. Altogether, our work links ACSS2-mediated local acetyl-CoA generation to purine metabolism through PAICS acetylation that dictates the functionality of SASP, and identifies ACSS2 as a potential senomorphic target to prevent senescence-associated diseases.

Fig. 1. ACSS2 regulates SASP during senescence.

a ER:RAS-expressing IMR90 cells were induced to senesce by adding 100 nM 4-OHT, expressing non-targeting shRNA (shCtrl) or ACSS2 targeted shRNA (shACSS2). The puromycin-selected cells were harvested and analysed for expression of the indicated proteins by immunoblots. b–d The indicated cells were examined by SA-β-gal staining and colony formation (b). SA-β-gal-positive cells were quantified (c), the integrated intensity of the colonies was quantified using NIH Image J software (d). e, f Heat map of the RNA-seq data of the genes for which expression was significantly changed among proliferating and senescent cells with or without ACSS2 knockdown (e), KEGG analysis of genes altered by ACSS2 knockdown is shown (fold change, FC ≥ 2, P < 0.05) (f). g Heat map of SASP genes in senescent cells with ACSS2 knockdown vs. control knockdown determined by RNA-seq analysis. h The secretion of SASP components under the indicated conditions was detected by antibody arrays. The relative FC in comparison to the senescent cell with control shRNA was shown (n = 2 independent repeats). i Average signal from ChIP-seq analysis of H3K27ac occupancy in proliferating and senescent cells with or without knockdown of ACSS2. j Overlap between genes with expression upregulated in senescence while decreased by ACSS2 knockdown and genes with increased H3K27ac signal while decreased by ACSS2 knockdown (FC ≥ 2). RNA-seq and ChIP-seq analysis, n = 3 biologically independent repeats. k Gene ontology analysis of the overlap genes from (j). l Tracks of H3K27ac distribution on the representative SASP genes in groups of proliferating and senescent cells with or without ACSS2 knockdown. Data represent the mean ± s.d. of three biologically independent experiments. The P values were calculated using One-way ANOVA (Tukey’s multiple-comparison test) (c, d) or two-tailed Fisher Exact test (j) or one-sided Fisher exact test with Benjamini-Hochberg adjustment (f, k). Representative blot of n = 3 independent experiments was shown (a). Source data are provided as a Source Data file.

Fig. 2. ACSS2 deficiency blunts CCF formation during senescence.

a, b IMR90 cells were induced to senesce by oncogene RAS (a) or Etoposide (b). The proliferating cells and senescent cells expressing non-targeting shRNA, ACSS2, or ACLY-targeted shRNAs were infected with cGAS-FLAG encoding lentivirus. All groups of cells were subjected to γH2AX and cGAS-FLAG immunostaining. CCFs were indicated by arrows. Scale bar = 20 μm. c, d the CCF-positive cells were quantified among groups as described in (a, b). Randomly imaged fields with over 100 cells were analysed. Data represent the mean values of six different fields ± s.d. The P values were calculated using One-way ANOVA (Tukey’s multiple-comparison test). Representative of n = 3 independent experiments were shown (c, d). Source data are provided as a Source Data file.

Fig. 3. ACSS2 deficiency increases purine biosynthesis in senescence.

a ER:RAS-expressing cells were induced to senesce by adding 100 nM 4-OHT. At day 8 post-induction, cells were infected with lentivirus encoding non-targeting shRNA, ACSS2, or ACLY-targeted shRNAs. The puromycin-selected cells were harvested to perform targeted metabolomics analysis. Hierarchical clustering of significantly altered metabolites among groups was shown (n = 3 each group). b Top-ranked metabolic pathways in senescent cells with ACSS2 knockdown compared to control shRNA knockdown. P values were calculated using hypergeometric distribution (MetaboAnalyst v6.0). c, d Heat map analysis of the metabolites involved in the pyrimidine pathway (c) and the purine pathway (d). e ER:RAS-expressing cells were induced to senesce by adding 100 nM 4-OHT. On day 8 post-induction, cells were infected with lentivirus-encoding non-targeting shRNA or ACSS2-targeted shRNA. The puromycin-selected cells were harvested to measure cellular dNTP levels. f–h ER:RAS-expressing cells were induced to senesce by adding 100 nM 4-OHT. At day 8 post-induction, cells were supplemented with or without nucleosides or purine-only nucleosides for 14 days. Cells were stained for γH2AX (f). Three randomly imaged fields were analysed (g). The SASP mRNA expression was examined by qRT-PCR (h). Scale bar = 20 μm. Data represent mean ± s.d. of three biologically independent experiments. The P values were calculated using an unpaired two-tailed Student’s t-test (e) or One-way ANOVA (Tukey’s multiple-comparison test) (g, h). Pro, proliferating cells; Sen, RAS induced senescence; NS, nucleosides; A + G, Adenosine+Guanosine. Source data are provided as a Source Data file.

Fig. 4. DNPB enzymes partially mediate the effects of ACSS2 knockdown on CCF formation and SASP gene expression.

a Volcano plot showing Log₂ fold changes in protein intensities on the X-axis and −Log₁₀ P values on the Y-axis. Significantly enriched proteins were in blue (P < 0.05) and non-significant in gray. P values were calculated using a two-tailed Student’s t-test. b–e ER:RAS-expressing cells were induced to senesce by adding 100 nM 4-OHT, and the proliferating cells and senescent cells expressing non-targeting shRNA, ACSS2 targeting shRNA together with PAICS or AK1 specific shRNAs were subjected to immunoblots for the indicated proteins (b), or CCF detection (c). Scale bar = 10 μm. d The CCF-positive cells were quantified among groups. Four Randomly imaged fields with over 100 cells were analysed. Bar graphs represent mean values of different fields ± s.d. e Proliferating and senescent cells, as described in (b), were subjected to qRT-PCR analysis for expression of the indicated SASP genes. f Proliferating and OIS cells with shCtrl or shACSS2 were harvested and the labeled ¹⁵N-Glutamine incorporation into purine intermediates were analysed by LC-MS/MS. Data represent mean ± s.d. of three biologically independent experiments. The P values were calculated using One-way ANOVA (Tukey’s multiple-comparison test) (d–f). Representative of n = 3 independent experiments were shown (b, d). Source data are provided as a Source Data file.

Fig. 5. ACSS2 promotes acetylation of PAICS and decreases its protein stability.

a Proliferating and OIS cells with or without ACSS2 knockdown were harvested. Co-immunoprecipitation analysis of endogenous PAICS and ACSS2 in extracts from the indicated IMR90 cells was performed and subjected to immunoblots of the indicated proteins. b Purified human protein of ACSS2 and PAICS was subjected to in vitro pull-down assay. Immunoblot of the indicated proteins was shown. c Proliferating and OIS cells were infected with HA-PAICS encoding lentivirus, and then treated with or without ACSS2i for 48 h. PAICS protein was immunoprecipitated, and the acetylation was analysed by immunoblot with the indicated antibody. d HEK293T cells were transfected with HA-tagged wildtype (WT) or K36R, K47R, and K36RK47R mutants of PAICS. HA-PAICS protein was immunoprecipitated, and the acetylation was analyzed by immunoblot with the indicated antibody. e, f ER:RAS-expressing cells were induced to senesce by adding 100 nM 4-OHT. On day 8 after the induction, the senescent cell was infected with lentivirus encoding non-targeting shRNA or ACSS2 targeted shRNA (e) or treated with ACSS2i for 48 h (f). The cells were harvested and analysed by immunoblot for expression of the indicated proteins. g, h Proliferating and senescent cells described in (e) were treated with CHX (100 μg/mL) for 0, 6, or 12 h, and the expression of the indicated proteins was analysed by immunoblot (g) and quantified (h). i OIS cells were subjected to co-immunoprecipitation analysis of endogenous PAICS and PCAF, immunoblots of the indicated proteins were shown. j, k OIS cells expressing non-targeting shRNA or PCAF-targeted shRNA were infected with HA-PAICS encoding lentivirus. PAICS protein was immunoprecipitated, and the acetylation was analyzed by immunoblot with the indicated antibody (j) and quantified (k). l Proliferating and OIS cells expressing non-targeting shRNA or PCAF-targeted shRNA were subjected to immunoblots of the indicated proteins. Data represent mean ± s.d. of three biologically independent experiments. The P values were calculated using an unpaired two-tailed Student’s t-test (k). Representative of n = 3 independent experiments were shown (a–g, i, j, l). Source data are provided as a Source Data file.

Fig. 6. PAICS interacts with LC3 and undergoes lysosomal degradation during cellular senescence.

a, b ER:RAS-expressing cells were induced to senesce by adding 100 nM 4-OHT. At day 8, senescent cells were treated with or without MG132 (0.1 μM) (a) or BafA1 (30 nM) for 48 h (b); the cells were harvested and analysed for expression of the indicated proteins by immunoblot. c OIS cells expressing non-targeting shRNA or ATG7 targeted shRNA were harvested for expression of the indicated proteins by immunoblot. d–f Co-immunoprecipitation analysis of endogenous PAICS and LC3 was performed in proliferating and OIS cells (d), and the PAICS or LC3 IP bands were normalized to the LC3 or PAICS IP and the input bands, respectively (e, f). g–i HA-tagged WT or mutant PAICS and FLAG-tagged LC3 were co-transfected into HEK293T cells and analysed by co-immunoprecipitation (g), the HA or FLAG IP bands were normalized to the FLAG or HA IP and the input bands respectively (h, i). j–l HA-tagged WT PAICS and FLAG-tagged WT or mutant LC3 were co-transfected into HEK293T cells, and analysed by co-immunoprecipitation (j), the HA or FLAG IP bands were normalized to the FLAG or HA IP and the input bands respectively (k, l). m Schematic of the three LIR motifs in PAICS protein. n–p HA-tagged WT or LIR-motif-mutant PAICS and FLAG-tagged LC3 were co-transfected into HEK293T and analysed by co-immunoprecipitation (n), the HA or FLAG IP bands were normalized to the FLAG or HA IP and the input bands respectively (o, p). q, r HA-tagged WT or LIR-motif-mutant PAICS was transfected into HEK293T. At 48 h post-transfection, cells were treated with CHX (100 μg/mL) for 0, 6, or 12 h, and the expression of the indicated proteins was analysed by immunoblot (q) and quantified (r). Data represent mean ± s.d. of three biologically independent experiments. The P values were calculated using an unpaired two-tailed Student’s t-test (e, f, h, i) or One-way ANOVA (Tukey’s multiple-comparison test) (k, l, o, p). Representative of n = 3 independent experiments were shown (a–d, g, j, n, q). Source data are provided as a Source Data file.

Fig. 7. ACSS2 is required for the in vivo function of SASP.

a, b TOV21G cells were co-injected with proliferating or senescent IMR90 cells into NCG mice. The tumor volume was measured (a) and the tumor weight was measured, (b). n = 5 mice per group. c A scheme of the experimental procedure. d–f One week after ionizing irradiation, Livers from vehicle or ACSS2i treated mice were analysed by immunohistochemistry for γH2AX (d), the mean values of γH2AX positive cells from nine randomly imaged fields were quantified per mouse (e), and SASP mRNA expression was analysed (f). Scale bar = 20 μm. n = 5 for non-IR group, n = 7 per IR group. g–j 18-month-old mice were administered with vehicle or ACSS2i for one week (g). Livers were harvested and analyzed by immunohistochemistry for γH2AX (h), and the mean values of γH2AX positive cells from ten randomly imaged fields were quantified per mouse (i) and SASP mRNA expression was analysed (j). Scale bar = 20 μm. n = 5 for the young or vehicle-treated old mice group, n = 7 for the treated old mice group. k Schematic model of the constructs and the experimental design. l–n Representative images of SA-β-gal staining of the liver tissues at day 7 and 14 post-injection (l), and quantified (m, n). n = 6 each group. o, p Representative images of CD45 immunofluorescence staining and NRas-dsRed at day 7 post-injection (o), and the mean value of the double positive clusters from five randomly imaged fields per mouse were calculated (p). Scale bar = 100 μm. q, r Representative images of CD45 IHC staining of the liver tissues at day 7 and 14 post-injection (q). Arrows point to examples of immune clusters. CD45⁺ cells in the indicated groups at day 7 post-injection were quantified (r). Scale bar = 100 μm. n = 6 each group. Data represent the mean ± s.d. The P values were calculated using One-way ANOVA (Tukey’s multiple-comparison test) (a, b, e, f, i, j) or an unpaired two-tailed Student’s t-test (m, n, p, r). Source data are provided as a Source Data file.