Nuclear mitochondrial acetyl-CoA acetyltransferase 1 orchestrates natural killer cell-dependent antitumor immunity in colorectal cancer

Abstract

Tumor metabolism often interferes with the immune microenvironment. Although natural killer (NK) cells play pivotal roles in antitumor immunity, the connection between NK cells and tumor metabolism remains unclear. Our systematic analysis of multiomics data and survival data from colorectal cancer (CRC) patients uncovered a novel association between mitochondrial acetyl-CoA acetyltransferase 1 (ACAT1) and NK cell infiltration that influences disease progression. ACAT1, a metabolic enzyme involved in reversible conversion of acetoacetyl-CoA to two molecules of acetyl-CoA, exhibits nuclear protein acetylation activity through its translocation. Under immune stimulation, mitochondrial ACAT1 can be phosphorylated at serine 60 (S60) and enters the nucleus; however, this process is hindered in nutrient-poor tumor microenvironments. Nuclear ACAT1 directly acetylates lysine 146 of p50 (NFKB1), attenuating its DNA binding and transcriptional repression activity and thereby increasing the expression of immune-related factors, which in turn promotes NK cell recruitment and activation to suppress colorectal cancer growth. Furthermore, significant associations are found among low nuclear ACAT1 levels, decreased S60 phosphorylation, and reduced NK cell infiltration, as well as poor prognosis in CRC. Our findings reveal an unexpected function of ACAT1 as a nuclear acetyltransferase and elucidate its role in NK cell-dependent antitumor immunity through p50 acetylation.

Fig. 1

ACAT1 promotes cytotoxic NK cell infiltration to suppress CRC growth. a Top-ranked proteins positively correlated with NK cell infiltration in CRC. The processes were as follows: first, metabolic enzymes whose gene expression levels were significantly correlated with CRC patient survival (OS, DFS) in TCGA-COADREAD cohort were identified (p_Chisq < 0.01; Supplementary Table 1); next, NK cell infiltration in each sample was estimated by MCPcounter on transcriptomic data from TCGA-COADREAD cohort and samples were divided into high and low NK cell infiltration groups based on the median estimated NK cell infiltration levels, after this, genes shown significant prognostic associations exclusively in the high NK cell infiltration group were selected (p_Chisq < 0.01; Supplementary Table 2); finally, genes/proteins expression analysis was conducted using transcriptomic (TCGA-COADREAD) (pvalue.wilcox < 0.001; Supplementary Table 3) and proteomic (CPTAC-COAD) (p < 0.01; Supplementary Table 4) datasets (referred to Suhas and colleagues study), and only differentially expressed proteins between normal and CRC tissues were selected. The final candidates were ranked based on p-values obtained from the proteomic differential analysis (p < 0.01; Supplementary Table 5). Kaplan-Meier survival analysis of ACAT1 expression for overall survival (b) or based on ACAT1 expression for patients with high-NK cell infiltration or low-NK cell infiltration (c, d) (as determined via MCPcounter) in the TCGA COADREAD cohort. Acat1-KO CT26 cells rescued with Acat1-Flag WT or empty vector (EV) were subcutaneously injected into BALB/c mice (e, 6 mice per group) and NSG mice (f, 6 mice per group). Tumor growth (left) and tumor weight (right) were measured. g, h Acat1-KO B16F10 cells rescued with Acat1-Flag WT or EV were subcutaneously injected into C57BL/6J mice (6 mice per group). Tumor growth (g) and tumor weight (h) were measured. i Luciferase-expressing Acat1-KO CT26 cells rescued with Acat1-Flag WT or EV were injected into the cecum of BALB/c mice (3 mice per group). Bioluminescence imaging was performed 14 days after injection, and representative images of tumor growth are shown (left). Bioluminescent quantification was calculated (right). j Representative H&E staining of tumors from mice in i. k NK cells were isolated from tumors 12 days after orthotopic injection and subjected to Smart-seq2. Heatmap of the mRNA expression levels of the genes related to NK cell cytotoxicity was shown. l GSEA of inflammatory response and leukocyte transendothelial migration signaling genes based on Smart-seq2 data. m Acat1-KO CT26 cells rescued with Acat1-Flag WT or EV were subcutaneously injected into BALB/c mice (up, 6 mice per group), and the percentages of tumor-infiltrating CD4⁺ T cells, CD8⁺ T cells, CD19⁺ B cells and NK (NKp46⁺) cells were calculated. The absolute numbers of NK cells (n), Perforin⁺, IFN-γ⁺, GzmB⁺ NK cells (o) in tumors from BALB/c mice subcutaneously injected with Acat1-KO CT26 cells re-expressing Acat1-Flag WT or not (5 mice per group). p Immunofluorescence staining with an anti-NCR1 (anti-NKp46 (red)) antibody and DAPI (blue) was performed in tumor tissues harvested from BALB/c mice treated as described in n, o. Representative images of IF staining are shown (p); DAPI, 4′,6-diamidino-2-phenylindole. q NKp46-positive cell numbers per field based on p were calculated by ImageJ. r BALB/c mice with subcutaneous CT26 tumor model (6 mice per group) were treated with rabbit serum or anti-ASGM1antibody, and tumor growth was measured. The data are shown as the means ± SDs. Two-tailed log-rank test (b–d), unpaired two-tailed t test (e, f (tumor weight), h, i, m–o), Welch’s t test (q) or two-way ANOVA (e–g, r (tumor volume))

Fig. 2

Nuclear ACAT1 engages activated NK cells to accumulate in the TME. a, b Immunofluorescence staining with anti-ACAT1 and anti-COX IV (mitochondrial marker) antibodies was performed in tumor tissues and peritumoral tissues from patients with colorectal cancer. Representative images of IF staining are shown (a). Semiquantitative scores of nuclear ACAT1 staining in tumor and peritumoral tissues were determined via HALO (b). Nuc, nuclear. c Semiquantitative scoring of IF and IHC staining was carried out with HALO, and the correlation between the nuclear ACAT1 (IF) and NCR1 (IHC) levels was analyzed. d, e Acat1-KO CT26 cells rescued with Acat1-Flag WT or ΔMTS were subcutaneously injected into BALB/c mice (6 mice per group), and tumors were excised 13 days after injection for flow cytometric analysis to calculate the percentage of tumor-infiltrating NK cells (d) and evaluate the expression of TNFα and Perforin in NK cells (e). f–h Acat1-KO MC38 cells rescued with Acat1-Flag WT, NLS, or NES were infected with lentivirus expressing luciferase and injected into the cecum of C57BL/6J mice. Bioluminescence imaging was performed 18 days after injection. Representative images of tumor growth and HE staining are shown (f, left); Bioluminescent quantification was calculated (3 mice per group) (f, right). Tumors were excised 13 days after injection for flow cytometric analysis to calculate the percentages of tumor-infiltrating NK1.1⁺ cells (g) and of IFN-γ⁺ and GzmB⁺ NK1.1⁺ cells (3 mice per group) (h). i, j Acat1-KO CT26 cells rescued with Acat1-Flag WT, NLS or NES were subcutaneously injected into BALB/c mice. Tumor growth and tumor volume were measured (6 mice per group) (i). Flow cytometric analysis was performed 13 days after injection to calculate the percentages of tumor-infiltrating NK cells, TNFα⁺ NK cells, and Perforin⁺ NK cells (6 mice per group) (j). k Acat1-KO CT26 cells rescued with Acat1-Flag WT, NLS or NLS-C123A were subcutaneously injected into BALB/c mice (6 mice per group). Flow cytometric analysis was performed 13 days after injection to calculate the percentages of tumor-infiltrating NK cells, IFN-γ⁺ NK cells, and GzmB⁺ NK cells. l Representative images of immunofluorescence staining for rACAT1-Flag WT and the indicated mutants in CT26 cells with anti-Flag, and anti-COX IV antibodies. Images were reconstructed by SIM (Structured Illumination Microscopy). COX IV, mitochondrial marker. The data are shown as the means ± SDs. Paired two-tailed t test (b), Pearson correlation test (c), one-way ANOVA (d, e, f, g–k (i, tumor weight)), or two-way ANOVA (i, tumor volume)

Fig. 3

Nuclear ACAT1 directly acetylates p50 at K146 a Mass spectrometry (MS) analysis to explore different ACAT1-binding proteins in nucleus of HCT116 cells. The representative candidates are listed. Nuclear factor NF-kappa-B p105 subunit (NFKB1) was identified. b HA-egfp-p50 (1-366aa) was immunoprecipitated from HEK293T cells to detect ACAT1-Flag bound to HA-egfp-p50 (1-366aa). c RKO cells stably expressing EV or ACAT1-Flag were subjected to subcellular fractionation assay for a subsequent immunoprecipitation (IP) analysis. d HCT116 cells and CT26 cells stably expressing EV or ACAT1-Flag/Acat1-Flag were subjected to subcellular fractionation assay for a subsequent IP analysis. e Detection of purified ACAT1-His (left) or ACAT1-Flag (right) bound to GST or GST-p50 (1-366aa) via a GST pulldown assay. f Endogenous IP with anti-p50 or anti-IgG antibodies was performed to determine the level of acetylated p50 in HEK293T cells transfected with EV or ACAT1-Flag. g HA-egfp-p50 (1-366aa) was immunoprecipitated from HCT116 cells co-transfected with ACAT1-Flag WT or C126A and HA-egfp-p50 (1-366aa). Immunoblot analysis was performed with the indicated antibodies. h The acetylation level of purified GST-p50 (1-366aa) was determined by incubation with or without purified ACAT1-His in the presence or absence of 10 µM acetyl-CoA (Ac-CoA) in vitro. i Subcellular fractionation assay and immunoprecipitation were performed to pull down nuclear HA-egfp-p50 (1-366aa) from HCT116 cells transfected with ACAT1-Flag NLS. K146 and K147 were found to be acetylated in the nucleus via mass spectrometry analysis. j IP was performed with anti-HA magnetic beads in HEK293T cells co-transfected with HA-egfp-p50 (1-366aa) WT/K146R/K147R and ACAT1-Flag WT or NLS. Immunoblot analysis was performed with the indicated antibodies. k Sequences of the acetylated peptides in the indicated species. l Purified GST-p50 (1-366aa) or K146R and purified ACAT1-His were incubated with or without 10 µM Ac-CoA in vitro. Immunoblot analysis was performed with the indicated antibodies. Immunoblots representative of three independent experiments are shown

Fig. 4

p50 K146 acetylation weakens DNA binding ability and thus promotes NK cell activation and infiltration. a, Schematic representation of p105/p50. b, c Subcellular fractionation assay and DNA pulldown assay were performed to detect endogenous nuclear p50 or exogenous HA-egfp-p50 (1-366aa) WT/K146R/K146Q bound to biotinylated dsDNA (5 µg) conjugated to streptavidin magnetic beads in HCT116 cells transfected with EV or ACAT1-Flag (b) and in HEK293T cells (c). d, e Molecular dynamics simulations of p50 homodimers bound to DNA. The total binding free energy (Gtotal, top) and electrostatic energy (Gele, bottom) were compared between p50 WT and K146ac (paired t test) (d). The potential hydrogen bonds (colored yellow) in WT-Cluster (top), and K146Ac-Cluster (bottom) are shown. p50 homodimers in WT-Cluster and K146Ac-Cluster are shown as cartoons, DNA is represented by an orange double helix, and the nucleotides bound to p50 are shown as sticks. K146 of p50 homodimer is colored magenta (chain A) or pink (chain B) in WT-Cluster and shown as sticks. K146Ac is colored in hot pink (chain A) or brown (chain B) in K146Ac-Cluster and shown as sticks, with the acetyl group labeled as green sticks (e). GSEA of antigen processing and presentation genes (f) and NK cell activation genes (g) based on RNA-seq data from NFKB1-depleted HCT116 cells rescued with HA-egfp-p50 K146Q/K146R. h Volcano plots showing the log₂-fold changes in gene expression and adjusted P values between HA-egfp-p50 K146Q versus K146R in NFKB1-depleted HCT116 cells. i, j Nfkb1-depleted CT26 cells rescued with mp50-HA (1-364aa) WT or K144Q were subcutaneously injected into BALB/c mice (6 mice per group). Tumor growth and tumor weight were measured (i). Tumors were excised 13 days after injection for flow cytometric analysis to calculate the percentage of tumor-infiltrating cytotoxic NK cells (j). Immunoblots representative of three independent experiments are shown. The data are shown as the means ± SDs. Paired t test (d), unpaired two-tailed t test (i (tumor weight), j), or two-way ANOVA (i (tumor volume))

Fig. 5

ACAT1 pS60 facilitates its nuclear translocation. a HCT116 cells were treated with or without IL18 for 12 h, and subcellular fractionation assay was subsequently performed. b HCT116 and RKO cells were cultured in RPMI-1640 medium with or without 10% fetal bovine serum (FBS) for 12 h, and subcellular fractionation assay was subsequently performed. c Subcellular fractionation assay was performed for IP analysis in HCT116 cells stably expressing ACAT1-Flag, and immunoblot analysis was then performed with the indicated antibodies. d Representative image of structure of S60 in human ACAT1 (PDB: 2F2S). Tetrameric ACAT1 is shown as surface pattern in blue, and S60 is shown as a red stick. e ACAT1-depleted HCT116 cells were rescued with ACAT1-Flag WT, S54A, or S60A and treated with or without 50 ng/ml IL18 for 12 h. Subcellular fractionation assay was performed for subsequent immunoblot analysis. f RKO cells transfected with ACAT1-Flag were cultured in RPMI-1640 medium supplemented with 10% dialyzed FBS (DFBS), or without DFBS and glucose separately for 12 h. ACAT1-Flag was immunoprecipitated with an anti-ACAT1 pS60 antibody to evaluate the phosphorylation level of S60. g HCT116 cells transfected with ACAT1-Flag were pretreated with or without 10 μM U0126 for 4 h before IL18 treatment for another 12 h. Immunoprecipitation assay was performed to detect the phosphorylation level of S60 with anti-ACAT1 pS60 antibody. h Representative images showing the interaction between endogenous PPM1A/Ppm1a and ACAT1-Flag/Acat1-Flag exogenously expressed in HCT116 cells and CT26 cells, as detected by the Duolink PLA assay. n = 3 biological replicates. i ACAT1-depleted HCT116 cells were rescued with ACAT1-Flag WT or S60A and infected with a lentivirus expressing shNC or shPPM1A and were then treated with IL18 or U0126. Subcellular fractionation assay was performed for subsequent immunoblot analysis. j HCT116 cells stably expressing ACAT1-Flag were infected with lentivirus expressing shNC or shPPM1A, followed by IL18 treatment or not. Immunoprecipitation assay was performed with an anti-ACAT1 pS60 antibody to evaluate the phosphorylation level of S60. k Subcellular fractionation assay was performed for IP in HCT116 cells stably expressing EV or ACAT1-Flag, and immunoblot analysis was then performed with the indicated antibodies to evaluate the interaction between ACAT1 and importin α3. l HCT116 cells stably expressing ACAT1-Flag were infected with lentivirus expressing shNC or shPPM1A, followed by IL18 treatment or not. IP and immunoblot analysis were performed with the indicated antibodies. m ACAT1-Flag was immunoprecipitated from HCT116 cells stably expressing ACAT1-Flag WT, S60A, or S60E. IP and immunoblot analysis were performed with the indicated antibodies. Immunoblots representative of three independent experiments are shown

Fig. 6

ACAT1 pS60 promotes NK cell activation and recruitment. a–c Acat1-KO CT26 cells rescued with Acat1-Flag WT, S58A or S58D were subcutaneously injected into BALB/c mice (6 mice per group). Tumor growth was measured (a). Flow cytometric analysis was performed 13 days after injection to calculate the percentage of tumor-infiltrating NK cells (b) and evaluate the expression of IFN-γ and GzmB (c) in the NK cells. d ACAT1-depleted HCT116 cells rescued with ACAT1-Flag WT, S60A, or S60E were co-transfected with HA-egfp-p50 (1-366aa), and HA-egfp-p50 was immunoprecipitated to evaluate its acetylation. e ACAT1-depleted HCT116 cells were rescued with ACAT1-Flag WT, S60A, S60E, C126A or EV (as a control). Subcellular fractionation assay and DNA pulldown assay were performed to detect nuclear p50 bound to biotinylated dsDNA (5 µg) conjugated to streptavidin magnetic beads. Each band was quantified via ImageJ. f ACAT1-depleted HCT116 cells were rescued with ACAT1-Flag WT or S60E and subjected to real-time qPCR analysis. n = 3 biological replicates. g–i Acat1-KO CT26 cells rescued with Acat1-Flag WT, S58A or S58D were cultured for 48 h, after which the supernatants were collected. Enzyme-linked immunosorbent assays (ELISAs) were then performed to quantify the secretion of mCCL5 (g), mCXCL10 (h), and mCXCL11 (i). n = 3 biological replicates. j–l Nfkb1-depleted CT26 cells rescued with HA-mp50 (1-364aa) or K144R were coinfected with lentiviruses expressing Acat1-Flag WT or S58D and subcutaneously injected into BALB/c mice (6 mice per group). Tumor growth was measured (j). Flow cytometric analysis was performed 13 days after injection to calculate the percentages of tumor-infiltrating IFN-γ⁺, GzmB⁺ NK cells (k) and CXCR3⁺, CCR5⁺ NK cells (l). Immunoblots representative of three independent experiments are shown. The data are shown as the means ± SDs. Two-way ANOVA (a, f, j) or one-way ANOVA (b, c, g–i, k, l)

Fig. 7

Clinical relevance of the ACAT1 pS60 level in CRC. IF analysis with anti-ACAT1 and anti-COX IV antibodies (a) and IHC analysis with anti-ACAT1 pS60, anti-PPM1A, and anti-NCR1 antibodies (b) of specimens from patients with colorectal cancer were performed. Representative images of IF and IHC staining of tumors from two patients with colorectal cancer are shown. c–e Semiquantitative scoring of IF and IHC staining was carried out via HALO, and the correlations between nuclear ACAT1 (IF) and ACAT1 pS60 (IHC) levels (c), NCR1 and ACAT1 pS60 levels (d), PPM1A and ACAT1 pS60 levels (e) were analyzed. f IHC analysis with anti-ACAT1 pS60 antibody was performed in tumor tissues of patients with colorectal cancer. Overall survival durations of 397 patients (stage1–4) with low (n = 207, blue curve) or high (n = 190, red curve) levels of ACAT1 pS60 were compared. g Mechanism through which ACAT1 pS60 promotes NK cell activation and infiltration to impair tumor growth. The data are shown as the means ± SDs. Pearson correlation test (c–e), two-tailed log-rank test (f)