Ceramide-mediated gut dysbiosis enhances cholesterol esterification and promotes colorectal tumorigenesis in mice

Abstract

Colorectal cancer (CRC) severely threatens human health and life span. An effective therapeutic strategy has not been established because we do not clearly know its pathogenesis. Here, we report that ceramide and sterol O-acyltransferase 1 (SOAT1) have roles in both spontaneous and chemical-induced intestinal cancers. We first found that miRNA-148a deficiency dramatically increased mouse gut dysbiosis through upregulating ceramide synthase 5 (Cers5) expression, which promoted ceramide synthesis afterward. The newly generated ceramide further promoted both azoxymethane/dextran sodium sulfate-induced (AOM/DSS-induced) and ApcMin/+ spontaneous intestinal tumorigenesis via increasing mouse gut dysbiosis. Meanwhile, increased level of ceramide correlated with the significant enhancements of both β-catenin activity and colorectal tumorigenesis in a TLR4-dependent fashion. Next, we found a direct binding of β-catenin to SOAT1 promoter to activate transcriptional expression of SOAT1, which further induced cholesterol esterification and colorectal tumorigenesis. In human patients with CRC, the same CERS5/TLR4/β-catenin/SOAT1 axis was also found to be dysregulated. Finally, the SOAT1 inhibitor (avasimibe) showed significant levels of therapeutic effects on both AOM/DSS-induced and ApcMin/+ spontaneous intestinal cancer. Our study clarified that ceramide promoted CRC development through increasing gut dysbiosis, further resulting in the increase of cholesterol esterification in a SOAT1-dependent way. Treatment with avasimibe to specifically decrease cholesterol esterification could be considered as a clinical strategy for effective CRC therapy in a future study.

Keywords: Cholesterol; Colorectal cancer; Gastroenterology; Metabolism; Oncogenes.

Figure 1. Depletion of miR-148a increases gut dysbiosis to enhance colorectal tumorigenesis.

(A) Representative images of colon tumors from WT and miR-148a^–/– mice under separately housed and cohoused situations. (B) Colon tumor numbers (left) and average tumor volumes (right) from A (n = 10–12/group). (C) The relative levels of indicated gut microbiota of mice from A after quantitative PCR (qPCR) analysis (n = 3/group). (A–C) Separately housed or cohoused WT and miR-148a^–/– mice were injected with AOM on day 0 and were treated with 3 rounds of 2.5% DSS in drinking water from day 0 to day 7 for 7 days followed by regular drinking water. (D) Representative images of colon tumors from WT and miR-148a^–/– mice treated with control (Ctrl) or antibiotics mix. (E–G) Colon tumor numbers (E), tumor sizes, (F) and average tumor volumes (G) from D (n = 10–13/group). (H) qPCR analyses on bacterial 16s rDNA of gut microbiota from D (n = 3/group). (Data were presented as mean ± SEM in B, and E–G.) *P < 0.05, **P < 0.01, ****P < 0.0001. Statistical significance was calculated by using 1-way ANOVA (B) or 2-tailed unpaired t test (E, F, and G).

Figure 2. Depletion of miR-148a upregulates CERS5 to increase colorectal carcinogenesis.

(A) Summary of bioinformatics screening results in CRCs. (B) qPCR analyses on the mRNA levels of Col4a1, Itga11, Cers5, Ltbp1, and Usp32 in WT and miR-148a^–/– mice colon tissues (n = 5/group). (C) Western blot assay obtained the CERS5 protein levels in CRC from indicated mice (n = 3/group). (D) Reverse transcription qPCR (RT-qPCR) assay was performed to quantitate the targeted mRNAs of miR-148a-3p that incorporated into the RISC derived from colon tissues of either WT or miR-148a^–/– mice. β-Actin was used as a control (n = 5/group). (E) Luciferase activity of the reporter vector containing WT or miR-148a-3p binding mutant 3′UTR of miR-148a-3p targets was determined in the HCT116 CRC cells transfected with indicated vectors. (F) CERS5 expression in normal colon and CRC tissues from TCGA data sets. (G) The colon C16:0 ceramide was detected by using mass spectrometry in WT mice and miR-148a^–/– mice (n = 5/group). (H–J) Colon tumor numbers (H), average volumes, (I) and sizes (J) in indicated mice (n = 10–12/group). (K) The colon C16:1 ceramide was detected by using mass spectrometry in CRCs of indicated mice (n = 6/group). (L) qPCR analyses on the indicated microbiota in mice from H (n = 3/group). (M) The endotoxin concentrations in miR-148a^–/– mice from H (n = 6/group). (Data were presented as mean ± SEM in B, D–K, and M.) *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Statistical significance was calculated by using 1-way ANOVA (E and H–K) or 2-tailed unpaired t test (B, D, F, G, and M). Data shown in E are representatives of 3 independent experiments. T/N, tumor/normal; FPKM, fragments per kilobase million; EU, endotoxin units.

Figure 3. Ceramide increases gut dysbiosis to promote colorectal tumorigenesis.

(A) Representative images of colon tumor from AOM/DSS-induced mice with indicated treatment. (B–D) The colon tumor numbers (B), average tumor volumes (C) and tumor sizes (D) from A (n = 11–13/group). (E) qPCR analyses on bacterial 16s rDNA in indicated gut microbiota from A (n = 3/group). (F) The relative mRNA levels of AMPs for mice from A by qPCR analysis (n = 5/group). (G–I) Colon tumor numbers (G), average tumor volumes, (H) and tumor sizes (I) from indicated mice (n = 10–12/group). (J) The small intestine tumor numbers of indicated mice (n = 10–12/group). (Data were presented as mean ± SEM in B–D and F–J.) *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Statistical significance was calculated by using 1-way ANOVA (B–D and G–J) or 2-tailed unpaired t test (F).

Figure 4. miR-148a depletion promotes CRC growth dependent on ceramide-mediated alterations in the gut microbiota to activate TLR4.

(A and B) Colon tumor numbers (A) and average tumor volumes (B) from Supplemental Figure 4A (n = 10/group). (C) qPCR analyses on bacterial 16s rDNA in indicated gut microbiota from Supplemental Figure 4A (n = 3/group). (D) Representative images of colon tumor induced by AOM/DSS from WT, miR-148a^–/–, Tlr4^–/–, and miR-148a^–/– Tlr4^–/– mice. (E–G) The colon tumor numbers (E), tumor sizes, (F) and average tumor volumes (G) in indicated mice from D (n = 10–13/group). (Data were presented as mean ± SEM in A, B, and E–G.) *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Statistical significance was calculated by using 1-way ANOVA (E–G) or 2-tailed unpaired t test (A and B).

Figure 5. SOAT1 is transcriptionally activated by β-catenin/TCF1 complex.

(A) qPCR analyzed lipid metabolism genes’ expression on CRCs from AOM/DSS-treated vehicle + control, ceramide + control, and ceramide + antibiotics mice (n = 3/group). (B and C) The mRNA (B) (n = 5/group) and protein (C) (n = 3/group) levels of SOAT1 in CRCs of WT and Tlr4^–/– mice with indicated treatment. (D) Results of Western blot assay for both β-catenin and SOAT1 protein levels in CRC from indicated mice (n = 3/group). (E) Schematic representation of the SOAT1 promoter (–5 kb to +5 kb of the transcription start site [TSS]). At the top, β-catenin binding sites are indicated. (F) Western blot detected β-catenin protein levels in CRCs of mouse with indicated treatment (n = 3/group). (G and H) The mRNA (G) and protein (H) levels of SOAT1 in HCT116 cells treated with different concentrations of β-catenin inhibitor (iCRT14). (I and J) Depletion of β-catenin repressed the mRNA (I) and protein (J) levels of SOAT1 in HCT116 cells. (K) ChIP qPCR analyses on the SOAT1 promoter with IgG and β-catenin antibodies in HCT116 cells. (L) Luciferase activity of SOAT1 promoter in HCT116 and SW480 cells in response to shRNA-mediated suppression of β-catenin. (Data were presented as mean ± SEM in B, G, I, K, and L.) *P < 0.05, **P < 0.01, ***P < 0.001. Statistical significance was calculated by using 1-way ANOVA (B) or 2-tailed unpaired t test (G, I, K, and L). Data shown in G–L are representatives of 3 independent experiments.

Figure 6. Soat1 loss attenuates intestinal carcinogenesis in miR-148a^–/– and Apc^Min/+ mice.

(A) The CRC cholesterol ester levels from WT mice with indicated treatments (n = 10/group). (B) Representative images of colon tumors induced by AOM/DSS from WT, miR-148a^–/–, Soat1^ΔIE, and miR-148a^–/– Soat1^ΔIE mice. (C–E) The colon tumor numbers (C), average tumor volumes, (D) and tumor sizes (E) in the indicated mice from B (n = 10–12/group). (F) The SOAT1 activity in CRCs from B (n = 10/group). (G) The colon cholesterol ester levels from indicated mice (n = 10/group). (H and I) The colon tumor numbers (H) and average tumor volumes (I) in indicated mice (n = 10–11/group). (J and K) The SOAT1 activity (J) and cholesterol ester (K) levels in CRCs from indicated mice (n = 10/group). (Data were presented as mean ± SEM in A and C–K.) *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Statistical significance was calculated by using 1-way ANOVA (C–G, J, and K) or 2-tailed unpaired t test (A, H, and I).

Figure 7. The CERS5/ceramide/β-catenin/SOAT1 signaling axis was dysregulated in human CRC samples.

(A–C) The CERS5 (A), β-catenin, (B) and SOAT1 (C) protein levels in CRC samples compared with normal colon tissues (n = 22/group). (D) The cholesterol ester in human CRC tissues compared with normal colon tissues (n = 22/group). (E and F) The correlation of SOAT1 protein levels with CERS5 protein (E) and β-catenin protein (F) levels in human colon tissues (n = 44/group). (G) Correlation of cholesterol ester levels with CERS5 protein levels in human colon tissues (n = 44/group). (H–J) Correlation of CERS5 protein (H), SOAT1 protein, (I) and cholesterol ester (J) levels with miR-148a-3p levels in CRC samples (n = 44/group). (Data were presented as mean ± SEM in A–D.) (A–D) Statistical significance was calculated by using 2-tailed matched-pair test. (E–J) Each circle represents an individual sample.

Figure 8. Targeting SOAT1 significantly suppresses spontaneous and chemical-induced colorectal carcinogenesis.

(A) Representative images of colon tumors subjected to the treatment with control or avasimibe after AOM/DSS induction for 120 days. (B) Survival rates of indicated mice subjected to the indicated treatment (n = 15/group). (C–E) Colon tumor numbers (C), average colon tumor volumes, (D) and tumor sizes (E) from A (n = 10–11/group). (F and G) The SOAT1 activity (F) and relative cholesterol ester level (G) from A (n = 10/group). (H) Survival rate of Apc^Min/+ mice subjected to the indicated treatment (n = 15/group). (I–K) Colon tumor numbers (I), average colon tumor volumes, (J) and tumor sizes (K) of Apc^Min/+ mice from indicated treatment (n = 11–12/group). (L) The small intestine tumor numbers of Apc^Min/+ mice from indicated treatment (n = 11–12/group). (M) The levels of cholesterol ester in CRCs from I (n = 10/group). (N) Schematic diagram. Depletion of miR-148a induced CERS5-mediated ceramide synthesis. Then, ceramide enhanced β-catenin activity through promoting gut dysbiosis–mediated Tlr4 activity. Next, β-catenin upregulated SOAT1 expression via directly binding to its promoter. Finally, SOAT1 enhanced cholesterol ester synthesis to promote colorectal tumorigenesis, and the SOAT1 inhibitor (avasimibe) had a significant level of therapeutic effect on intestinal cancer. (Data were presented as mean ± SEM in C–G and I–M.) *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Statistical significance was calculated by using 2-tailed unpaired t test.