HIF-P4H-2 inhibition enhances intestinal fructose metabolism and induces thermogenesis protecting against NAFLD

Abstract Non-alcoholic fatty liver disease (NAFLD) parallels the global obesity epidemic with unmet therapeutic needs. We investigated whether inhibition of hypoxia-inducible factor prolyl 4-hydroxylase-2 (HIF-P4H-2), a key cellular oxygen sensor whose inhibition stabilizes HIF, would protect from NAFLD by subjecting HIF-P4H-2-deficient (Hif-p4h-2gt/gt) mice to a high-fat, high-fructose (HFHF) or high-fat, methionine-choline-deficient (HF-MCD) diet. On both diets, the Hif-p4h-2gt/gt mice gained less weight and had less white adipose tissue (WAT) and its inflammation, lower serum cholesterol levels, and lighter livers with less steatosis and lower serum ALT levels than the wild type (WT). The intake of fructose in majority of the Hif-p4h-2gt/gt tissues, including the liver, was 15–35% less than in the WT. We found upregulation of the key fructose transporter and metabolizing enzyme mRNAs, Slc2a2, Khka, and Khkc, and higher ketohexokinase activity in the Hif-p4h-2gt/gt small intestine relative to the WT, suggesting enhanced metabolism of fructose in the former. On the HF-MCD diet, the Hif-p4h-2gt/gt mice showed more browning of the WAT and increased thermogenesis. A pharmacological pan-HIF-P4H inhibitor protected WT mice on both diets against obesity, metabolic dysfunction, and liver damage. These data suggest that HIF-P4H-2 inhibition could be studied as a novel, comprehensive treatment strategy for NAFLD. Key messages • HIF-P4H-2 inhibition enhances intestinal fructose metabolism protecting the liver. • HIF-P4H-2 inhibition downregulates hepatic lipogenesis. • Induced browning of WAT and increased thermogenesis can also mediate protection. • HIF-P4H-2 inhibition offers a novel, comprehensive treatment strategy for NAFLD. Electronic supplementary material The online version of this article (10.1007/s00109-020-01903-0) contains supplementary material, which is available to authorized users.


Introduction
The prevalence of non-alcoholic fatty liver disease (NAFLD) is constantly increasing, currently affecting a quarter of people worldwide [1]. It is considered the hepatic manifestation of metabolic syndrome, strongly linked to obesity and insulin resistance (IR), which further predisposes for diabetes and cardiovascular diseases [2]. NAFLD is characterized by hepatic triglyceride accumulation and if untreated, can lead to steatohepatitis (NASH), cirrhosis, and hepatocellular carcinoma (HCC) [1]. Western diet, rich in saturated fat and carbohydrates, especially fructose, is considered to be one of the major causes of the NAFLD epidemic [3]. No effective cure is currently available [4].
The first-in-class small-molecule HIF-P4H inhibitor that activates the hypoxia response has been approved for treatment of renal anemia and several others are in clinical trials [10,11]. Recent data suggest that besides anemia, HIF-P4H inhibition and hypoxia are powerful tools for promoting metabolic health [10,[12][13][14][15][16][17]. We have previously shown that genetic HIF-P4H-2 deficiency protects the Hif-p4h-2 gt/gt mice against metabolic disorder-related hepatic steatosis and chemically induced HCC [14,18], and against alcoholic liver disease (ALD) by downregulating hepatic lipogenesis and improving the elimination of harmful ethanol metabolites and reactive oxygen species [19]. Treatment of wild-type (WT) mice with a pan-HIF-P4H inhibitor FG-4497 phenocopied the protection against ALD [19]. However, the role of HIF-P4H-2 in NAFLD has not been studied systemically before. We therefore subjected the Hif-p4h-2 gt/gt mice, and FG-4497-treated WT mice, to two diet-induced rodent NAFLD models: a high-fat, high-fructose (HFHF) diet [20] and a methionine-choline-deficient high-fat diet (HF-MCD) [21] that mimic the human disease. The HF-MCD can additionally lead to NASH [22]. Our data show significantly less steatosis and liver damage in the Hif-p4h-2 gt/gt mice compared with the WT. Treatment with FG-4497 phenocopied most of these effects.
Further methods are described in the supplementary material.

Results
HIF-P4H-2-deficient mice were protected from fructose diet-induced weight gain, but the diet did not induce NAFLD Hif-p4h-2 gt/gt mice and their WT littermates were fed a standard rodent diet with a 30% fructose solution for 16 weeks. Although the daily intake of the fructose solution was similar between the genotypes (Fig. S1a), the Hif-p4h-2 gt/gt mice had a~20% lower body weight than the WT at sacrifice, had gained less weight during the diet, and had > 50% less gonadal white adipose tissue (WAT) (Fig. S1b-d). The Hif-p4h-2 gt/gt mice also showed a slight trend towards better glucose tolerance compared with the WT (Fig. S1e). The Hif-p4h-2 gt/gt livers were 21% lighter than the WT livers (Fig. S1f), suggesting more fructose-induced hepatic steatosis in the WT, since no baseline difference exits between the genotypes [14]. However, the diet only induced visible steatosis and increased the serum alanine aminotransferase (ALT) levels in a few WT mice, and no liver inflammation in either genotype ( Fig. S1g-i), thus not being potent enough to induce NAFLD. Hence, we next combined the 30% fructose solution with a high-fat diet (HFHF) to better mimic Western diet.
HIF-P4H-2-deficient mice were protected from obesity and retained a healthier serum lipid profile than the WT on a HFHF diet The Hif-p4h-2 gt/gt and WT mice were fed the HFHF diet for 8 weeks, during which the former did not gain any weight, whereas the latter's body weight increased by~10%, resulting in an almost 30% higher body weight at sacrifice (Fig. 1a-c). Consistently with this, the Hif-p4h-2 gt/gt mice had 40% less WAT and also less brown adipose tissue (BAT) than the WT (Fig. 1d, e), as well as smaller adipocytes (Fig. 1f). Additionally, the Hif-p4h-2 gt/gt mice had less inflammatory macrophage aggregates in their WAT than the WT (Fig. 1g), and furthermore, their serum leptin levels were lower (Fig. 1h). The HFHF diet induced elevation of the serum total cholesterol, HDL, and LDL levels in the WT, which the Hif-p4h-2 gt/gt mice were protected against (Fig. 1i). There were no differences in serum triglyceride and free fatty acid (FFA) levels between the genotypes (Fig. 1j).
HIF-P4H-2-deficient mice had lower hepatic fructose levels and better glucose tolerance on the HFHF diet than the WT We next studied the 14 C-labeled fructose uptake by tissues 15 min after oral administration. Majority of radioactivity was found in the intestine in both genotypes (Fig. 3a) while no difference was seen in its secretion to feces or urine between the genotypes (Fig. S3a). Interestingly, the 14 C levels measured in most tissues were 15-35% lower in the Hif-p4h-2 gt/gt mice than in the WT, the reduction in the liver, kidney, and serum reaching significance, unlike the induction of~20% in the intestine, pancreas, and BAT (Fig. 3a). The sum of radioactivity in all tissues was similar in both genotypes (Fig. 3b), but the sum excluding the intestine was significantly lower in the Hif-p4h-2 gt/gt mice, suggesting that the lower amount of fructose found in the liver, kidney, and serum would correspond to increased uptake by the Hif-p4h-2 gt/gt intestine (Fig. 3a). The small intestine metabolizes most of dietary fructose ketohexokinase (KHK) dependently, and if its clearance capacity is exceeded by high fructose concentration, the excess fructose spills over into the liver and other organs, as reported previously [26]. Therefore, we studied the mRNA levels of KHK isoenzymes and fructose transporters and found upregulation of Slc2a2 (GLUT2) mRNA levels in the Hif-p4h-2 gt/gt small intestine relative to the WT, the downregulation of Hif-p4h-2 mRNA being 65% in the intestine in the former (Fig. 3c). However, no difference was seen in the immunostaining of GLUT2 (Fig.  S3b). Interestingly, the Khka and Khkc mRNA levels were significantly higher in the Hif-p4h-2 gt/gt small intestine than in the WT (Fig. 3c) as was the measured catalytic KHK activity (Fig. 3d). Altogether, these data suggest enhanced fructose metabolism in the Hif-p4h-2 gt/gt intestine, which would result in less fructose being transported to the liver and other tissues.
Furthermore, the HFHF-fed Hif-p4h-2 gt/gt mice maintained a better glucose tolerance than the WT and had lower serum insulin levels and HOMA-IR scores, indicating better insulin sensitivity ( Fig. 3e-g). A similar reduction was seen in circulating levels of the harmful byproducts of fructose metabolism, uric acid, and lactate (Fig. 3h, i), supporting differences in its metabolism between genotypes. This, and the lower lactate/glucose ratio in the Hif-p4h-2 gt/gt mice (Fig. 3j), agrees with their previously characterized enhanced lactate clearance [27]. The hepatic Fig. 3 Changes in fructose metabolism associated with protection from high-fat, high-fructose (HFHF) diet-induced NAFLD in the HIF-P4H-2deficient mice. Wild-type (wt) and Hif-p4h-2 gt/gt (gt/gt) males after e-g 6 or c, h-k 8 weeks on a HFHF diet (n = 8-10/group). a, b 14 C-fructose uptake test. After 2 weeks on the HFHF diet and a 12-h fast, wt and gt/gt females received 14 C-fructose orally and were sacrificed after 15 min and tissue samples were measured for 14 C radioactivity as disintegration/min/mg (DPM) (n = 12/group). a DPM per mg of tissue in indicated tissues. b Sum radioactivity detected in all tissues measured including and excluding the small intestine. c qPCR analysis of the mRNA levels of glucose and fructose transporters and metabolizing enzymes in the small intestine of the gt/gt mice relative to wt. Gene expression was studied relative to β-actin mRNA. d Ketohexokinase activity in the small intestine of females after 4 weeks on a HFHF diet (n = 6-8/group). e Oral glucose tolerance test (GTT). The value for 0 min was determined after a 12-h fast. f Serum insulin levels determined from the 0-min GTT samples. g HOMA-IR scores determined from the 0-min values. h Serum uric acid levels. i Blood lactate levels. j Blood lactate-to-glucose ratio. k qPCR analysis of liver mRNA levels of glucose and fructose transporters and metabolizing enzymes in gt/gt mice relative to wt, studied relative to TATA-box-binding protein mRNA. Data are means ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001. The p value for d was calculated from log-transformed values. b, blood; BAT, brown adipose tissue; HOMA-IR, homeostatic model assessment-insulin resistance; KHK, ketohexokinase; s, serum; SLC2A 1-5, solute carrier family-2 member 1-5; WAT, white adipose tissue mRNA level of the fructose-uptaking Slc2a2 was significantly lower in the Hif-p4h-2 gt/gt than in the WT, supporting the reduced hepatic 14 C-fructose uptake, while there was no difference between the genotypes in the levels of Slc2a5 or Khks (Fig. 3k). Of note, the expression level of Slc2a2 mRNA was~65-fold compared Slc2a5 (data not shown), being in agreement with the role of GLUT2 as the main hepatic fructose transporter.
HIF-P4H-2-deficient mice were protected from obesity and retained a healthier serum lipid profile when challenged with an HF-MCD diet Next, the Hif-p4h-2 gt/gt and WT mice were fed the HF-MCD diet, inducing a more severe form of NAFLD than the HFHF diet. The weight gain of the Hif-p4h-2 gt/gt mice was significantly lower than that of the WT during the 7-week diet, and at sacrifice, they had a 22% lower body weight (Fig. 4a-c). In agreement, the Hif-p4h-2 gt/gt mice had 45% less WAT than the WT with significantly smaller adipocytes and a trend towards lower BAT weight (Fig. 4d-f). Also, less macrophage aggregates were detected in their WAT (Fig. 4g). Additionally, the Hif-p4h-2 gt/gt mice retained a healthier serum lipid profile than the WT, their total cholesterol level being significantly lower (Fig. 4h), the same applying to their glucose levels (Fig. S4). No differences in serum triglyceride and FFA levels were detected between the genotypes on this diet, either (Fig. 1i).
Hif-p4h-2 gt/gt mice were protected against HF-MCD diet-induced steatosis and liver damage The liver weights of the Hif-p4h-2 gt/gt mice were > 20% lower than those of the WT after the HF-MCD diet (Fig. 5a). Consistently, the Hif-p4h-2 gt/gt livers had a significantly lower hepatic triglyceride content (Fig. 5b) and histological scores for steatosis severity,~35% of the Hif-p4h-2 gt/gt livers against 70% of WT scored as "very severe" (Fig. 5c). Only mild inflammation (Fig. 5d) and fibrosis (Fig. S5a) were detected to similar extents in both genotypes. Serum ALT and AST were elevated to a pathological level in both genotypes, but they were significantly~30% lower in the Hif-p4h-2 gt/gt than in the WT mice (Fig. 5e, f). Additionally, the serum albumin levels were significantly higher for the Hif-p4h-2 gt/gt mice than for the WT (Fig. 5g), and the Hif-p4h-2 gt/gt livers had more proliferating cells than the WT (Fig. 5h) with no difference in apoptosis (Fig. S5b), suggesting enhanced regeneration in the Hif-p4h-2 gt/gt livers. Both genotypes' livers showed similar dispersed expression of pericentral zonation marker glutamine synthetase (Fig. S5c). In agreement with less fat accumulation and damage detected in the Hif-p4h-2 gt/gt livers than in the WT, the hepatic expression levels of several lipid metabolism and oxidative stress mRNAs were lower, while the Hif-p4h-2 mRNA downregulation was similar to that on the HFHF diet (Figs. 5i and S5d). Additionally, the levels of all cholesterol synthesis precursors but squalene were reduced in the Hif-p4h-2 gt/gt liver compared with that in the Fig. 4 HIF-P4H-2-deficient mice are protected against HF-MCD dietinduced obesity, adipose tissue inflammation, and higher cholesterol levels. Wild-type (wt) and Hif-p4h-2 gt/gt (gt/gt) females were studied while on a 7-week high-fat, methionine-choline-deficient (HF-MCD) diet (n = 10-12/group). a Body weight development during the 7 weeks. b Body weight at sacrifice. c Weight gain at the end of the diet relative to weights on the day before the diet started. d Weight of gonadal WAT. e Cross-sectional area of WAT adipocytes. Scale bar = 100 μm. f Weight of BAT. g Number of macrophage aggregates in WAT. *Adipocytes surrounded by macrophage aggregates. Scale bar = 100 μm. h Serum total cholesterol, HDL cholesterol, and LDL cholesterol levels. i Serum TG and FFA levels (n = 6-9/group). Data are means ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001. BAT, brown adipose tissue; FFA, free fatty acids; s, serum; TG, triglycerides; WAT, white adipose tissue WT (Table S2), which suggests downregulated cholesterol synthesis in the former, likely contributing to the observed lower serum cholesterol levels (Fig. 4h). Additionally, serum uric acid levels were significantly lower in the Hif-p4h-2 gt/gt mice, while no differences were detected in the FGF21 levels (Fig. S5e, f). These data indicate that the Hif-p4h-2 gt/ gt mice showed protection against the HF-MCD dietinduced hepatic fat accumulation and damage compared with the WT.
Increased heat production in the HIF-P4H-2-deficient mice on an HF-MCD diet is associated with protection from NAFLD Browning of WAT has been shown to be a protective mechanism against NAFLD in mice on MCD diet [28,29]. Therefore, we next studied the expression of uncoupling protein 1 (UCP1) in the WAT of the HF-MCD diet-fed mice. Surprisingly, the Hif-p4h-2 gt/gt mice showed significantly more UCP1 expression (83.3%) than the WT (33.3%) (Fig. 6a), verified by Western blotting (Fig. 6b). No difference between the genotypes in BAT UCP1 levels was detected (data not shown). In agreement, the mRNA levels of the browning markers Ucp1, Ppara, Pparg, Pparg2, Cebpa, Prdm16, Vegfa, Ppargc1a, Acsl1, and Lipe were significantly upregulated in the Hif-p4h-2 gt/ gt WAT relative to the WT, the downregulation of Hif-p4h-2 mRNA being similar to that on normal chow [14] (Fig. 6c). Since UCP1 expression relates to heat production, we submitted the mice on the HF-MCD diet for indirect gas calorimetry analysis, which showed significantly increased heat production by the Hif-p4h-2 gt/gt mice compared with the WT (Fig. 6d). There was no difference between the genotypes in physical activity, Fig. 5 HIF-P4H-2 deficiency protects mice from HF-MCD diet-induced fatty liver disease. Wild-type (wt) and Hif-p4h-2 gt/gt (gt/gt) females were studied while on a 7-week high-fat, methionine-choline-deficient (HF-MCD) diet (n = 10-12/group). a Liver weight. b Hepatic triglyceride content. c Scoring of steatosis and H&E-stained liver sections. Steatosis grading: "Moderate" corresponds to score 2, "Severe" to 3, and "Very severe" to 4. "Very severe" and "severe" images are representative of scoring for wt, and "moderate" for gt/gt. Scale bar = 500 μm for × 4 and 200 μm for × 10. d Scoring of inflammation from H&E-stained liver sections. "No" corresponds to scores 0-1 and "Yes" to 2-4. Arrows indicate neutrophil clusters in wt. Scale bar = 100 μm. e Serum ALT levels. f Serum AST levels. g Serum albumin levels (n = 9-10/group). h Number of Ki67-positive proliferating cells in the liver sections. Arrows indicate Ki67-positive cells. Scale bar = 50 μm. i qPCR analysis of liver mRNA levels of gt/gt mice relative to wt, studied relative to TATA-boxbinding protein mRNA. a, b, e-i Data are means ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001. Alb, albumin; ALT, alanine aminotransferase; AST, aspartate aminotransferase; CHREBP, carbohydrate-responsive element-binding protein; GPX1, glutathione peroxidase 1; LPIN, lipin; MTTP, microsomal triglyceride transfer protein; PPARg2, peroxisome proliferator-activated receptor gamma 2; s, serum; TG, triglycerides respiratory exchange ratio, or food intake ( Fig. 6e-g), but an increase in water intake was observed in the Hif-p4h-2 gt/gt mice (Fig. 6h). These differences were not observed in previous measurements on normal chow [14], neither in the baseline measurements preceding the HF-MCD diet feeding, and no browning was detected in either genotype on the HFHF diet (data not shown). These data suggest that the Hif-p4h-2 gt/gt mice used advanced browning of WAT as a mechanism of protection against HF-MCD diet-induced NAFLD. Fig. 6 Heat production via UCP1 activation in the WAT is associated with protection from HF-MCD diet-induced NAFLD in HIF-P4H-2deficient mice. Wild-type (wt) and Hif-p4h-2 gt/gt (gt/gt) females were studied while on a 7-week high-fat, methionine-choline-deficient (HF-MCD) diet. a Scoring of UCP1-stained WAT sections. "No" corresponds to < 5% of UCP1 staining/field and "Yes" to 5-25% (n = 10-12/group). Images are representative of scoring for wt. Scale bar = 500 μm for × 4 and 200 μm for × 10. b Western blotting and densitometric quantification of WAT UCP1 levels (n = 3-4/group.). β-Actin was used as a loading control. c qPCR analysis of the mRNA levels of browning markers in the WAT of the gt/gt mice relative to wt (n = 10-12/group). Gene expression was studied relative to β-actin mRNA. d-h Mice were analyzed in metabolic home cages for the last week of the diet (n = 7-8/group). d Heat production and AUC. e Total physical activity and AUC. f Respiratory exchange ratio (VCO 2 /VO 2 /kg) and AUC. g Food intake and AUC. h Water intake and AUC. b-h Data are means ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001. ACSL1, acyl-CoA synthetase long-chain family member 1; AUC, area under the curve; CEBPA, CCAAT-enhancer-binding protein alpha; CIDEA, cell death-inducing DFFA-like effector A; LIPE, hormone-sensitive lipase; PPAR a/g, peroxisome proliferator-activated receptor alpha/gamma; PPARGC1A, PPARg coactivator-1 alpha; PRDM16, PR/SET domain-16; UCP1, uncoupling protein 1; VEGFA, vascular endothelial growth factor A; WAT, white adipose tissue Fig. 7 Pharmacological inhibition of HIF-P4Hs ameliorates diet-induced obesity, metabolic dysfunction, and liver damage. Wild-type (wt) females were fed a high-fat, high-fructose (HFHF) diet for e 4 or a-d, f-l 6 weeks or a high-fat, methionine-choline-deficient (HF-MCD) diet for m-s 3 weeks and simultaneously given vehicle (VEH) or 60 mg/kg of FG-4497 (FG) on days 1, 3, and 5 of each week (n = 8-10/group). a Western blot analysis of hepatic HIF1α and HIF2α protein levels. β-Actin was used as a loading control. b Blood hemoglobin levels. c Weight gain at the end of the diet relative to weights on the day before the diet started. d Weight of gonadal WAT. e Oral glucose tolerance test. The value for 0 min was determined after a 12-h fast. f Liver weight. g Scoring of hepatic steatosis. Steatosis grading: "None" corresponds to scores 0-2, "Moderate" to 3, and "Severe" to 4. h Scoring of liver inflammation. "No" corresponds to score 0 and "Yes" to 1-2. i qPCR analysis of liver mRNA levels of FG-treated mice relative to VEH-treated, studied relative to TATA-box-binding protein mRNA. j Serum ALT levels. k Blood lactate levels. l Serum uric acid levels. m Weight gain at the end of the HF-MCD diet relative to weights on the day before the diet started. n Serum total cholesterol, HDL cholesterol, and LDL cholesterol levels. o Liver weight. p Scoring of hepatic steatosis. Steatosis grading: "Moderate" corresponds to score 2 and "Severe" to 3. q Serum ALT levels. r Scoring of UCP1-stained WAT sections. "No" corresponds to < 5% of UCP1 staining/field and "Yes" to 5-25%. Images are representative of scoring for the VEH group. Scale bar = 200 μm. s qPCR analysis of WAT mRNA levels of FG-treated mice relative to VEH-treated group, studied relative to peptidylprolyl isomerase A mRNA. b-f, i-o, q, s Data are means ± SEM. *p ≤ 0.05, **p < 0.01, ***p < 0.001. ACSL1, acyl-CoA synthetase long-chain family member 1; ALT, alanine aminotransferase; b, blood; CEBPA, CCAAT-enhancer-binding protein alpha; CHREBP, carbohydrate-responsive element-binding protein; DGAT1, diacylglycerol o-acyltransferase 1; FASN, fatty acid synthase; GPAM, mitochondrial glycerol-3-phosphate acyltransferase; LPIN, lipin; PPAR a/g, peroxisome proliferator-activated receptor alpha/gamma; PRDM16, PR/SET domain-16; s, serum; SCD1, stearoyl-CoA desaturase 1; SLC2A2, solute carrier family-2 member 2; SREBF1c, sterol regulatory element-binding protein 1c; UCP-1, uncoupling protein 1; VEGFA, vascular endothelial growth factor A; WAT, white adipose tissue Pharmacological inhibition of HIF-P4Hs ameliorates diet-induced obesity, metabolic dysfunction, and NAFLD WT mice were fed the HFHF diet for 6 weeks and simultaneously treated with a pharmacological pan-HIF-P4H inhibitor FG-4497 or vehicle. The FG-4497 treatment stabilized HIF1α and HIF2α in the livers (Fig. 7a) and elevated the blood hemoglobin levels (Fig. 7b). The FG-4497-treated mice did not gain weight (Fig. 7c) and had less WAT (Fig. 7d) than the vehicle-treated mice. They also showed a trend towards a better glucose tolerance (Fig. 7e). However, no significant differences in liver weight or steatosis were detected between groups (Fig. 7f, g), liver inflammation being detected only in the vehicle-treated group (Fig. 7h). The inability of the FG-4497 treatment to protect against steatosis compared with the Hif-p4h-2 gt/gt livers associated with only a few lipogenic mRNAs and Slc2a2 being downregulated, and conversely with Pparg mRNA upregulation (Fig. 7i). Although not significant, the serum ALT levels were~40% lower in the FG-4497 group (Fig. 7j), indicating less liver damage, possibly stemming from lower intake of fructose due to Slc2a2 downregulation. Moreover, systemic levels of lactate and uric acid were significantly lower in the FG-4497-treated group (Fig. 7k, l). In conclusion, the FG-4497 treatment stabilized HIF, protecting from the HFHF diet-induced obesity and hepatic inflammation, but not from steatosis.
We next fed WT mice the HF-MCD diet for 3 weeks with simultaneous FG-4497 or vehicle treatment. The FG-4497 treatment was protected from weight gain (Fig. 7m) and the mice had a healthier serum lipid profile than the vehicletreated group (Fig. 7n), but no difference in liver weight or steatosis was detected between the groups (Fig. 7o, p). Nevertheless, the FG-4497-treated mice had lower serum ALT levels (Fig. 7q). Lastly, similar expression levels of WAT UCP1 and browning marker mRNAs were detected in both groups (Fig. 7r, s). In conclusion, the FG-4497 treatment on the HF-MCD diet was protected from obesity, hypercholesterolemia, and liver damage, but could not induce browning of WAT to the extent seen in the Hif-p4h-2 gt/gt mice and thus, was unable to protect against steatosis.

Discussion
Obesity and WAT function play key roles in the progression of NAFLD, since obesity and IR induce adipose tissue lipolysis and the liver then accumulates circulating FAs in a concentration-dependent manner [30]. Fructose contributes to NAFLD by providing a substrate for increased FA synthesis and activating hepatic de novo lipogenesis, which it promotes directly by upregulating SREBP1c [31], and indirectly by uric acid production, which reduces mitochondrial β-oxidation resulting in oxidative stress, hepatic IR, and endoplasmic reticulum (ER) stress [32,33]. Absorbed fructose is transported via the portal vein, thus reaching the liver in higher concentrations than the other tissues.
The roles of HIFs and HIF-P4Hs in NAFLD are complex and not yet fully understood. The data available until this study is mainly from liver/hepatocyte-specific settings and suggest that stabilization of HIF2α, or simultaneous inhibition of all HIF-P4Hs, promotes NAFLD while restricting the inhibition to selected isoenzymes may have beneficial effects [14,[34][35][36][37][38]. We have reported earlier that normal chow-fed Hif-p4h-2 gt/gt mice have hepatic stabilization of HIF2α but not of HIF1α with upregulation of Irs2 and downregulation of Srebf1c mRNA, resulting in lower acetyl-CoA levels, reduced de novo lipogenesis, and lower IR than the WT [14,17]. Furthermore, we recently showed that mouse embryonic fibroblasts isolated from Hif-p4h-2 gt/gt mice show~50% reduced ATP production compared with WT cells, likely resulting in the baseline differences in adiposity [14,18]. We now show that Hif-p4h-2 gt/gt mice were able to resist weight gain and adiposity on a HFHF diet and had less hepatic steatosis and liver damage and no inflammation or fibrosis. These features associated with downregulation of lipogenic mRNAs, lower levels of hepatic acetyl-CoA and serum insulin, and better glucose tolerance. Moreover, differences in fructose metabolism in favor of the Hif-p4h-2 gt/gt mice were detected, including 20-33% less fructose found in the liver, kidney, and serum than in the WT, and lower levels of serum uric acid and lactate. Interestingly, the mRNA levels and enzymatic activity of KHKs, which are responsible for the cleavage of fructose, were upregulated in the small intestine of the Hif-p4h-2 gt/gt mice. The small intestine is the major site of fructose metabolism when ingested in low doses, whereas fructose spills over into the liver at high doses [26]. Our data suggest that inhibition of HIF-P4H-2 increases the capacity of the intestine to metabolize high doses of dietary fructose and therefore provides protection from its spillover into the liver and other tissues.
Moreover, the Hif-p4h-2 gt/gt mice on a HFHF diet expressed hepatic mRNAs for Pnpla2, Pnpla3, and Cyp2e1 at lower levels than the WT. Inhibition of the adipose ligase PNPLA2 has been associated with protection against ER stress in mice [39,40], while activating genetic variants in PNPLA3 are associated with human NAFLD [41]. Moreover, increased expression of hepatic CYP2E1, which carries out the omega-hydroxylation of FAs, has been associated with NASH in mice and humans [42].
Additionally, the Hif-p4h-2 gt/gt mice on the HF-MCD diet had less WAT, and interestingly, they showed improved browning of WAT, which was associated with increased heat production. This likely contributed to less hepatic steatosis in the Hif-p4h-2 gt/gt mice, as WAT browning protecting against NAFLD has been reported on an MCD diet [28,29]. However, this is the first report associating HIF-P4H inhibition with induced browning of WAT and increased thermogenesis, while earlier reports associate environmental hypoxia with HIF pathway activation and WAT browning [43,44]. This likely involves the observed upregulation of adipose tissue PPARγ, which is a HIF target gene and essential in WAT browning [45][46][47].
Even though fructose is also required to develop liver injury on the HF-MCD diet, which contains sucrose [48], this diet additionally prevents hepatic VLDL export further compromising liver metabolism. This likely explains the lower degree of downregulation of hepatic lipid metabolism genes in the Hif-p4h-2 gt/gt mice on the HF-MCD diet compared with the HFHF diet.
NAFLD treatment is currently based on diet and exercise, while several therapeutics are being used with varying results. We show here that beneficial effects could be obtained with HIF-P4H-2 inhibition, as its genetic deficiency counteracted the symptoms of NAFLD. However, treatment with pan-HIF-P4H inhibitor was unable to reduce steatosis, as it demonstrated reduced ability to downregulate hepatic lipogenic gene expression in the HFHF diet and failure to significantly induce WAT browning in the HF-MCD diet. These data suggest that treatment of NAFLD by HIF-P4H inhibition should target only isoenzyme 2. This is supported by previous studies that associate the knockout of all HIF-P4Hs with steatosis, while the presence of a single HIF-P4H-1 or HIF-P4H-3 allele minimizes steatosis [49]. In conclusion, our findings demonstrate important underlying mechanisms of HIF-P4H-2 interference involving organ crosstalk in the pathogenesis of NAFLD, which offers potential for developing HIF-P4H-2 inhibitors for its treatment.