YIPF6 controls sorting of FGF21 into COPII vesicles and promotes obesity

Significance

Fibroblast growth factor 21 (FGF21) is an endocrine hormone that regulates glucose, lipid, and energy homeostasis. To date, little is known about the regulation of trafficking and secretion of FGF21. Here, we show that mice with a mutation in the Yipf6 gene (Klein–Zschocher [KLZ]; Yipf6^KLZ/Y) on a high-fat diet (HFD) have higher plasma levels of FGF21 than control mice due to the increased FGF21 secretion from their hepatocytes. Yipf6^KLZ/Y mice are thus resistant to HFD-induced features of the metabolic syndrome. The regulation of YIPF6 on FGF21 secretion was also conserved in nonalcoholic fatty liver disease (NAFLD) patients. YIPF6 is therefore a newly identified regulator of FGF21 secretion during development of obesity and could be a target for treatment of obesity and NAFLD.

Abstract

Fibroblast growth factor 21 (FGF21) is an endocrine hormone that regulates glucose, lipid, and energy homeostasis. While gene expression of FGF21 is regulated by the nuclear hormone receptor peroxisome proliferator-activated receptor alpha in the fasted state, little is known about the regulation of trafficking and secretion of FGF21. We show that mice with a mutation in the Yip1 domain family, member 6 gene (Klein–Zschocher [KLZ]; Yipf6^KLZ/Y) on a high-fat diet (HFD) have higher plasma levels of FGF21 than mice that do not carry this mutation (controls) and hepatocytes from Yipf6^KLZ/Y mice secrete more FGF21 than hepatocytes from wild-type mice. Consequently, Yipf6^KLZ/Y mice are resistant to HFD-induced features of the metabolic syndrome and have increased lipolysis, energy expenditure, and thermogenesis, with an increase in core body temperature. Yipf6^KLZ/Y mice with hepatocyte-specific deletion of FGF21 were no longer protected from diet-induced obesity. We show that YIPF6 binds FGF21 in the endoplasmic reticulum to limit its secretion and specifies packaging of FGF21 into coat protein complex II (COPII) vesicles during development of obesity in mice. Levels of YIPF6 protein in human liver correlate with hepatic steatosis and correlate inversely with levels of FGF21 in serum from patients with nonalcoholic fatty liver disease (NAFLD). YIPF6 is therefore a newly identified regulator of FGF21 secretion during development of obesity and could be a target for treatment of obesity and NAFLD.

According to the World Health Organization, ∼39% of adults were overweight worldwide in 2016; among these, 13% were obese (1). Obesity-related deaths are predominantly due to the development of cardiovascular disease and nonalcoholic fatty liver disease (NAFLD) (1–3). NAFLD has become a leading cause of chronic liver disease with the growing obesity pandemic (3), and liver cells contribute to the progression of obesity. The hepatocyte secretome changes during obesity. Steatotic hepatocytes produce factors that promote or protect against inflammation and insulin resistance (4).

Fibroblast growth factor 21 (FGF21) belongs to the endocrine subfamily of FGFs. Unlike the majority of FGFs, which bind heparan sulfate and become predominantly entrapped within the extracellular matrix of their site of synthesis, FGF21 enters circulation upon secretion due to its diminished heparan sulfate binding, and instead relies on a tertiary receptor complex composed of beta-klotho and FGF receptor 1 (FGFR1) for tissue targeting and signal transduction (5–7). Circulating FGF21 is a primarily hepatocyte-derived hormone in response to fasting or starvation, which exerts its function in a variety of tissues, including white and brown adipose tissue, liver, pancreas, and brain (8–10).

Administration of recombinant FGF21 improves metabolic homeostasis in obese and diabetic mouse models, decreasing plasma glucose and triglycerides (11), increasing insulin sensitivity (12), reducing hepatic steatosis (13), increasing energy expenditure, and decreasing body weight (14, 15). These results suggest that FGF21 might be a promising antidiabetic therapy in humans. Indeed, LY2405319, an engineered variant of FGF21, reduces body weight and dyslipidemia and alters atherogenic apolipoprotein profiles of patients with obesity and type 2 diabetes (16). While delivering engineered recombinant FGF21 variants is a dominant strategy to exploit the therapeutic potential of FGF21, efforts are also underway to boost endogenous FGF21 level or efficacy as an alternative therapeutic strategy, such as using fibroblast-activating protein inhibitors (17) and gene therapy-mediated FGF21 overexpression in liver via adeno-associated virus serotype 8 (AAV8) (18). FGF21 has an amino-terminal signal peptide (leader sequence) and is delivered to the cell surface through the endoplasmic reticulum (ER) to the Golgi (19). However, the mechanisms of FGF21 secretion remain elusive.

Yip1 domain family, member 6 (YIPF6) belongs to a family of proteins comprising YIPF1 through YIPF7. Little is known about the functions of these proteins in mammalian cells (20). The yeast homologs of mammalian YIPFs, named Yips, are membrane receptors for YPT and RAB guanosine triphosphatases (GTPases) and associate with secretory vesicles between the ER and Golgi (21–23). Secretory vesicles with the coat protein complex II (COPII) mediate transport of newly synthesized proteins from the ER (24–26). Yipf6 is located on the X chromosome. The protein it encodes colocalizes with the SEC31 homolog A (SEC31A), a COPII coat complex component, so YIPF6 might be involved in trafficking proteins from the ER to the Golgi (20).

We used mice with germline disruption of Yipf6 (Klein–Zschocher [KLZ]; Yipf6^KLZ/Y mice) by exposure of mice to N-ethyl-N-nitrosourea (20). In Yipf6^KLZ/Y mice, only the Yipf6 gene is disrupted; when these mice are crossed with transgenic mice that carry a bacterial artificial chromosome that contains the full-length Yipf6, the progeny do not have the phenotype of Yipf6^KLZ/Y mice (20). We demonstrate that Yipf6^KLZ/Y mice are protected from diet-induced obesity, hepatic steatosis, and insulin resistance via increased secretion of FGF21 from hepatocytes. We show that YIPF6 binds to FGF21 in the ER lumen and selectively transfers it into COPII vesicles by interaction with SEC23A. Hepatocyte-specific deletion of Yipf6 increased the FGF21 level in plasma and reduced diet-induced obesity, hepatic steatosis, and insulin resistance. Moreover, we observed an inverse correlation between the level of YIPF6 in liver and serum levels of FGF21 in overweight patients with NAFLD.

Results

Yipf6^KLZ/Y Mice Are Protected from Diet-Induced Obesity and Metabolic Syndrome.

Yipf6^KLZ/Y mice had lower body weights after 16 wk on the high-fat diet (HFD) compared with mice without disruption of Yipf6 (wild type [WT]) (Fig. 1 A–C). Body weights of Yipf6^KLZ/Y mice fed the HFD did not differ significantly from those of WT or Yipf6^KLZ/Y mice fed a normal chow (control) diet (Fig. 1A). Yipf6^KLZ/Y mice fed the HFD also had significantly lower weights of subcutaneous (s.c.) white adipose tissue, epididymal white adipose tissue, brown adipose tissue, and liver than WT mice fed the HFD (Fig. 1D). There were no differences in food intake or intestinal fat absorption (SI Appendix, Fig. S1 A and B).

Fig. 1.

Yipf6 mutation protects mice from diet-induced obesity. (A) Body weight (n = 4 to 5 for control diet and n = 8 to 10 for HFD). (B) Representative images of WT and Yipf6^KLZ/Y mice placed on the HFD for 16 wk. (C) Body weights of groups (n = 70 WT mice fed the HFD and n = 69 Yipf6^KLZ/Y mice fed the HFD). (D, Top) Weight of s.c. white adipose tissue (SubQ WAT; n = 8 to 11 for control diet and n = 18 to 22 for HFD), epididymal white adipose tissue (Epi WAT; n = 8 to 11 for control diet and n = 18 to 23 for HFD), brown adipose tissue (BAT; n = 5 for control diet and n = 5 to 7 for HFD), and liver (n = 8 to 11 for control diet and n = 18 to 23 for HFD). (D, Bottom) Representative images of SubQ WAT, Epi WAT, BAT, and liver from WT and Yipf6^KLZ/Y mice following the HFD. (E) Total, HDL, and LDL cholesterol in plasma (n = 4 to 6 for control diet and n = 5 for HFD). (F) Ratio of high-molecular-weight (HMW) adiponectin to total adiponectin (Top) and leptin (Bottom) in plasma from WT and Yipf6^KLZ/Y mice (n = 4 to 5 for control diet and n = 14 to 16 for HFD). (G) Blood glucose concentrations during the GTT and ITT (n = 8 to 9 for control diet and n = 16 to 18 for HFD). i.p., intraperitoneal. Data are presented as mean ± SEM. *P < 0.05.

Consistent with decreased adiposity, Yipf6^KLZ/Y mice on the HFD had better metabolic and endocrine profiles than WT mice on the HFD, based on significantly lower serum levels of total, high-density lipoprotein (HDL), and low-density lipoprotein (LDL) cholesterol (Fig. 1E); an increased ratio of high-molecular-weight adiponectin to total adiponectin; and decreased plasma leptin levels (Fig. 1F). Adiponectin messenger RNA (mRNA) in white adipose tissue was higher in Yipf6^KLZ/Y mice fed the HFD than in WT mice fed the HFD (SI Appendix, Fig. S1C). Moreover, Yipf6^KLZ/Y mice had significantly increased glucose tolerance and insulin sensitivity compared with WT mice fed the HFD, as detected by a glucose tolerance test (GTT) and an insulin tolerance test (ITT) (Fig. 1G). Yipf6^KLZ/Y and WT mice fed the control diet had comparable glucose tolerance and insulin sensitivity (Fig. 1G). These findings indicate that disruption of Yipf6 protects mice from diet-induced obesity and insulin resistance.

Attenuated Liver Steatosis and Injury in Yipf6^KLZ/Y Mice.

To assess the role of YIPF6 in steatohepatitis, a more advanced disease than hepatic steatosis (3), mice were additionally fed a high-fat, high-cholesterol diet (HFCD). Yipf6^KLZ/Y mice fed the HFCD did not have a significant increase in body weight compared with Yipf6^KLZ/Y mice or WT mice fed a control diet (SI Appendix, Fig. S1D). Yipf6^KLZ/Y mice fed the HFCD had lower weights of s.c. white adipose tissue, epididymal white adipose tissue, brown adipose tissue, and liver than WT mice fed the HFCD (SI Appendix, Fig. S1E). Glucose tolerance was improved in Yipf6^KLZ/Y mice fed the HFCD compared with WT mice fed this diet (SI Appendix, Fig. S1F).

Liver steatosis and steatohepatitis are hepatic manifestations of obesity (3). HFD-induced hepatic steatosis and triglyceride and cholesterol accumulation were reduced in Yipf6^KLZ/Y mice (SI Appendix, Fig. S2 A and B). Yipf6^KLZ/Y mice also had reduced liver injury after 16 wk on the HFCD, based on their lower plasma level of alanine aminotransferase (SI Appendix, Fig. S2C). Expression of lipogenesis markers, such as peroxisome proliferator-activated receptor gamma (Pparg) and cluster of differentiation 36 (Cd36) mRNAs, were decreased in livers of Yipf6^KLZ/Y mice compared with livers from WT mice (SI Appendix, Fig. S2D). Notably, the mRNA expression of peroxisome proliferator-activated receptor alpha (Ppara), which can directly regulate FGF21 transcription, was up-regulated by the HFD, whereas no difference was observed between HFD-fed WT and Yipf6^KLZ/Y mice (SI Appendix, Fig. S2E). Functional analyses of liver transcriptomes revealed differences in triglyceride and steroid biosynthesis and PPAR signaling pathways (SI Appendix, Fig. S2F). Levels of mRNAs, such as fatty acid synthase (Fasn), elongation of long chain fatty acids family member 6 (Elovl6), lipin1 (Lpin1), stearyl CoA desaturase 1 (Scd1), stearyl CoA desaturase 2 (Scd2), lipoprotein lipase (Lpl), and fatty acid binding protein 4 (Fabp4), were all lower in Yipf6^KLZ/Y mice than WT mice when both strains were fed the HFD (SI Appendix, Fig. S2F). Disruption of Yipf6 therefore appears to reduce liver steatosis and injury following the HFD or HFCD.

Increased Lipolysis in White Adipose Tissue of Yipf6^KLZ/Y Mice.

Adipocytes in Yipf6^KLZ/Y mice fed the HFD were smaller than adipocytes in WT mice fed the HFD (Fig. 2A). We found no significant differences in mRNA levels of lipogenesis markers, such as Fasn, Pparg, Cd36, or sterol regulatory element-binding protein 1c (Srebp1c), in the adipose tissue (SI Appendix, Fig. S3A). However, mRNA levels of lipolysis markers, including adipose triglyceride lipase (Atgl) and hormone sensitive lipase (Hsl), were significantly increased in adipose tissue from Yipf6^KLZ/Y mice fed the HFD vs. WT mice fed the HFD (Fig. 2B). HSL protein was also higher in the white adipose tissue of Yipf6^KLZ/Y mice than WT mice fed the HFD (Fig. 2C). RALA, which is a Ras family small guanosine triphosphate (GTP)-binding protein, was used as a loading control (27). Consistent with increased lipolysis in Yipf6^KLZ/Y mice, HFD-fed Yipf6^KLZ/Y mice had higher systemic levels of free fatty acids than HFD-fed WT mice (Fig. 2D). These results demonstrate that loss of YIPF6 increases fat lipolysis in white adipose tissue, which might contribute to the reduced body weight and fat mass in Yipf6^KLZ/Y mice.

Fig. 2.

Disruption of Yipf6 increases lipolysis in white adipose tissue and thermogenesis in brown adipose tissue. (A) Representative images of hematoxylin/eosin (H&E) staining of epididymal fat from HFD-fed WT and Yipf6^KLZ/Y mice. (B) Atgl and Hsl mRNA levels in white adipose tissue (n = 3 to 5 for control diet and n = 8 for HFD). (C) Immunoblot for HSL and RALA in epididymal fat from WT and Yipf6^KLZ/Y mice (n = 5 per group). (D) Plasma free fatty acid (FFA) levels (n = 4 to 5 for control diet and n = 11 to 13 for HFD). (E) Rectal temperatures of 16 wk HFD-fed WT and Yipf6^KLZ/Y mice at room temperature (n = 4 to 5 for control diet and n = 8 to 9 for HFD). (F) Representative images of H&E-stained brown adipose tissue. Levels of Ucp1, Dio2, and Adrb3 mRNAs (G) and Hsl and Atgl mRNAs (H) are shown in brown adipose tissue (n = 4 to 5 for control diet and n = 7 to 10 for HFCD). (I) Immunoblot for HSL, UCP1, and RALA in brown adipose tissue from WT and Yipf6^KLZ/Y mice fed the control diet or HFCD for 16 wk (n = 5 per group). Data are presented as mean ± SEM. *P < 0.05. (Scale bars, 50 μm.)

Increased Energy Expenditure and Thermogenesis in Brown Adipose Tissue of Yipf6^KLZ/Y Mice.

We compared metabolic rates of WT and Yipf6^KLZ/Y mice on the HFD. Oxygen consumption and carbon dioxide production rates were significantly higher in Yipf6^KLZ/Y mice than WT mice (SI Appendix, Fig. S3B). There was a lower respiratory exchange ratio in Yipf6^KLZ/Y mice, indicating increased use of fat for energy production (SI Appendix, Fig. S3C). Yipf6^KLZ/Y mice were more active than WT mice, as measured by the ambulatory horizontal and vertical counts, potentially contributing to higher energy expenditure by Yipf6^KLZ/Y mice (SI Appendix, Fig. S3D). In line with increased energy expenditure, Yipf6^KLZ/Y mice generated more heat, with an ∼0.64 °C higher core body temperature and a higher body temperature after cold exposure (4 °C) (Fig. 2E and SI Appendix, Fig. S3E), indicating augmented adaptive thermogenesis.

Consistent with increased energy expenditure and thermogenesis, lipid accumulation in brown adipose tissue was reduced in HFD-fed and HFCD-fed Yipf6^KLZ/Y mice compared with WT mice on the same diet (Fig. 2F and SI Appendix, Fig. S3F). Gene expression analysis showed strong induction of uncoupling protein 1 (Ucp1), type 2 deiodinase (Dio2), Hsl, and Atgl, which are involved in thermogenesis and lipolysis, respectively (Fig. 2 G and H). Expression of the adrenergic receptor, beta 3 (Adrb3), a major mediator of adrenergic signaling in adipose tissue, was also consistently increased (Fig. 2G). We confirmed increased protein levels of HSL and UCP1 in brown adipose tissue of Yipf6^KLZ/Y mice compared with WT mice (Fig. 2I). Loss of YIPF6 therefore increases energy expenditure and thermogenesis in brown adipose tissue, which might contribute to the reduced body weight in Yipf6^KLZ/Y mice.

Increased Hepatic Secretion of FGF21 Mediates Protective Effects of Yipf6 Deficiency.

Yipf6^KLZ/Y mice have a phenotype similar to that of mice expressing an Fgf21 transgene, characterized by lower body weight gain and increased lipolysis and expression of UCP1, indicating increased thermogenesis and energy expenditure (11, 15, 28). Considering the role of YIPF family members in vesicle trafficking (20, 22, 23), we measured plasma levels of FGF21, which we found to be higher in Yipf6^KLZ/Y mice compared with WT mice when both were fed the HFD (Fig. 3A). Hepatocytes are the main producers of FGF21 (29). Expression of Fgf21 mRNA in liver is higher than in adipose tissue (white adipose tissue and brown adipose tissue) and muscle (Fig. 3B). However, hepatic levels of Fgf21 mRNA and FGF21 protein did not differ significantly between WT and Yipf6^KLZ/Y mice fed the HFD for 16 wk (Fig. 3 B and C). In addition, primary hepatocytes isolated from WT and Yipf6^KLZ/Y mice fed the HFD had similar Fgf21 mRNA levels (SI Appendix, Fig. S4A); thus, loss of YIPF6 might affect the plasma level of FGF21 after its transcription and translation. Although primary hepatocytes isolated from Yipf6^KLZ/Y mice fed the control diet secreted less FGF21 than hepatocytes from WT mice fed the control diet (SI Appendix, Fig. S4B), hepatocytes isolated from HFD-fed Yipf6^KLZ/Y mice secreted more FGF21 into supernatants than hepatocytes from HFD-fed WT mice (Fig. 3D), while cellular levels of FGF21 protein did not differ between WT and Yipf6^KLZ/Y hepatocytes (Fig. 3 E, Left). When we used cycloheximide to block protein synthesis, we found that cellular FGF21 protein was similar in hepatocytes isolated from HFD-fed WT and Yipf6^KLZ/Y mice (Fig. 3 E, Right). This indicates that YIPF6 does not affect protein degradation or stability.

Fig. 3.

Disruption of Yipf6 increases FGF21 secretion by hepatocytes. (A–C) Mice were fed the HFD for 16 wk. (A) Serum level of FGF21 (n = 5 per group). N.D., not detected. (B) Fgf21 mRNA in liver, white adipose tissue (WAT), brown adipose tissue (BAT), and muscle (n = 4 to 5 for control diet and n = 8 to 12 for HFD). (C) FGF21 protein in liver extracts (n = 3 for control [Ctrl] diet and n = 4 for HFD). (D and E) WT and Yipf6^KLZ/Y mice were fed the HFD diet for 16 wk. (D) Hepatocytes were isolated, and levels of FGF21 in supernatants were measured (n = 6 per group). (E) Levels of FGF21 were detected in hepatocytes incubated with or without cycloheximide. (F and G) AML12 cells were transfected with empty vector (Ctrl) or vectors that expressed FGF21 along with shRNA control (ShCtrl) or shRNA Yipf6 (ShYipf6) for 48 h. Levels of FGF21 were measured in cell supernatants (F) and in cell extracts (G, immunoblot). (H) AML12 cells were transfected with empty vector (Ctrl) or vectors that expressed FGF21 or YIPF6 for 24 h, and FGF21 levels were measured in supernatants (n = 3). (I) Mice were fed the HFD for 16 wk, and levels of A1AT were measured in plasma (Left, n = 5 per group) and in supernatants of hepatocytes isolated from mice (Right, n = 6 per group). (J–N) WT, Fgf21^ΔHep, Yipf6^KLZ/Y, and Yipf6^KLZ/Y/Fgf21^ΔHep mice were fed the HFD for 16 wk. (J) Plasma levels of FGF21 (n = 6 for WT and Fgf21^ΔHep mice, n = 5 for Yipf6^KLZ/Y and Yipf6^KLZ/Y/Fgf21^ΔHep mice). (K) Body weights (n = 9 for WT mice, n = 11 for Fgf21^ΔHep mice, n = 8 for Yipf6^KLZ/Y mice, n = 9 for Yipf6^KLZ/Y/Fgf21^ΔHep mice). (L) Blood glucose levels in the GTT and ITT (n = 10 for WT mice, n = 10 to 11 for Fgf21^ΔHep mice, n = 5 to 6 for Yipf6^KLZ/Y mice, n = 6 to 11 for Yipf6^KLZ/Y/Fgf21^ΔHep mice). i.p., intraperitoneal. (M) Rectal temperatures (n = 7 for WT mice, n = 5 for Fgf21^ΔHep mice, n = 10 for Yipf6^KLZ/Y mice, n = 13 for Yipf6^KLZ/Y/Fgf21^ΔHep mice). (N) Levels of UCP1 protein in brown adipose tissue (n = 3 per group). Data are presented as mean ± SEM. *P < 0.05.

Knockdown of YIPF6 with small hairpin RNA (shRNA) increased secretion of FGF21 by the mouse hepatocyte cell line AML12 (Fig. 3F) but did not affect cellular levels of FGF21 protein (Fig. 3G). Furthermore, actinomycin D, which is an inhibitor of transcription and replication (30), was utilized to block Fgf21 mRNA expression (SI Appendix, Fig. S4C). Knockdown of YIPF6 increased the secreted FGF21 level in the supernatant compared with controls in the presence of actinomycin D (SI Appendix, Fig. S4D). Alternatively, overexpression of YIPF6 significantly reduced secretion of FGF21 by these cells (Fig. 3H). It is important to note that AML12 cells secrete an endogenous level of FGF21 that is below the limit of detection (or close to the lowest concentration that can be detected), so we had to express FGF21 from a transgene for these experiments (Fig. 3 F–H and SI Appendix, Fig. S4 C and D) to resemble the increased FGF21 expression during obesity.

FGF21 has an N-terminal signal peptide; we confirmed its transport via COPII vesicles by demonstrating the colocalization of the FGF21 and COPII marker protein SEC31A (19) (SI Appendix, Fig. S4E). To test whether loss of YIPF6 causes an overall increase in COPII vesicle-mediated transport and secretion, we measured α1-antitrypsin (A1AT or SERPINA1), a protein secreted via the COPII vesicle pathway (31). Levels of A1AT did not differ significantly between plasma from HFD-fed WT vs. Yipf6^KLZ/Y mice or in supernatants of isolated hepatocytes from HFD-fed WT vs. Yipf6^KLZ/Y mice (Fig. 3I). Loss of YIPF6 promotes secretion specifically of FGF21 from hepatocytes.

To demonstrate that increased FGF21 secretion from hepatocytes protects Yipf6^KLZ/Y mice from the effects of the HFD, we generated Yipf6^KLZ/Y mice with hepatocyte-specific disruption of Fgf21 (Yipf6^KLZ/Y/Fgf21^ΔHep). FGF21 was not detectable in plasma from Yipf6^KLZ/Y/Fgf21^ΔHep mice fed the HFD (Fig. 3J), demonstrating the disruption of the gene and confirming that hepatocytes are the main source of systemic FGF21. The Yipf6^KLZ/Y/Fgf21^ΔHep mice lost the protection from the effects of the HFD observed in Yipf6^KLZ/Y mice and became obese, had body weight gain, and developed glucose intolerance and insulin resistance (Fig. 3 K and L and SI Appendix, Fig. S4F). Yipf6^KLZ/Y/Fgf21^ΔHep mice had normal body temperatures (Fig. 3M) and reduced production of UCP1 protein in brown adipose tissue, compared with Yipf6^KLZ/Y mice, after 16 wk on the HFD (Fig. 3N).

SEC proteins affect selection of proteins for transport via the COPII-mediated secretory pathway from the ER to the Golgi apparatus (25, 32, 33). Mice with a disrupted Sec24a gene are not able to secrete proprotein convertase subtilisin/kexin type 9 (PCSK9) (34). PCSK9 is secreted into the plasma by hepatocytes, where it binds to the LDL receptor (LDLR) and promotes its endocytosis and degradation. PCSK9 is therefore a negative regulator of the LDLR (35). Hepatic levels of LDLR are increased in SEC24A-deficient mice, which lowers total plasma cholesterol (34). WT, Yipf6^KLZ/Y, and Yipf6^KLZ/Y/Fgf21^ΔHep mice had comparable plasma levels of PCSK9 following the HFD (SI Appendix, Fig. S4G). In addition, hepatocytes isolated from HFD-fed WT and Yipf6^KLZ/Y mice secreted similar amounts of PCSK9 (SI Appendix, Fig. S4H). So, loss of YIPF6 seems to protect mice from HFD-induced obesity by specifically increasing secretion of FGF21 by hepatocytes.

YIPF6 Limits Sorting of FGF21 into COPII Vesicles.

To determine the mechanism by which YIPF6 modulates FGF21 secretion, we measured hepatic gene expression of YIPF1 to YIPF7. Although hepatic expression of most Yipf genes, including Yipf6, increased in mice on the HFD vs. the control diet (Yipf2 to Yipf6) (Fig. 4 A and B and SI Appendix, Fig. S5A), no significant differences in expression of these genes were observed in HFD-fed WT vs. Yipf6^KLZ/Y mice (Fig. 4A and SI Appendix, Fig. S5A). Yipf7 was not expressed in liver (36).

Fig. 4.

YIPF6 binds SEC23A and controls packaging of FGF21 into COPII vesicles. (A–C) WT and Yipf6^KLZ/Y mice were fed the control diet or the HFD for 16 wk. (A) Levels of Yipf6 mRNA in liver. (B, Upper) Hepatocytes were isolated from WT and Yipf6^KLZ/Y mice fed the control diet, and cellular YIPF6 protein was detected by immunoblot. (B, Lower) YIPF6 protein in livers of WT mice fed the control (Ctrl) diet and HFD. (C) Sec12, Sec13, Sec31a Sec16a, Sec23a, Sec24a, Sar1a, and Rab1a mRNAs in liver (n = 5 per group). (D) WT and Yipf6^KLZ/Y mice were fed the HFD for 16 wk. Primary hepatocytes were isolated. FGF21 subcellular localization was analyzed in hepatocytes by fractionation of the ER and Golgi. (E) In vitro COPII budding assays with primary hepatocytes from WT or Yipf6^KLZ/Y mice each fed the HFD for 16 wk. (F) HEK293T cells overexpressing YIPF6, shYipf6, or FGF21 for 48 h, providing donor membranes. Liver cytosol from Fgf21^ΔHep mice provided COPII components. The respective medium-speed samples (Medium) and high-speed samples (High, contains COPII vesicles) collected from the budding reaction were resolved on sodium dodecyl sulfate polyacrylamide gel electrophoresis and processed for immunoblotting with SEC22B, FGF21, and YIPF6 antibodies. (G, Left) Immunoprecipitated (IP) endogenous YIPF6 in AML12 cells and immunoblots for SEC23A, SEC24A, SAR1A/B, SEC13, and YIPF6. (G, Right) HEK293T cells were transfected with Yipf6-GFP and Sec23a-Flag constructs for 48 h. Immunoprecipitated SEC23A with Flag beads in HEK293T cells and immunoblots for YIPF6 and SEC23A. (H, Left) HEK293T cells were transfected with Yipf6-GFP and Fgf21-Flag constructs for 48 h. FGF21 was immunoprecipitated from HEK293T cells with Flag beads and immunoblot for YIPF6 and FGF21. (H, Right) HEK293T cells were transfected with Yipf6-GFP, Fgf21-Flag, and Sec23a-hemagglutinin (HA) for 48 h. FGF21 was immunoprecipitated with Flag beads from HEK293T cells; immunoblot for SEC23A, FGF21, and YIPF6. (I) HEK293T cells were transfected with Fgf21-Flag, Yipf6-GFP, and Sec23a-HA for 48 h. Immunofluorescent staining was performed with FGF21 and HA antibodies. Confocal images of FGF21-Flag, YIPF6-GFP, and SEC23A-HA. (Scale bar, 10 μm.) Data are presented as mean ± SEM. *P < 0.05.

FGF21 is secreted via the COPII vesicle pathway, so we investigated hepatic expression of COPII components. Although expression of most proteins involved in the COPII early secretory pathway increased in mice on the HFD compared with the control diet, no significant differences were found between WT and Yipf6^KLZ/Y mice (Fig. 4C). This means that the increase in FGF21 secretion observed in Yipf6^KLZ/Y mice cannot be attributed to increases in specific COPII components. Confocal imaging showed that YIPF6 localized to the ER and Golgi in mouse hepatocytes, based on colocalization with the ER marker protein calnexin and cis-Golgi matrix protein GM130 (SI Appendix, Fig. S5B), consistent with previous reports (37, 38). We studied ER and Golgi fractions to determine the subcellular distribution of FGF21. A higher amount of FGF21 was present in the ER of hepatocytes isolated from HFD-fed WT mice compared with hepatocytes isolated from HFD-fed Yipf6^KLZ/Y mice (Fig. 4D), although the levels of FGF21 in Golgi fractions were similar. These findings indicate that YIPF6 affects FGF21 secretion at the ER stage.

To explore the role of YIPF6 in secretion of FGF21, we used an in vitro COPII vesicle budding assay. Primary hepatocytes from HFD-induced WT and Yipf6^KLZ/Y mice, supplied donor membranes, and liver cytosol from Fgf21^ΔHep mice provided the COPII components (34, 36, 39, 40). In the budding assay, we detected more FGF21 in Yipf6^KLZ/Y membrane-derived COPII vesicles (Fig. 4E), which is consistent with the enzyme-linked immunosorbent assay (ELISA) result in Fig. 3D. We performed a second COPII vesicle budding assay, in which permeabilized 293T cells expressing YIPF6, shYipf6, or FGF21 supplied the membrane fraction and liver cytosol from Fgf21^ΔHep mice provided COPII components. In agreement with results from an ELISA (Fig. 3 F and H), the COPII budding assay demonstrated that overexpression of YIPF6 decreased FGF21 packaging into COPII vesicles, whereas shRNA knockdown of YIPF6 promoted this process (Fig. 4F). We also demonstrated that YIPF6 and FGF21 are incorporated into the COPII vesicles and are part of the early secretory pathway (Fig. 4F).

We investigated the roles of YIPF6 as a transmembrane protein in the ER (20). To search for YIPF6 interaction partners, YIPF6 was immunoprecipitated from AML12 hepatocytes; the immunoprecipitant was screened with antibodies against COPII coat proteins and RAB1 proteins, which have been reported to mediate COPII vesicle trafficking (41, 42). Although we could not detect an interaction between YIPF6 and SEC24A, SEC13, SAR1A/B, or RAB1A/B (Fig. 4G and SI Appendix, Fig. S5 C and D), YIPF6 interacted with SEC23A (Fig. 4G). Furthermore, YIPF6 interacted with FGF21 (Fig. 4 H and I). FGF21 interacted with SEC23A, which required the presence of YIPF6 (Fig. 4 H, Right). Collectively, these results indicate that YIPF6 selects FGF21 for sorting into COPII vesicles, functioning as a cargo receptor. Through its interaction with SEC23A, YIPF6, together with FGF21, is pulled into budding vesicles. FGF21 sorting and packaging are more efficient in the absence of YIPF6.

Hepatocyte-Specific Deletion of Yipf6 Increases Systemic FGF21 and Reduces Diet-Induced Obesity and Metabolic Syndrome.

To confirm the regulation of FGF21 secretion by YIPF6 in hepatocytes in vivo, we generated a Yipf6 conditional knockout mouse (Yipf6^flox/Y). LoxP sites were inserted between exons 1 and 2 and between exons 4 and 5 (SI Appendix, Fig. S6 A and B). To generate a hepatocyte-specific Yipf6 deletion, AAV8, which has high tropism for hepatocytes, overexpressing Cre recombinase under the hepatocyte-specific albumin promoter was injected into the tail vein of Yipf6^flox/Y mice. AAV8-expressing GFP was used as a control vector. AAV8-mediated deletion of Yipf6 in hepatocytes (Yipf6^ΔHep; Fig. 5A) increased plasma levels of FGF21 (Fig. 5B), but did not significantly affect Fgf21 mRNA expression in the liver after 22 wk on the HFD (Fig. 5C). Yipf6^ΔHep mice generated more heat with a higher core body temperature (Fig. 5D) and increased UCP1 expression in brown adipose tissue (Fig. 5E) compared with Yipf6^flox/Y mice injected with AAV8 expressing GFP after 22 wk on HFD. Yipf6^ΔHep mice showed significantly lower body weight gain (Fig. 5F), lower weight of epididymal white adipose tissue (Fig. 5G), and attenuated hepatic steatosis (Fig. 5H) despite similar food intake (Fig. 5I). Moreover, HFD-fed Yipf6^ΔHep mice had significantly increased glucose tolerance and insulin sensitivity compared with Yipf6^flox/Y mice injected with AAV8-expressing GFP, as detected by the GTT and ITT (Fig. 5J). These findings indicate that deletion of Yipf6 specifically in hepatocytes protects mice from diet-induced obesity and features of the metabolic syndrome by increasing FGF21 secretion.

Fig. 5.

Hepatocyte-specific deletion of Yipf6 protects mice from diet-induced obesity and features of the metabolic syndrome. Yipf6^flox/Y mice were fed the HFD for 5 wk, injected with AAV8 encoding GFP (WT; n = 9 to 16) or Cre recombinase (Yipf6^ΔHep; n = 15 to 20) under the hepatocyte-specific albumin promoter through the tail vein. After injection, mice were fed the HFD for an additional 22 wk. (A) Yipf6 mRNA in liver. (B) Plasma level of FGF21. (C) Fgf21 mRNA in liver. (D) Rectal temperature. (E) UCP1 protein expression in brown adipose tissue (n = 3 for WT mice, n = 3 for Yipf6^ΔHep mice). (F) Percentage of body weight gain. (G) Weight of epididymal white adipose tissue (Epi WAT). (H) Representative hematoxylin/eosin (H&E)-stained liver sections. (Scale bar, 50 μm.) (I) Food intake. (J) Blood glucose concentrations during the GTT and ITT (n = 7 to 8 for WT mice, n = 8 for Yipf6^ΔHep mice). i.p., intraperitoneal. Data are presented as mean ± SEM. *P < 0.05.

Inverse Correlation between Level of YIPF6 in Liver and Serum Level of FGF21 in Patients with NAFLD.

We examined the association between hepatic levels of YIPF6 and serum levels of FGF21 in patients with NAFLD. We obtained liver tissues from 16 patients with biopsy-proven NAFLD (liver biopsies are not available from patients without liver disease) (SI Appendix, Tables S1 and S2). Levels of YIPF6 in liver tissues correlated inversely with patients’ serum level of FGF21 (Fig. 6 A and B). Furthermore, hepatic levels of YIPF6 correlated with the extent of liver steatosis (Fig. 6 A and C).

Fig. 6.

Relationships among hepatic YIPF6, serum FGF21, and steatosis in patients with NAFLD. (A) Representative images of hepatic tissues from patients with NAFLD, with immunohistochemistry for immunoglobulin G (IgG; control) or YIPF6. (Scale bar, 50 μm.) (B) Inverse correlation between hepatic level of YIPF6 and serum level of FGF21 in patients with NAFLD. (C) Correlation of hepatic YIPF6 with steatosis in patients with NAFLD. (D) Model for YIPF6 as a cargo receptor for FGF21 budding. (Left Upper) Under baseline conditions with low FGF21 (blue circle) and YIPF6 protein (red) in the ER of WT hepatocytes, the small amount of FGF21 is bound to YIPF6 in the ER lumen and is secreted via COPII vesicles. YIPF6 is using SEC23A (light blue) as a guide into budding vesicles. (Right Upper) In the absence of YIPF6, the small amount of FGF21 cannot easily enter the budding vesicle without the selection and sorting of YIPF6. In both situations, secreted FGF21 levels are low in the systemic circulation under healthy conditions. (Left Lower) During steatosis and obesity, YIPF6 and FGF21 increase in the ER of WT hepatocytes. YIPF6 binds FGF21 in the ER lumen and interacts with SEC23A, keeping FGF21 from an unrestricted flow into COPII vesicles. The selection of FGF21 by YIPF6 and packaging of FGF21 into COPII vesicles determine the amount of secreted FGF21. (Right Lower) In Yipf6^KLZ/Y hepatocytes, the absence of YIPF6 allows FGF21 to escape retention by YIPF6 and facilitates FGF21 entering into COPII vesicles at prevailing luminal ER concentrations, leading to increased secretion of FGF21.

Discussion

We found Yipf6^KLZ/Y mice to have increased lipolysis, thermogenesis, and energy expenditure and to be protected from diet-induced features of the metabolic syndrome, including obesity and hepatic steatosis. We investigated the role of YIPF family members in vesicle transport between the ER and Golgi apparatus (20, 22, 23) and found that increased hepatocyte secretion of FGF21 mediates the protection of Yipf6^KLZ/Y mice from the effects of the HFD. Hepatocyte-specific deletion of Yipf6 confirmed our findings.

Using ER-Golgi fractionation and COPII budding assays, we demonstrated that FGF21 is secreted via COPII vesicles and that YIPF6 is also present in COPII vesicles. YIPF6 is a transmembrane-spanning protein that contains a C-terminal domain with 5 transmembrane domains and a soluble, cytoplasmic N-terminal domain (20). We provided evidence for the interaction between YIPF6 and FGF21. Our findings indicate that the ER transmembrane protein YIPF6 might function as a trafficking receptor for FGF21 because it selects FGF21 for packaging into COPII vesicles in the ER.

By screening proteins involved in the early secretory COPII pathway, we found YIPF6 to interact with SEC23A, but not with other proteins in this pathway. SEC23A is the GTPase-activating protein (GAP) for SAR1, which is a small GTPase that initiates COPII coat assembly (25). SEC23A regulates coat assembly and vesicle trafficking by interacting with a variety of proteins (43, 44). The interaction between YIPF6 and SEC23A provides a physical connection between YIPF6 and COPII vesicles. Immunofluorescence analyses confirmed the colocalization of FGF21, YIPF6, and SEC23A. The interaction between FGF21 and SEC23A could not be detected in the absence of YIPF6, indicating that YIPF6 functions as a transmembrane bridge between the cytosol and the ER lumen. Interestingly, YIPF5, another member of mammalian YIPFs, also interacts with SEC23 (36), indicating the possible overlap in functions of YIPF family members.

Our results indicate that under baseline, nonobese conditions, with low levels of FGF21 and YIPF6 protein in the ER of hepatocytes, the small amount of FGF21 is bound to YIPF6 in the ER lumen and secreted via COPII vesicles. YIPF6 uses SEC23A as a guide into budding vesicles. Secreted levels of FGF21 are low in the systemic circulation of healthy individuals (45). During obesity, when YIPF6 and FGF21 increase in the ER, FGF21 bound to YIPF6 is secreted into COPII vesicles under the guidance of SEC23A, which is also increased in obesity. Therefore, the rate-limiting step for FGF21 secretion is the selection of FGF21 by YIPF6 and packaging of FGF21 into budding vesicles with the guidance of SEC23A. In the absence of YIPF6, FGF21 packaging into COPII transport vesicles is increased during obesity, as we have shown using the COPII budding assay. Because of the higher lumen concentration of unbound FGF21 in the ER, the gradient from the ER to budding vesicles increases, resulting in a flow from the ER lumen into the vesicles; this likely allows for increased secretion of FGF21 (Fig. 6D). Alternatively, other adaptor proteins might be more efficient to transport FGF21 into budding vesicles in the absence of YIPF6. Additional studies are needed to further evaluate this possibility.

We confirmed that YIPF6 is not required for general COPII vesicle formation, because secretion of other proteins via the early COPII pathway (A1AT and PCSK9) does not increase in the absence of YIPF6. It will also be important to determine the fate of YIPF6 in COPII vesicles, how YIPF6 releases FGF21 into the Golgi, and how it eventually returns to the ER via the COPI pathway (46). Importantly, this regulatory mechanism seems to be conserved in humans, because we found an inverse correlation between levels of YIPF6 on liver biopsies and FGF21 in serum from patients with NAFLD.

FGF21 has attracted much attention in the field of obesity research, and lately in NAFLD research, and it might be developed as a treatment for these diseases (10, 47, 48). LY2405319, an engineered variant of FGF21, reduced features of obesity and related metabolic syndrome in patients with type 2 diabetes (16, 49). Furthermore, a long-acting FGF21 analog, PF-05231023, reduced body weight and improved dyslipidemia in obese patients with type 2 diabetes (50).

However, the potential side effects of FGF21 therapy, such as bone loss, are a concern (50). A phase 2 clinical trial found that treatment with pegylated FGF21 (BMS-986036) decreased the hepatic fat fraction, improved biomarkers of fibrosis, and reduced liver injury in patients with progressive nonalcoholic steatohepatitis (51). No treatment-related side effects were reported. This is consistent with our observed correlation between hepatic YIPF6 and steatosis in human liver, although we could not find a correlation between hepatic YIPF6 and body mass index (SI Appendix, Fig. S7A). One reason for this could be that our study required liver tissue to measure YIPF6 expression. Therefore, our study enrolled patients with liver disease, who had a medical indication to undergo liver biopsy. Although most of the included patients met the criteria for being overweight, only 3 patients were obese and 2 patients were diagnosed with lean NAFLD.

We demonstrate that mice with disruption of Yipf6 are resistant to diet-induced features of the metabolic syndrome, including obesity and hepatic steatosis, through increased secretion of FGF21 by hepatocytes. We report that YIPF6 is a cargo receptor that becomes a rate-limiting molecule in the FGF21 secretory pathway during development of obesity. Considering the ability of FGF21 analogs to reduce features of obesity in humans, strategies to reduce or disrupt the activity of YIPF6 might be developed for treatment of obesity and NAFLD. Agents that block YIPF6 could promote release of endogenous FGF21 from hepatocytes to reduce the metabolic effects of obesity.

Materials and Methods

All animal studies were reviewed and approved by the Institutional Animal Care and Use Committee of the University of California, San Diego and the Institutional Animal Care and Use Committee of China Pharmaceutical University, Nanjing, China. The human study protocol was approved by the Ethics Committee of the Beijing Ditan Hospital, Capital Medical University, China, and a written informed consent was signed by each participant after the nature and possible consequences of the studies were explained. All of the methods, including those involving mice, human samples, cell lines and primary hepatocytes, triglycerides and cholesterol measurement in liver and feces, plasma assays, GTT and ITT, indirect calorimetry, histology, immunohistochemistry, immunofluorescence analysis, immunoprecipitation assay, coimmunoprecipitation assay, immunoblotting, protein stability experiments, actinomycin D treatment, subcellular fractionation, COPII budding assay, real-time PCR, RNA sequencing, quantification, and statistical analysis, are described in SI Appendix, Supplementary Methods.