神经酰胺合成促进CRC的再生状态与肝转移

Fig. 1. High-fat diet feeding promotes colorectal cancer liver metastasis. (A) Schematic depicting experimental setup. Wild-­type (WT) C57BL/6J mice fed a CD or HFD for 8 weeks were transplanted with engineered naive AKP CRCOs. Survival was followed up to 15 weeks, and primary tumor burden (mass) and liver metastasis (number and frequency) were analyzed at the end point. (B) Survival curves in CD and HFD-­fed WT C57BL/6J mice. HFD-­fed mice had significantly lower survival. Log-­rank (Mantel-­Cox) test; n = 25 (CD) and 18 (HFD) mice per group. (C) Representative brightfield (BF) and fluorescent images of metastatic tumors in the liver. Scale bars, 2000 μm. (D) Number of liver metastasis (>100 μM) per mouse from CD and HFD mice, represented as violin plots. Each dot represents a mouse; n = 26 (CD) and 22 (HFD) mice per group, two-­tailed Student’s t test. The three dashed lines in the violins from top to bottom represent top quartile (75th), median, and bottom quartile (25th), respectively. Dashed horizontal line at y = 0 reference line; y axis extends below zero to display the full kernel density estimate of the violin plot. All observed values are ≥0. (E) Frequency of liver metastasis in CD and HFD cohorts; n = 26 (CD) and 22 (HFD) mice per group, Fisher’s exact test. (F) Representative brightfield and fluorescent images of primary tumor in the colon. Scale bars, 2000 μm. (G) Survival curves of CD and HFD-­fed Rag2−/− mice transplanted with AKP-­tdT CRCOs; 75,000 CRCO-­derived cells per mouse, similar to the experiment described in (A). Survival was followed up to 9 weeks, and primary tumor burden (mass) and liver metastasis (number and frequency) were analyzed at the end point. HFD-­fed mice had significantly lower survival; n = 14 (CD) and 16 (HFD) mice per group, log-­rank (Mantel-­Cox) test. (H) Number of liver metastasis (>100 μM) per mouse from CD and HFD mice, represented as violin plots. Each dot represents a mouse; n = 12 (CD) and 16 (HFD) mice per group, two-­tailed Student’s t test. The three dashed lines in the violins from top to bottom represent top quartile (75th), median, and bottom quartile (25th), respectively. Dashed horizontal line at y = 0 reference line; y axis extends below zero to display the full kernel density estimate of the violin plot. All observed values are ≥0. (I) Frequency of liver metastasis in CD and HFD cohorts; n = 12 (CD) and 16 (HFD) mice per group, Fisher’s exact test. *P < 0.05, **P < 0.01.

Fig. 2. High-fat diet engages a YAP-­DR program in CRC. (A) Uniform manifold approximation and projection (UMAP) plots of epithelial cell populations in normal colons and AKP-­injected colon tumors revealed by scRNA-­seq. The following analysis in (B to I) was performed on tumor cells from 3 mice per group. (B) Bar plot of enriched pathways in colon tumors from HFD-­fed mice compared with the CD group by GSEA. (C) Enrichment scores for genes in the regenerative signature gene set, hereafter referred to as YAP-­DR. (D) UMAP plot of reclustered tumor cells from epithelial cells in (A). (E) UMAP plots of expression of EpiHR, YAP-­DR, YAP target signatures, and Lgr5high CSCs in tumor cells. (F) Stacked bar plot showing increased YAP-­DRhigh cell frequency in colon tumors from HFD-­fed mice compared with the CD group. Fisher’s exact test. (G) UMAP plots of Basp1 and Ccn2 (a YAP target) expression in tumor cells. (H) Violin plot showing increased expression level of Basp1 in colon tumors from HFD-­fed mice compared with the CD group (P < 2.2 × 10−16). (I) Box plot showing increased expression level of YAP targets in colon tumors from HFD-­fed mice compared with the CD group [two-­tailed Student’s t test for (H) and (I)]. (J) Representative images (scale bars, 20 μm) and quantification of in situ hybridization (ISH) for Basp1 in colon tumors from CD-­ and HFD-­fed mice. Two-­tailed Student’s t test. Each dot represents a field quantified as percent positive cells of total cells per 400 μm2 (n = 15 fields from 3 mice for CD group and 20 fields from 4 mice for HFD group). (K) Representative images (scale bars, 20 μm) and quantification of immunohistochemistry (IHC) for active YAP nuclear localization in colon tumors from CD-­ and HFD-­fed mice [one-­tailed Student’s t test for (J) and (K)]. Each dot represents a field quantified as percent positive cells of total cells per 400 μm2 (n = 15 fields from 3 mice for both CD and HFD groups). The height of bars indicate mean, and error bars indicate SD. (L) Dot heatmap of individual YAP targets in tumor cells of scRNA-­seq data from lean and obese CRC patients (35). (M) Violin plots of YAP-­DR signature and YAP target scores in tumors from obese and lean CRC patients (4). Two-­tailed Student’s t test. (N) Box plot of YAP-­DR signature expression (BASP1, CLU, CCND1, CCND2, CXADR, ANXA1, ANXA3, and CD44) in tumors from lean (n = 11) and obese (n = 9) CRC patients, TCGA stage IV. P = 0.0250, one-­tailed Student’s t test. In box plots, the middle lines indicate median, and the top and bottom edges of boxes indicate the third (75th) and first (25th) quartiles. *P < 0.05, ***P < 0.001, ****P < 0.0001.

Fig. 3. Basp1high CSCs are the origin of liver metastasis. (A) Schematic of generation of naïve Basp1-­eGFP-­DTR; AKP-­tdT organoids. (B) Schematic showing fluorescence-­ activated cell sorting (FACS) of GFP+ and GFP− cells isolated from Basp1-­eGFP-­DTR; AKP-­tdT–injected tumors. (C) Quantitative RT-­PCR demonstrating significantly higher Basp1 expression in GFP+ cells, indicating that GFP+ cells are Basp1high cells. Each dot represents a mouse, n = 4 mice per group, two-­tailed Student’s t test. The height of bars indicates mean, and error bars indicate SD. (D) Significantly higher expression of Basp1 and YAP targets in GFP+ cells as detected by transcriptomic analysis. Each dot represents a mouse, n = 12 (GFP−) and 11 (GFP+) per group. In box plots, the middle lines indicate mean and the top and bottom edges of boxes indicate the third (75th) and first (25th) quartiles. (E) Schematic showing generation of GFP− and GFP+ CRCOs used subsequently in experiments shown in (F to H). GFP− and GFP+ cells were sorted from Basp1-­eGFP-­DTR; AKP-­tdT–injected tumors and propagated as CRCOs. GFP− and GFP+ CRCOs were then orthotopically transplanted into mice, 200,000 CRCO-­derived cells per mouse, and analyzed at 6 weeks posttransplantation to ascertain resulting primary tumor and metastatic potential. (F) GFP+ CRCOs tend to retain a significantly higher proportion of GFP+ cells as detected by flow cytometry. n = 4 (GFP− CRCO) and 5 (GFP+ CRCO) per group, two-­tailed Student’s t test. The height of bars indicates mean, and error bars indicate SD. (G) Liver metastasis per mouse in GFP− CRCO–injected and GFP+ CRCO–injected mice. GFP+ CRCOs showed significantly higher metastatic potential. n = 4 (GFP− CRCO–injected) and 6 (GFP+ CRCO–injected) mice per group, two-­tailed Student’s t test. The height of bars indicates mean, and error bars indicate SD. (H) Primary tumor mass in GFP− CRCO–injected and GFP+ CRCO–injected mice. No significant difference was observed. n = 4 (GFP− CRCO–injected) and 6 (GFP+ CRCO–injected) mice per group, two-­tailed Student’s t test. The height of bars indicates mean, and error bars indicate SD. (I) Schematic of depletion of Basp1high tumor cells in vivo leveraging diphtheria toxin receptor (DTR) expression in Basp1high cells. Mice injected with Basp1-­eGFP-­DTR; AKP-­tdT, 100,000 CRCO-­derived cells per mouse, were allowed to establish a primary tumor for 4 weeks, followed by treatment with control phosphate-­buffered saline or diphtheria toxin (DT) 16.7 μg/kg, 3 times a week intraperitoneally for 4 weeks. Mice were analyzed 8 weeks posttransplantation to ascertain primary tumor and liver metastatic burden. (J) Flow cytometric confirmation that DT treatment (DTX) depletes GFP+ (Basp1high) cells. n = 6 (Control) and 3 (DTX) mice per group, two-­tailed Student’s t test. The height of bars indicates mean, and error bars indicate SD. (K) Liver metastasis per mouse, Control and DT-­treated cohorts (DTX). DTX significantly decreased metastasis. n = 10 (Control) and 20 (DTX) mice per group, two-­tailed Student’s t test. The height of bars indicates mean, and error bars indicate SD. (L) Primary tumor mass in Control and DTX cohorts. No significant difference was observed. n = 10 (Control), 15 (DTX), two-­tailed Student’s t test. The height of bars indicates mean, and error bars indicate SD. *P < 0.05, **P < 0.01, ***P < 0.001; ns, not significant.

Fig. 4. Comprehensive lipidomic profiling reveals tumor and diet induced enrichment of ceramide metabolism. (A) Schematic of spatial lipidomic analysis of primary tumors, generation of primary tumor-­derived AKP organoids (primary CD and HFD CRCOs), followed by analysis of metastatic potential (in CD recipient mice) and lipidomics, and generation of secondary CD→CD and HFD→CD CRCOs followed by lipidomic analysis. (B) Liver metastasis burden per mouse of CD recipient mice injected with primary CD and HFD CRCOs. n = 14 (primary CD CRCO) and 13 (primary HFD CRCO) mice per group, two-­tailed Student’s t test. The height of bars indicates mean, and error bars indicate SD. (C) Schematic of enrichment of GFP+ cells from CD and HFD primary tumors, followed by serial transplantation of equal number of GFP+ cells (50,000 per mouse) into secondary CD recipients and analysis of metastasis at 4 weeks. (D) Liver metastasis per mouse of CD recipient mice injected with CD and HFD-­derived GFP+ cells. Each dot represents a mouse. n = 18 (CD-­derived GFP+) and 19 (HFD-­derived GFP+) mice per group, two-­tailed Student’s t test. The height of bars indicates mean, and error bars indicate SD. (E) Principal components analysis (PCA) plot of all individual lipids quantified by mass spectrometry from primary CD and HFD CRCOs. Each dot represents a primary (in vivo tumor-­derived) CRCO, n = 6 per group. (F) Volcano plot of up-­regulated lipid species in primary HFD CRCOs organoids compared with the CD group. Each dot represents a lipid; lipids with log2FoldChange (log2FC) >1 or <–1 and adjusted P value <0.05 are highlighted. (G) Schematic of the de novo ceramide synthesis pathway in cells. Degs1, catalyzing conversion of dihydroceramide to ceramide, is highlighted. (H) PCA plot of ceramides in primary CD and HFD CRCOs. Each dot represent a CRCO, n = 6 per group. (I) PCA analysis of spatial metabolomics results for CRC gut tissue from colon of CRC-­transplanted mice. (J) Volcano plot of colon of CRC-­transplanted mice comparing tumor-­containing and non–tumor-­containing tissue regions. Ceramides are significantly up-­regulated in tumor regions. Log2FC > 1.5 and Padj < 0.05. (K) Quantification of specific ceramide species in tumor and nontumor regions. Each dot represents a region of interest, n = 155 (nontumor) and 88 (tumor) per group for (I) and (K). Two-­tailed Student’s t test. (L) Comparison of spatial lipidomics results for [Cer 42:2;O2+Na]+ in representative CD and HFD samples with hematoxylin and eosin (H&E) stains of the same tissue sections. Total ion current normalized intensities are indicated, and scale bars are 400 μm. (M) H&E stain and spatial lipidomics of the same HFD gut section. The ceramide [Cer 42:2;O2+Na]+ (m/z 670.6109, red) is compared with [CerP 44:2;O2+Na]+ (m/z 778.6085, blue) and [PA O-­40:6+Na]+ (m/z 757.5143, green). Scale bars are 400 μm (top) and 250 μm (bottom) in the magnified views. (N) PCA plot of all individual lipids quantified by mass spectrometry from secondary CD→CD and HFD→ CD CRCOs. Secondary HFD→CD CRCOs still retain distinct lipid profiles from secondary CD→CD CRCOs. Each dot represents a CRCO, n = 11 (CD→CD) and 8 (HFD→CD) per group. (O) Box plots showing levels of specific ceramide species in the secondary CD→CD and HFD→CD CRCOs. Each dot represents a mouse, n = 11 (CD→CD) and 8 (HFD→CD) per group, two-­tailed Student’s t test. The middle lines of the boxes indicate mean, and the top and bottom edges of boxes indicate the third (75th) and first (25th) quartiles. *P < 0.05, **P < 0.01.

Fig. 5. High-fat diet increases ceramide synthesis in YAP-­DRhigh CSCs. (A) UMAP plots of Basp1, Degs1, and Degs2 expression in tumor cells. (B) Violin plot showing increased Degs1 expression in colon tumors from HFD-­fed mice compared with the CD group, two-­tailed Student’s t test. (C) Representative images and quantification of ISH for Degs1 in colon tumors from CD-­ and HFD-­fed mice. Scale bars, 20 μm. Each dot represents a field quantified as percent positive cells of total cells per 400 μm2 (n = 15 fields from 3 mice for CD group and 20 fields from 4 mice for HFD group), two-­tailed Student’s t test. (D) Violin plot (left) and box plot (right) of DEGS1 expression in CRC tumors from scRNA-­seq [two-­tailed Student’s t test (4)] and bulk RNA-­seq (one-­tailed Student’s t test, n = 11 and 9 per group, TCGA stage IV) of lean and obese CRC patients. (E) Representative images and (F) quantification of ISH showing increased expression of DEGS1 in colon tumors of obese CRC patients compared with lean patients. Scale bars, 50 μm. Each dot represents the average number of fluorescent puncta per cell (n = 17 fields from 3 lean patients and 14 fields from 3 obese patients), two-­tailed Student’s t test. (G) Schematic showing transcriptomic and open chromatin analysis of primary CD and HFD CRCOs. (H) Box plots showing increased expression (RNA-­seq) and normalized ATAC-­seq tracks showing increased open chromatin regions of Basp1 and (I) Degs1 in primary HFD CRCOs compared with CD group. Each dot represents a CRCO; n = 3 (CD) and 5 (HFD) per group, one-­tailed Student’s t test. In bar plots, the height of bars indicates mean, and error bars indicate SD. In box plots, the middle lines indicate median, and the top and bottom edges of boxes indicate the third (75th) and first (25th) quartiles, respectively. *P < 0.05, **P < 0.01, ****P < 0.0001.

Fig. 6. Ceramides activate the YAP-­DR program to promote liver metastasis. (A) Part of ceramide de novo synthesis pathway highlighting Degs1 and Asah1. (B) schematic of generation of Scrambled control and Degs1−/− naive AKP-­tdT organoids through CRISPR-­editing. (C) Functional validation of loss-­of-­function deletion of DEGS1 in Degs1−/− AKP-­tdT CRCOs, as seen by significant decrease in production of ceramides (Cer d18:1). Each dot represents a CRCO; n = 5 (Scrambled) and 6 (Degs1−/−) per group, one-­tailed Student’s t test. (D) Box plots showing expression of YAP targets and regenerative marker Basp1 in Degs1−/− AKP-­tdT CRCOs. Knockout of Degs1 significantly decreases YAP activity and Basp1 marker. Each dot represents a CRCO; n = 12 (Scrambled) and 18 (Degs1−/−) per group, two-­tailed Student’s t test. (E) Scrambled and Degs1−/− AKP-­tdT CRCOs were orthotopically transplanted into mice on regular chow to assess metastatic potential after 8 weeks. (F) Liver metastasis per mouse. Degs1−/− AKP-­tdT CRCOs give rise to significantly less metastasis. Each dot represents a mouse; n = 13 (Scrambled) and 12 (Degs1−/−) mice per group, two-­tailed Student’s t test. The height of bars indicates mean, and error bars indicate SD. (G) Scrambled and Degs1−/− AKP-­tdT CRCOs were orthotopically transplanted into mice fed CD or HFD to assess impact on tumor progression. (H) Liver metastasis per mouse. Degs1−/− reversed the ability of HFD to increase metastasis, as seen in Scrambled control. Each dot represents a mouse; n = 16 (Scrambled, CD), 25 (Scrambled HFD), 18 (Degs1−/− CD), and 19 (Degs1−/− HFD) mice per group, 2-­way ANOVA. The three dashed lines in the violins from top to bottom represent top quartile (75th), median, and bottom quartile (25th), respectively. Dashed horizontal line at y = 0 reference line; y axis extends below zero to display the full kernel density estimate of the violin plot. (I) Schematic showing generation of tdTomato-­labeled, DEGS1−/− human CRC patient–derived organoids (patient-­derived material or PDM) followed by orthotopic transplantation. (J) Functional validation of loss-­of-­function deletion of DEGS1 in DEGS1−/− PDMs, as seen by significant decrease in production of ceramides. PDM5: n = 12 (Scrambled) and 3 (DEGS1−/−), one-­tailed Student’s t test. PDM7: n = 6 (Scrambled) and 12 (DEGS1−/−), one-­tailed Student’s t test. PDM9: n = 9 (Scrambled) and 6 (DEGS1−/−), one-­tailed Student’s t test. (K) Representative brightfield and fluorescent images of primary tumor in the colon and metastatic tumors in the liver when mice were transplanted with patient-­derived organoids. (L) Liver metastasis per mouse. DEGS1−/− CRCOs give rise to significantly less metastasis across PDM5, 7, and 9. Each dot represents a mouse. PDM5: n = 9 (Scrambled and DEGS1−/−) mice per group, two-­tailed Student’s t test. PDM7: n = 8 (Scrambled) and 7 (DEGS1−/−) mice per group, two-­tailed Student’s t test. PDM9: n = 5 (Scrambled) and 4 (DEGS1−/−) mice per group, two-­tailed Student’s t test. The height of bars indicates mean, and error bars indicate SD. (M) Loss of function of acid ceramidase Asah1 in Asah1−/− AKP-­tdT CRCOs resulted in significantly increased ceramides. n = 8 (Scrambled) and 15 (Asah1−/−) per group, one-­tailed Student’s t test, ± SD. (N) Box plots showing expression of YAP targets and regenerative marker Basp1 in Asah1−/− AKP-­tdT CRCOs in RNA-­seq. Knockout of Asah1 significantly increases YAP activity and Basp1 expression in two independent clones. n = 3 (Scrambled) and 5 (Asah1−/−) mice per group, one-­tailed Student’s t test. (O) Liver metastasis per mouse at 4 weeks. Asah1−/− AKP-­tdT CRCOs give rise to significantly higher metastasis. Each dot represents a mouse; n = 9 (Scrambled) and 19 (Asah1−/−) mice per group, two-­tailed Student’s t test. (P) Immunoblot- ting of active YAP levels in AKP CRCOs treated with ceramide C16 and C24:1 liposomes (LP), 5 μM for 5 days. C24:1-­LP—but not C16-­LP—significantly increased active YAP. Representative images and quantification of n = 6 for control, C16-­LP and C24:1-­LP groups, one-­way ANOVA, ± SD. (Q) Box plots showing expression of YAP-­DR marker Basp1 and (R) YAP targets in Control and Degs1 overexpressing (Degs1 OE) cells, treated with or without PP2A inhibitor LB100 (5 μM). LB100 treatment reverses Degs1 OE– mediated increase in YAP-­DR markers and YAP targets. n = 3 (Control), 5 (Degs1 OE), 3 (Control LB100), and 3 (Degs1 OE LB100) replicates per group, one-­tailed Student’s t test. (S) Immunoblotting and quantification showing coimmunoprecipitation of PP2A catalytic subunit with total YAP in Control and Degs1 OE. Increased levels of PP2A coimmunoprecipitated with total YAP protein in Degs1 OE cells, indicating increased interaction. n = 3 (Control and Degs1 OE), one-­tailed Student’s t test, ± SD. (T) Schematic summarizing overall observations. Ceramide synthesis, further enhanced by exposure to dietary fat, can activate YAP to drive YAP-­DR programs which are responsible for metastatic initiation from the primary tumor. In bar plots, the height of bars indicates mean, and error bars indicate SD. In box plots, the middle lines indicate median, and the top and bottom edges of boxes indicate the third (75th) and first (25th) quartiles, respectively. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.