The cholesterol biosynthesis enzyme FAXDC2 couples Wnt/β-catenin to RTK/MAPK signaling

Abstract

Wnts, cholesterol, and MAPK signaling are essential for development and adult homeostasis. Here, we report that fatty acid hydroxylase domain containing 2 (FAXDC2), a previously uncharacterized enzyme, functions as a methyl sterol oxidase catalyzing C4 demethylation in the Kandutsch-Russell branch of the cholesterol biosynthesis pathway. FAXDC2, a paralog of MSMO1, regulated the abundance of the specific C4-methyl sterols lophenol and dihydro-T-MAS. Highlighting its clinical relevance, FAXDC2 was repressed in Wnt/β-catenin-high cancer xenografts, in a mouse genetic model of Wnt activation, and in human colorectal cancers. Moreover, in primary human colorectal cancers, the sterol lophenol, regulated by FAXDC2, accumulated in the cancerous tissues and not in adjacent normal tissues. FAXDC2 linked Wnts to RTK/MAPK signaling. Wnt inhibition drove increased recycling of RTKs and activation of the MAPK pathway, and this required FAXDC2. Blocking Wnt signaling in Wnt-high cancers caused both differentiation and senescence; and this was prevented by knockout of FAXDC2. Our data show the integration of 3 ancient pathways, Wnts, cholesterol synthesis, and RTK/MAPK signaling, in cellular proliferation and differentiation.

Keywords: Cancer; Cholesterol; Metabolism; Oncology; Phosphotyrosine.

Figure 1. Wnt signaling represses the cholesterol biosynthesis enzyme FAXDC2.

(A) Wnt inhibition upregulates genes regulating cholesterol biosynthesis. Pharmacologic Wnt inhibition with ETC-159 in HPAF-II orthotopic xenografts upregulates 4,350 genes that are grouped into temporal clusters (n = 4–6 mice per group). GO Biological Process and Reactome analysis of the Wnt-repressed genes highlights processes including cholesterol biosynthesis, vesicle-mediated transport, and EGFR/VEGF signaling (hypergeometric test, FDR < 10%). (B) Wnt signaling represses genes encoding enzymes in the cholesterol biosynthesis pathway. Left: Wnt inhibition with ETC-159 increases expression of multiple cholesterol biosynthesis pathway genes in 3 independent systems: HPAF-II and AsPC-1 orthotopic xenografts and cells in culture and colorectal PDX (14) (n = 4–6 tumors per group). Boxed genes show significant change in expression (FDR < 0.10). Right: Cholesterol biosynthesis pathway. (C) Wnt inhibition increases FAXDC2 expression in multiple Wnt-addicted cancer models: ETC-159–treated HPAF-II, AsPC-1 orthotopic tumors, and Wnt-addicted colorectal and pancreatic cancer PDX models have 2- to 10-fold higher FAXDC2 expression compared with control tumors. Relative expression from RNA-Seq in TPM or quantitative reverse transcription PCR (qRT-PCR) is shown. Each data point represents an independent tumor, n = 5–6 per group (hypergeometric test, FDR < 10%, or Mann-Whitney test). (D) ETC-159 treatment increases FAXDC2 protein abundance in tumors. Protein lysates from HPAF-II and AsPC-1 orthotopic xenografts from vehicle- or ETC-159–treated mice were probed with FAXDC2 or GAPDH antibodies. Each lane represents tumor lysate from an individual mouse. (E) Stabilized β‑catenin suppresses FAXDC2 expression despite upstream Wnt inhibition by ETC-159. Mice bearing xenografts from control HPAF-II cells or cells expressing stabilized β‑catenin were treated with ETC-159 or vehicle for 56 hours. FAXDC2 and AXIN2 mRNA were quantitated by qRT-PCR and normalized to both ACTB and EPN1. AXIN2 upregulation is a control for stabilized β‑catenin activity. (F) Genetic inhibition of Wnt signaling by TCF7L2 knockout in HT29 and HCT116 colon cancer xenografts increases FAXDC2 expression 2- to 8-fold in comparison with WT controls. (E and F) Each data point represents an independent tumor. Unpaired t test was used to calculate P values.

Figure 2. FAXDC2 is a C4-demethylase in the KR branch of the cholesterol biosynthesis pathway.

(A) Postlanosterol cholesterol biosynthesis via Bloch and KR pathways, highlighting key steps and associated enzymes. *Proposed location of FAXDC2 in the pathway. (B) Conversion of T-MAS to zymostenol, highlighting reduction at C24 and demethylation at C4 involving sterol methyl oxidase (SMO). *SMO indicates either MSMO1 or FAXDC2 sterol methyl oxidase. (C) MSMO1 and FAXDC2 structures predicted by AlphaFold show considerable homology (root-mean-square deviation of atomic positions 1.36 Å). (D) Like MSMO1, FAXDC2 colocalizes with NSDHL in the SER. Epitope-tagged MSMO1, NSDHL, and FAXDC2 constructs were expressed in HeLa cells, and their localization was visualized by staining with fluorescence-tagged anti-epitope antibodies. (E–G) Combined knockdown of FAXDC2 and MSMO1 reduces total cholesterol levels and prevents cellular proliferation. HPAF-II cells, HPAF-II cells with doxycycline-inducible (DOX-inducible) single-guide (isg) RNAs targeting FAXDC2 or MSMO1 alone, and FAXDC2-KO cells with DOX-isg targeting MSMO1 were cultured for 10 days. (E) Representative images of crystal violet staining from 3 independent experiments. (F) Crystal violet dye was solubilized, and absorbance was measured at 570 nm. (G) Total cholesterol levels were significantly reduced in the FAXDC2 and MSMO1 double-KO cells compared with the single knockouts. (H and I) ETC-159 treatment reduces C4-methyl sterols in HPAF-II orthotopic tumors with high FAXDC2 expression. Sterol levels in mice treated with vehicle or ETC-159 were measured by GC-MS. Data from individual tumor samples (n = 4–5 per group) from 2 independent experiments were combined to calculate P values, controlling for batch effects. (J–L) Stabilized β-catenin, which represses FAXDC2, increased C4-methyl sterol levels independent of upstream Wnt signaling inhibition. C4-methyl sterol levels in HPAF-II tumors with and without stabilized β-catenin from mice treated with ETC-159 or vehicle were analyzed by GC-MS. Each point denotes an individual tumor, n = 4–5 per group. (M) C4-methyl sterols are reduced in TCF7L2-KO HCT116 xenografts compared with control tumors. Sterol levels were measured by GC-MS. Each point depicts an individual tumor sample. P values were calculated by Mann-Whitney U test (G and M) and unpaired t test (J–L). *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001.

Figure 3. FAXDC2 expression correlates with and drives changes in C4-methyl sterols.

(A–C) Primary Wnt-high colorectal cancers have repressed FAXDC2 expression and high levels of lophenol. (A) FAXDC2 is repressed in primary colorectal cancers compared with corresponding normal tissues. Expression of FAXDC2 was compared in normal tissues versus tumor tissues using GEPIA 2.0 (TCGA and GTEx data sets). (B) H3K4me3, a marker of active transcription, marks the FAXDC2 genomic locus in the normal colon but is absent in cancer cell lines. Data from Cistrome Data Browser. (C) Lophenol is markedly elevated in primary colorectal cancers. Sterol abundance was measured in primary colorectal cancers and adjacent normal tissue by GC-MS. Each point represents lophenol levels in an individual tumor, n = 12 tumors per group. (D–H) Manipulation of FAXDC2 expression regulates C4-methyl sterol abundance. (D) Relative expression of FAXDC2 in HPAF-II, FAXDC2-KO, and FAXDC2-OE xenografts as measured by qRT-PCR. (E and F) Knockout of FAXDC2 abrogates the ETC-159 treatment–induced change in the abundance of C4-methyl sterols. Sterol abundance was measured in HPAF-II and FAXDC2-KO xenografts from control and ETC-159–treated mice. Each point represents the level of indicated methyl sterols measured using GC-MS in an individual tumor sample, n = 4–5 per group. (G and H) FAXDC2 overexpression reduces lophenol and T-MAS to levels comparable to those in the Wnt-inhibited tumors. Sterol abundance was measured in HPAF-II and FAXDC2-OE xenografts from control and ETC-159–treated mice. Each point represents the abundance of methyl sterols measured using GC-MS in an individual tumor sample, n = 4–5 per group. P values were calculated by Mann-Whitney U test for all graphs in this figure. ***P ≤ 0.001.

Figure 4. FAXDC2 is a downstream effector of Wnt signaling, and FAXDC2-regulated genes are enriched for RTK signaling.

(A) Expression of a subset of Wnt-regulated genes depends on FAXDC2 (FDR < 10%). Heatmap of fold changes of 3,560 FAXDC2-dependent Wnt-regulated genes (interaction test, FDR < 10%). Left: Gene expression changes (log₂ fold changes) between WT and FAXDC2-KO tumors at 0 hours. Right: Gene expression changes following treatment with ETC-159 in both parental HPAF-II tumors (56 vs. 0 hours) and FAXDC2-KO tumors (56 vs. 0 hours). This set of differentially responding genes was partitioned into 6 distinct clusters based on similarities in their response to ETC-159, including 3 clusters of Wnt-repressed genes (A, B, and C) and 3 three clusters of Wnt-activated genes (D, E, and F). (B) Representative genes from 4 clusters in part A show that FAXDC2 knockout blunts or reverses the response to Wnt inhibition (hypergeometric test, FDR < 10%). TPM, transcripts per million. (C) FAXDC2-regulated Wnt-repressed genes in clusters A–C show an enrichment of AP1 family TFBS motifs in their promoters (hypergeometric test, FDR < 10%). (D) RTK signaling pathways are enriched in FAXDC2-regulated Wnt-repressed genes. Data from representative GO Biological Process and Reactome enrichment of Wnt-repressed genes from clusters A–C that were identified in part A (hypergeometric test, FDR < 10%). (E) GO Biological Processes and Reactome enrichment analysis of Wnt-repressed genes in ETC-159–treated AsPC-1 and colorectal cancer PDX tumors shows an upregulation of RTK signaling pathways (hypergeometric test, FDR < 10%). (F) Pharmacologic Wnt inhibition with ETC-159 in HPAF-II xenografts increases protein tyrosine phosphorylation. HPAF-II orthotopic tumor protein lysates from vehicle- or ETC-159–treated mice were separated on a 10% SDS gel. Blots were probed with phosphotyrosine antibodies. Each lane represents tumor lysate from an individual mouse. Asterisks indicate prominent phospho-tyrosine bands. (G and H) Genetic inhibition of Wnt/β-catenin signaling in TCF7L2-KO HT29 and HCT116 xenografts increases protein tyrosine phosphorylation. Protein lysates from WT and TCF7L2-KO HT29 and HCT116 tumors were analyzed for abundance of phosphotyrosine proteins as in F. Each lane contains tumor lysate from an individual mouse.

Figure 5. Increased RTK recycling in Wnt-addicted cancers upon Wnt inhibition.

(A and B) ETC-159 treatment of HPAF-II cells increases abundance of EGFR on the cell surface. HPAF-II cells were treated with 100 nM ETC-159 for 72 hours. (A) Cells were stained with Alexa Fluor 488–conjugated anti-EGFR antibody and analyzed by flow cytometry. Each histogram represents ~50,000 cells. Data are representative of 3 independent experiments (P = 0.009). (B) Endogenous EGFR levels on non-permeabilized cells were assessed by indirect immunofluorescence microscopy. Scale bars: 20 μm. (C and D) Partial knockdown of RAB11B or EHD1 blunts the ETC-159–induced EGFR increase on the surface of HPAF-II cells. HPAF-II cells were transfected with 2 independent siRNAs (si6 and si7) against RAB11B (C) or EHD1 (D) for 24 hours, followed by treatment with ETC-159 for 72 hours. EGFR levels on the cell surface were assessed by flow cytometry. Each histogram represents ~50,000 cells from 1 replicate. Data are representative of 3 independent experiments. Average median fluorescence intensity (MFI) of the technical replicates is shown. Av MFI, average MFI of the technical replicates from the same experiment. (E) EPHB2 and EPHB4 levels are increased by Wnt inhibition. HPAF-II cells were treated with 100 nM ETC-159 for 72 hours, and the levels of endogenous EPHB2 and EPHB4 on non-permeabilized cells were assessed by indirect immunofluorescence microscopy. Scale bars: 20 μm. (F–I) Wnt inhibition increases activation of EPHA2 and EGFR receptor tyrosine kinases and protein abundance of multiple receptor tyrosine kinases in HPAF-II xenografts and pancreatic PDX models. Protein lysates from HPAF-II orthotopic xenografts or pancreatic PDX from vehicle- or ETC-159–treated mice were analyzed for expression of p-EPHA2 and p-EGFR and abundance of indicated RTKs by Western blots. Each lane represents tumor lysate from an individual mouse. (H) The protein lysates were prepared as a master mix and loaded on independent gels. Only 1 blot was probed for the load control, which is shared with Figure 4F.

Figure 6. FAXDC2 is downstream of Wnt/β-catenin in the regulation of RTK signaling.

(A) Stabilized β-catenin prevents multiple RTKs from moving to the cell surface upon treatment with ETC-159. Cell surface abundance of EGFR, ERBB2, and EPHA2 was assessed as before in HPAF-II cells with or without stabilized β-catenin. Cells were treated with DMSO or 100 nM ETC-159 for 72 hours before staining with fluorescent tagged antibodies. Each histogram represents ~50,000 cells. Data are representative of 3 independent experiments, and 2 replicates are shown. (B) Knockdown of FAXDC2 prevents Wnt inhibition–mediated increase in EGFR cell surface abundance. HPAF-II cells were transfected with a pool of 4 siRNAs or 2 independent siRNAs (si6 and si7) against FAXDC2 for 24 hours, followed by treatment with ETC-159 for 72 hours. The cells were then stained with Alexa Fluor 488–conjugated anti-EGFR antibody and analyzed by flow cytometry. Data are representative of 3 independent experiments, and 2 replicates are shown. (C and D) Knockout of FAXDC2 in HPAF-II tumors prevents Wnt inhibition–mediated increase in tyrosine phosphorylation and p-EPHA2 levels. Protein lysates from HPAF-II or FAXDC2-KO tumor xenografts from vehicle- or ETC-159–treated mice were separated on 10% SDS gels and transferred to a PVDF membrane. Membranes were probed with anti-phosphotyrosine or –p-EphA2 antibodies. Each lane represents lysate from an individual tumor. (E) Overexpression of FAXDC2 in FAXDC2-KO tumors rescues the Wnt inhibition–mediated increase in tyrosine phosphorylation. Protein lysates from FAXDC2-KO or FAXDC2-OE tumor xenografts from vehicle- or ETC-159–treated mice were probed with phosphotyrosine antibodies. Each lane represents lysate from an individual tumor. (F–H) FAXDC2 knockout prevents the Wnt inhibition–mediated increase in EGFR and Eph family receptor tyrosine kinase abundance. Protein lysates from HPAF-II or FAXDC2-KO tumors from mice treated with vehicle or ETC-159 were analyzed as in C. Data are shown from 2 independent FAXDC2-KO clones, clone 3 (F and G) and clone 12 (H). Each lane represents lysate from an individual tumor.

Figure 7. FAXDC2 knockout prevents Wnt inhibition–induced MAPK activation, cellular differentiation, and senescence.

(A and B) FAXDC2 is required for MAPK activation. Xenografts from 2 independent FAXDC2-KO clones had reduced ERK phosphorylation with no further increase upon ETC-159 treatment. Protein lysates were prepared as a master mix and loaded on independent gels. Load control of A is shared with Figure 6D and of B with Figure 6H. (C) FAXDC2 overexpression mediates sustained ERK activation. Western blot analysis of protein lysates from indicated xenografts. Each lane represents tumor lysate from an individual mouse. (D) Knockout of FAXDC2 reduced the growth of HPAF-II subcutaneous xenografts, with further reduction upon Wnt inhibition. n = 5–6 mice per group. (E) Overexpression of FAXDC2 delayed implantation and reduced growth of HPAF-II xenografts. n = 5–6 mice per group. (F and G) Wnt inhibition caused an increase in senescence-associated β-galactosidase in HPAF-II tumors that was diminished in the FAXDC2-KO tumors. (F) Representative images are shown. Scale bars: 100 μm. (G) Percentage of positively stained area (blue) was quantitated. Each dot represents quantitation of a tumor section from an individual mouse. P values were calculated by Mann-Whitney U test. (H and I) FAXDC2 knockout blunts the differentiation response to Wnt inhibition. (H) Expression of selected differentiation markers in tumors assessed by RNA-Seq. Each data point represents an individual tumor (hypergeometric test, FDR <10%). (I) FAXDC2 knockout blunts the Wnt inhibition–mediated increase in mucin expression as assessed by Alcian blue staining. Scale bars: 50 μm. (J–L) The Wnt/MAPK axis is present in non-malignant mouse pancreas. (J and K) Genetic activation and subsequent pharmacologic inhibition of Wnt signaling in mouse pancreas led to reciprocal regulation of Faxdc2 expression. Pancreas from control or ETC-159–treated WT and Ptf1α^Cre Rnf43^–/– Znrf3^–/– mice was analyzed for (J) Tcf7 and (K) Faxdc2 (Gm12248) expression by qRT-PCR. Data are from 2 independent biological experiments; each dot represents an individual mouse. P values were calculated by the Mann-Whitney U test. (L) Wnt activation reduced p-ERK levels in the mouse pancreas as assessed by IHC. Representative image of anti–p-ERK–stained pancreas. Scale bars: 100 μm.