Hepatic PTP4A1 ameliorates high-fat diet-induced hepatosteatosis and hyperglycemia by the activation of the CREBH/FGF21 axis

Precise regulation of kinases and phosphatases is crucial for human metabolic homeostasis. This study aimed to investigate the roles and molecular mechanisms of protein tyrosine phosphatase type IVA1 (PTP4A1) in regulating hepatosteatosis and glucose homeostasis. Method: Ptp4a1-/- mice, adeno-associated virus encoding Ptp4a1 under liver-specific promoter, adenovirus encoding Fgf21, and primary hepatocytes were used to evaluate PTP4A1-mediated regulation in the hepatosteatosis and glucose homeostasis. Glucose tolerance test, insulin tolerance test, 2-deoxyglucose uptake assay, and hyperinsulinemic-euglycemic clamp were performed to estimate glucose homeostasis in mice. The staining, including oil red O, hematoxylin & eosin, and BODIPY, and biochemical analysis for hepatic triglycerides were performed to assess hepatic lipids. Luciferase reporter assays, immunoprecipitation, immunoblots, quantitative real-time polymerase chain reaction, and immunohistochemistry staining were conducted to explore the underlying mechanism. Results: Here, we found that deficiency of PTP4A1 aggravated glucose homeostasis and hepatosteatosis in mice fed a high-fat (HF) diet. Increased lipid accumulation in hepatocytes of Ptp4a1-/- mice reduced the level of glucose transporter 2 on the plasma membrane of hepatocytes leading to a diminution of glucose uptake. PTP4A1 prevented hepatosteatosis by activating the transcription factor cyclic adenosine monophosphate-responsive element-binding protein H (CREBH)/fibroblast growth factor 21 (FGF21) axis. Liver-specific PTP4A1 or systemic FGF21 overexpression in Ptp4a1-/- mice fed an HF diet restored the disorder of hepatosteatosis and glucose homeostasis. Finally, liver-specific PTP4A1 expression ameliorated an HF diet-induced hepatosteatosis and hyperglycemia in wild-type mice. Conclusions: Hepatic PTP4A1 is critical for regulating hepatosteatosis and glucose homeostasis by activating the CREBH/FGF21 axis. Our current study provides a novel function of PTP4A1 in metabolic disorders; hence, modulating PTP4A1 may be a potential therapeutic strategy against hepatosteatosis-related diseases.

Hepatosteatosis is a metabolic disorder with an accumulation of fat in the liver of at least 5% more, which is a clinical hallmark of non-alcoholic fatty liver disease (NAFLD) [11,12]. NAFLD is the most common liver disorder worldwide and is an expanding health problem estimating a global prevalence of 25% [13]. NAFLD is categorized into non-alcoholic fatty liver and non-alcoholic steatohepatitis, which are histologically distinguished by inflammation with hepatocyte ballooning injury [14]. As the liver is an essential organ for maintaining glucose homeostasis, NAFLD is closely associated with the dysregulation of glucose uptake and utilization by the liver, accompanying the augmentation of gluconeogenesis and impairment of hepatic glucose transporter 2 (GLUT2) translocation to the plasma membrane [15][16][17][18]. Previous studies prove that NAFLD is associated with liver-related mortality or morbidity and an increased risk of extrahepatic diseases; however, there are currently no approved therapeutics [19]. Early development of therapeutics for NAFLD has been focused on preventing the pathogenesis of fibrosis, which is closely linked to the progression of NAFLD to cirrhosis and hepatocellular carcinoma [20]. Because of limited clinical efficacy in anti-fibrotic therapeutics, the approaches for developing therapeutics against NAFLD have been replaced with the repression of hepatosteatosis [21,22]. Fibroblast growth factor 21 (FGF21), a hormone expressed primarily by the liver, has been considered a promising therapeutic agent in decreasing hepatosteatosis and hepatocyte injury [21,23]. Administration of FGF21 protein, FGF21 analogs, or delivery of FGF21 by adenovirus into rodent models of NAFLD prevented hepatosteatosis with the suppression of de novo lipogenesis and the increased fat oxidation in the liver [24][25][26][27]. In addition, FGF21, as an endocrine factor, exerts inhibitory effects against NAFLD by regulating diverse extrahepatic tissues, including adipose tissue, pancreas, and skeletal muscle [24][25][26][27][28]. FGF21 expression is regulated by changes in a nutritional state, including fasting, high-carbohydrate diets, or low-protein diets through the activation of peroxisome proliferator-activated receptor (PPAR) α and/or cyclic adenosine monophosphate-responsive element-binding protein H (CREBH) [27,29].
Here, we identified PTP4A1 as a negative regulator in the pathogenesis of hepatic steatosis through activation of the CREBH/FGF21 axis. Deficiency of PTP4A1 in mice accelerates hepatic steatosis in mice fed a high-fat (HF) diet and leads to the disruption of glucose homeostasis with the reduction of GLUT2 on the plasma membrane of hepatocytes. PTP4A1 prevented hepatosteatosis through the activation of the transcription factor CREBH/FGF21 axis. Liver-specific PTP4A1 or systemic FGF21 overexpression in Ptp4a1 -/mice fed an HF diet restored PTP4A1 deficiency-mediated upregulation of hepatosteatosis and blood glucose. In addition, liver-specific PTP4A1 expression ameliorated HF diet-induced hepatosteatosis and hyperglycemia in wild-type (WT) mice. Therefore, controlling PTP4A1 may be a potential candidate for therapeutic strategy against hepatosteatosis-related diseases.

The deficiency of PTP4A1 in mice exacerbates HF diet-induced hyperglycemia and NAFLD
The mRNA levels of Ptp4a1, but not other isoforms, were significantly increased in the liver of mice fed an HF diet compared to those of mice fed a normal chow (NC) diet ( Figure S1A). Also, the alteration of hepatic fat contents by fasting and refeeding increased the Ptp4a1 mRNA levels in mice fed an NC diet ( Figure S1B). Consistent with mouse data, human hepatoma cells treated with bovine serum albumin-oleic acid (BSA-OA) tended to increase PTP4A1 mRNA levels compared to controls ( Figure S1C). In addition, we identified that the levels of PTP4A1 expression were significantly increased in human NASH samples than in healthy controls in the dataset (GSE63067) from the Gene Expression Omnibus database ( Figure S1D).
To evaluate whether PTP4A1 affects the regulation of obese-mediated metabolic diseases such as type 2 diabetes and NAFLD, we generated Ptp4a1 -/mice by CRISPR/Cas9 system ( Figure S2) and placed Ptp4a1 -/mice and WT littermates on an HF diet for 12 weeks. Body weight (BW) gain and food intake were comparable between the two groups on an HF diet for 12 weeks (Figure 1A and S3A). Dual-energy X-ray absorptiometry (DEXA) revealed that both lean and fat mass were similar between the two groups ( Figure  1B), and the ratio of epididymal white adipose tissue (eWAT) mass to BW was also comparable between the two groups ( Figure S3B). However, the blood glucose levels in Ptp4a1 -/mice were significantly higher than in WT mice in a postprandial state after 12 weeks of an HF diet feeding ( Figure 1C). Concomitantly, plasma insulin levels were markedly increased in Ptp4a1 -/mice compared to WT mice fed an HF diet ( Figure 1D). In fasting conditions, homeostatic model assessment-insulin resistance (HOMA-IR) was significantly increased in Ptp4a1 -/mice compared to WT mice after an HF diet ( Figure 1E). Glucose tolerance test (GTT) and insulin tolerance test (ITT) revealed that deficiency of PTP4A1 in mice fed an HF diet exacerbated glucose tolerance and reduced insulin sensitivity ( Figure 1F-G). After an HF diet for 12 weeks, Ptp4a1 -/mice showed an increase in fat accumulation in the liver compared to those of WT mice, which was proved by oil red O staining, hematoxylin and eosin (H&E) staining, and hepatic triglyceride (TG) analysis ( Figure 1H-I). In addition, Ptp4a1 -/mice increased the levels of alanine aminotransferase (ALT) and aspartate transaminase (AST) in plasma compared to those of WT mice fed an HF diet ( Figure 1J). By immunohistochemistry assay for cluster of differentiation 45 (CD45), we verified the enhanced immune cell accumulation in the liver of Ptp4a1 -/mice compared to WT mice fed an HF diet ( Figure 1K-L). Consistently, inflammatory markers, including F4/80, Mcp1, and Mip1a, but not Cd11c and KC, were significantly increased in the liver of Ptp4a1 -/mice compared to WT mice fed an HF diet ( Figure  S3C).

Lack of PTP4A1 reduces glucose uptake by decreasing GLUT2 on the plasma membrane in hepatocytes
To verify whether the alteration of gluconeogenesis contributes to an increase in blood glucose in Ptp4a1 -/mice fed an HF diet compared to controls, we performed the pyruvate tolerance test (PTT) and glycerol tolerance test (GlyTT) in WT and Ptp4a1 -/-mice fed an HF diet. After starvation, both groups were injected with pyruvate or glycerol as a substrate for glucose production. The blood glucose levels were comparable between the two groups in PTT and GlyTT analyses ( Figure S4A-B). In addition, the mRNA levels of gluconeogenesis, including Foxo1, Pck1, and G6pc, were comparable between the two groups ( Figure S4C). Next, we performed the hyperinsulinemic-euglycemic clamp study and 2-deoxyglucose (2-DG) uptake assay in WT and Ptp4a1 -/mice fed an HF diet. Basal hepatic glucose production (HGP), clamp HGP, and glucose infusion rate were comparable between the two groups; however, glucose uptake tended downward in Ptp4a1 -/mice compared to WT fed an HF diet in the hyperinsulinemic-euglycemic clamp study ( Figure  S4D). In the 2-DG uptake assay, we verified that deficiency of PTP4A1 in mice fed an HF diet decreased the levels of 2-DG in livers but not in eWATs and skeletal muscles ( Figure 2A).
As hepatic steatosis induces a decrease in plasma membrane GLUT2 levels in the liver of mice fed an HF diet has been reported [16], we verified whether the levels of GLUT2 on the plasma membrane of hepatocytes were reduced by feeding an HF diet compared to an NC diet in mice. Primary hepatocytes revealed that plasma membrane GLUT2 was markedly decreased in the hepatocytes of mice fed an HF diet compared to those fed an NC diet ( Figure  S4E). In line with increased hepatic TG levels in Ptp4a1 -/mice fed an HF diet, isolated primary hepatocytes of Ptp4a1 -/mice showed a higher lipid accumulation than those of WT mice after feeding an HF diet ( Figure 2B and S4F). Along with the increased lipid accumulation, deficiency of PTP4A1 in hepatocytes of mice fed an HF diet lessened the levels of GLUT2 on the plasma membrane ( Figure 2C and S4F) and the uptake of 2-(N-(7-Nitrobenz-2-oxa-1,3-diazol-4-yl)Amino)-2-Deoxyglucose (2-NBDG), a fluorescent glucose analog ( Figure 2D). In addition, an inverse correlation between fat accumulation and surface GLUT2 levels in hepatocytes was verified ( Figure S4G). Overexpression of PTP4A1 in Hep3B reduced the lipid accumulation after incubation with BSA-OA compared to controls, leading to augmentation of the uptake of 2-NBDG more than controls ( Figure 2E-F). In contrast, PTP4A1 down-regulation in Hep3B by PTP4A1-specific short hairpin RNA (shRNA)-expressing lentivirus decreased the uptake of 2-NBDG after incubation with BSA-OA compared to controls ( Figure S4H). The representative images for CD45 positive cells in the liver of WT and Ptp4a1 -/-mice fed a normal chow (NC) or an HF diet (n = 6). Scale bar, 300 μm. The magnified images in the red dot boxes are presented on the right. (L) The quantitative graph for CD45 positive colonies per field in the liver of WT and Ptp4a1 -/-mice fed an HF diet for 12 weeks (n = 6). Data are presented as the mean ± standard error of the mean. *P < 0.05, **P < 0.01, n.s., not significant (Mann-Whitney U test for B-E, G, I, and J, two-way ANOVA for F and L). Representative images (left) and quantification for relative fluorescence intensity (RFI, right) (n = 4). Scale bar, 500 μm. Data are presented as the mean ± standard error of the mean. **P < 0.01, n.s., not significant (Mann-Whitney U test for A; twotailed Student's t-test for F).

PTP4A1 regulates the expression of FGF21 through an increase in CREBH activity
To investigate how PTP4A1 deficiency in mice induces hepatic steatosis, we analyzed the levels of transcripts involved in lipid metabolisms. The mRNA levels involved in fatty acids oxidation-related genes such as Pparα, Acox1, Cpt1a, Cpt2, Acadvl, and Acadm were significantly decreased; however, the mRNA levels associated with lipogenic genes were not increased in the liver of Ptp4a1 -/mice fed an HF diet compared to controls ( Figure S5A-B). Interestingly, the mRNA levels of Fgf21 in the liver of Ptp4a1 -/mice fed an HF diet were significantly mitigated, and plasma FGF21 levels were also diminished in Ptp4a1 -/mice fed an HF diet compared to WT mice ( Figure  3A-B). Adenoviral overexpression of PTP4A1 in human hepatoma cells increased the mRNA levels of FGF21 compared to controls ( Figure 3C), and shRNA-mediated PTP4A1 knock-down in human hepatoma cells reduced the mRNA levels of FGF21 compared to controls ( Figure 3D). As FGF21 expression is regulated by transcription factor PPARα and/or CREBH, we tested whether PTP4A1 physi-cally interacts with PPARα or CREBH in human embryonic kidney 293T (HEK293T) cells. By co-immunoprecipitation, we identified that PTP4A1 physically interacted with the active form of CREBH (CREBH(N)) and the inactive form of CREBH (CREBH(F)) but not PPARα ( Figure 3E and S5C). PTP4A1 synergistically increased CREBH-mediated luciferase activity in an FGF21-luciferase reporter assay ( Figure 3F). However, mutation of PTP4A1 at the residues of 72 (D to A) and 104 (C to S) amino acids did not increase CREBH-mediated luciferase activity in an FGF21-luciferase reporter assay because mutant of PTP4A1 failed to interact with CREBH ( Figure 3E-F). To verify the molecular mechanism for the modulatory role of PTP4A1 in the CREBH-FGF21 axis, we first confirmed the levels of CREBH in the liver of WT and Ptp4a1 -/mice. The mRNA levels of the Creb3l3 gene (encoding CREBH) were comparable between the liver of WT and Ptp4a1 -/mice ( Figure  S5D). To confirm the protein levels of CREBH(N) in the liver, we isolated the nuclear fraction and identified the similar levels of CREBH(N) between the two groups ( Figure 3G). Next, we tested whether PTP4A1 may increase the CREBH binding activity on the FGF21 promoter region by chromatin immunoprecipitation (ChIP) assay. Consistent with the luciferase reporter gene assay, we verified that PTP4A1 could enhance the activity of CREBH on the FGF21 gene expression ( Figure 3H).

PTP4A1 deficiency in mice disturbs FGF21 expression, blood glucose, and hepatic TG in fasting conditions
As FGF21 expression is induced by fasting, we investigated the levels of FGF21 in WT and Ptp4a1 -/mice after feeding and fasting. The plasma FGF21 levels were comparable between two groups fed an NC diet; however, those were significantly mitigated in Ptp4a1 -/mice after fasting compared to WT mice ( Figure 4A). The blood glucose levels were higher in Ptp4a1 -/mice than WT mice after fasting but not fed an NC diet ( Figure 4B). However, plasma insulin levels of the two groups were comparable ( Figure 4C). After fasting, Ptp4a1 -/mice revealed impaired glucose tolerance and insulin sensitivity by GTT and ITT, respectively ( Figure 4D-E). Consistently, the levels of hepatic TG were significantly augmented in Ptp4a1 -/mice compared to WT mice by fasting; however, those were comparable between the two groups after feeding and refeeding an NC diet ( Figure  4F-G). During feeding an NC diet from 8 to 22 weeks old, the BW of WT and Ptp4a1 -/mice were comparable, and the ratios of eWAT/BW and liver/BW were not different ( Figure S6A-C).  . Data are presented as the mean ± standard error of the mean. *P < 0.05, **P < 0.01, n.s., not significant (two-way ANOVA for A-C, D (left), E (left) and G; Mann-Whitney U test for D (right) and E (right)).

Liver-specific PTP4A1 or systemic FGF21 overexpression ameliorates hyperglycemia and NAFLD in Ptp4a1 -/-mice fed an HF diet
To evaluate the role of PTP4A1 in the liver tissue, Ptp4a1 -/mice were administrated with adeno-associated virus (AAV)-aat-control (Ctrl) or AAV-aat-Ptp4a1 under the control of the liver-specific promoter alpha-1-antitrypsin (aat) and were fed an HF diet for 12 weeks ( Figure S7A). The liver-specific expression of PTP4A1 in mice injected by AAV-aat-Ptp4a1 was validated by western blot analysis for Flag-tag ( Figure  S7B). After an HF diet for 12 weeks, Ptp4a1 -/mice injected AAV-aat-Ctrl and AAV-aat-Ptp4a1 showed similar BW, fat mass, and lean mass (Figure 5A-B). PTP4A1 in the liver of Ptp4a1 -/mice injected with AAV-aat-Ptp4a1 was highly expressed (Figure 5C), and the mRNA levels of hepatic Fgf21 and plasma FGF21 levels were increased ( Figure 5D). AAV-aat-Ptp4a1 administration diminished blood glucose and plasma insulin levels compared to AAV-aat-Ctrl in Ptp4a1 -/mice fed an HF diet for 12 weeks ( Figure 5E). The glucose tolerance was markedly improved, and the insulin sensitivity was slightly enhanced by administration of AAV-aat-Ptp4a1 in Ptp4a1 -/mice fed an HF diet (Figure 5F and S7C). By histological analyses, we verified that AAV-aat-Ptp4a1 significantly reduced lipid accumu-lation in the liver of Ptp4a1 -/mice fed an HF diet ( Figure 5G-H). Finally, the decreased levels of ALT and AST in plasma of Ptp4a1 -/mice injected with AAV-aat-Ptp4a1 were identified compared to controls after feeding an HF diet for 12 weeks ( Figure 5I).
Next, we tested whether FGF21 overexpression in Ptp4a1 -/mice could ameliorate an HF diet-induced hyperglycemia and NAFLD. Ptp4a1 -/mice fed an HF diet for 11 weeks were administrated with adenovirus (Ad)-Ctrl or Ad-Fgf21 through the tail vein and kept both groups on an HF diet for an additional one week ( Figure 6A). The levels of hepatic Fgf21 mRNA and plasma FGF21 were markedly increased in the Ad-Fgf21 group compared to the Ad-Ctrl group ( Figure 6B). Overexpression of FGF21 reduced blood glucose levels and plasma insulin levels in Ptp4a1 -/mice fed an HF diet ( Figure 6C). In addition, Ad-Fgf21 administration decreased the accumulation of lipids in the liver (Figure 6D-E).

AAV-aat-Ptp4a1 delivery ameliorates HF diet-induced hyperglycemia and NAFLD in WT mice
Next, we investigated whether hepatic PTP4A1 overexpression by AAV-aat-Ptp4a1 could reveal metabolic benefits, including improved glucose homeostasis and NAFLD in WT mice fed an HF diet. After an HF diet for 12 weeks, DEXA analysis revealed that WT mice injected with AAV-aat-Ctrl and AAV-aat-Ptp4a1 showed similar BW, fat mass, and lean mass (Figure 7A-B). AAV-aat-Ptp4a1 administration highly expressed Flag-PTP4A1 in the liver tissue in WT mice fed an HF diet for 12 weeks ( Figure 7C). Hepatic PTP4A1 overexpression increased mRNA levels of hepatic Fgf21 and plasma FGF21 ( Figure 7D). AAV-aat-Ptp4a1 administration diminished blood glucose and plasma insulin levels compared to AAV-aat-Ctrl in WT mice after feeding an HF diet for 12 weeks (Figure 7E). Improved glucose tolerance was identified in AAV-aat-Ptp4a1 injected WT mice compared to AAV-aat-Ctrl injected mice; however, the insulin sensitivity was comparable between the two groups ( Figure 7F and S8). AAV-aat-Ptp4a1 significantly reduced lipid accumulation in the liver of WT mice fed an HF diet without the difference in the levels of plasma ALT and AST (Figure 7G-I).  . (A and B) Representative images (A) and the graphs (B) for body weight, fat, and lean mass of Ptp4a1 -/-mice administrated with adeno-associated virus (AAV)-aat-control (Ctrl) or AAV-aat-Ptp4a1 after feeding an HF diet for 12 weeks in dual-energy X-ray absorptiometry analysis (n = 5). (C) Immunoblot analysis in the liver lysates of Ptp4a1 -/-mice administrated with AAV-aat-Ctrl or AAV-aat-Ptp4a1. HSP90 was used as a loading control. (D) The levels of Fgf21 mRNA in the liver and FGF21 in plasma of Ptp4a1 -/-mice administrated with AAV-aat-Ctrl or AAV-aat-Ptp4a1 after feeding an HF diet (n = 5). (E) Blood glucose levels (left) and plasma insulin levels (right) in Ptp4a1 -/-mice administrated with AAV-aat-Ctrl or AAV-aat-Ptp4a1 after feeding an HF diet (n = 5). (F) Glucose tolerance test (GTT) and the area under the curve (AUC) of GTT in Ptp4a1 -/-mice administrated with AAV-aat-Ctrl or AAV-aat-Ptp4a1 after feeding an HF diet (n = 5). (G) The representative images for hematoxylin&eosin (H&E) staining and oil red O staining in the liver sections of Ptp4a1 -/-mice administrated with AAV-aat-Ctrl or AAV-aat-Ptp4a1 after feeding an HF diet (n = 4). Scale bar, 200 μm. (H) Hepatic triglyceride (TG) levels in Ptp4a1 -/-mice administrated with AAV-aat-Ctrl or AAV-aat-Ptp4a1 after feeding an HF diet (n = 6). (I) The levels of alanine aminotransferase (ALT) and aspartate transaminase (AST) in plasma of Ptp4a1 -/-mice administrated with AAV-aat-Ctrl or AAV-aat-Ptp4a1 after feeding an HF diet (n = 5). Data are presented as the mean ± standard error of the mean. *P < 0.05, **P < 0.01, n.s., not significant (Mann-Whitney U test for B, D (right), E, F (right), H, and I; two-tailed Student's t-test for D (left); two-way ANOVA for F (left)).

Figure 6. FGF21 overexpression in Ptp4a1 -/-mice decreases a high-fat (HF) diet-induced hyperglycemia and NAFLD. (A) The experimental design for administering adenovirus on HF diet-fed Ptp4a1 -/-mice. (B)
The levels of Fgf21 mRNA in the liver and FGF21 in plasma of Ptp4a1 -/-mice administrated with Ad-Ctrl or Ad-Fgf21 after feeding an HF diet (n = 7). (C) Blood glucose levels (left) and plasma insulin levels (right) in Ptp4a1 -/-mice administrated with Ad-Ctrl or Ad-Fgf21 after feeding an HF diet (n = 7). (D) The representative images for hematoxylin&eosin (H&E) staining and oil red O staining in the liver sections of Ptp4a1 -/-mice administrated with Ad-Ctrl or Ad-Fgf21 after feeding an HF diet (n = 7). Scale bar, 200 μm. (E) Hepatic triglyceride (TG) levels in Ptp4a1 -/-mice administrated with Ad-Ctrl or Ad-Fgf21 after feeding an HF diet (n = 7). Data are presented as the mean ± standard error of the mean. **P < 0.01 (two-tailed Student's t-test for B (left); Mann-Whitney U test for B (right), C, and E).

Discussion
NAFLD is, to date, the most common chronic liver disease leading to a significant health problem worldwide, which affects up to 30% of adults in the general population and 70% of patients with type 2 diabetes [30,31]. Despite understanding the epidemiology and the pathogenic mechanisms in the progress of NAFLD, there are no approved pharmacological therapies for NAFLD [32], providing the need for a novel therapeutic target for the treatment of NAFLD. In the present study, we provided PTP4A1 as a novel regulator in the pathogenesis of NAFLD. PTP4A1 ameliorated hepatosteatosis through the activation of the transcription factor CREBH/FGF21 axis. Liver-specific PTP4A1 expression, followed by augmentation of FGF21 levels, reduced HF diet-induced hepatosteatosis and hyperglycemia in WT mice. Thus, controlling PTP4A1 may be a potential candidate for therapeutic strategy against hepatosteatosis-related diseases.
Protein tyrosine kinases and phosphatases control the phosphorylation of tyrosine residues within proteins that regulate physiological signaling cascades in cells [33]. Disruption of the regulation of protein phosphorylation is closely associated with various diseases, including cancers, autoimmune diseases, cardiovascular diseases, and metabolic diseases [34,35]. Recently, protein tyrosine phosphatase 1B (PTP1B), a classical non-transmembrane tyrosine phosphatase, is emerging as a critical regulator in type 2 diabetes, obesity, and liver diseases, suggesting a promising therapeutic target in metabolic disorders [36]. PTP1B-null mice resisted Fas-induced extensive hepatocyte apoptosis and promoted hepatocyte proliferation in response to liver damage [36]. Thus, PTP1B deficiency in mice revealed a rapid and synchronized compensatory liver regeneration after partial hepatectomy [37]. In contrast, PTP4A1 is significantly induced during liver regeneration and is required for proper timing of liver regeneration after partial hepatectomy with modulating expression of cell cycle regulators [10]. PTP4A1 constitutes a unique subfamily of protein tyrosine phosphatases and plays a pivotal role during cell development and tissue regeneration [33]. Interestingly, both PTP1B and PTP4A1 localized to the endoplasmic reticulum (ER) in non-mitotic cells [38]. Based on the opposite phenotype and co-localization in the subcellular compartment between PTP1B and PTP4A1, the beneficial potential of PTP4A1 in metabolic disorders such as NAFLD might be considerable. It may be necessary to test whether PTP4A1 could antagonize PTP1B in various cellular responses.
A previous report verified that PTP4A1 physically interacted with the ATF5/7 containing a basic helix-loop-helix leucine zipper (bZIP) domain [8]. In the current study, we proved that PTP4A1 physically interacted with CREBH, a transcription factor containing the bZIP domain, and enhanced the transcriptional activity of CREBH on the Fgf21 gene. PTP4A1, but not in mutant PTP4A1, synergistically increased CREBH-mediated luciferase activity in an FGF21-luciferase reporter assay. A previous report revealed that proteolytic cleavage of CREBH was modulated by glycogen synthase kinase 3β-mediated phosphorylation [39]. As PTP4A1 is localized in ER, plasma membrane, and nucleus, PTP4A1 may enhance both the transcriptional activity of CREBH in the nucleus and proteolytic cleavage of CREBH in ER. Indeed, we verified the synergistic effects in CREBH(N) or CREBH(F) (data not shown) with PTP4A1 in the FGF21 luciferase assay and confirmed interaction in CREBH(N) or CREBH(F) with PTP4A1. Consistent with a previous report [40], a decreased CREBH activity in Ptp4a1 -/mice may result in the down-regulation of fatty acid oxidation-related gene expressions, including Pparα and Cpt1a. In a future study, the identification of the PTP4A1-mediated dephosphorylation site of CREBH would be evaluated to understand detailed mechanisms.

Figure 7. Liver-specific PTP4A1 overexpression reduces hyperglycemia and NAFLD after feeding a high-fat (HF) diet in wild-type (WT) mice. (A and B)
Representative images (A) and the graphs (B) for body weight, fat, and lean mass of WT mice administrated with adeno-associated virus (AAV)-aat-control (Ctrl) or AAV-aat-Ptp4a1 after feeding an HF diet for 12 weeks in dual-energy X-ray absorptiometry analysis (n = 5). (C) Immunoblot analysis in the liver lysates of WT mice administrated with AAV-aat-Ctrl or AAV-aat-Ptp4a1. HSP90 was used as a loading control. (D) The levels of Fgf21 mRNA in the liver (n = 4) and FGF21 in plasma (n = 5) of WT mice administrated with AAV-aat-Ctrl or AAV-aat-Ptp4a1 after feeding an HF diet. (E) Blood glucose levels (left, n = 4) and plasma insulin levels (right, n = 5) in WT mice administrated with AAV-aat-Ctrl or AAV-aat-Ptp4a1 after feeding an HF diet. (F) Glucose tolerance test (GTT) and the area under the curve (AUC) of GTT in WT mice administrated with AAV-aat-Ctrl or AAV-aat-Ptp4a1 after feeding an HF diet (n = 5). (G) The representative images for hematoxylin&eosin (H&E) staining and oil red O staining in the liver sections of WT mice administrated with AAV-aat-Ctrl or AAV-aat-Ptp4a1 after feeding an HF diet (n = 4). Scale bar, 200 μm. (H) Hepatic triglyceride (TG) levels in WT mice administrated with AAV-aat-Ctrl or AAV-aat-Ptp4a1 after feeding an HF diet (n = 6). (I) The levels of alanine aminotransferase (ALT) and aspartate transaminase (AST) in plasma of WT mice administrated with AAV-aat-Ctrl or AAV-aat-Ptp4a1 after feeding an HF diet (n = 5). Data are presented as the mean ± standard error of the mean. *P < 0.05, n.s., not significant (Mann-Whitney U test for B, D (right), E, F (right), H, and I; two-tailed Student's t-test for D (left); two-way ANOVA for F (left)).
FGF21 is an inducible metabolic hormone by fasting or stress, produced mainly in the liver, and regulates glucose and lipid homeostasis via a heterodimeric receptor complex, FGF receptor 1 and β-klotho [41]. FGF21 derivatives or FGF21 receptor agonists have been tried as therapeutic agents for various metabolic diseases, including type 2 diabetes, obesity, and NAFLD, for the past decade [21]. However, the limitations of FGF21 as a clinically valuable medicine include its poor pharmacokinetics and biophysical properties [41]. The current study provided an option to adopt FGF21 therapy in metabolic diseases. Although the oncogenic property of PTP4A1 overexpression and the restricted beneficial effects of PTP4A1-mediated FGF21 remain a hurdle to developing a therapeutic target, PTP4A1-mediated CREBH activation might be considered one of the strategies to elevate FGF21 levels in metabolic disorders, including NAFLD.
In the current study, we identified that Ptp4a1 -/mice fed an HF diet revealed hyperglycemia and fatty liver compared to controls. In GTT and ITT analysis, Ptp4a1 -/mice fed an HF diet showed impaired glucose tolerance and insulin sensitivity. However, hepatic glucose output in the clamp study and glucose productions in the pyruvate and glycerol tolerance test were similar between WT and Ptp4a1 -/mice after feeding an HF diet. The mRNA levels of gluconeogenic genes were not different between the two groups. In addition, glucose uptakes by skeletal muscle and eWAT in the 2-DG uptake assay were similar between the two groups. The levels of phosphorylated Akt in the liver, skeletal muscle, and eWAT were comparable between the two groups (data not shown). Therefore, we could not conclude that PTP4A1 depletion in mice fed an HF diet develops insulin resistance. Instead of insulin resistance, we guessed that PTP4A1 deficiency in mice induced fat accumulation in the liver by feeding an HF diet or fasting, followed by decreased GLUT2 translocation on the hepatocyte surface, leading to hyperglycemia.
Previous reports have shown an HF diet-induced GLUT2 internalization in rat and mouse hepatocytes [16,42]. Also, hepatic GLUT2 location is regulated by feeding states; GLUT2 levels with the insulin receptor in plasma membrane fractions from rat livers are down-regulated after feeding [43]. In the current study, we also verified that increased fat accumulation in hepatocytes reduced the levels of GLUT2 on the plasma membrane. Primary hepatocytes from Ptp4a1 -/mice revealed significant differences in fat accumulation and surface GLUT2 levels compared to WT controls. Although increased fat content is inversely correlated with surface GLUT2 levels in hepatocytes, the detailed mechanism of down-regulating the GLUT2 levels on the plasma membrane of hepatocytes by PTP4A1 deficiency will be clarified in a further study.
In summary, our results establish hepatic PTP4A1-mediated activation of the CREBH/FGF21 axis as a novel therapeutic strategy in NAFLD and hyperglycemia. Lacking PTP4A1 in mice fed an HF diet exhibited hepatosteatosis and hyperglycemia, which were recovered by liver-specific PTP4A1 or systemic FGF21 overexpression. Finally, we proved that liver-specific PTP4A1 expression ameliorated HF diet-induced hepatosteatosis and hyperglycemia in WT mice. Therefore, modulating hepatic PTP4A1 may be a potential therapeutic target against hepatosteatosis-related diseases.

Animal experiments
This study followed the Guidelines on the Care Use of Laboratory Animals (National Institutes of Health Publication no. 85-23, revised 1996). Animal study protocols were approved by the Institutional Animal Care and Committee of the Korea Research Institute of Bioscience and Biotechnology (KRIBB-AEC-18203). Ptp4a1 -/mice were generated using the CRISPR/Cas9 system and backcrossed onto a C57BL/6 background. The primers used for Ptp4a1 genotyping were as follows: forward 5'-TCCGCAG GCTGCCTCCTCTC-3' and reverse 5'-CAGAACA GTAGCAACAAAAT-3'. Mice were kept in a controlled environment with a 12-hour light/dark cycle in a specific pathogen-free facility and fed an NC diet and water for the study. To investigate the effects of PTP4A1 on hepatic steatosis, 8-week-old male mice were given an HF diet (60% calories from fat; D12492, Research DIET) for 12 weeks. Mouse body composition was identified using DEXA (Lunar, GE Lunar Corp.). Body fat and the bone area were demonstrated with a DEXA scan using a dedicated densitometer. After the study, animals were anesthetized using isoflurane inhalation (3%) plus 1 L/min O2 and euthanized by exsanguination.

Measurement of metabolites
Plasma insulin was measured using a mouse insulin enzyme-linked immunosorbent assay (ELISA) kit (Crystal Chem). Plasma AST and ALT were analyzed using a colorimetric assay kit (BioVision). Plasma FGF-21 was measured using a mouse/rat FGF-21 ELISA kit (R&D systems), and hepatic glucose was determined using a glucose assay kit (Sigma). Hepatic TG was measured by Triglyceride Determination Kit (Sigma). 2-DG uptake of mice was analyzed using a 2-DG uptake measurement kit (Cosmo Bio Co.). Blood glucose levels were determined from tail vein blood using an Accu-Check Active blood glucose meter (Roche).

Tolerance test
Mice were starved for 16 hours and then given an intraperitoneal injection with glucose (1 g/kg BW on HF diet, 2 g/kg BW on NC diet, Sigma), pyruvate (1 g/kg BW on HF diet, 1.5 g/kg BW on NC diet, Sigma), or glycerol (1 g/kg BW on HF diet, 2 g/kg BW on NC diet, Sigma). Mice were starved for six hours and then injected intraperitoneally with mouse insulin (1 unit/kg BW on HF and NC diet, Sigma). Blood glucose levels were measured from tail vein blood collected at the designated times.

Virus
Liver-specific adeno-associated virus (AAV8.2-hAAT-mFlagPTP4A1-pA) was purchased from Sirion (Martinsried, Germany). For PTP4A1 overexpression in the liver, mice were injected with a concentration of 2.0×10 11 vg AAV per mouse through the tail vein. PTP4A1 expression levels were determined by immunoblot analysis.
Recombinant adenoviruses expressing FGF21 or PTP4A1 were generated using the AdEasy adenoviral vector system described previously [44]. Recombinant adenoviruses were expanded in HEK 293AD cells and purified using an Adeno-X Maxi Purification kit (631533, Clontech). For animal experiments, mice were injected with recombinant adenovirus (1×10 8 plaque-forming unit per mouse) intravenously through the tail vein. Adenovirus-mediated gene expressions were analyzed by quantitative real time polymerase chain reaction (qRT-PCR) and ELISA.

Luciferase reporter gene assay
Hep3B cells were seeded in 48-well plates (4 × 10 4 cells/well) using DMEM and 10% FBS. Cells were transfected with 50 ng of luciferase reporter, 100 ng of effector, and 5 ng of pRL-TK (Renilla TK) plasmids using 0.5 µl of the transfection reagent Lipofectamine 2000 (Invitrogen). Cells were lysed and moved into 96-well plates 48 hours later for luciferase assays using the Dual-Luciferase Reporter Assay kit (E1960, Promega). Luciferase activity was measured on a fluorescence spectrophotometer (GloMax 96, Promega), and relative transcriptional activity was normalized by the Renilla activity.

Histology and Immunohistochemistry
Liver pieces were fixed in 10% (v/v) phosphate-buffered formalin solution overnight at room temperature and then placed in 30% sucrose overnight at 4 °C. The fixed tissues were embedded in paraffin or OCT (3801480, Leica) and sectioned. H&E staining was applied to frozen and paraffinembedded tissue sections. Oil-red O staining was used on frozen tissue sections. After staining, images were captured under a light microscope (BX53F2, Olympus Corp).

RNA isolation and qRT-PCR
Total RNA was isolated using TRIZOL Reagents (15596026, Life Technologies) according to the manufacturer's recommendation. Complementary DNA was synthesized from target RNA using the M-MLV Reverse Transcriptase kit (N1705, C1101, N2515, U1518; Promega). The qRT-PCR was performed using the quantiMix SYBR kit (QS105; PKT Co.). The mRNA expression levels were normalized using human GAPDH or mouse 18s expression. The qRT-PCR primers are listed in Supplementary Table 1.

Statistics
Data are expressed as the mean ± standard error of the mean. We subjected two-tailed Student's t-tests, one-way ANOVA, two-way ANOVA, the Mann-Whitney U test, and linear regression analysis to test statistical significance where appropriate. Linear correction index R square and P value were calculated. Statistical tests are described in Figure  Legends for each experiment. P values less than 0.05 were considered significant.