FAM13A Represses AMPK Activity and Regulates Hepatic Glucose and Lipid Metabolism.

Obesity commonly co-exists with fatty liver disease with increasing health burden worldwide. Family with Sequence Similarity 13, Member A (FAM13A) has been associated with lipid levels and fat mass by genome-wide association studies (GWAS). However, the function of FAM13A in maintaining metabolic homeostasis in vivo remains unclear. Here, we demonstrated that rs2276936 in this locus has allelic-enhancer activity in massively parallel reporter assays (MPRA) and reporter assay. The DNA region containing rs2276936 regulates expression of endogenous FAM13A in HepG2 cells. In vivo, Fam13a-/- mice are protected from high-fat diet (HFD)-induced fatty liver accompanied by increased insulin sensitivity and reduced glucose production in liver. Mechanistically, loss of Fam13a led to the activation of AMP-activated protein kinase (AMPK) and increased mitochondrial respiration in primary hepatocytes. These findings demonstrate that FAM13A mediates obesity-related dysregulation of lipid and glucose homeostasis. Targeting FAM13A might be a promising treatment of obesity and fatty liver disease.

Introduction

Incidence of obesity and nonalcoholic fatty liver disease (NAFLD) has been increasing rapidly and is now the most common cause of chronic liver diseases worldwide (Younossi, 2019, Younossi et al., 2018). NAFLD may develop into nonalcoholic steatohepatitis and increase the risk of cardiovascular diseases and mortality; however, efficient treatment to block or slow down progression of NAFLD is still lacking (Lefere et al., 2019, Younossi et al., 2018). Genome-wide association studies (GWAS) have been instrumental to pinpoint important susceptible genes and possible therapeutic targets (Cannon and Mohlke, 2018) in complex traits. Genetic variants in the FAM13A (Family with Sequence Similarity 13, Member A) locus has been reported to be associated with HDL cholesterol and body mass index (BMI)-adjusted fasting insulin levels (Lundback et al., 2018, Willer et al., 2013), two relevant phenotypes for liver diseases. Furthermore, FAM13A locus has also been identified for its significant association with waist-to-hip ratio (WHR) adjusted for body mass index (WHRadjBMI) (Shungin et al., 2015) and visceral to subcutaneous fat ratio (Ji et al., 2019).

Recently, FAM13A was reported to interact and stabilize IRS1 from degradation, thereby enhancing insulin signaling in adipocytes (Wardhana et al., 2018). In contrast, another group reported that Fam13a is dispensable for adipose development and insulin sensitivity in adipose tissue (Tang et al., 2019), suggesting complex regulation of glucose metabolism by FAM13A in adipose tissue. However, whether FAM13A regulates hepatic insulin sensitivity and lipid metabolism and their roles in fatty liver development are not explored yet.

In this study, we find that rs2276936 in the FAM13A locus, associated with multiple metabolic traits including HDL levels, demonstrated allelic enhancer activity. Furthermore, we find that Fam13a mice showed improved hepatic insulin sensitivity and increased hepatic AMP-activated protein kinase (AMPK) activity, a major regulator in maintaining energy homeostasis and preventing against hepatic steatosis (Coughlan et al., 2014, Woods et al., 2017). These changes may collectively contribute to ameliorated high-fat diet (HFD)-induced obesity and nonalcoholic fatty liver in Fam13a mice.

Results

RS2276936 within FAM13A Gene Associated with HDL and Fat Mass Exerts Allelic Activity

Previously, we performed massively parallel reporter assays (MPRA) to assess allele-specific enhancer activity of SNPs within FAM13A GWAS locus (Castaldi et al., 2019) and identified 45 SNPs with significant allelic effects. Given that the FAM13A locus associated with multiple metabolic traits including HDL levels, we further queried the association of these 45 MPRA SNPs with HDL levels using GWAS data from Global Lipids Genetics Consortium (Willer et al., 2013) (http://csg.sph.umich.edu/abecasis/public/lipids2013/). We found that three tightly correlated MPRA SNPs (r2> 0.9) demonstrated genome-wide (rs2276936, p < 10−8) or sub-genome-wide significance (rs2167750 and rs7695177, p < 10−5) of association with HDL levels (Table S1). Furthermore, these three SNPs are tightly correlated with previously published top GWAS SNPs associated with lipid traits including rs3822072 (Yaghootkar et al., 2016) and rs9991328 (Yaghootkar et al., 2016) (r2> 0.9 for all pairwise comparison). Major and minor haplotypes of these five SNPs comprised of >90% population of all ethnic groups (Machiela and Chanock, 2015, Machiela and Chanock, 2018) (Figure 1A). We also evaluated the association of these three SNPs, rs2276936, rs2167750, and rs7695177 with over 2000 complex traits from UK Biobank (Ge et al., 2017). They were associated with body fat distribution at genome-wide significance (p < 5 × 10−8) (Table S2).

Figure 1

Identification of Functional Variants at the FAM13A Locus Associated with Lipid Levels

(A)LD matrix of SNPs (rs2276936, rs2167750, rs7695177) with previously frequently reported top GWAS SNPs (rs3822072 and rs9991328). The first two columns represent the major (red frame) and minor (blue frame) haplotypes of these five SNPs.

(B–D) (B) Top: the diagram shows the ATAC-Seq and DNase-Seq signals around SNP rs2276936, rs2167750, and rs7695177 in liver-relevant cell and tissues profiled by Roadmap epigenomics project and ENCODE project. Bottom: reporter assays of FAM13A promoter with or without DNA regions spanning rs2276936, rs2167750, and rs7695177 in HepG2 cells. For: forward orientation; Rev: reverse orientations. Data are presented as mean ± SEM from three biological replicates with triplicate wells in each repeat. ∗p < 0.05 by unpaired Student's t test. Expression of FAM13A was assessed in HepG2 cells transfected with gRNA targeting rs2276936 for regional deletion (C) or indel deletion (D). Con: Control. Rs227-1 and rs227-2 are from two independent repeats. Data are presented as mean ± SEM from two biological replicates with triplicate wells in each repeat. ∗p < 0.05 by unpaired Student's t test.

Furthermore, we analyzed open chromatin status nearby three SNPs, rs2276936, rs2167750, and rs7695177 in primary human hepatic tissues and hepatocytes available from Roadmap epigenomics and ENCODE (encyclopedia of DNA elements) projects (Inoue et al., 2017, Kundaje et al., 2015). The DNA region nearby rs2276936 showed highest open chromatin scores among three SNPs, suggesting active regulatory element nearby this SNP in human liver tissues or cells (Figure 1B). To further investigate the regulatory effects of these three SNPs on the expression of FAM13A, we performed reporter assays by cloning the endogenous promoter of FAM13A gene and preceding SNP regions with one of two opposing alleles for each SNP into the luciferase constructs. The primers used for molecular cloning of reporter constructs are shown in Table S3. After transfection into HepG2 cells, rs2276936 demonstrated significant allelic effects (Figure 1B), with A-allele at forward orientation associated with higher luciferase activity. Furthermore, we engineered the CRISPR/Cas-9 constructs with either gRNA pairs to generate a ∼100bp deletion (Figure 1C) or a single gRNA to generate an indel (Figure 1D) spanning rs2276936. Sequence of gRNAs used for deletion generation nearby the rs2276936 is shown in Table S4. After transfection of these CRISPR constructs into HepG2 cells, we found that deletion of rs2276936 region led to 30% reduction in the expression of FAM13A, suggesting an enhancer activity of the DNA region spanning rs2276936.

Collectively, these data strongly suggested rs2276936 associated with HDL-regulated expression of FAM13A in an allele-specific manner with rs2276936C allele associated with lower enhancer activity in HepG2 cells.

Fam13a Deficiency Ameliorates HFD-induced Body Weight Gain

Given that FAM13A was associated with multiple metabolic phenotypes in GWAS, we investigated metabolic phenotypes of Fam13a knockout mice that we have previously generated (Jiang et al., 2016). Consistent with the previous report, Fam13a mice demonstrated similar body weights, serum-free fatty acid levels, and triglycerides levels as compared with wild-type littermate control mice (Fam13a) (Figures S1A–S1D) when fed on normal chow diet. However, total cholesterol levels are reduced in Fam13a mice compared with Fam13a (Figure S1E), suggesting that Fam13a is required to maintain normal cholesterol levels.

HFD-treated mice chronically develop multiple metabolic changes, seen in fatty liver disease, an increasing global health burden (Perlemuter et al., 2007). To determine roles of FAM13A in this process, we treated mice with HFD for four months. In contrast to Fam13a mice, Fam13a mice demonstrated less body weight gain starting at two months of HFD treatment (Figures 2A and S2A) and increased lean mass and reduced fat mass after four months of HFD feeding (Figures 2B and S2B), likely attributable to decreased mass of inguinal fat (Figures 2C and S2C) rather than brown or epididymal fat (Figures S2E–S2H). Furthermore, liver mass is also reduced in HFD-treated Fam13a mice (Figures 2D and S2D). However, oxygen consumption (VO2) (Figures S3A and S3B), energy expenditure (EE) (Figure S3C), respiratory exchange rate (Figure S3D), and food intake (Figure S3E) were comparable in Fam13a mice and Fam13a mice after seven weeks of HFD treatment.

Figure 2

Body Weight Gain in Mice after High-Fat diet (HFD) Treatment for Four Months

(A) Gain of body weight measured during HFD treatment for four months in Fam13a (n = 7) and Fam13a mice (n = 5).

(B–D) (B) Fat and lean mass of mice (A) measured by MRI after four months of HFD treatment. Inguinal fat (C) and liver (D) mass normalized to body weight in mice treated with HFD (n = 6 for Fam13a; n = 5 for Fam13a). Data are presented as mean ± SEM. ∗p < 0.05 or ∗∗p < 0.01 by unpaired Student's t test.

Fam13a Mice Display Improved Hepatic Insulin Sensitivity and Reduced Hepatic Glucose Production

To assess glucose homeostasis, we performed glucose tolerance test (GTT) and insulin tolerance test (ITT) on Fam13a mice and wild-type littermates after 14 weeks of HFD treatment. Despite similar fasting glucose levels (Figure S2I) and glucose tolerance (Figure 3A), Fam13a mice demonstrated increased insulin sensitivity as suggested by ITT assay (Figure 3B), indicating that Fam13a deficiency sensitizes mice to insulin response. However, we detected reduced insulin levels in serum from Fam13a mice (Figure 3C), indicating a plausible feedback due to increased insulin sensitivity.

Figure 3

Fam13a Mice Showed Improved Systemic and Hepatic Insulin Sensitivity after HFD Treatment

(A and B) (A) Glucose tolerance test (GTT) and (B) insulin tolerance test (ITT) were performed after overnight fasting or 6-h fasting in female Fam13a and Fam13a mice fed with HFD for 14 weeks, respectively.

(C) Serum insulin was measured in female Fam13a and Fam13a mice fed with HFD for four months.

(D) Immunoblotting of phosphorylation of Akt (Thr308) in liver from HFD-fed Fam13a and Fam13a mice for four months with or without insulin injection via portal vein. Mice were harvested five minutes after insulin treatment. Mean ± SEM shown represent the densitometry of bands averaged from 3 to 4 mice/group. ∗p < 0.05 or ∗∗p < 0.01 by unpaired Student's t test.

(E) Akt phosphorylation at Thr308 was measured in primary hepatocytes cultured in the presence or absence of insulin at the concentration of 100 nM for 6 h. INS, insulin.

(F) Hepatic glucose production was measured and normalized to total cellular protein amount in hepatocytes isolated from Fam13a and Fam13a mice. (Mean ± SEM from three mice/genotypes). ∗p < 0.05, unpaired Student's t test.

Because liver is critical in systemic glucose homeostasis, we analyzed hepatic insulin signaling pathway via portal vein injection of insulin into HFD-fed Fam13a mice. Increased phosphorylation of Akt (p-Akt) (Thr308) was observed in Fam13a mice compared with Fam13a controls five minutes after insulin stimulation (Figure 3D). Furthermore, primary Fam13a hepatocytes also demonstrated increased levels of p-Akt after insulin treatment ex vivo compared with Fam13a cells (Figure 3E), suggesting increased hepatic insulin sensitivity in Fam13a hepatocytes. Additionally, hepatic glucose production was also significantly decreased in isolated primary Fam13a hepatocytes, compared with Fam13a cells (Figure 3F), suggesting that deficiency of Fam13a may lead to intrinsic reduction of gluconeogenesis in hepatocytes.

Partially consistent with recent findings that Fam13a may increase IRS-1 levels in white adipose tissues (Wardhana et al., 2018), we found a trend toward reduced levels of IRS-1 in epididymal fat (Figure S4A) whereas slightly increased IRS-1 levels in both inguinal fat (Figure S4B) and liver tissues (Figure S4C) compared with Fam13a control mice, suggesting that the regulation of Fam13a on insulin signaling in adipocytes is depot dependent.

FAM13A Regulates Lipid Metabolism and Fam13a Mice Are Resistant to HFD-Induced Fatty Liver

Given that the FAM13A locus is associated with HDL levels in GWAS, we also measured lipid levels in HFD-fed mice. Fam13a mice showed significantly reduced levels of LDL/VLDL and total cholesterol (Figure 4A) in serum. Although levels of HDL (Figure 4A) and triglycerides (Figure S5A) in serum are comparable between Fam13a and Fam13a mice, expression of the apolipoprotein A-I (Apoa1) gene that encodes major component of HDL was significantly increased in isolated hepatocytes (Figure 4B) and showed a trend toward increased levels in liver samples (Figure S5B) from Fam13a mice compared with Fam13a littermates.

Figure 4

HFD-fed Fam13a Mice Showed Improved Hepatic Steatosis

(A) Levels of total cholesterol, HDL, and LDL/VLDL were determined in serum from Fam13a and Fam13a mice after HFD treatment for four months.

(B–E) (B) Expression of Apoa1 was measured in primary hepatocytes from Fam13a and Fam13a mice using q-PCR. Free fatty acid (C), triglycerides (D), and cholesterol (E) were measured in murine liver samples. Data are presented as mean ± SEM. ∗p < 0.05 or ∗∗p < 0.01, unpaired Student's t test.

(F) Representative images of HE staining of livers are shown. (n = 6 for Fam13a; n = 5 for Fam13a mice). Scale bar: 20 μm.

Chronic HFD treatment results in lipid accumulation in the liver, leading to fatty liver disease (Perlemuter et al., 2007). However, significantly lower levels of free fatty acid (Figure 4C), triglycerides (Figure 4D), and total cholesterol (Figure 4E) were detected in liver samples from Fam13a mice compared with Fam13a control mice fed with HFD. We observed much less lipid accumulation in the livers of Fam13a mice as compared with HFD-fed Fam13a mice (Figure 4F). Collectively, these data suggest that Fam13a deficiency ameliorates HFD-induced fatty liver. However, expression of inflammatory cytokines was comparable between these two genotypes (Figures S5C–S5I).

Increased Mitochondrial Function and Activation of the AMPK Signaling in Hepatocytes from Fam13a Mice

To determine the mechanism for improved fatty liver in Fam13a mice after HFD treatment, we assessed several possibilities including lipolysis, hepatic lipogenesis, mitochondrial function such as oxidative phosphorylation, and fatty acid oxidation in hepatocytes. First, lipolysis induced by injection of isoproterenol, a β3-adrenergic agonist, stimulated a significant increase of free fatty acid release in both Fam13a and Fam13a mice with negligible genotypic difference (Figure S6A). Secondly, expression of lipogenic genes including sterol regulatory element-binding transcription factor (Srebf1), a critical transcription factor to activate lipogenic gene (Hagiwara et al., 2012), and its target, fatty acid synthase (Fasn) (Radenne et al., 2008), were comparable in liver samples from Fam13a and Fam13 mice (Figures S6B and S6C). Lastly, expression of critical genes for fatty acid β-oxidation, medium-chain acyl-coenzyme A dehydrogenase (Acadm), was increased in liver samples from HFD-fed Fam13a mice (Figure S6D), in contrast to carnitinepalmitoyltransferase 1A (Cpt1a) (Figure S6E).

We then assessed mitochondrial function in primary hepatocytes. Indeed, basal and maximal mitochondrial respiration were significantly increased in Fam13a hepatocytes, compared with Fam13a hepatocytes as measured by Seahorse assay (Figures 5A and 5B), suggesting increased mitochondrial oxidative phosphorylation in primary hepatocytes under Fam13a deficiency. Consistently, expression of peroxisome proliferative activated receptor, gamma, coactivator 1 alpha (Ppargc1α), the master regulator of mitochondrial biogenesis and oxidative phosphorylation (Jun et al., 2014), is increased in Fam13a hepatocytes, compared with Fam13a hepatocytes (Figure S6F).

Figure 5

FAM13A Regulates Mitochondrial Function in Primary Hepatocytes

(A) Mitochondrial respiration was measured by Seahorse assay in hepatocytes isolated from Fam13a and Fam13a mice. Arrows indicate the sequential addition of oligomycin, carbonyl cyanide 4-(trifluoromethoxy) phenylhydrazone (FCCP), and rotenone/antimycin treatments. Representative results of one pair of mice from two biological replicates.

(B–E) (B) The basal oxygen consumption rate (OCR), maximal mitochondrial respiration, ATP production, and non-mitochondrial respiration from A are shown. Data are presented as mean ± SD. ∗∗p < 0.01 by unpaired Student's t test. Phosphorylation of AMPK (Thr172) was measured in the liver samples from HFD-fed Fam13a and Fam13a mice (C, n = 4–5 mice/genotype), in primary hepatocytes (D) and HepG2 cells transfected with small interfering RNA (siRNA) targeting FAM13A (E). Data are presented as mean ± SEM. ∗p < 0.05, ∗∗p < 0.01 by unpaired Student's t test.

Furthermore, we determined the activity of AMPK, a critical regulator in energy homeostasis, in Fam13a liver samples. Accompanied by improved fatty liver in Fam13a mice, AMPK activation as indicated by phosphorylation of AMPK (Thr172) was significantly enhanced in liver from Fam13a mice fed with HFD (Figure 5C). Increased activation of AMPK was also detected in primary hepatocytes from Fam13a mice fed on normal chow, suggesting an intrinsic inhibition on AMPK in hepatocytes by FAM13A (Figure 5D). This was further confirmed by increased AMPK activation detected in HepG2 cells transfected with FAM13A siRNA (Figure 5E). These data indicate that deficiency of Fam13a led to hepatic activation of AMPK pathway that may contribute to less body weight gain and ameliorated fatty liver in Fam13a mice under HFD treatment.

FAM13A Regulates Mitochondrial Function through AMPK Signaling

Furthermore, overexpression of FAM13A reduced basal and maximal mitochondrial respiration, and ATP production (Figure 6A), accompanied by decreased phosphorylation level of AMPK (Thr172) (Figure 6B) in HepG2 cells. To determine whether FAM13A regulates mitochondrial function through AMPK, stable knockdown of AMPK or FAM13A was established in HepG2 cells by shRNA targeting AMPKα and/or FAM13A, as opposed to scramble RNA control. Although knockdown of FAM13A led to increased mitochondrial respiration (Figures 6C–6E), this was abolished by silencing of AMPK. Taken together, these data suggest that FAM13A regulates mitochondrial function through dephosphorylation and inactivation of AMPK and thus possibly promotes lipid accumulation in hepatic cells (Figure 7A).

Figure 6

FAM13A Regulates Mitochondrial Respiration through AMPK Signaling

(A) The basal mitochondrial respiration, maximal mitochondrial respiration, ATP production, and non-mitochondrial respiration were shown in HepG2 cells transfected with the plasmid expressing FAM13A or empty vector. Data are presented as mean ± SEM. ∗p < 0.05, ∗∗p < 0.01 by unpaired Student's t test.

(B) Phosphorylation of AMPK (Thr172) was determined in HepG2 cells with overexpression of Myc-tagged FAM13A. V: Vector control; F: Fam13a.

(C) Immunoblotting showing the knockdown efficiency. Combo shRNA: combination of FAM13A shRNA and AMPK shRNA.

(D) Bioenergetics assays measured by Seahorse in HepG2 cells with stable knockdown of FAM13A and/or AMPK by shRNA. Arrows indicated sequential addition of oligomycin, carbonyl cyanide 4-(trifluoromethoxy) phenylhydrazone (FCCP), and rotenone/antimycin.

(E) The basal oxygen consumption rate (OCR), maximal mitochondrial respiration, ATP production, and non-mitochondrial respiration were shown. Data are presented as mean ± SD. ∗p < 0.05, ∗∗p < 0.01 by unpaired Student's t test.

Figure 7

Schematic Illustration of FAM13A Regulating lipid and Glucose Metabolism

(A) Mechanism of FAM13A regulating Hepatic steatosis and insulin resistance through AMPK. TG: triglycerides. HFD: high-fat diet.

(B) Metabolic phenotypes observed in human subjects with opposing genotypes for SNP rs2276936 and Fam13a mice (Ji et al., 2019).

Discussion

Our current studies provided in vivo and in vitro evidence supporting important roles of FAM13A, a gene consistently associated with multiple metabolic phenotypes in GWAS, in regulating hepatic glucose and lipid metabolism. By reporter and CRISPR/Cas-9 assays, we pinpointed rs2276936 as the possible functional variant regulating hepatic FAM13A expression. Furthermore, we demonstrated that Fam13a mice are protected from HFD-induced fatty liver and lipid toxicity with improved hepatic insulin sensitivity. Increased AMPK activation associated with increased mitochondrial respiration was detected in hepatocytes from Fam13a mice, which likely contributes to ameliorated HFD-induced fatty liver in Fam13a mice. Importantly, most of the directional effects of risk allele rs2276936C were recapitulated in knockout mouse model, such as fat mass, insulin sensitivity, and total cholesterol levels (Figure 7B). Admittedly, some of human GWAS traits were not confirmed with genotypic difference in mouse models such as HDL levels, further highlighting the complexity of using mouse model to recapitulate human GWAS findings (Small et al., 2018).

From GWAS to Function

Identification of functional variants in the GWAS locus is the crucial and necessary first step to translate genetic discoveries to disease understanding and ultimately, therapy. The FAM13A locus has been associated with BMI-adjusted fasting insulin level (Lundback et al., 2018), HDL level (Willer et al., 2013), and waist-to-hip ratio (Shungin et al., 2015) in GWAS in multiple replicated studies. However, functional variants in this region and function of FAM13A gene in regulating metabolism in vivo remain incompletely understood yet. Here, we identified rs2276936 with allelic enhancer activities is located in the first intron of FAM13A, about ∼18kb away from its transcription start site. Interestingly, this SNP is also located nearby multiple enhancer marks including H3K4Me1, H3K27Ac, and H3K9Ac as well as DNase hypersensitive sites in liver, suggesting regulatory function of rs2276936. We primarily focused on roles of rs2276936 in liver here, given the importance of liver in regulating HDL, lipid, and glucose metabolism. Further investigations are warranted to determine roles of rs2276936 in adipose tissue and other metabolic active tissues. It is not surprising that pleiotropic functional variants may have effects on multiple tissues. For example, the significant BMI GWAS SNPs in the fat mass and obesity-associated (FTO) locus may regulate Iroquois homeobox 3 (Irx3) and Iroquois homeobox 5 (Irx5), the major causal genes (Smemo et al., 2014), in both brain (Smemo et al., 2014) and adipose tissues (Claussnitzer et al., 2015). Functional variants regulating lipid GWAS gene sortilin 1 (Sort1) demonstrated greater allelic effects in the liver sample (Musunuru et al., 2010) despite abundant expression of Sort1 in adipose tissues. These examples highlight the challenges and complexity of functional variants identification in GWAS loci in effect cell types.

It is noteworthy that rs2276936C, tightly correlated with previous top GWAS SNP rs9991328T (Figure 1A), is associated with reduced fat mass in human subjects (Table S2) and reduced enhancer activity for Fam13A (Figure 1B), which is consistent with reduced body and fat weight in Fam13a mice (Figures 2A and 2C) in mouse models as we summarized in Figure 7.

Insulin Sensitivity in Fam13a Mice

Through ITT measurements, we observed increased insulin sensitivity in Fam13a mice than Fam13a mice fed with HFD. However, no glucose tolerance difference was detected by GTT assay. This may result from compensatory reduction of insulin levels in Fam13a mice (Figure 3C).

Hepatic glucose production is also crucial for maintaining glucose homeostasis that is mainly controlled by gluconeogenesis through insulin and glucagon pathways. Reduced hepatic glucose production may result from increased insulin sensitivity in Fam13a hepatocytes through inhibition on the gluconeogenesis by insulin. Given that increased insulin sensitivity upon Fam13a deficiency was detected in primary hepatocytes isolated from normal chow- and HFD-fed mice, FAM13A likely intrinsically inhibits hepatic insulin sensitivity.

In contrast, the regulation of FAM13A on insulin sensitivity in adipose tissues seems more complex depending on the depots of white adipose tissue, which may explain previously inconsistent results ranging from reduced (Wardhana et al., 2018) to similar (Tang et al., 2019) insulin sensitivity in adipose tissues from Fam13a mice fed with HFD. We found variable changes of IRS-1 in different types of adipose tissues from Fam13a mice with a trend toward decreased levels of IRS-1 in epididymal fat (Figure S4A), similar as previously reported (Wardhana et al., 2018), whereas slightly increased IRS-1 levels in inguinal fat (Figure S4B).

Other reasons for some of the discrepancies between two previous studies and ours may include differences in the composition of diet, HFD treatment age and duration, genetic backgrounds of Fam13a mice, and technical details, which may complicate results interpretation. For example, an HFD containing 35% fat, 25.3% carbohydrates, and 23% protein was used previously to feed six-week-old Fam13a mice (C57BL/6N background) for 14 weeks compared with wild-type C57BL/6N mice in the studies (Wardhana et al., 2018). Herein, four- to five-month-old Fam13a mice after three generations of backcrossing toward C57BL/6J background and Fam13a littermates were fed with HFD containing 36% fat, 35.7% carbohydrates, and 20.5% protein for four months. Furthermore, previous ITT assay was performed in mice without fasting (Wardhana et al., 2018), in contrast to our ITT assay with mice after 6 h of fasting. Therefore, experimental condition is critical to interpret metabolic phenotypes in mice, and more stringently controlled studies are required to elucidate how FAM13A modulates insulin sensitivity in the adipose tissue.

Ameliorated Fatty Liver and Increased AMPK Activation in the Fam13a Mice

Insulin sensitivity, body weight gain, and fatty liver can influence each other. Reduced body weight can possibly improve fatty liver and insulin resistance, whereas HFD-induced hepatic insulin resistance may lead to fatty liver due to failed inhibition on gluconeogenesis combined with selectively enhanced lipogenesis in liver by insulin (Petersen et al., 2017). In the present study, given reduced gluconeogenesis was found in isolated primary hepatocytes from Fam13a mice, it is likely ameliorated fatty liver resulted from improved hepatic insulin sensitivity. Furthermore, in HFD-fed Fam13a mice, despite increased hepatic insulin sensitivity, improved fatty liver was found without significantly increased lipogenesis by insulin pathway, possibly due to hepatic activation of AMPK. Admittedly, future studies using liver-specific Fam13a mice will be helpful to conclusively exclude the possibility of improved fatty liver resulting from less body weight gain in Fam13a mice.

AMPK is activated by reduced ATP/AMP ratio due to prolonged energy deprivation, thus facilitating utilization of various sources of nutrients to generate ATP. Consequently, upon AMPK activation, ATP-consuming pathways, including cholesterol and fatty acid synthesis, are repressed and ATP-generating pathways, including fatty acid oxidation and lipolysis, are activated (Woods et al., 2017) through regulatory enzymes or regulatory proteins involved in anabolic pathways (Hardie et al., 2012). For example, AMPK represses cholesterol synthesis by inhibiting sterol regulatory element-binding factor 2 (Srebf2) and hydroxymethylglutaryl-CoA (HMG-CoA) reductase (Li et al., 2011). Therefore, activation of AMPK in primary hepatocytes isolated from Fam13a mice may explain reduced cholesterol levels in serum from Fam13a mice fed with normal chow (Figure S1E). Furthermore, increased mitochondrial respiration in primary Fam13a hepatocytes is also consistent with enhanced expression of Ppargc1α, a key gene essential for mitochondrial synthesis and regulated by AMPK (Hardie et al., 2012).

AMPK was recently suggested as a therapeutic target for the treatment of obesity, fatty liver disease (Garcia et al., 2019), and type 2 diabetes (Coughlan et al., 2014) due to its importance as a master regulator in maintaining glucose and lipid homeostasis. For example, genetic or pharmacological activation of AMPK in murine livers decreased lipogenesis, ameliorated high-fructose-induced hepatic triglyceride accumulation and thus improved hepatic steatosis (Woods et al., 2017), similar as we observed in HFD-fed Fam13a mice. Notably, hepatic activation of AMPK by overexpression of trancated AMPKα1 induced by doxycycline treatment improved HFD-induced weight gain and white fat depot accumulation, even after obesity and fatty liver disease have been established (Garcia et al., 2019). In addition, such hepatic activation of AMPK usually does not reduce levels of blood glucose (Cokorinos et al., 2017, Woods et al., 2017), as we observed as well (Figure S2I). Admittedly, hepatic role of AMPK activation in regulating body weight gain is far more complex. Despite knockin of activating mutations in the AMPKγ2 subunit reduced fat mass and liver steatosis upon diet-induced obesity (Yang et al., 2016), liver-specific constitutive activation of AMPK protected mice against only high-fructose diet-induced liver steatosis (Woods et al., 2017), in contrast to HFD treatment.

It is likely that hepatic activation of AMPK pathway may not explain all the metabolic phenotypes that we observed in Fam13a mice, and future studies using hepatocyte-specific and/or adipocyte-specific Fam13a knockout mice may help explain the pleotrophic regulation of FAM13A on glucose and lipid metabolism. Nonetheless, our study provides several lines of evidence to reveal that a common DNA variant rs2276936 associated with four metabolic traits regulates hepatic expression of FAM13A. Fam13a mice are protected from HFD-induced weight gain, hepatic lipid accumulation, and insulin resistance, possibly by increasing the AMPK activity. More importantly, in human subjects, FAM13A locus is associated with lipid metaoblism and insulin sensitivity and increased expression of FAM13A was found in liver cirrhosis tissues compared with normal liver samples (Zhang et al., 2019). Given protective effects of Fam13a deficiency in diet-induced fatty liver, our results suggest FAM13A as a possible interventional target for the treatment of obesity and fatty liver disease.

Limitations of the Study

We demonstrated that deficiency of FAM13A protected mice from HFD-induced obesity and fatty liver using a global Fam13a knockout mouse line. To conclusively determine whether FAM13A deficiency in liver or fat contributes to amolriated body weight gain and lipid accumulation in liver after HFD treatment, future studies with hepatocyte-specific Fam13a mice and adipose-specific Fam13a mice are warranted. Secondly, despite phosphorylation of AMPK at Thr172 has been widely used to indicate AMPK activity, future more compreheisve evalautation on AMPK acitivty measured by AMPK activity kits including tissue-speicific regulation of AMPK by FAM13A would strengthen our findings. Lastly, future work will also include testing the silencing of Fam13a as a possible treatment strategy in diet-induced obesity and fatty liver mouse models.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.