The mechanisms underlying the pathogenesis of steatosis and insulin resistance in nonalcoholic fatty liver disease remain elusive. Increased phosphorylation of hepatic p38 has long been noticed in fatty liver; however, whether the activation of hepatic p38 is a cause or consequence of liver steatosis is unclear. Here, we demonstrate that hepatic p38 activation by MKK6 overexpression in the liver of mice induces severe liver steatosis, reduces fat mass, and elevates circulating fatty acid levels in a hepatic p38α- and FGF21-dependent manner. Mechanistically, through increasing FGF21 production from liver, hepatic p38 activation increases the influx of fatty acids from adipose tissue to liver, leading to hepatic ectopic lipid accumulation and insulin resistance. Although hepatic p38 activation exhibits favorable effects in peripheral tissues, it impairs the hepatic FGF21 action by facilitating the ubiquitination and degradation of FGF21 receptor cofactor β-Klotho. Consistently, we show that when p38 phosphorylation and FGF21 expression are increased, β-Klotho protein levels are decreased in the fatty liver of both mice and patients. In conclusion, our study reveals previously undescribed effects of hepatic p38 activation on systemic metabolism and provides new insights into the roles of hepatic p38α, FGF21, and β-Klotho in the pathogenesis of nonalcoholic fatty liver disease.
Introduction
Nonalcoholic fatty liver disease (NAFLD) is a common complication of obesity, which has been linked to pathological metabolic conditions in both obese and nonobese subjects (1,2). Liver is an endocrine organ that controls systemic lipid homeostasis through secreting hepatokines (3,4). Emerging evidence indicates that the secretion of hepatokines is dysregulated in NAFLD (3,5). However, whether hepatokine-mediated mechanisms are involved in the development of NAFLD remains elusive (3). Understanding of the crosstalk between liver and other metabolic tissues will provide novel insights into the pathogenesis of NAFLD.
The p38 mitogen-activated protein kinase (MAPK) family has four members (p38α, β, γ, and δ). The activation of p38 is mediated by upstream cascade components MAP2K and MAP3K (6,7). Increased phosphorylation of hepatic p38 has been observed either upon fasting or in genetic and diet-induced mouse models of NAFLD (8–11). Available evidence indicates that hepatic p38 may have a profound effect on NAFLD, depending on the disease stage, activation potency, and upstream signaling (6,12–16). However, the contribution of hepatic p38 activation to liver lipid accumulation in mouse models of NAFLD is not fully understood.
Fibroblast growth factor 21 (FGF21) has favorable effects on hepatic lipid metabolism and insulin sensitivity (17,18). FGF21 is considered a hepatokine, and liver is the major contributor to circulating levels of FGF21, while white adipose tissue (WAT) is the major target of FGF21, contributing greatly to the metabolic effects of FGF21 (19–23). Serum FGF21 levels are positively associated with the severity of NAFLD, baseline FGF21 is an independent predictor of NAFLD, and “FGF21 resistance” has been suggested in obesity, insulin resistance, type 2 diabetes, and NAFLD (19,24–26); however, the underlying mechanisms are poorly understood (22,27,28).
Here, we demonstrate that hepatic p38 activation by overexpression of MKK6, a major upstream MAP2K of p38, in the liver redistributes fat from WAT to liver. Mechanistically, hepatic p38 activation increases the release of FGF21 from liver, which in turn stimulates WAT lipolysis, thereby enhancing the influx of fatty acids (FAs) into liver. Meanwhile, although the increase in circulating FGF21 improves insulin action in peripheral tissues, hepatic p38 activation impairs local FGF21 action by downregulating FGF21 receptor cofactor β-Klotho (KLB), which further favors hepatic lipid accumulation.
Research Design and Methods
Animal Study and Human Specimens
Protocols were approved by the institutional animal care and use committee (SIBS-2019-YH-1) and ethics committee of Zhongshan Hospital (B2020-127R). FGF21 flox/flox mice were from The Jackson Laboratory (JAX 022361). Mice were rendered steatotic by feeding a high-fat diet (HFD) (D12492; Research Diets) for 12 weeks. To evaluate insulin responsiveness, mice were fasted for 16 h and injected with 0.75 units/kg i.p. insulin. Mice were injected with anisomycin (ANS) at a dose of 20 mg/kg/day for 7 days.
Measurement of RNA, Protein Levels, and Metabolic Parameters
Quantitative PCR, Western blot, and immunohistochemistry analyses were performed as described before (8,29,30). Antibodies and kits for measurement of triglyceride (TG), nonesterified fatty acid (NEFA), insulin, FGF21, and phospho-p38 (p-p38) levels are shown in Supplementary Table 1.
Adenoviruses, Plasmids, Oligonucleotide, and Reagents
Adenoviruses were generated and administered to mice via tail vein injection (5 × 108 to 1 × 109 plaque-forming units), as described previously (29). Flag-KLB and the coding regions of human p38α, MKK6, and TRIM25 were cloned into pcDNA3.1 plasmid. Mouse FGF21 promoter fragment (−3100/+3) was cloned into PGL3-Basic plasmid. siRNAs, primers, and other reagents are shown in Supplementary Table 2.
Cell Lines and Treatment
Cas9-expressing HEK293T cells were transfected with p38α-targeting single guide RNA to delete p38α. To examine protein stability, cells were harvested or incubated with cycloheximide (CHX) (100 μg/mL) 24 h after transfection for the indicated time points or treated with ANS (25 ng/mL) for 2 h. For coimmunoprecipitation (co-IP) analysis, cells were harvested or treated with ANS (25 ng/mL) 24 h after transfection or infection in the presence or absence of MG132, chloroquine, or bafilomycin A1 for 6 h. To identify putative E3 ubiquitin ligase, cells were harvested 24 h after Flag-KLB transfection and lysed, followed by IP with anti-FLAG-beads (A2220; Sigma) overnight at 4°C. Immune complexes were sent to Majorbio for liquid chromatography tandem mass spectrometry analysis. For luciferase assay, HEK293T cells were cotransfected with FGF21-promoter-Luc and pRL-TK as an internal control with or without the spliced form of XBP1 (XBP1s) or MKK6 expression plasmid.
Statistical Analysis
Results are expressed as mean ± SD. Student t test was performed to assess statistically significant differences. GraphPad Prism 8.0.2 software was used for the statistical analyses. P < 0.05 was considered significant.
Data and Resource Availability
All data generated or analyzed during this study are included in the published article and its online supplementary files.
Results
Hepatic p38 Activation Induces Severe Liver Steatosis in Mice
In agreement with previous observations, we found that excessive hepatic lipid accumulation induced by HFD feeding or in a genetic mouse model of liver steatosis was accompanied by a two- to threefold increase in hepatic p-p38 levels in mice (Supplementary Fig. 1A and B). Accordingly, the phosphorylation of p38 upstream kinases MKK3/6 and TAK1 was also increased (Supplementary Fig. 1C and D). To define the consequence of p38 phosphorylation, adenovirus-mediated overexpression of MKK6, an upstream MAP2K that plays a major role in p38 activation, was used. Adenoviral MKK6 (Ad-MKK6) infection increased mRNA expression of MKK6 in liver but not in inguinal WAT (iWAT), epididymal WAT (eWAT), or skeletal muscle of mice (Fig. 1A). A 1.5-fold increase in p-p38 levels was observed in the liver of mice after Ad-MKK6 infection (Fig. 1B). Interestingly, mice infected with Ad-MKK6 displayed severe liver steatosis, as evident from enlargement and whitish appearance of liver, increased liver mass, accumulation of hepatocellular lipid droplets, and elevated hepatic TG levels, which was accompanied by reduced body weight and normal food intake (Fig. 1C–F and Supplementary Fig. 1E–H).
Hepatic p38 activation by Ad-MKK6 results in severe liver steatosis. A: Relative mRNA levels of MKK6 in the liver, iWAT, eWAT, and gastrocnemius muscle of mice infected with Ad-MKK6 (n = 4). B: Western blot and densitometry analysis of p-p38 in the liver of mice infected with Ad-MKK6 (n = 3). C and D: Representative gross liver morphology, ratio of liver weight to body weight (LW/BW), and BW of mice infected with Ad-MKK6 (n = 5). E and F: Representative images of hematoxylin-eosin (HE) staining and oil-red O staining of liver sections and liver TG levels of mice infected with Ad-MKK6 (n = 5). G: Western blot analysis of p-p38 in the liver of p38α-LKO and control floxed mice infected with Ad-MKK6 as indicated. H: Representative gross liver morphology of p38α-LKO and control floxed mice infected with Ad-MKK6, as indicated. I: LW/BW and BW of p38α-LKO and control floxed mice infected with Ad-MKK6, as indicated (n = 5). J and K: Representative images of HE staining and oil-red O staining of liver sections and liver TG levels of p38α-LKO and control floxed mice infected with Ad-MKK6 as indicated (n = 5). Data are mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001. Ctrl, control; NS, not significant.
Hepatic p38 activation by Ad-MKK6 results in severe liver steatosis. A: Relative mRNA levels of MKK6 in the liver, iWAT, eWAT, and gastrocnemius muscle of mice infected with Ad-MKK6 (n = 4). B: Western blot and densitometry analysis of p-p38 in the liver of mice infected with Ad-MKK6 (n = 3). C and D: Representative gross liver morphology, ratio of liver weight to body weight (LW/BW), and BW of mice infected with Ad-MKK6 (n = 5). E and F: Representative images of hematoxylin-eosin (HE) staining and oil-red O staining of liver sections and liver TG levels of mice infected with Ad-MKK6 (n = 5). G: Western blot analysis of p-p38 in the liver of p38α-LKO and control floxed mice infected with Ad-MKK6 as indicated. H: Representative gross liver morphology of p38α-LKO and control floxed mice infected with Ad-MKK6, as indicated. I: LW/BW and BW of p38α-LKO and control floxed mice infected with Ad-MKK6, as indicated (n = 5). J and K: Representative images of HE staining and oil-red O staining of liver sections and liver TG levels of p38α-LKO and control floxed mice infected with Ad-MKK6 as indicated (n = 5). Data are mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001. Ctrl, control; NS, not significant.
Because Ad-MKK6 infection increased the phosphorylation of both p38α and p38β (Supplementary Fig. 1I), to test whether hepatic p38α, the predominant isoform in liver (Supplementary Fig. 1J), was responsible for the liver steatosis after Ad-MKK6 infection, we used p38α liver-specific knockout (p38α-LKO) mice (8,31). The p-p38 levels were very low in the liver of p38α-LKO mice, even after Ad-MKK6 infection (Fig. 1G and Supplementary Fig. 1K). Ad-MKK6 infection neither changed the appearance and weight of liver nor induced hepatic steatosis in p38α-LKO mice (Fig. 1H–K and Supplementary Fig. 1L–N), suggesting that hepatic p38α is primarily responsible for the occurrence of liver steatosis induced by Ad-MKK6.
Hepatic p38 Activation Affects Hepatic Lipid Metabolism and Insulin Signaling
We then analyzed the expression of key genes involved in FA uptake, TG synthesis, FA oxidation, and TG export. CD36 expression was elevated in the liver of mice infected with Ad-MKK6, indicating an increase in FA uptake (Fig. 2A and B). The expression of acyl-CoA synthetase long chain family member 1 (ACSL1), glycerol-3-phosphate acyltransferase (GPAT), and diacylglycerol acyltransferase (DGAT) 1 and 2 were increased, indicating an increased esterification of FAs to TG (Fig. 2C). Meanwhile, the mRNA expression of peroxisome proliferator–activated receptor-α (PPARα), medium-chain acyl-CoA dehydrogenase (MCAD), and carnitine palmitoyltransferase 1α (CPT1α) was increased, indicating an increased FA oxidation (Fig. 2D). The expression levels of microsomal TG transfer protein (MTTP) and apolipoprotein B (apoB) were increased, indicating an increase in TG export (Fig. 2E and F). Surprisingly, Ad-MKK6 infection did not alter the expression of all above-mentioned genes in primary hepatocytes (Supplementary Fig. 2A–D), indicating that the regulation by hepatic p38 activation involves interorgan communication.
Hepatic p38 activation by Ad-MKK6 affects hepatic lipid metabolism and insulin signaling. A and B: Relative mRNA levels (n = 5) and Western blot and densitometry analysis (n = 4–5) of CD36 in the liver of mice infected with Ad-MKK6. C: Relative mRNA levels of ACSL1, GPAT, DGAT1, and DGAT2 in the liver of mice infected with Ad-MKK6, as indicated (n = 5). D: Relative mRNA levels of PPARα, MCAD, and CPT1α in the liver of mice infected with Ad-MKK6 as indicated (n = 5). E and F: Relative mRNA levels of MTTP and apoB (n = 5) and Western blot and densitometry analysis of MTTP, apoB-100, and apoB-48 (n = 4) in the liver of mice infected with Ad-MKK6 as indicated. G: Western blot analysis of phosphorylated insulin receptor (p-IR) and p-Akt in the liver of mice infected with Ad-MKK6 after insulin administration. H and I: Western blot analysis of p-IR and p-Akt in the eWAT, iWAT, and gastrocnemius muscle of mice infected with Ad-MKK6 after insulin administration, as indicated. Data are mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001. Ctrl, control.
Hepatic p38 activation by Ad-MKK6 affects hepatic lipid metabolism and insulin signaling. A and B: Relative mRNA levels (n = 5) and Western blot and densitometry analysis (n = 4–5) of CD36 in the liver of mice infected with Ad-MKK6. C: Relative mRNA levels of ACSL1, GPAT, DGAT1, and DGAT2 in the liver of mice infected with Ad-MKK6, as indicated (n = 5). D: Relative mRNA levels of PPARα, MCAD, and CPT1α in the liver of mice infected with Ad-MKK6 as indicated (n = 5). E and F: Relative mRNA levels of MTTP and apoB (n = 5) and Western blot and densitometry analysis of MTTP, apoB-100, and apoB-48 (n = 4) in the liver of mice infected with Ad-MKK6 as indicated. G: Western blot analysis of phosphorylated insulin receptor (p-IR) and p-Akt in the liver of mice infected with Ad-MKK6 after insulin administration. H and I: Western blot analysis of p-IR and p-Akt in the eWAT, iWAT, and gastrocnemius muscle of mice infected with Ad-MKK6 after insulin administration, as indicated. Data are mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001. Ctrl, control.
Additionally, the insulin-induced phosphorylation of Akt was decreased in the liver after Ad-MKK6 infection, indicating an impaired hepatic insulin action (Fig. 2G). However, Ad-MKK6 infection could not affect the phosphorylation of Akt in vitro (Supplementary Fig. 2E), suggesting that interorgan communication is also involved here. Surprisingly, insulin levels were decreased while systemic insulin sensitivity was not impaired in Ad-MKK6–infected mice (Supplementary Fig. 2F and G), suggesting that the attenuated hepatic insulin signaling might be compensated by increases in insulin sensitivity in other insulin-responsive peripheral tissues. Indeed, we found that insulin-induced phosphorylation of Akt was increased in WAT and skeletal muscle after Ad-MKK6 infection (Fig. 2H and I), suggesting that insulin sensitivity was improved in these peripheral tissues. Notably, the insulin levels in p38α-LKO mice were not altered after Ad-MKK6 infection, further suggesting a predominant role of hepatic p38α (Supplementary Fig. 2H).
Hepatic p38 Activation Increases Delivery of FAs From WAT to Liver
Normally, clarifying the sources of FAs is helpful to understand the development of liver steatosis. We found that serum free FA (FFA) levels were significantly elevated after Ad-MKK6 infection (Fig. 3A). Together with the finding of increased expression of hepatic CD36 after Ad-MKK6 infection (Fig. 2A and B), these results suggest that FA supply to the liver was increased. The mRNA levels of adipose TG lipase (ATGL) and hormone-sensitive lipase (HSL) were increased in WAT after Ad-MKK6 infection, indicating increased lipolysis (Fig. 3B and C). Accordingly, WAT weights were reduced in Ad-MKK6–infected mice (Fig. 3D and E and Supplementary Fig. 3A). These data suggest that hepatic p38 activation might enhance WAT lipolysis, thereby increasing the delivery of FAs from WAT to liver, which would promote ectopic hepatic lipid accumulation via a substrate push mechanism. In contrast, increased serum FFA levels, reduced WAT mass, and elevated mRNA levels of HSL and ATGL in WAT could not be observed in p38α-LKO mice after Ad-MKK6 infection (Fig. 3F–H and Supplementary Fig. 3B–D), suggesting that hepatic p38α is required for increased fat mobilization induced by Ad-MKK6 infection.
Hepatic p38 activation by Ad-MKK6 enhances lipolysis in WAT. A: Serum NEFA levels in mice infected with Ad-MKK6 (n = 5). B and C: Relative mRNA levels of ATGL and HSL in the eWAT and iWAT of mice infected with Ad-MKK6, as indicated (n = 4). D and E: Representative gross morphology of the liver, eWAT, and iWAT and ratio of eWAT and iWAT weight to body weight (FW/BW) of mice infected with Ad-MKK6, as indicated (n = 4). F: Serum NEFA levels in p38α-LKO and control floxed mice infected with Ad-MKK6 as indicated (n = 5). G and H: Representative gross morphology of the liver, eWAT, and iWAT and FW/BW of p38α-LKO and control floxed mice infected with Ad-MKK6, as indicated (n = 5). Data are mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001. Ctrl, control; NS, not significant.
Hepatic p38 activation by Ad-MKK6 enhances lipolysis in WAT. A: Serum NEFA levels in mice infected with Ad-MKK6 (n = 5). B and C: Relative mRNA levels of ATGL and HSL in the eWAT and iWAT of mice infected with Ad-MKK6, as indicated (n = 4). D and E: Representative gross morphology of the liver, eWAT, and iWAT and ratio of eWAT and iWAT weight to body weight (FW/BW) of mice infected with Ad-MKK6, as indicated (n = 4). F: Serum NEFA levels in p38α-LKO and control floxed mice infected with Ad-MKK6 as indicated (n = 5). G and H: Representative gross morphology of the liver, eWAT, and iWAT and FW/BW of p38α-LKO and control floxed mice infected with Ad-MKK6, as indicated (n = 5). Data are mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001. Ctrl, control; NS, not significant.
Liver-Derived FGF21 Mediates the Metabolic Effects of Hepatic p38 Activation
As loss of hepatic p38α could abolish the effect of hepatic MKK6 overexpression, we then used Ad-MKK6–infected mice as a model system for investigating the effect of hepatic p38 activation. Because an increase in systemic FGF21 levels can promote WAT lipolysis and reduce WAT mass (19,21,32) and increased FGF21 signaling was observed in iWAT after Ad-MKK6 infection, as evident by increased phosphorylation of FRS2α (p-FRS2α) and elevated mRNA levels of egr-1 and c-fos, two FGF21 target genes (Supplementary Fig. 4A and B), we tested whether FGF21 was involved in the regulation by hepatic p38 activation. We found that FGF21 transcription and secretion were increased in the liver and primary hepatocytes after Ad-MKK6 infection (Fig. 4A and B), suggesting that hepatic p38 activation promotes FGF21 transcription and secretion in a cell-autonomous manner.
Liver-derived FGF21 mediates the metabolic effects of hepatic p38 activation. A: Relative FGF21 mRNA levels in the liver (n = 5) and serum FGF21 levels (n = 4) of mice infected with Ad-MKK6. B: Relative FGF21 mRNA levels in primary hepatocytes (PHs) and FGF21 protein levels in culture medium of PHs after Ad-MKK6 infection (n = 4). C: Relative XBP1s/total XBP1 (XBP1t) mRNA ratios and FGF21 mRNA levels in the liver of p38α-LKO and control floxed mice infected with Ad-MKK6 (n = 5). D: Relative FGF21 mRNA levels in the liver and serum FGF21 levels of FGF21-LKD and control floxed mice infected with Ad-MKK6, as indicated (n = 5). E and F: Representative gross liver morphology and ratio of liver weight to body weight (LW/BW) of FGF21-LKD and control floxed mice infected with Ad-MKK6, as indicated (n = 5). G–I: Representative images of hematoxylin-eosin (HE) staining and oil-red O staining of liver sections, liver TG levels (n = 5), and serum NEFA levels (n = 4) of FGF21-LKD and control floxed mice infected with Ad-MKK6 (n = 5), as indicated. J and K: Representative gross morphology of the eWAT and iWAT and ratio of eWAT and iWAT weight (FW) to BW of FGF21-LKD and control floxed mice infected with Ad-MKK6 (n = 5) as indicated. Data are mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001. Ctrl, control; NS, not significant.
Liver-derived FGF21 mediates the metabolic effects of hepatic p38 activation. A: Relative FGF21 mRNA levels in the liver (n = 5) and serum FGF21 levels (n = 4) of mice infected with Ad-MKK6. B: Relative FGF21 mRNA levels in primary hepatocytes (PHs) and FGF21 protein levels in culture medium of PHs after Ad-MKK6 infection (n = 4). C: Relative XBP1s/total XBP1 (XBP1t) mRNA ratios and FGF21 mRNA levels in the liver of p38α-LKO and control floxed mice infected with Ad-MKK6 (n = 5). D: Relative FGF21 mRNA levels in the liver and serum FGF21 levels of FGF21-LKD and control floxed mice infected with Ad-MKK6, as indicated (n = 5). E and F: Representative gross liver morphology and ratio of liver weight to body weight (LW/BW) of FGF21-LKD and control floxed mice infected with Ad-MKK6, as indicated (n = 5). G–I: Representative images of hematoxylin-eosin (HE) staining and oil-red O staining of liver sections, liver TG levels (n = 5), and serum NEFA levels (n = 4) of FGF21-LKD and control floxed mice infected with Ad-MKK6 (n = 5), as indicated. J and K: Representative gross morphology of the eWAT and iWAT and ratio of eWAT and iWAT weight (FW) to BW of FGF21-LKD and control floxed mice infected with Ad-MKK6 (n = 5) as indicated. Data are mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001. Ctrl, control; NS, not significant.
Since p38 regulates mRNA stability of XBP1s, a transcriptional regulator of FGF21 (33,34), we hypothesized that XBP1 might be involved in the regulation of FGF21 by p38 activation. Consistent with previously reported in vitro data (34) and findings in obese mice (33), Ad-XBP1s infection increased hepatic FGF21 mRNA levels, while Ad-MKK6 infection elevated hepatic XBP1s mRNA levels both in vivo and in vitro (Supplementary Fig. 4C–F). Accordingly, overexpression of either XBP1s or MKK6 increases luciferase activity of the reporter containing a mouse FGF21 promoter (Supplementary Fig. 4G). Again, the effect of Ad-MKK6 infection on the expression of XBP1s and FGF21 was abolished in p38α-LKO mice, suggesting that hepatic p38α is the major contributor (Fig. 4C).
To clarify the role of liver-derived FGF21 in the development of liver steatosis, we used liver-specific FGF21 knockdown (FGF21-LKD) mice by infecting FGF21 flox/flox (floxed) mice with Cre-recombinase adenovirus. As expected, Ad-MMK6 infection could not increase FGF21 transcription and secretion in FGF21-LKD mice (Fig. 4D). Interestingly, the effects of Ad-MKK6 infection on the appearance and weight of liver, hepatic lipid content, serum insulin and FFA levels, WAT mass, and body weight were all abolished in FGF21-LKD mice, suggesting that liver-derived FGF21 is required for the metabolic effects of hepatic p38 activation observed in this study (Fig. 4E–K and Supplementary Fig. 4H–L).
Activation of Hepatic p38 Attenuates Local FGF21 Action in Liver
To explore the role of hepatic FGF21 in the liver with p38 activation, we infected mice with Ad-FGF21. Ad-FGF21 infection increased hepatic FGF21 mRNA levels, serum FGF21 and FFA levels, and HSL and ATGL mRNA levels in WAT; decreased WAT weights; and increased FGF21 signaling in iWAT (Fig. 5A and B and Supplementary Fig. 5A–F), which are very similar to those observed after Ad-MKK6 infection (Fig. 3A–E and Supplementary Fig. 3A). Furthermore, Ad-FGF21 infection resulted in decreased serum insulin levels and increased Akt phosphorylation in WAT and skeletal muscle (Fig. 5C and D and Supplementary Fig. 5G), which are also similar to that observed after Ad-MKK6 infection (Fig. 2H and I and Supplementary Fig. 2F). These results indicate that increased hepatic FGF21 expression might contribute to the alteration of serum FFA and insulin levels, WAT lipolysis, WAT mass, and insulin responsiveness in peripheral tissues after hepatic p38 activation.
Activation of hepatic p38 attenuates the local FGF21 action in liver. A–D: Serum NEFA levels, ratio of eWAT and iWAT weight to body weight (FW/BW), serum insulin levels (n = 5), and insulin-induced phosphorylation of insulin receptor (p-IR) and Akt in the eWAT and iWAT of mice infected with Ad-FGF21. E–H: LW/BW, BW (n = 5), hematoxylin-eosin (HE) and oil-red O staining of liver sections, liver TG levels (n = 5), and insulin-induced phosphorylation of IR and Akt in the liver of mice infected with Ad-FGF21. I–K: Relative FGF21 mRNA levels, p-FRS2α protein levels, and relative egr-1 and c-fos mRNA levels in the liver of mice infected with Ad-MKK6 under fasted state (n = 4). L and M: Relative KLB mRNA levels (n = 5) and Western blot and densitometry analysis of KLB (n = 3) in the liver of mice infected with Ad-MKK6. N and O: Relative KLB mRNA levels (n = 4) and KLB protein levels in the liver of mice infected with Ad-FGF21 and Ad-shKLB. P–S: LW/BW, liver TG levels, relative hepatic FGF21 mRNA levels, and serum NEFA levels of mice infected with Ad-FGF21 and Ad-shKLB (n = 4). Data are mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001. Ctrl, control; NS, not significant.
Activation of hepatic p38 attenuates the local FGF21 action in liver. A–D: Serum NEFA levels, ratio of eWAT and iWAT weight to body weight (FW/BW), serum insulin levels (n = 5), and insulin-induced phosphorylation of insulin receptor (p-IR) and Akt in the eWAT and iWAT of mice infected with Ad-FGF21. E–H: LW/BW, BW (n = 5), hematoxylin-eosin (HE) and oil-red O staining of liver sections, liver TG levels (n = 5), and insulin-induced phosphorylation of IR and Akt in the liver of mice infected with Ad-FGF21. I–K: Relative FGF21 mRNA levels, p-FRS2α protein levels, and relative egr-1 and c-fos mRNA levels in the liver of mice infected with Ad-MKK6 under fasted state (n = 4). L and M: Relative KLB mRNA levels (n = 5) and Western blot and densitometry analysis of KLB (n = 3) in the liver of mice infected with Ad-MKK6. N and O: Relative KLB mRNA levels (n = 4) and KLB protein levels in the liver of mice infected with Ad-FGF21 and Ad-shKLB. P–S: LW/BW, liver TG levels, relative hepatic FGF21 mRNA levels, and serum NEFA levels of mice infected with Ad-FGF21 and Ad-shKLB (n = 4). Data are mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001. Ctrl, control; NS, not significant.
Notably, liver weight and hepatic lipid content were reduced after Ad-FGF21 infection (Fig. 5E–G and Supplementary Fig. 5H) rather than increased as seen after Ad-MKK6 infection (Fig. 1C–F). Moreover, the insulin-induced phosphorylation of Akt was increased after Ad-FGF21 infection (Fig. 5H) rather than suppressed as observed in Ad-MKK6–infected mice (Fig. 2G). Given that Ad-MKK6–infected liver did not respond to the increased availability of FGF21 as normal liver did, we speculated that hepatic FGF21 action was impaired. As expected, although FGF21 mRNA levels were higher under the fasted state, FGF21 signaling was decreased in Ad-MKK6–infected liver (Fig. 5I–K). Consistently, the repressive effect of Ad-MKK6 infection on egr-1 and c-fos expression could not be observed in the liver of p38α-LKO mice (Supplementary Fig. 5I).
Because FGF21 action depends on KLB, while loss of KLB in hepatocytes has been linked to hepatic lipid accumulation (35), we then tested whether KLB expression was altered after hepatic p38 activation. Interestingly, KLB mRNA levels were not changed, while KLB protein levels were decreased after Ad-MKK6 infection (Fig. 5L and M), suggesting that the regulation of KLB expression by p38 activation might occur at a posttranscriptional level. Additionally, the effect of Ad-MKK6 infection on KLB protein levels was abolished in the liver of p38α-LKO mice (Supplementary Fig. 5J), suggesting that hepatic p38α is the primary mediator.
To test whether the downregulation of hepatic KLB could have any effect on local FGF21 action, we repressed KLB expression by Ad-shKLB in Ad-FGF21–infected liver (Fig. 5N and O and Supplementary Fig. 5K). Interestingly, knockdown of KLB expression increased liver weight and TG levels without changing hepatic FGF21 expression and serum NEFA levels in Ad-FGF21–infected mice infected (Fig. 5P–S). These data suggest that a defect in hepatic FGF21 action as a result of KLB downregulation could promote lipid accumulation, contributing to the formation of liver steatosis after hepatic p38 activation.
Notably, liver weight, hepatic TG content, and iWAT weight were not significantly altered in p38α-LKO mice on normal chow diet (Supplementary Fig. 5L). Consistent with a previous report (15), a similar body weight was observed between floxed and p38α-LKO mice after HFD feeding; however, the differences in liver weight, hepatic TG content, and iWAT weight were not significant (Supplementary Fig. 5M). Decreases in hepatic FGF21 mRNA expression and serum FGF21 levels were observed in HFD-fed p38α-LKO mice, but the difference either was small or did not reach statistical significance (Supplementary Fig. 5L and M). The protein levels of hepatic KLB were increased in HFD-fed but not chow diet–fed p38α-LKO mice (Supplementary Fig. 5N and O). Also worth noting is that hepatic p38 activation induced by short-term ANS treatment could result in a phenotype similar to that observed in mice infected with Ad-MKK6 (Supplementary Fig. 5P–X), which further substantiates our notion that hepatic p38 activation could result in fatty liver by affecting hepatic expression of FGF21 and KLB. These findings suggest that although hepatic p38 activation is sufficient to induce fatty liver, hepatic p38α is dispensable for maintaining normal lipid homeostasis under a normal chow diet condition and hepatic p38α deficiency cannot prevent mice from HFD-induced liver steatosis. On the basis of our results, we speculate that HFD feeding has multiple mechanisms of action that induce liver steatosis.
p38 Activation Facilitates the Ubiquitination and Degradation of KLB
In agreement with the in vivo results, p38 activation by MKK6 overexpression decreased the Flag-KLB protein levels (Fig. 6A and Supplementary Fig. 6A), while inhibition of p38 by a dominant-negative form of p38α (p38αAF) attenuated the effect of MKK6 overexpression in cultured cells (Fig. 6B). Accordingly, ANS treatment reduced Flag-KLB protein levels in a time- and dose-dependent manner (Fig. 6C and D and Supplementary Fig. 6B). Furthermore, in the presence of CHX, the Flag-KLB protein was less stable in cells overexpressing MKK6 or treated with ANS but more stable in cells infected with Ad-p38αAF, suggesting that the degradation process for KLB protein was accelerated after p38 activation (Fig. 6E and Supplementary Fig. 6C). KLB mRNA stability was not altered by ANS treatment in the presence of actinomycin, further suggesting that the regulation did not occur at a transcriptional level (Supplementary Fig. 6D). Notably, treatment of MG132 increased KLB protein levels, indicating that KLB could be degraded by the proteasome pathway (Fig. 6F and Supplementary Fig. 6E).
Activation of p38 facilitates the degradation of KLB. A: Western blot and densitometry analysis of KLB and p-p38 in HEK293T cells transfected with Flag-KLB and infected with Ad-MKK6. B: Western blot analysis of KLB and p-p38 in primary hepatocytes transfected with Flag-KLB and infected with Ad-MKK6 and Ad-p38αAF as indicated. C and D: Western blot analysis of Flag-KLB in Flag-KLB–transfected HEK293T cells treated with ANS with indicated doses or for indicated time periods. E: Western blot and linear regression analysis of the rate of decay of Flag-KLB in Flag-KLB–transfected HEK293T cells infected with Ad-MKK6 or Ad-p38αAF for the indicated time periods in the presence of CHX. F: Western blot analysis of KLB in the HepG2 cells transfected with Flag-KLB for 12 h and then treated with MG132 for 12 h. Data are mean ± SD. *P < 0.05. Ctrl, control.
Activation of p38 facilitates the degradation of KLB. A: Western blot and densitometry analysis of KLB and p-p38 in HEK293T cells transfected with Flag-KLB and infected with Ad-MKK6. B: Western blot analysis of KLB and p-p38 in primary hepatocytes transfected with Flag-KLB and infected with Ad-MKK6 and Ad-p38αAF as indicated. C and D: Western blot analysis of Flag-KLB in Flag-KLB–transfected HEK293T cells treated with ANS with indicated doses or for indicated time periods. E: Western blot and linear regression analysis of the rate of decay of Flag-KLB in Flag-KLB–transfected HEK293T cells infected with Ad-MKK6 or Ad-p38αAF for the indicated time periods in the presence of CHX. F: Western blot analysis of KLB in the HepG2 cells transfected with Flag-KLB for 12 h and then treated with MG132 for 12 h. Data are mean ± SD. *P < 0.05. Ctrl, control.
To investigate the effect of p38 activation on KLB protein stability, we used HEK293T cells without p38α (Supplementary Fig. 7A). Intriguingly, ANS treatment increased p-p38 levels and decreased Flag-KLB protein levels only in wild-type (WT) cells but not in p38α knockout (KO) cells (Fig. 7A). Because ubiquitination of Flag-KLB was decreased in KO cells compared with WT cells (Fig. 7B and Supplementary Fig. 7B), we speculated that p38α is involved in the regulation of KLB ubiquitination.
Activation of p38 promotes ubiquitination (Ub) of KLB through TRIM25. A: Western blot analysis of KLB and p-p38 in the p38α WT and KO HEK293T cells transfected with Flag-KLB and treated with ANS. B: IP and immunoblot (IB) analysis with indicated antibodies after p38α WT and KO cells were transfected with plasmids, as indicated. C–E: IP and IB analysis with indicated antibodies after HEK293T cells were transfected with plasmids and infected with Ad-p38αAF or Ad-MKK6 or treated with ANS, as indicated. F: IP and IB analysis with indicated antibodies after p38α WT and KO cells were transfected with plasmids and treated with ANS, as indicated. G: Western blot analysis of p-p38 and KLB in Flag-KLB–transfected cells treated with ANS after administration of siRNA, as indicated. A pool of three siRNAs targeting different regions of the corresponding gene was used. H and I: Co-IP analysis of the interaction of Myc-TRIM25 and endogenous TRIM25 with Flag-tagged KLB in HEK293T cells, as indicated. J and K: IP and IB analysis with indicated antibodies after HEK293T cells were cotransfected with Flag-KLB and indicated siRNA and then treated with Ad-MKK6 and ANS. A pool of three siRNAs targeting different regions of TRIM25 was used. L and M: Co-IP analysis of the interaction of endogenous TRIM25 and Myc-TRIM25 with Flag-KLB after HEK293T cells were infected with Ad-MKK6.
Activation of p38 promotes ubiquitination (Ub) of KLB through TRIM25. A: Western blot analysis of KLB and p-p38 in the p38α WT and KO HEK293T cells transfected with Flag-KLB and treated with ANS. B: IP and immunoblot (IB) analysis with indicated antibodies after p38α WT and KO cells were transfected with plasmids, as indicated. C–E: IP and IB analysis with indicated antibodies after HEK293T cells were transfected with plasmids and infected with Ad-p38αAF or Ad-MKK6 or treated with ANS, as indicated. F: IP and IB analysis with indicated antibodies after p38α WT and KO cells were transfected with plasmids and treated with ANS, as indicated. G: Western blot analysis of p-p38 and KLB in Flag-KLB–transfected cells treated with ANS after administration of siRNA, as indicated. A pool of three siRNAs targeting different regions of the corresponding gene was used. H and I: Co-IP analysis of the interaction of Myc-TRIM25 and endogenous TRIM25 with Flag-tagged KLB in HEK293T cells, as indicated. J and K: IP and IB analysis with indicated antibodies after HEK293T cells were cotransfected with Flag-KLB and indicated siRNA and then treated with Ad-MKK6 and ANS. A pool of three siRNAs targeting different regions of TRIM25 was used. L and M: Co-IP analysis of the interaction of endogenous TRIM25 and Myc-TRIM25 with Flag-KLB after HEK293T cells were infected with Ad-MKK6.
Accordingly, Flag-KLB ubiquitination was reduced by Ad-p38αAF infection, while either Ad-MKK6 or ANS administration increased the ubiquitination of Flag-KLB (Fig. 7C–E and Supplementary Fig. 7C–E). Importantly, ubiquitination of Flag-KLB was abolished in KO cells with ANS treatment (Fig. 7F and Supplementary Fig. 7F), further suggesting that p38α is required for the increased ubiquitination of KLB by p38 activation.
The effect of either chloroquine or bafilomycin A1 treatment on the ubiquitination of Flag-KLB was less potent than MG132 treatment, indicating that only a small proportion of ubiquitinated KLB would undergo lysosome-dependent degradation (Supplementary Fig. 7G). Because ANS treatment increased both K48-linked (Fig. 7C–E and Supplementary Fig. 7C–E) and K63-linked (Supplementary Fig. 7H) ubiquitination of Flag-KLB, we speculated that ubiquitinated KLB might be degraded by different pathways.
To identify potential E3 ubiquitin ligases, Flag-KLB binding proteins were purified by co-IP and then subjected to liquid chromatography tandem mass spectrometry analysis. Among the potential interacting proteins, four are E3 ligases, including ZNF598, TRIM25, MKRN2, and CHIP. Because the effect of ANS treatment on KLB protein levels was greatly attenuated by the transfection of the specific siRNA for TRIM25 but not the specific siRNA for ZNF598, MKRN2, or CHIP, we hypothesized that TRIM25 might be the primary E3 ubiquitin ligase responsible for the KLB ubiquitination and degradation after p38 activation (Fig. 7G and Supplementary Fig. 7I). Consistently, the physical interaction of either exogenous or endogenous TRIM25 with Flag-KLB could be detected by co-IP analysis (Fig. 7H and I). Moreover, knockdown of TRIM25 attenuated the effect of p38 activation by either Ad-MKK6 infection or ANS treatment on the ubiquitination and stability of KLB (Fig. 7J and K and Supplementary Fig. 7J and K). Notably, Ad-MKK6 treatment enhanced the interaction of endogenous and exogenous TRIM25 with Flag-KLB (Fig. 7L and M). Because mRNA and protein levels of TRIM25 were both increased in the liver of mice and HepG2 cells infected with Ad-MKK6, we speculated that increased TRIM25 expression after p38 activation could partially contribute to the enhanced interaction between TRIM25 and KLB, thereby facilitating ubiquitination and degradation of KLB (Supplementary Fig. 7L and M). Accordingly, increased TRIM25 expression was observed in the liver of either HFD-fed mice or ob/ob mice (Supplementary Fig. 7N–P).
KLB Protein Levels Are Downregulated in the Fatty Liver of Mice and Patients
To assess the physiopathological relevance of our findings, we examined the expression of KLB in mouse models of and human subjects with NAFLD. Interestingly, we found that the increase in phosphorylation of hepatic p38 was accompanied by a decrease in protein levels but not mRNA levels of KLB and an increase in FGF21 mRNA levels in the liver of HFD-fed or ob/ob mice (Fig. 8A–C, Supplementary Fig. 1B and D, and Supplementary Fig. 8A). More interestingly, increased phosphorylation of hepatic p38 and downregulation of hepatic KLB protein levels but not mRNA levels were observed in patients with NAFLD with elevated hepatic FGF21 mRNA levels (Fig. 8D–H). Furthermore, hepatic FGF21 action was impaired in HFD-fed and ob/ob mice (Supplementary Fig. 8B and C). In line with the results in mouse models of NAFLD, increased phosphorylation of MKK3/6, reduced levels of p-FRS2α, decreased mRNA levels of egr-1 and c-fos, and increased TRIM25 levels were observed in patients with NAFLD (Supplementary Table 3 and Supplementary Fig. 8D–F).
KLB protein levels are downregulated in the fatty liver of mice and patients. A and B: Western blot and densitometry analysis of KLB in the liver of HFD-fed and ob/ob mice, as indicated. C: Relative FGF21 mRNA levels in the liver of HFD-fed and ob/ob mice (n = 5). D and E: Relative mRNA levels of FGF21 and KLB in the normal and fatty liver tissues from human subjects (n = 5). F: Western blots and densitometry analysis of p-p38 and KLB in human subjects with normal or fatty liver (n = 5), as indicated. G and H: Detection of p-p38 levels in the liver lysate by ELISA, hematoxylin-eosin (HE) staining and immunohistochemistry of KLB (original magnification 20×, inset 60×), and quantification of immunohistochemical staining of KLB in human subjects with normal or fatty liver (n = 5). I: Schematic of the effect of hepatic p38 activation on systemic metabolism. Data are mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001. Ctrl, control; NS, not significant.
KLB protein levels are downregulated in the fatty liver of mice and patients. A and B: Western blot and densitometry analysis of KLB in the liver of HFD-fed and ob/ob mice, as indicated. C: Relative FGF21 mRNA levels in the liver of HFD-fed and ob/ob mice (n = 5). D and E: Relative mRNA levels of FGF21 and KLB in the normal and fatty liver tissues from human subjects (n = 5). F: Western blots and densitometry analysis of p-p38 and KLB in human subjects with normal or fatty liver (n = 5), as indicated. G and H: Detection of p-p38 levels in the liver lysate by ELISA, hematoxylin-eosin (HE) staining and immunohistochemistry of KLB (original magnification 20×, inset 60×), and quantification of immunohistochemical staining of KLB in human subjects with normal or fatty liver (n = 5). I: Schematic of the effect of hepatic p38 activation on systemic metabolism. Data are mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001. Ctrl, control; NS, not significant.
Together, we demonstrate that hepatic p38 activation can modulate whole-body homeostasis through FGF21-mediated interorgan communication and could be a casual driver of liver steatosis.
Discussion
Because p38 is activated in fatty liver, it has been implicated in the pathogenesis of liver steatosis (8,9,11). Either activation of ASK1, a MAP3K, or KO of MKP5, a MAPK phosphatase, promotes HFD-induced liver steatosis, which is accompanied by increased hepatic p38 activation (12,13). In contrast, loss of MKP1, another MAPK phosphatase, which results in hyperactivation of both hepatic p38 and c-Jun N-terminal kinase, prevents lipid accumulation in the liver (14). Deletion of hepatic p38α attenuates liver injury and fibrosis in an HFD and a chemokine-induced NASH model with a strong hepatic p38 activation and favors hepatic lipid accumulation upon HFD challenge (15,16). Although evidence suggests that p38 could be crucial in HFD-induced hepatic steatosis, whether hepatic p38 activation is a cause or sequela of liver steatosis is still unclear. Here, we uncovered previously undescribed roles that sustained hepatic p38 activation is a causal driver of liver steatosis and hepatic insulin resistance.
With recent progress in understanding how tissues communicate by secreting factors, we have an expanded appreciation for the role of interorgan crosstalk in the pathogenesis of metabolic diseases. As a hepatokine dysregulated in fatty liver, FGF21 has favorable effects (19,24,26). Here, we show that FGF21-mediated interorgan crosstalk is required for liver steatosis formation induced by hepatic p38 activation. Mechanistically, hepatic p38 activation promotes fat mobilization from WAT to liver through increasing hepatic FGF21 expression and secretion, thereby driving hepatic TG accumulation through a substrate push mechanism. Thus, our study provides new insight into the role of liver-derived FGF21 in the pathogenesis of NAFLD (3,5).
As the role of FGF21 was usually explored in the context of diet-induced obesity, the physiological action of FGF21 is not fully understood (22,27,28). Moreover, the diversity of upstream signals and the tissue-specific induction and action have made generating a unifying physiological model for FGF21 function complicated (36). Although FGF21 has potent lowering effects on hepatic TG, whether the liver is an important and direct site of FGF21 action is debatable. Studies in seasonal mammals (Siberian hamsters, ground squirrels, etc.), which allow investigation of the physiological role of FGF21, suggested that FGF21 acts directly on the hepatocyte, promoting FA rather than carbohydrate oxidation (37). Besides FA oxidation and ketogenesis (36,37), a recent study suggested that FGF21 regulates hepatic lipid metabolism by promoting autophagy-mediated lipid degradation (38). Nevertheless, delineating the intracellular signaling pathways that mediate FGF21 action within hepatocytes will be helpful in fully understanding the physiological role of FGF21 in regulating hepatic lipid metabolism.
The hypothesis of FGF21 resistance has been proposed for many years (19,26); however, whether paradoxically increased hepatic and circulating FGF21 levels are a consequence or cause of NAFLD is unclear. Here, we found that hepatic p38 activation displays a hepatic FGF21-resistant phenotype in mouse models. Because hepatic p38 activation increases hepatic FGF21 expression and secretion through XBP1 and impairs local FGF21 action by promoting KLB degradation, we speculated that hepatic p38 activation can contribute to the development of FGF21 resistance. Additionally, NAFLD is usually accompanied by central obesity, and a large proportion of patients with NAFLD are lean. Because mice with hepatic p38 activation display not only hepatic steatosis but also reduced adiposity, which is similar to those observed in lean NAFLD, our finding may also provide a plausible explanation for the lean phenotype observed in NAFLD.
This article contains supplementary material online at https://doi.org/10.2337/figshare.16811332.
W.L., C.S., and Y.Y. contributed equally to this work.
Article Information
Acknowledgments. The authors thank Prof. Yi Liu (Fudan University, Shanghai, China), Prof. Yong Liu (Wuhan University, Wuhan, China), and Prof. Ronggui Hu (Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai, China) for guidance and expertise in adenoviral technology and ubiquitination assay. Virus and plasmid of XBP1s were gifts from Prof. Yong Liu’s laboratory (34); plasmids of WT and mutant HA-ubiquitin were gifts from Prof. Ronggui Hu’s laboratory (36). The authors also thank Zhonghui Weng and colleagues from the Institutional Center for Shared Technologies and Facilities (ICSTF) of Shanghai Institute of Nutrition and Health, Chinese Academy of Sciences, for technical assistance.
Funding. This work was supported by the Chinese Ministry of Science and Technology (2017YFA0102800), National Natural Science Foundation of China (91957205, 82070821, 81970748, 31900841, 32071166), Laboratory for Marine Drugs and Bioproducts of Pilot National Laboratory for Marine Science and Technology (Qingdao) (LMDBKF-2019-04), Youth Innovation Promotion Association, Chinese Academy of Sciences (2021261), Pujiang Talent Program from Science and Technology Commission of Shanghai Municipality (21PJ1416100), Young Elite Scientists Sponsorship Program by China Association for Science and Technology (2020QNRC001), and National Health Commission Key Laboratory of Food Safety Risk Assessment (2020K02).
Duality of Interest. No potential conflicts of interest relevant to this article were reported.
Author Contributions. W.L., C.S., Y.Y., H.C., and Z.N. performed the experiments. W.L., C.S., Y.Y., H.C., Z.N., S.S., S.L., Y.W., J.J., and H.Y. analyzed and interpreted the data. W.L., C.S., Y.Y., J.J., and H.Y. wrote the manuscript. W.L., C.S., Y.Y., and H.Y. designed the study. Ya.L., L.H., Yu.L., L.Z., C.H., and Q.D. contributed to the discussion. All authors reviewed and edited the manuscript. H.Y. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.