Cancer-associated fibroblast-derived acetate promotes pancreatic cancer development by altering polyamine metabolism via the ACSS2-SP1-SAT1 axis

Abstract

The ability of tumour cells to thrive in harsh microenvironments depends on the utilization of nutrients available in the milieu. Here we show that pancreatic cancer-associated fibroblasts (CAFs) regulate tumour cell metabolism through the secretion of acetate, which can be blocked by silencing ATP citrate lyase (ACLY) in CAFs. We further show that acetyl-CoA synthetase short-chain family member 2 (ACSS2) channels the exogenous acetate to regulate the dynamic cancer epigenome and transcriptome, thereby facilitating cancer cell survival in an acidic microenvironment. Comparative H3K27ac ChIP-seq and RNA-seq analyses revealed alterations in polyamine homeostasis through regulation of SAT1 gene expression and enrichment of the SP1-responsive signature. We identified acetate/ACSS2-mediated acetylation of SP1 at the lysine 19 residue that increased SP1 protein stability and transcriptional activity. Genetic or pharmacologic inhibition of the ACSS2-SP1-SAT1 axis diminished the tumour burden in mouse models. These results reveal that the metabolic flexibility imparted by the stroma-derived acetate enabled cancer cell survival under acidosis via the ACSS2-SP1-SAT1 axis.

Fig. 1. PSCs support cancer cell growth by the secretion of acetate under acidosis.

a, Schematic of NMR-based metabolomics to identify PSC-derived metabolites. b, Metabolites identified in CAF-0911 CM: yellow, increased; blue, decreased; white, unchanged; acetate, red). c, Acetate levels in the interstitial fluid of human PDAC tumours (n = 13) and normal pancreas (n = 3). d, Relative PDAC cell survival with the indicated treatments (n = 3 in each group, from independent biological replicates). e, Immunoblot of ACLY in HPS cells. Representative image of two independent experiments. f, Acetate levels in shScr and shACLY HPS cell–CM compared to serum-free medium (SFM; n = 3 in each group, from independent biological replicates). g, Relative survival of S2-013 cells treated with CM derived from shScr and shACLY HPS cells in normal and low-pH conditions (n = 3 in each group, from independent biological replicates). h,i, Mean organoid diameters (i) and representative images (h) of PA417 and PA901 organoids cultured alone or in combination with shScr or shACLY HPS cells. Scale bars, 100 µm (PA417), 250 µm (PA901) (n = 3 in each group from independent biological replicates). j, Relative survival of PDAC cells treated with CM derived from shScr and shACLY HPS cells under low pH (pH 7.0) with 5 mM acetate (n = 3 in each group, from independent biological replicates). k–o, Tumour weights (k) and volumes (l), Ki-67 IHC staining images (m), Ki-67 quantification in three different fields from three tumours of each (n), and αSMA and ACLY co-expression by IHC (o) upon necropsy from mice implanted with S2-013 cells alone (n = 9) or co-implanted with shScr (n = 7) and shACLY (n = 9) HPS cells. Scale bars, 100 µm (m), 250 µm (o). Images are representative of three tumours of each group. p, Representative images and quantification of αSMA and ACLY colocalization coefficient in normal human pancreas and PDAC tumours. Scale bars, 50 µm. Image are representative of at least two different fields from n = 4 normal human pancreas and n = 23 tumours of each group. Unpaired two-tailed t-test, mean ± s.e.m. (c); two-way analysis of variance (ANOVA) with Šídák’s post-hoc test, mean ± s.e.m. (d); two-way ANOVA with Tukey’s post-hoc test, mean ± s.e.m. (g,j); one-way ANOVA with Bonferroni’s post-hoc test; mean ± s.e.m. (f,i,k,l,n). Source data

Fig. 2. Acetate-mediated histone acetylation and ACSS2 activation control pancreatic cancer growth under acidosis.

a, Relative expression of SREBF1, ACACA and FASN genes in PDAC cells treated with 5 mM acetate under acidosis (n = 4 in each group, from independent biological replicates). b,c, Representative fluorescent images (b) and quantification (c) of Nile red staining in PDAC cells treated with 5 and 10 mM acetate. Scale bars, 100 µm (n = 5 in each group, from independent biological replicates). d, Relative survival of PDAC cells treated with vehicle or 100 µM orlistat, without or with acetate (n = 5 in each group, from independent biological replicates). e, Immunoblots of acetylated histones in PDAC cells treated with acetate (SE, short exposure; LE, long exposure). Images are representative of two independent experiments. f, Immunoblots of acetylated histones in PDAC cells treated with CM from shScr, shACLY-a and shACLY-b HPS cells. Images are representative of two independent experiments. g, ¹⁴C-acetate incorporation (counts per minute (c.p.m.)) in total histones extracted from S2-013 and HPAF-II cells treated with acetate in the absence and presence of 10 µM C646 for 6 h under acidosis (n = 3 in each group, from independent biological replicates). h, Relative survival of PDAC cells treated with 10 µM C646 inhibitor, without or with acetate (n = 5 in each group, from independent biological replicates). i, Relative mRNA expression of ACSS2 in PDAC cells treated with acetate under acidosis (n = 4 in each group, from independent biological replicates). j, Immunoblot of ACSS2 in PDAC cells upon treatment with acetate for 24 h under acidosis. Representative of two independent experiments. k,l, Representative immunofluorescent images (k) and quantitation (l) showing expression of ACSS2 protein in acidic tumour regions, as imaged by staining with pHLIP in tumour sections from 20-week-old KPC mice, and AT-nude mice orthotopically implanted with S2-013 cancer cells. Scale bars, 100 µm (KPC, normal (n = 24) and acidic (n = 24) pH; S2-013, normal (n = 21) and acidic (n = 168) pH). Unpaired two-tailed t-test, mean ± s.e.m. (a,g,i,l); one-way ANOVA with Bonferroni’s post-hoc test, mean ± s.e.m. (c); two-way ANOVA with Tukey’s post-hoc test, mean ± s.e.m. (d,h). Source data

Fig. 3. ACSS2 expression in PDAC cells regulates acetate-mediated growth of pancreatic cancer cells in vitro and in vivo under acidosis.

a, Relative survival of PDAC cells upon treatment with 20 µM ACSS2 inhibitor, without or with acetate (n = 5 (S2-013), n = 4 (HPAF-II) in each group, from independent biological replicates). b, ¹⁴C-acetate incorporation in total histones extracted from PDAC cells treated with 5 mM acetate in the presence of vehicle control or 20 µM ACSS2 inhibitor for 6 h (n = 4 in each group (S2-013, HPAF-II), n = 3 (CFPAC-1, T3M4), from independent biological replicates). c, Representative immunoblots showing the levels of ACSS2, acetylated histones and total H3 protein in scrambled control (siScr) and ACSS2 knockdown (siACSS2) S2-013 and HPAF-II cells upon treatment with 5 mM acetate. Images are representative of two independent experiments. d, Relative survival of siScr and siACSS2 PDAC cells cultured upon acetate treatment (n = 3 for S2-013, n = 5 for HPAF-II in each group, from independent biological replicates). e, Relative survival of control (shScr) and ACSS2 knockdown (shACSS2) S2-013 cells cultured without or with acetate (n = 3 in each group, from independent biological replicates). f, Relative oxygen consumption rate (OCR) in siScr and siACSS2 S2-013 cells upon acetate treatment (n = 8 in each group, from independent biological replicates). 2,4-DNP, 2,4-dinitrophenylhydrazine; 2-DG, 2-deoxyglucose. g,h, Representative images (g) and quantification of mean organoid diameter (h) of PA417 and PA901 organoids cultured −/+ HPS cells upon treatment with 50 µM ACSS2 inhibitor. Scale bars, 100 µm. PA417, n = 5, 7, 6, 10); PA901, n = 3 in each group, from independent biological replicates. i,j, Tumour weights (i) and volumes (j), upon necropsy for mice implanted with shScr or shACSS2 S2-013 cells alone or co-implanted with HPS cells (shScr − HPS (n = 8), shScr + HPS (n = 8), shACSS2 − HPS (n = 7), shACSS2 + HPS (n = 11)). k,l, Representative IHC images for Ki-67 staining (k) in tumour sections from mice implanted with shScr or shACSS2 S2-013 cells alone or co-implanted with HPS cells, along with quantitation (l) counted in three different fields from three tumour sections of each group (n = 9 in each group). Scale bars, 100 µm. Unpaired, two-tailed t-test, mean ± s.e.m. (b); two-way ANOVA with Tukey’s post-hoc test, mean ± s.e.m. (a,d,e); one-way ANOVA with Bonferroni’s post-hoc test, mean ± s.e.m. (h–j,l). Source data

Fig. 4. H3K27Ac ChIP–seq and RNA–seq analyses identify SAT1 as a critical regulator of acetate-mediated effects.

a, Distribution of H3K27 acetylation (H3K27Ac) ChIP–seq reads from vehicle-treated (blue) and acetate- treated (red) S2-013 cells within ±10 kb of the TSS. b, Density of ChIP–seq reads for H3K27ac ±10 kb from the TSS in vehicle- and acetate-treated S2-013 cells. c, DAVID-based pathway enrichment analysis of differentially H3K27-acetylated gene promoters in vehicle and acetate-treated S2-013 cell line (statistical analysis using Fisher’s exact test). d, Volcano plot depicting differentially regulated genes (1.5-fold change cutoff) in S2-013 cells upon acetate treatment. e, Venn diagram of ChIP–seq and RNA–seq data showing 282 genes that are upregulated in RNA–seq data with differentially acetylated H3K27 in their gene promoters. f, Heatmap showing 48 differentially regulated genes upregulated by acetate in an ACSS2-dependent manner. The colour bar shows the ratio of fragments per kilobase of transcript per million mapped reads (FPKM) values of siScr, siScr + acetate, siACSS2, or siACSS2 + acetate to that of siScr. g, Volcano plot showing the Cox coefficient and −log(P value) of the 48 regulated genes extracted from the OncoLnc database for survival of patients with PDAC. Prospective oncogenes and tumour suppressors are shown in red and green, respectively. h, Principal component analysis (PCA) plot of S2-013 cells treated with acetate under acidosis relative to untreated cells as determined by LC–MS/MS-based metabolomics (n = 5 biological replicates per group). i, Heatmap of top 25 altered metabolites from S2-013 cells treated with acetate under acidosis and displayed with row Z-score normalization, as determined by LC–MS/MS-based metabolomics (n = 5 biological replicates per group). j, H3K27ac at putative enhancer regions proximal to the SAT1 gene. The y axis shows reads per bin per million. k, Relative mRNA levels of the SAT1 gene in siScr and siACSS2 S2-013 cells in the presence or absence of acetate for 24 h under acidosis (n = 4 in each group, from independent biological replicates; one-way ANOVA with Bonferroni’s post-hoc test, mean ± s.e.m.). l, Relative levels of SAT1 and ACSS2 proteins in scrambled control (siScr) and ACSS2 knockdown (siACSS2) S2-013 cells in the presence and absence of 5 mM acetate for 24 h under acidosis. Immunoblots are representative of two independent experiments. Source data

Fig. 5. SAT1 expression in PDAC cells is critical for acetate-mediated growth of pancreatic cancer cells in vitro and in vivo.

a,b, Relative levels of SAT1 mRNA and protein in control (shScr) and SAT1 knockdown (shSAT1-a and shSAT1-b) S2-013 (a) and HPAF-II (b) cells under acidosis (n = 4 in each group, from independent biological replicates). Images are representative of two independent experiments. c,d, Relative survival of shScr and SAT1 knockdown S2-013 (c) and HPAF-II (d) cells in the presence and absence of acetate under acidosis (n = 3 in each group, from independent biological replicates). e,f, Relative levels of polyamine biosynthetic pathway metabolites in shScr and shSAT1 S2-013 (e) and HPAF-II (f) cells in the presence and absence of 5 mM acetate under acidosis. Metabolites are presented as normalized row Z-scores from n = 6 (S2-013) and n = 4 (HPAF-II) biological replicates per group. g,h, Representative images (g) of PA417 and PA901 organoids cultured alone or in combination with HPS cells upon treatment with 25 µM pentamidine, along with mean organoid diameters (h). Scale bars, 100 µm. PA417, n = 4 (vehicle), n = 4 (+HPS), n = 6 (pentamidine), n = 4 (HPS + pentamidine), and PA901, n = 4 in each group from independent biological replicates. i,j, Tumour weights (i) and volumes (j) upon necropsy of mice implanted with control and SAT1 knockdown S2-013 cells alone or co-implanted with HPS cells (n = 12 (shScr), n = 12 (shScr + HPS), n = 10 (shSAT1), n = 10 (shSAT1 + HPS) mice). k,l, Representative IHC images for Ki-67 staining (k) in tumour sections from mice implanted with shScr or shSAT1 S2-013 cells alone or co-implanted with HPS cells along with the quantitation of percent positive cells (l). Scale bars, 100 µm. Ki-67 positive cells were counted in three different fields from three tumour sections of each group (n = 9 in each group). One-way ANOVA with Bonferroni’s post-hoc test, mean ± s.e.m. (a,b,h–j,l); two-way ANOVA with Tukey’s post-hoc test, mean ± s.e.m. (c,d). Source data

Fig. 6. SP1-mediated SAT1 regulation is critical for tumour cell survival under acidosis.

a, Consensus sequence of the SP1 binding motif. b, Putative binding sites of SP1 in the promoter region of the SAT1 gene. c, Fold enrichment of SP1 binding in SAT1 gene promoter in S2-013 cells upon acetate treatment normalized to IgG control (n = 4 from independent biological replicates). d,e, Relative levels of SP1 mRNA (d) and protein (e) in PDAC cells upon acetate treatment (n = 4 from independent biological replicates; representative immunoblot images of two independent experiments). f,g, Relative SP1 and SAT1 mRNA levels in scrambled control (siScr) and SP1 knockdown (siSP1) S2-013 (f) and HPAF-II (g) cells upon acetate treatment (n = 4 from independent biological replicates). h, Immunoblots of SP1 and SAT1 in siScr and siSP1 S2-013 and HPAF-II cells, upon treatment with acetate. Images are representative of two independent experiments. i, Relative SAT1 and SP1 mRNA levels in SP1-overexpressing S2-013 cells upon acetate treatment (n = 4 from independent biological replicates). j, SAT1 and SP1 immunoblots in SP1-overexpressing S2-013 cells upon acetate treatment. Images are representative of two independent experiments. k, Immunoblot of lysine-acetylated proteins in S2-013 and HPAF-II cells treated with 0 and 5 mM acetate for 6 h. Images are representative of two independent experiments. l, Schematic representation of acetyl-lysine-modified proteome analysis of S2-013 cells treated with 5 mM acetate for 6 h. m, Venn diagram showing differentially acetylated proteins in S2-013 cells upon acetate treatment. n, LC–MS/MS analysis of acetylated peptide corresponding to SP1 (n = 2 from independent biological replicates). o, Peptide sequence alignment of SP1 (amino acids 1–31) in various species, highlighting the prospective acetylation site K19. p, Detection and quantitation (n = 5) of HA-tagged SP1 wild-type (WT) or K19R mutant in HEK293T cells treated with acetate, without or with MG132. Images are representative of five independent experiments. q,r, Representative immunofluorescent images (q) and quantitation (r) showing expression of SP1 in acidic tumour regions, as imaged by staining with pHLIP in tumour sections from 20-week-old KPC mice. Scale bars, 100 μm (n = 12 for normal pH, n = 24 for acidic pH). Unpaired, two-tailed t-test, mean ± s.e.m. (d,r); one-way ANOVA with Bonferroni’s post-hoc test, mean ± s.e.m. (c,f,g,i,p). Source data

Fig. 7. The ACSS2–SP1–SAT1 metabolic axis is critical for pancreatic cancer progression.

a,b, Representative images (a) of PA417 and PA901 organoids cultured alone or in combination with HPS cells and treated with 100 and 25 µM mithramycin (MTA), respectively, and mean organoid diameters, represented as bar charts (b). Scale bars, 100 µm. PA417, n = 5 and PA901, n = 4 in each group from independent biological replicates. c,d, Tumour weights (c) and volumes (d) following necropsy of mice implanted with control (sgScr) or SP1 knockout (sgSP1) S2-013 cells alone or co-implanted with HPS cells (n = 12 mice in each group). e,f, Representative IHC images for Ki-67 (e) in tumour sections from mice implanted with sgScr or sgSP1 S2-013 cells alone or co-implanted with HPS cells, along with the quantitation of percent positive cells (f). Scale bars, 100 µm. Ki-67⁺ cells were counted in three different fields from three tumour sections of each group (n = 9 in each group). g, Meta-analysis of gene expression of SAT1 in normal pancreas and pancreatic tumour samples (GENT U133A-normal (n = 62), cancer (n = 174); GENT U133plus2-normal (n = 8), cancer (n = 13); GDS4103-normal (n = 39), cancer (n = 39). h, Comparison of Kaplan–Meier survival curves for human patients with PDAC, with all high and low SAT1 expression in the The Cancer Genome Atlas (TCGA) cohort stratified by highest quartile (n = 43) and lowest quartile (n = 43) of SAT1 levels. i, Kaplan–Meier survival plot showing the relative survival of PANC137 PDX tumour-bearing mice upon treatment with vehicle (n = 9) or 10 mg kg⁻¹ pentamidine (n = 10) (median survival: vehicle, 61 days; pentamidine, 82 days). j, Kaplan–Meier survival curves for stage IV human patients with PDAC, comparing patients (male, 16; female, 10; median age, 66 years; age range, 36–92 years) with above-median (high; n = 13) and below-median (low; n = 13) plasma levels of N¹-acetylspermidine. k, Schematic illustration of the overall findings of the study. OAA, oxaloacetate; HAT, histone acetyl transferase; KAT, lysine acetyl transferase. Two-way ANOVA with Tukey’s post-hoc test, mean ± s.e.m. (b); one-way ANOVA with Bonferroni’s post-hoc test, mean ± s.e.m. (c,d,f); unpaired, two-tailed t-test, mean ± s.e.m. (g); paired two-tailed t-test, mean ± s.e.m. (g, GDS4103); Mantel–Cox log-rank test (h–j). Source data

Extended Data Fig. 1. Additional characterization of pancreatic stellate cell-mediated support of cancer cell growth by secretion of acetate.

a, Relative cell survival of pancreatic cancer cells (S2-013, HPAF-II, CFPAC-1, T3M4, and MIAPaCa-2) upon treatment with stellate cell (CAF-0911 and HPS)-derived conditioned medium (CM) (n = 4 for S2-013, n = 4 for HPAF-II, n = 6 for CFPAC-1, n = 4 for T3M4, n = 4 for MIAPaCa-2 in each group from independent biological replicates). b,c, Representative images and quantitation of diameter of PA417 (b) and PA901 (c) organoids cultured in the presence or absence of HPS cells. Scale bar = 100 µm (PA417), 250 µm (PA901) (n = 6 in each group from independent biological replicates). d, Immunofluorescent images showing the distribution of HPS cells stably expressing LeGO-dKatushka2 plasmid when co-cultured with multiple pancreatic cancer organoids (PA417, PA901, and PA137) labeled with CellTrace Violet dye for 5 days. Scale bar = 100 µm. Representative image of two independent experiments. e, Relative fold change in labeled metabolite levels in CM from CAF-0911 stellate cells (CM) or double conditioned medium (DCM) derived from CAF-0911 cells pre-conditioned with S2-013 tumor cell-conditioned media (n = 3 in each group from independent biological replicates). f, Levels of acetate in the interstitial fluid of pancreatic cancer tissues of KPC mice (female mice at 20-22 weeks of age) and control C57BL/6 J mice (n = 5 in each group from independent biological replicates). g-i, Relative cell survival of pancreatic cancer cells, S2-013 (g), HPAF-II (h), and CFPAC-1 (i) treated with increasing doses of acetate (0.1–10 mM) for 72 hrs cultured in normoxic, hypoxic, low glucose, and low glutamine conditions. The cell survival is normalized to the respective untreated controls (n = 6 in each group from independent biological replicates). One-way ANOVA with Tukey’s post-hoc test; mean ± s.e.m. (a,g,h,i); unpaired, two-tailed t-test; mean ± s.e.m. (b,c,e,f). Source data

Extended Data Fig. 2. Pancreatic stellate cell ACLY contributes to acetate secretion and tumor burden.

a, Relative mRNA levels of ACLY in human pancreatic stellate (HPS) cells expressing scrambled control (shScr) and shACLY plasmids (n = 3 in each group from independent biological replicates). b, Experiment scheme and immunoblots showing the expression of ACLY protein in PDAC cells upon treatment with CM-derived from HPS and CAF-0911 cells. Representative image of two independent experiments. c, Experiment scheme and immunoblots showing the expression of ACLY protein in S2-013, HPAF-II, CFPAC-1, and T3M4 cells upon co-culture with HPS and CAF-0911 cells for 24 hrs. Representative image of two independent experiments. d, Representative images of tumors excised from athymic-nude mice implanted with S2-013 cells alone (n = 9) or co-implanted with control (shScr; n = 7) and ACLY knockdown (shACLY; target a, n = 9; target b, n = 9) HPS cells. e,f, Representative IHC images for Ki-67 staining, nuclear staining (DAPI), cytokeratin 19 (epithelial/tumor cells), and merge in tumor sections from athymic-nude mice implanted with S2-013 cells alone or co-implanted with shScr or shACLY HPS cells (along with the quantitation of percent Ki-67 and CK19 dual positive cells (f). Scale bar = 100 µm. Ki-67 and CK19 positive cells were counted manually in three different fields from 3 tumor sections of each group (n = 9 in each group). g, Tumor volumes, upon necropsy of athymic-nude mice implanted with CFPAC-1 cells alone (n = 7) or co-implanted with shScr (n = 8) and shACLY (shACLY-a, n = 7; shACLY-b, n = 6) HPS cells. h,i, Representative IHC images for Ki-67 staining (h) in tumor sections from athymic-nude mice implanted with CFPAC-1 cells alone or co-implanted with shScr or shACLY HPS cells along with the quantitation of percent positive cells (i). Scale bar = 100 µm. Ki-67 positive cells were counted manually in three different fields from 3 tumor sections of each group (n = 9 in each group). j, Representative immunofluorescent images showing co-expression of αSMA and ACLY proteins in tumor sections from athymic-nude mice implanted with CFPAC-1 cells alone or co-implanted with shScr or shACLY HPS cells. Scale bar = 250 µm. k, Levels of acetate in the interstitial fluid of tumor tissues from athymic-nude mice implanted with S2-013 cells alone (-) or co-implanted with shScr or shACLY HPS cells. l,m, Tumor weights (l) and tumor volumes (m), upon necropsy from athymic-nude mice implanted with S2-013 cells alone or co-implanted with shScr or shACLY HPS cells. For k,l,m, n = 9 (S2-013 alone), 9 (co-implanted with shScr HPS), 8 (co-implanted with shACLY HPS) mice. One-way ANOVA with Bonferroni’s post-hoc test; mean ± s.e.m. (a,f,g,i,k,l,m). Source data

Extended Data Fig. 3. Characterizing acetate secretion in different CAF subtypes.

a,b Representative images showing expression of αSMA (a) or Vimentin (b) by immunofluorescence staining in HPS and CAF-0911 cell lines and primary CAFs (CAF-0906, CAF-1003, and CAF-1016), as imaged by staining cells with fluorescently-tagged phalloidin for F-actin. Nuclei were stained with DAPI. Scale bar = 20 µm. c, Relative mRNA expression of iCAF markers (IL6, LIF) and myCAF markers (ACTA2, MYL1) in Huff1 (human foreskin fibroblast cell line; used as a control), CAF-0911, CAF-0906, CAF-1003, and CAF-1016 cells. The gene expression is normalized to Huff1 cells (n = 4 in each group from independent biological replicates). d, Acetate levels in Huff1 control cells and CAF-0911, CAF-0906, CAF-1003, and CAF-1016 cancer-associated fibroblast cells (n = 3 in each group from independent biological replicates). e,f, Acetate levels in HPS and CAF-0911 cells upon treatment with IL-1β (e) or Tgfβ1 (f) for 48 hrs (n = 3 in each group from independent biological replicates). g,h, Acetate estimation in CAF-0906 and CAF-1003 cells upon treatment with IL-1β (g) or Tgfβ1 (h) for 48 hrs. (n = 3 in each group from independent biological replicates). One-way ANOVA with Bonferroni’s post-hoc test; mean ± s.e.m. (c,d); unpaired, two-tailed t-test; mean ± s.e.m. (e,f,g,h). Source data

Extended Data Fig. 4. Effect of acetate treatment on lipid biosynthesis, histone acetylation, and ACSS2 expression in pancreatic cancer cells under acidosis.

a, Relative mRNA expression of SREBF1, ACACA and FASN genes in CFPAC-1 and T3M4 cells upon treatment with 5 mM acetate for 12 hrs (n = 4 in each group from independent biological replicates). b,c, Representative fluorescent images (b) showing Nile red staining in CFPAC-1 and T3M4 cells upon acetate treatment for 96 hrs. The bar charts show relative fluorescence levels of Nile red staining (c). Total lipid fluorescence is normalized to the cell count. Scale bar = 100 µm. (n = 5 in each group from independent biological replicates). d,e, Relative quantification of Nile red fluorescence in PDAC cells treated with acetate for 96 hrs. The cells were cultured in a DMEM medium containing normal and delipidated FBS in a 3:1 (d) or 1:3 (e) ratio. The total Nile red fluorescence was normalized to the cell count (n = 8 in each group from independent biological replicates). f,g, Relative mRNA expression of FASN (f) and SREBF1 (g) genes in PDAC cell lines upon treatment with CM from HPS and CAF-0911 cells for 12 hrs (n = 4 in each group from independent biological replicates). h-k, Representative fluorescent images showing the Nile red (h) or BODIPY (j) staining in PDAC cell lines upon treatment with CM from HPS and CAF-0911 cells for 96 hrs. Scale bar = 100 µm. The bar charts show relative fluorescence levels of Nile red (i) or BODIPY (k) staining. Total lipid fluorescence is normalized to the cell count (n = 6 in each group for ‘i’ and n = 4 in each group for ‘k’ from independent biological replicates). l, Relative survival of CFPAC-1 and T3M4 cells upon treatment with control or 100 µM orlistat, without or with acetate (n = 3 in each group from independent biological replicates). m, Relative survival of CFPAC-1 and T3M4 cells upon treatment with 10 µM C646 inhibitor, without or with acetate (n = 3 in each group from independent biological replicates). n, Relative mRNA expression of ACSS2 in S2-013, HPAF-II, CFPAC-1, and T3M4 cells upon treatment with 5 mM acetate for 12 hrs under normal pH conditions (n = 4 in each group from independent biological replicates). o, Representative immunoblots showing expression of ACSS2 in S2-013, HPAF-II, CFPAC-1, and T3M4 cells upon treatment with 5 mM acetate under normal pH conditions. Representative image of two independent experiments. Unpaired, two-tailed t-test; mean ± s.e.m. (a,n); one-way ANOVA with Bonferroni’s post-hoc test; mean ± s.e.m. (c,d,e,f,g,i,k,); two-way ANOVA with Tukey’s multiple comparisons test; mean ± s.e.m.(l,m). Source data

Extended Data Fig. 5. Role of ACSS2 in acetate-mediated epigenetic reprogramming, survival under acidosis, and tumor burden.

a, Relative survival of CFPAC-1 and T3M4 cells upon treatment with 20 µM ACSS2 inhibitor, without or with acetate (n = 3 in each group from independent biological replicates). b, Representative immunoblots (of two independent experiments) showing the levels of ACSS2, acetylated histones, and total H3 protein in scrambled control (siScr) and ACSS2 knockdown (siACSS2) CFPAC-1 and T3M4 cells upon treatment with 5 mM acetate. c, Relative survival of siScr and siACSS2 CFPAC-1 and T3M4 cells in the presence and absence of acetate (n = 3 in each group from independent biological replicates). d, Immunoblots (representative of two independent experiments) showing the levels of ACSS2 protein in control (shScr) and ACSS2 knockdown (shACSS2- a and shACSS2-b) S2-013, HPAF-II, and CFPAC-1 cells under acidosis. e, Relative survival of control and ACSS2 knockdown HPAF-II and CFPAC-1 cells in the presence, and absence of acetate (n = 6 for HPAF-II and n = 3 for CFPAC-1 in each group from independent biological replicates). f, Representative immunofluorescent images showing co-expression of stellate cell marker (αSMA) and ACSS2 in tumor sections from athymic-nude mice implanted with shScr or shACSS2 S2-013 cells alone or co-implanted with HPS cells. Scale bar = 250 µm. g, Representative images of tumors excised from athymic-nude mice implanted with shScr or shACSS2 CFPAC-1 cells alone or co-implanted with HPS cells. h, Tumor volumes, upon necropsy, of athymic-nude mice implanted with control and ACSS2 knockdown CFPAC-1 cells alone or co-implanted with HPS cells (n = 7, 8, 8, 7, 9, 9 mice in indicated groups). i,j, Representative IHC images for ACSS2 and Ki-67 staining (j) in tumor sections from athymic-nude implanted with shScr or shACSS2 CFPAC-1 cells alone or co-implanted with HPS cells along with the quantitation of percent positive Ki-67 cells (i). Scale bar = 100 µm. Ki-67 positive cells were counted manually in three different fields from 3 tumor sections of each group (n = 9 in each group). k, Representative IHC staining for cleaved caspase-3 protein in tumor sections from athymic-nude mice implanted with shScr or shACSS2 S2-013 cells alone or co-implanted with HPS cells. Scale bar = 100 µm. l,m, Tumor weights (l) and tumor volumes (m), upon necropsy for athymic-nude mice implanted with shScr or shACSS2 (shACSS2-a and shACSS2-b) S2-013 cells. All groups were compared to shScr cohort. (n = 9 mice in each group). n,o, Representative IHC images for Ki-67 staining (n) in tumor sections from athymic-nude mice implanted with shScr or shACSS2 S2-013 cells along with the quantitation of percent Ki-67 positive cells (o). Scale bar = 100 µm. Ki-67 positive cells were counted manually in three different fields from 3 tumor sections of each group (n = 9 in each group). Two-way ANOVA with Tukey’s post-hoc test; mean ± s.e.m. (a,c,e); one-way ANOVA with Bonferroni’s post-hoc test; mean ± s.e.m. (h,i,l,m,o). Source data

Extended Data Fig. 6. Additional characterization of the role of ACSS2 in tumor cell survival under normal pH.

a, Relative survival of S2-013, HPAF-II, CFPAC-1, and T3M4 cells upon treatment with 20 µM ACSS2 inhibitor, without and with acetate, for 72 hrs under normal pH conditions. The cell survival is normalized to the respective untreated controls. (n = 3 for S2-013, n = 6 for HPAF-II, n = 8 for CFPAC-1, n = 3 for T3M4 in each group from independent biological replicates). b, The relative survival of scrambled control (siScr) and ACSS2 knockdown (siACSS2) S2-013, HPAF-II, CFPAC-1, and T3M4 cells cultured in the presence and absence of acetate for 72 hrs under normal pH conditions. The cell survival is normalized to the respective untreated controls. (n = 3 for S2-013, n = 6 for HPAF-II, n = 3 for CFPAC-1, n = 3 for T3M4 in each group from independent biological replicates). c, The relative survival of scrambled control (shScr) and ACSS2 knockdowns (shACSS2-a and shACSS2-b) S2-013, HPAF-II, and CFPAC-1 cells cultured in the presence and absence of acetate for 72 hrs under normal pH conditions. The cell survival is normalized to the respective untreated controls. (n = 3 for S2-013, n = 3 for HPAF-II, n = 8 for CFPAC-1 in each group from independent biological replicates). Two-way ANOVA with Tukey’s post-hoc test; mean ± s.e.m. (a,b,c). Source data

Extended Data Fig. 7. Acetate-mediated regulation of chromatin accessibility, gene expression, metabolic reprogramming, and SAT1 expression via ACSS2.

a, Analysis of total acetylated H3K27 genomic loci in acetate-treated cells. The inner circle includes the differentially acetylated gene promoters upon acetate treatment. b, Heatmap shows the expression of 282 commonly upregulated genes in vehicle control- and acetate-treated S2-013 cells identified through ChIP-Seq and RNA-Seq. The scale bar represents log2 (fold change) values (n = 1 sample per group). c, PCA plot of HPAF-II cells treated with acetate under acidosis relative to vehicle control-treated cells as determined by LC–MS/MS-based metabolomics (n = 6 in both groups from independent biological replicates). d, Heatmap of top 25 regulated metabolites of HPAF-II cells treated with acetate under acidosis, presented as normalized row Z-score, as determined by LC–MS/MS-based metabolomics (n = 6 in each group from independent biological replicates). e, Relative mRNA levels of ACSS2 gene in scrambled control (siScr) and ACSS2 knockdown (siACSS2) S2-013 cells in the presence and absence of acetate for 24 hrs under acidosis (n = 4 in each group from independent biological replicates). f, Relative mRNA levels of ACSS2 and SAT1 genes in scrambled siScr and siACSS2 HPAF-II cells upon acetate treatment for 24 hrs under acidosis (n = 4 in each group from independent biological replicates). g, Relative mRNA levels of ACSS2 and SAT1 genes in scrambled siScr and siACSS2 CFPAC-1 cells upon acetate treatment for 24 hrs under acidosis (n = 4 in each group from independent biological replicate). h, Relative mRNA levels of ACSS2 and SAT1 genes in scrambled siScr and siACSS2 T3M4 cells upon acetate treatment for 24 hrs under acidosis (n = 4 in each group from independent biological replicates). i, The relative levels of SAT1 and ACSS2 proteins in siScr and siACSS2 HPAF-II, CFPAC-1, and T3M4 cells upon acetate treatment for 24 hrs under acidosis. Immunoblots are representative of 2 independent experiments. One-way ANOVA with Bonferroni’s post-hoc test; mean ± s.e.m. (e,f,g,h). Source data

Extended Data Fig. 8. Validation of SAT1 knockdown in orthotopically implanted tumors.

Representative IHC staining of SAT1 protein in the formalin-fixed tumor sections from athymic-nude mice implanted with control (shScr) and SAT1 knockdown (shSAT1) S2-013 cells alone or co-implanted with HPS cells. Scale bar = 100 µm. Representative image of images taken from three different fields from 3 tumor sections of each group.

Extended Data Fig. 9. SP1 mediates acetate-induced regulation of SAT1 in an ACSS2-dependent manner.

a, Transcription factor motif search for H3K27ac-bound peaks. b, Fold enrichment of SP1 binding on the promoter regions of SAT1 gene in HPAF-II cells upon acetate treatment. The fold change values are normalized to IgG control (n = 4 in each group from independent biological replicates). c, Relative mRNA levels of SP1 and SAT1 genes in scrambled control (siScr) and SP1 knockdown (siSP1) CFPAC-1 cells upon acetate treatment for 24 hrs (n = 4 in each group from independent biological replicates). d, Relative mRNA levels of SP1 and SAT1 genes in siScr and siSP1 T3M4 cells upon acetate treatment (n = 4 in each group from independent biological replicates). e, Immunoblots showing the levels of SP1 and SAT1 proteins in siScr and siSP1 CFPAC-1 and T3M4 cells, upon treatment with 5 mM acetate. Representative images of two independent experiments. f, g Relative mRNA (f) and protein (g) levels of SAT1 and SP1 in HPAF-II cells overexpressing SP1 in the absence and presence of 5 mM acetate for 24 hrs under acidosis (f: n = 4 in each group from independent biological replicates; g: Representative images of two independent experiments). h, Immunoblots showing the levels of ACSS2 and SP1 proteins in siScr and siACSS2 PDAC cell lines upon treatment with 5 mM acetate for 24 hrs in acidosis. Representative image of two independent experiments. i, GSEA enrichment plot of SP1 signaling pathway in S2-013 cells between control cells (siScr) in presence and absence of acetate, siScr vs siACSS2 cells, and siScr vs siACSS2 both in the presence of acetate (statistics calculated using Permutation analysis through GSEA). j, Relative levels of lysine-acetylated proteins in CFPAC-1 and T3M4 cells upon treatment with 5 mM acetate for 6 hrs under acidosis. Representative images of two independent experiments. k, Pathway enrichment analysis of differentially regulated lysine-acetylated proteins in vehicle control- and acetate-treated S2-013 cells (statistics calculated using the Binomial test). l,m, Plot of RMSD for wild type SP1 (l) and SP1 acetylated at K19 position (m) for 250 nsec of simulations. One-way ANOVA with Bonferroni’s post-hoc test; mean ± s.e.m. (b,c,d,f). Source data

Extended Data Fig. 10. SP1/SAT1 inhibition or knockdown diminishes acetate-mediate survival advantage under acidosis and CAF-induced tumor burden in vivo.

a, Relative survival of pancreatic cancer cells upon treatment with SP1 inhibitor mithramycin (MTA) in vehicle control- and acetate-treated cells for 72 hrs under acidosis. S2-013, HPAF-II, CFPAC-1, and T3M4 cells treated with 50 nM (S2-013, CFPAC-1, T3M4) and 100 nM (HPAF-II) mithramycin (n = 3 in each group from independent biological replicates). b, Relative survival of siScr and siSP1 PDAC cell lines upon acetate treatment (n = 3 in each group from independent biological replicates). c, Representative IHC images for SP1 staining in tumor sections from athymic-nude mice implanted with sgScr or sgSP1 S2-013 cells alone or co-implanted with HPS cells. Scale bar = 100 µm. Images were captured from three different fields from 3 tumor sections of each group. d,e, Post necropsy tumor weights (d) and tumor volumes (e) from athymic-nude mice implanted with S2-013 cells alone (n = 10) or co-implanted with HPS cells upon treatment with vehicle control (n = 11) and MTA (tumor cells alone: n = 9, HPS co-implanted: n = 11). All groups were compared to S2-013 cells alone-implanted group treated with vehicle control or as indicated. f,g, Representative immunofluorescent images showing the expression of Ki-67 and αSMA proteins (f) in tumor sections from athymic-nude mice implanted with S2-013 cells, alone or co- implanted with HPS cells, upon treatment with vehicle control and MTA. Scale bar = 250 µm. Ki- 67 positive cells (g) were counted manually in three different fields from 3 tumor sections of each group (n = 9 in each group). h, Representative IHC staining of SAT1 in tumor sections from athymic-nude mice implanted with S2-013 cells alone or co-implanted with HPS cells upon MTA treatment from three tumor sections of each group. Scale bar = 100 µm. Representative image of images taken from three different fields from 3 tumor sections of each group. i,j, Kinetics of tumor growth (i) and body weight changes (j) in athymic-nude mice orthotopically implanted with PANC137 human PDX tissue upon treatment with vehicle control or 10 mg/kg pentamidine (n = 9 (vehicle), 10 (pentamidine-treated) mice in group). k, Forest plot indicating Cox-proportional hazard regression analysis for stage IV PDAC patients for above and below median plasma N1-acetylspermidine levels, age, gender, and chemotherapy. Hazard ratios and 95% confidence intervals are presented and indicated on the right side. Cox-proportional hazard regression analysis. Two-way ANOVA with Tukey’s post-hoc test; mean ± s.e.m. (a,b); one-way ANOVA with Bonferroni’s post-hoc test; mean ± s.e.m. (d,e,g); two-way ANOVA with Bonferroni’s post-hoc test; mean ± s.e.m. (i,j). Source data