AhR and SHP regulate phosphatidylcholine and S-adenosylmethionine levels in the one-carbon cycle

Abstract

Phosphatidylcholines (PC) and S-adenosylmethionine (SAM) are critical determinants of hepatic lipid levels, but how their levels are regulated is unclear. Here, we show that Pemt and Gnmt, key one-carbon cycle genes regulating PC/SAM levels, are downregulated after feeding, leading to decreased PC and increased SAM levels, but these effects are blunted in small heterodimer partner (SHP)-null or FGF15-null mice. Further, aryl hydrocarbon receptor (AhR) is translocated into the nucleus by insulin/PKB signaling in the early fed state and induces Pemt and Gnmt expression. This induction is blocked by FGF15 signaling-activated SHP in the late fed state. Adenoviral-mediated expression of AhR in obese mice increases PC levels and exacerbates steatosis, effects that are blunted by SHP co-expression or Pemt downregulation. PEMT, AHR, and PC levels are elevated in simple steatosis patients, but PC levels are robustly reduced in steatohepatitis-fibrosis patients. This study identifies AhR and SHP as new physiological regulators of PC/SAM levels.

Fig. 1

FGF19 inhibition of hepatic PC production is SHP dependent. a Schematic diagram of the 1C cycle. b From liver ChIP-seq data for mice treated with FGF19, positions of SHP binding peaks at the promoter regions of key genes involved in the 1C cycle are shown. c, d Effects of FGF19 treatment for 2 h on SHP occupancy determined by ChIP at 1C genes in C57BL/6 mice (c), and comparison of occupancy of SHP at Pemt and Gnmt genes in C57BL/6 and SHP-KO mice (d). e, f C57BL/6 or SHP-KO mice were treated with FGF19 for 6 h, and mRNA levels of 1C cycle genes (e) or 1C cycle metabolites levels (f) were determined by qRT-PCR or LC–MS, respectively. c–f Means ± SD are shown (n = 5 mice), and statistical significance was measured using the two-way ANOVA with the FDR post-test and *P < 0.05, **P < 0.01, NS, not statistically significant

Fig. 2

FGF15 and SHP are physiological repressors of PC production in the fed state. C57BL/6 mice and SHP-KO mice (a, b) or FGF15-KO and WT littermate (c, d) mice were fasted for 12 h (Fs) or fasted and refed for 6 h (Fd). Hepatic mRNA levels of selected 1C cycle genes (a, c) and 1C cycle metabolites (b, d) were measured by qRT-PCR and LC–MS analysis, respectively. Means ± SD are shown (n = 5 mice), and statistical significance was measured using the two-way ANOVA with the FDR post-test and *P < 0.05, **P < 0.01, NS, not statistically significant

Fig. 3

AhR transactivates 1C cycle genes and increases hepatic PC levels. a Potential binding sites for transcription factors within SHP peak regions at 1C genes were identified using the JASPAR database, and the abundance of representative factor binding sites is shown (left). An AhR/Arnt binding motif is shown (right). b Mice (n = 5 mice per group) were fasted overnight (Fs) or fasted and refed for 6 h (Fd), and AhR occupancy within the SHP binding peak regions at the 1C genes was determined by liver ChIP assays. c Hepa1c1c7 cells were transfected with a Pemt-luc or Gnmt-luc construct containing the WT or mutated AhR binding site in the Pemt promoter along with expression plasmids as indicated and treated with FGF19 for 2 h, and luciferase activity was measured (n = 5). d–g AhR transgenic mouse studies: Liver tissues were collected from WT littermate and CA-AhR transgenic mice (n = 4–7 mice per group), or C57BL/6 (WT control) and AhR-KO mice (n = 5–10 mice per group) and d, e mRNA levels of selected 1C genes and f, g 1C cycle metabolites were measured by qRT-PCR and LC–MS analysis, respectively. Means ± SD are shown, and statistical significance was measured using d–g Student’s t-test or b, c two-way ANOVA with the FDR post-test. *P < 0.05, **P < 0.01, NS, not statistically significant

Fig. 4

SHP inhibits AhR transactivation of Pemt in response to feeding or FGF19 treatment. a, b Hepa1c1c7 cells were grown in low-glucose and serum-free media for 12 h, and transferred for 15 min to complete medium or treated for 15 min with insulin, FGF19, or CDCA, and then cells were harvested. Cells were pre-treated with a PKB inhibitor, PKB124005, or an ERK inhibitor, PD98059, for 30 min prior to insulin treatment as indicated. Levels of AhR and SHP in the cytoplasmic and nuclear fractions were determined by IB. Consistent results were observed from two independent studies. Full size immunoblots are in Supplementary Figure 12a. c Hepa1c1c7 cells were treated with insulin for 15 min or 45 min, and pre-mRNA levels of Pemt and Cyp1a1 were measured (n = 6). d Mice were fasted for 12 h (Fs) or fasted and refed for 4 h (Fd) or fasted 4 h and treated with FGF19 for 2 h, liver nuclear extracts were prepared, and the interaction of SHP and AhR was examined by CoIP. e, f Mice were fasted overnight (Fs) or fasted and refed for 6 h (Fd). e Cellular localization of AhR and SHP in mouse liver sections was examined by IF. Scale bar: 100 µM. f Liver chromatin was immunoprecipitated with AhR antibody, then eluted, and re-precipitated with ARNT or SHP antibody to examine the occupancy of SHP at AhR-bound chromatin at the Pemt promoter (n = 5 mice). g Hepa1c1c7 cells were transfected with the indicated plasmids, and then treated with FGF19 for 2 h and luciferase activity was determined (n = 5). h, i Mice (n = 5 mice per group) were fasted for 12 h and then refed for the indicated times, and the occupancy of AhR, ARNT, and SHP at the Pemt promoter (h) or Pemt pre-mRNA levels (i) were determined by liver ChIP assay or by qRT-PCR, respectively. Means ± SD are shown, and statistical significance was measured using the c, g, i one- or f, h two-way ANOVA with the FDR post-test. *P < 0.05, **P < 0.01, NS, statistically not significant

Fig. 5

AhR induces its own expression and SHP inhibits the AhR function. a SHP binding peak at the AhR promoter from published ChIP data and potential binding sites for transcription factors (bottom) within SHP peak region (top) were identified using the JASPAR database. b Mice (n = 5 mice per group) were refed for 6 h after fasting overnight. Liver chromatin was immunoprecipitated with AhR antibody first and then eluted and re-precipitated with SHP antibody. c Hepa1c1c7 cells were transfected with indicated plasmids. After 2 days, cells were treated with FGF19 for 1 h, and luciferase activities were measured and normalized to β-galactosidase activity (n = 5). d Mice (n = 5 mice per group) were fasted overnight or fasted and refed for the indicated times, and AhR pre-mRNA levels were determined by qRT-PCR. e, f C57BL/6 control mice and SHP-KO mice (n = 3–5 mice per group) were fasted overnight (Fs) or fasted and refed for 6 h (Fd). Hepatic mRNA (e) and proteins (f) levels were determined. Full size immunoblots are in Supplementary Figure 12b. Means ± SD are shown, and statistical significance was measured using the c, d one- or b, e, f (right) two-way ANOVA with the FDR post-test. *P < 0.05, **P < 0.01, NS, statistically not significant

Fig. 6

FGF19-induced phosphorylation of SHP at T55 is required for functional interaction with AhR. a Experimental outline: SHP-KO mice (n = 5 mice per group) were injected via the tail vein with control Ad-GFP, Ad-SHP-WT or p-defective Ad-T55A-SHP, and 2 weeks later, mice were treated with FGF19 for 2 h, and livers were collected. b Protein levels of SHP-WT or T55A-SHP in liver extracts were detected by IB. c The interaction between SHP-WT or T55A-SHP with AhR was determined by CoIP using whole cell liver extracts. AhR and SHP protein levels in input samples are shown. Full size immunoblots are in Supplementary Figure 12c. d mRNA levels of Pemt, Gnmt, and AhR were determined by qRT-PCR. e Selected 1C cycle metabolites levels were determined by LC–MS and the ratios of SAH/SAM and PC/PE were calculated. Means ± SD (n = 5) are shown, and statistical significance was measured using the d, e one-way ANOVA with the FDR post-test. *P < 0.05, **P < 0.01, NS, not statistically significant

Fig. 7

Downregulation of Pemt or overexpression of SHP blunts AhR-mediated increases in PC levels and liver steatosis in mice fed HF/HF. a Experimental outline: Mice (n = 5 mice per group) were fed a HF/HF diet for 8 weeks and infected with the indicated adenovirus. For the Ad-CA-AhR injection, Ad-GFP was added so that the same number of viral particles were injected for each group. At 3 weeks after adenoviral injection, livers were collected for further analyses. b Samples from three mice per each group were randomly selected and analyzed by IB (left) and quantification of the bands is shown (right). Full size immunoblots are in Supplementary Figure 12d. c Representative images of livers from the experimental groups. d Oil Red-O staining of frozen liver sections. Scale bar: 100 µM. e Liver TG levels. f mRNA levels of Pemt, Gnmt, and Cyp1a1 as determined by qRT-PCR. g Ratios of SAH/SAM and PC/PE as calculated from levels of these metabolites determined by LC–MS (n = 5). Means ± SD are shown (n = 5 mice), and statistical significance was measured using the b- (right), e–g one-way ANOVA with the FDR post-test. *P < 0.05, **P < 0.01, NS, not statistically significant

Fig. 8

Elevated AHR, PEMT, and PC levels in livers of NAFLD steatosis and NASH patients. a mRNA levels of PEMT, GNMT, and AHR in liver samples of 15 normal, 15 simple steatosis, and 15 severe NASH-fibrosis patients were determined by qRT-PCR. b Protein levels of PEMT, AHR, phosphorylated T55-SHP, and SHP in liver extracts from 5 samples each randomly pooled from 3 individuals (total 15 individuals) were analyzed by IB analysis (top) and the band intensities were quantified (bottom). The same normal samples were analyzed as controls in the two separate gels. Full size immunoblots are in Supplementary Figure 12e. c Hepatic PC and PE levels were determined by LC–MS and the ratios of PC/PE calculated. Means ± SD are shown (n = 15), and statistical significance was measured using the b (bottom) Student’s t-test or a, c one-way ANOVA with the FDR post-test. *P < 0.05, **P < 0.01, NS, not statistically significant. d Model: Temporal transcriptional regulation of the 1C cycle genes by a postprandial AhR–SHP axis. Nuclear levels of AhR are increased early after feeding by insulin/PKB signaling and nuclear AhR mediates transcriptional induction of 1C cycle genes, including Pemt and AhR, which results in increased hepatic PC and decreased SAM levels. In the late fed state, SHP blocks this AhR action through FGF15/19 signal-induced phosphorylation at T55