Fig 2: Acidosis-induced TGF-ß2 supports cancer progression by stimulating both EMT and lipid storage in tumor cells.Acidosis triggers TGF-ß2 upregulation/activation in a TSP1-dependent manner (1) and promotes an autocrine signaling pathway through TGF-ßRI receptor and subsequent Smad2/3 phosphorylation (2) in cancer cells. TGFß2-mediated signaling pathways induce a partial EMT gene programme (3) as well as a PKC?-dependent translocation of CD36 fatty acid translocase to the plasma membrane (4). TGF-ß2 autocrine signaling, in acidosis-adapted cancer cells, favors the uptake of long-chain fatty acids (5), the latter being oxidized in mitochondria for energy production (6) but also stored, as neutral lipids (triacylglycerols) in lipid droplets, in a DGAT1-dependent manner (7); the resulting increase in the cellular acetyl-CoA pool accounts for an overall increase in the acetylation of proteins, including Smad2 (thereby increasing its transcriptional activity). Both TGFß2-induced partial EMT and lipid storage participate to increase survival (i.e., anoikis resistance) and invasion capacities that support in vivo metastatic spreading, ATP needs being fulfilled by FAO upon preferred hydrolysis of TG stored into LD (8). Inhibitors able to block different steps of the acidosis-induced TGF-ß2/DGAT1 axis are indicated in red. ACSL1: acyl-CoA synthetase long chain family member 1; ATGL: adipose triglyceride lipase; CPT1: carnitine palmitoyltransferase 1; DAG: diacylglycerol; endo: endosomes; FA: fatty acids; PKC?: protein kinase C ?; PLIN2: perilipin 2; TAG: triacylglycerol; TSP-1: thrombospondin-1.

Fig 3: TGF-ß2 promotes FA uptake and TG accumulation into LD.a–c Abundance of neutral lipids (NL), phospholipids (PL) and free fatty acids (FFA) (a), abundance of saturated and monounsaturated fatty acids (SFA and MUFA, respectively) in the neutral lipid fraction (b), and LD content (c) in native SiHa cells after treatment with 4 ng/ml TGF-ß2 for 6 h. d, e 14C-palmitate uptake for 10 min in SiHa cells after treatment with 4 ng/ml TGF-ß2 for 6 h (d) and in acidosis-adapted SiHa cells following treatment with 10 µM Trabedersen for 7 days in absence or presence of 4 ng/ml TGF-ß2 for 24 h (e). f–h Representative immunoblotting for cell surface-localized CD36 and total biotinylated proteins in native and acidosis-adapted SiHa cells following treatment with 4 ng/ml TGF-ß2 for 6 h and 24 h (f), following treatment with 10 µM Trabedersen for 7 days or 2 µM SB431542 for 24 h (g) or with 4 ng/ml TGF-ß2 and 10 µM PKC-? pseudo-substrate inhibitor for 24 h (h). i, j Quantification of surface-localized CD36 in native and acidosis-adapted SiHa cells treated as indicated in (h). k, l mRNA (k) and protein expression of DGAT1 (l) in native and acidosis-adapted tumor cells. m–o Representative immunoblotting for DGAT1 in native and acidosis-adapted SiHa cells following treatment with 10 µM Trabedersen for 7 days or 2 µM SB431542 for 24 h (m), with 10 µM PKC-? pseudo-substrate inhibitor for 24 h (n) or with 10 µM GW6471 for 48 h (o). p mRNA expression of PLIN2 in native and acidosis-adapted SiHa cells following treatment with 10 µM TGFß2-specific antisense oligonucleotide Trabedersen for 7 days. q Co-expression analysis of TGFB2, PLIN1, PLIN2, and PLIN3 genes in human healthy volunteers and colorectal cancer patient samples. Data are represented as mean ± SEM of three independent experiments (with =6 technical replicates). Significance was determined by Student’s t-test (c, d, j), by one-way ANOVA (e–i) or two-way ANOVA (a, k, p) with Bonferroni multiple-comparison analysis. *p < 0.05; **p < 0.01; ***p < 0.001; ns, not significant. Source data are provided as a Source Data file.

Fig 4: GW4064 and CDCA reduce CD36 expression and function.(a) HepG2 cells were treated with the indicated concentrations of GW4064 or chenodeoxycholate (CDCA) in media containing lipoprotein-deficient serum (lpds) for 24 hours and gene expression was analyzed by qRT-PCR (n = 3). (b) Cells were incubated with 10 µM GW4064 or 100 µM CDCA in media containing lpds for 24 hrs and protein expression was determined by western blot analysis and results were quantitated by densitometry (n = 3). (c) Fatty-acid uptake was determined after treatment with 10 µM GW4064 or 100 µM CDCA as described in the methods section (n = 3).

Fig 5: EMT in acidosis-adapted cancer cells is driven by TGF-ß2.a Representative immunofluorescence pictures for EMT-related protein markers in native and acidosis-adapted SiHa cells. Scale bar: 20 µm. b mRNA expression for epithelial (CDH1) and mesenchymal (CDH2, SNAIL1, SNAIL2, and ZEB1) genes in native and acidosis-adapted SiHa cells. c Representative immunoblotting for some epithelial (ZO-1 and E-cadherin) and mesenchymal (N-cadherin, vimentin, ZEB1, Snail) protein markers in native and acidosis-adapted cancer cell lines. d, e mRNA expression for SNAIL1 (d) and ZEB1 (e) in native and acidosis-adapted SiHa cells following treatment with 10 µM Trabedersen for 7 days or 2 µM SB431542 for 24 h. f Representative immunoblotting for mesenchymal protein markers in native and acidosis-adapted SiHa cells following treatment as above. g, h mRNA expression for SNAIL1 (g) and ZEB1 (h) in native and acidosis-adapted SiHa cells following transfection of Smad2/3-targeting (or control) siRNA for 72 h. i LD content in native and acidosis-adapted SiHa cells following transfection with ZEB1-targeting (or control) siRNA for 72 h. j, k Palmitate-dependent oxygen-consumption rate (OCR) in native and acidosis-adapted SiHa cells after treatment with 4 ng/ml TGF-ß2 for 6 h (j) or following treatment with 10 µM Trabedersen for 7 days or 2 µM SB431542 for 24 h (k). l Levels of cellular acetyl-CoA in native and acidosis-adapted SiHa cells. m, n Representative immunoblotting for acetylated Smad2 in native and acidosis-adapted cancer cells (with or without treatment with 4 ng/ml TGF-ß2 for 6 h) (m) and following treatments with 30 µM etomoxir (CPT1 inhibitor), 2 µg/ml JC63.1 or 1 µg/ml FA6-152 (anti-CD36 blocking antibodies) for 24 h (n). Data are represented as mean ± SEM of three independent experiments (with =6 technical replicates). Significance was determined by Student’s t-test (l) or two-way ANOVA (b, d, e, g–k) with Bonferroni multiple-comparison analysis. **p < 0.01; ***p < 0.001; ns, not significant. Source data are provided as a Source Data file.