Epoxygenase Cyp2c44 Regulates Hepatic Lipid Metabolism and Insulin Signaling by Controlling FATP2 Localization and Activation of the DAG/PKCδ Axis

Hepatic insulin action requires phosphorylation/dephosphorylation of proteins downstream of the insulin receptor (IR)-β, including tyrosine phosphorylation of insulin receptor substrates (IRS) (1). Serine phosphorylation of IRS (e.g., Ser307) leads to decreased IRS-1 function or expression, thus attenuating insulin signaling, leading to insulin resistance (2). Insulin resistance results from several altered pathways, including insulin signaling, glucose homeostasis, lipid metabolism, and free fatty acid (FFA) overload (3). The mechanisms by which lipid levels are regulated and how they induce insulin resistance is unclear.

The cytochrome P450 (CYP) epoxygenases, which metabolize arachidonic acid (AA) to epoxyeicosatranoic acids (EETs), are regulators of insulin signaling. CYP epoxygenase loss of function leads to insulin resistance in humans, and mice lacking Cyp2c44 have hepatic insulin resistance with decreased IR-β activation and downstream signaling (4,5). In addition to controlling insulin signaling, CYP epoxygenases and EETs regulate lipid metabolism. CYP2C19 loss-of-function polymorphisms are linked with increased serum lipid levels in individuals with ischemic stroke (6), and decreased expression and activity of various cytochrome CYP enzymes are associated with nonalcoholic fatty liver disease (7). In high-fat diet (HFD)-fed mice, reduced adipose-derived EETs lead to increased adipogenesis, which is mitigated by administration of EET analogs (8). Inhibition or deletion of soluble epoxyhydrolase, which converts EETs into less biologically active dihydroxyeicosatrienoic acids, ameliorates HFD-induced hepatic steatosis (9).

The beneficial contribution of EETs against HFD-induced lipid accumulation was primarily investigated in pathological settings, and the physiological role of Cyp2c in regulating cross-talk between lipid metabolism and insulin signaling is less understood. Because insulin resistance leads to increased hepatic lipid accumulation and hyperlipidemia, and Cyp2c44^−/− mice show hepatic insulin resistance on standard chow diet (SCD) (5), we investigated whether Cyp2c44 regulates hepatic lipid levels and metabolism in lean, healthy and HFD-fed mice.

Sexually mature male mice were used because they manifest impaired glucose and lipid metabolic disease better than females (10). Experiments were approved by the Vanderbilt Institutional Animal Care and Use Committee (Nashville, TN). The 129/SvJ wild-type (WT) and Cyp2c44^−/− mice littermates were generated (11) and housed in rooms on a 12 light/12 dark cycle with temperature set at 68–79°F. Mice were fed SCD (∼14% fat Purina Laboratory Rodent 5001) up to 6 weeks of age and then fed HFD (60% calories from fat) (F3282; Bio-Serv) or continued on SCD. After 6 weeks, mice were fasted for 5 h and sacrificed. In some experiments, 6-week-old mice were fed HFD. After 2 weeks, mice were fed HFD + the EET analog EET-A (0.125 mg/mL in drinking water) or HFD + water, and, after 4 weeks, they were sacrificed. EET-A, synthetized as described previously (12), ameliorates glucose intolerance in SCD-fed Cyp2c44^−/− mice (5). In some experiments, 8-week-old Cyp2c44^−/− male mice were fed SCD or switched to a 10% low-fat diet (LFD) (D12450B; Research Diets, Inc.) and sacrificed after 3 weeks.

Liver structure was examined on hematoxylin-eosin (H&E) (Sigma-Aldrich)-stained paraffin sections. Neutral lipid droplets were visualized on liver frozen sections stained with Oil Red O (O0625; Sigma-Aldrich). Bright-field images (40×) were taken, and Oil Red O quantification was done using ImageJ.

Mouse livers (100 mg) were homogenized in 500 μL PBS. Homogenates were spiked with 10 μL SPLASH-lipidomics internal standard mix (Avanti Polar Lipids/Croda), and lipids were extracted as described previously (13). Lipid extracts were evaporated and resuspended in 100 μL methanol and chloroform (9:1 ratio). Discovery lipidomics, chromatographic separation, high-resolution mass spectrometry analysis, and statistical calculations were performed as described in Supplementary Fig. 1.

Lipids were extracted from livers (50 mg) and plasma (25 μL) using the Abcam Lipid Extraction Kit (ab211044). Total unsaturated FA were measured using the Abcam Lipid Assay Kit (ab242305) and expressed as microgram per milligram liver or milligrams per deciliter plasma.

To quantify hepatic diacylglycerol (DAG), homogenized livers (50 mg) were mixed with methanol, NaCl (1 mol/L) and chloroform (MeOH/NaCl/CHCl₃) at a 1.5:2.25:2.5 ratio (v/v/v) and separated into two phases by centrifugation. The lower phase was mixed with MeOH/NaCl/CHCl₃ at a 5:4.5:5 ratio (v/v/v) and separated into two phases. The lower phase was dried and suspended in Abcam-provided buffer, which solubilizes DAG. DAG content (nanograms per milligram) was measured using the Abcam DAG Assay Kit (ab242293) and a SpectraMax iD3 fluorimeter (excitation 530–560 nm, emission 585–595 nm).

Hepatic lipids were extracted as described previously (13). DAG was isolated by thin-layer chromatography plates using silica gel 60-Å plates developed in petroleum ether, ethyl ether, and acetic acid (80:20:1) and visualized with rhodamine 6G. DAG ester bands were scraped from plates and methylated using BF₃/methanol (14). The inclusion of lipid standards with odd-chain FA permits quantification. The methylated FAs were analyzed using an Agilent 7890A gas chromatography (GC) system with Agilent 7693 autoinjector, flame ionization detectors, and Chemstation software. The total DAG amount and the FA profile associated with the lipid class were quantified.

Hepatic EET levels were analyzed as described previously (15). Livers (20 mg) were homogenized in 1 mL 0.15 mol/L KCl, and lipids were extracted into acidified CHCl₃/MeOH (2/1, v/v) with the internal standards D11-dihydroxyeicosatrienoic acids (DHETs) and D3-EETs and then dried. The dried lipids were hydrolyzed in 0.4 N KOH, 80% MeOH/H₂O. The DHETs and EETs were separated by SiO₂ chromatography, and the EETs further hydrated to DHETs in 9.5 N acetic acid before being analyzed by ultraperformance liquid chromatography–negative electrospray ionization (ESI)/mass spectrometry (MS)/MS. Additionally, the ions (charge/mass ratio, selective reaction monitoring (SRM): 127, 167, 207) originating from charge/mass ratio 348 were monitored as D11-DHETs.

Hepatocytes, isolated as described previously (5), were suspended in PBS (50,000 cells per 50 μL) together with 50 μL of 20 μmol/L Bodipy-BSA (C1-BODIPY 500/510 C1, C12) (D3823; ThermoFisher). After 2 min, hepatocytes were washed with 0.1% BSA in PBS, fixed with 4% paraformaldehyde, resuspended in PBS, and loaded into black 96-well plates (3603; Corning). Cells were analyzed using a SpectraMax iD3 fluorimeter (excite 500–650 nm, emission 510–665 nm) and subsequently lysed for protein concentration. Lipid uptake was expressed as relative fluorescence unit per microgram of protein.

In some experiments, hepatocytes (50,000 cells per well) were plated in serum-free medium onto Collagen-I–coated (50 μg/mL) (Corning Collagen I, Rat Tail, 354236; Corning) six-well plates (3516; Corning) and incubated with vehicle (DMSO) or the FATP2 inhibitor Grassofermata (50 μmol/L) (NAV-2729; Axon Med Chem). After 12 h, C1-BODIPY assay was performed.

An intraperitoneal glucose tolerance test (IPGTT) and intraperitoneal insulin tolerance test (IPITT) were conducted following the National Mouse Metabolic Phenotyping Centers glucose and insulin tolerance test protocols (5). Mice were fasted for 5 h in the morning with free access to water. After measuring baseline blood glucose level, mice received an intraperitoneal injection of 20% w/v glucose (final 2 g/kg body weight) or 1 unit/kg insulin (Novolin R). Blood glucose was measured at 15, 30, 45, 60, and 90 min. For IPGTT, the area under the curve (AUC) for glucose above baseline was calculated using the trapezoidal method. For IPITT, the decrease in glucose from baseline to t = 30 was considered the primary measure of insulin sensitivity, calculated as AUC below baseline glucose x − 1, such that a larger positive number indicates increased insulin sensitivity (16).

Plasma membrane–rich fractions of livers or primary hepatocytes, prepared as described previously (5), were lysed in radioimmunoprecipitation assay buffer (Sigma Aldrich, #R0278-50ML) supplemented with EDTA-free Protease Inhibitor (Roche), phosphatase inhibitor cocktails (Sigma-Aldrich), and 2 nmol/L Na₃VO₄. Serum-starved hepatocytes were incubated with vehicle (water) or insulin (100 nmol/L, Novolin R) for 30 min, and then lysed for analysis. In some experiments, hepatocytes were pretreated for 17 h with the DAG kinase inhibitor I (10 μmol/L) (D5919; Sigma-Aldrich) or for 2 h with the PKCδ inhibitor delcasertib (2 μmol/L δv1-1 PKC inhibitor, MedChemExpress HY-106262B) prior to insulin treatment. Five-hour-fasted mice were injected with insulin (Novolin R, 5 units) or vehicle (PBS) into the inferior vena cava, as described previously (17). After 15 min, the mice were sacrificed, and livers were processed for analysis of insulin signaling. The antibodies used are listed in Supplementary Table 1.

Protein bands were detected on Odyssey CLx (LI-COR Biosciences) and quantified with Image Studio Lite (LI-COR Biosciences). Values are expressed as phosphorylated protein-to-total protein or total protein-to-loading control ratio.

Data are presented as mean ± SD. Between-group comparisons were made using one-way ANOVA followed by multiple comparisons with P < 0.05 defined as significant. Post hoc comparisons between groups were performed with Wilcoxon rank-sum test, and results with two-sided P < 0.05 were considered significant. Data analysis and graphs were prepared using GraphPad Prism 9 or R version 4.3.1.

The graphical abstract was created with BioRender.com.

The hepatic lipid profile is available at the National Institutes of Health Common Fund's National Metabolomics Data Repository website, the Metabolomics Workbench (18), https://www.metabolomicsworkbench.org. Data (Study ID ST003038) are available at https://dx.doi.org/10.21228/M8DQ64. Other data sets generated and/or analyzed during the current study as well as Cyp2c44^−/− mice are available upon request.

Cyp2c44^−/− SCD-fed mice have hepatic insulin resistance (5), which leads to fatty liver (19). H&E and Oil Red O staining revealed increased hepatic lipid vacuoles and content in Cyp2c44^−/− SCD-fed mice versus WT mice (Fig. 1A–C). Lipidomic analysis identified two lipid clusters in livers from SCD-fed WT and Cyp2c44^−/− mice (Supplementary Fig. 1A) and a total of 1,565 lipids (695 in negative mode and 870 in positive mode) that passed quality control filters (25% for quality control relative SD and 10% for quality control-to-blank ratio). Heatmap analysis revealed a hierarchical clustering of significant differences in lipid species between the two genotypes (Supplementary Fig. 1B). Volcano plot of the 695 lipids in negative mode (containing FA) revealed 160 lipid species upregulated, 61 downregulated, and 931 not significantly changed in Cyp2c44^−/− livers (Supplementary Fig. 1C); 146 FA (49 upregulated, 3 downregulated, and 94 unchanged) were identified in Cyp2c44^−/− livers (Supplementary Table 2). Arachidonic, palmitic, oleic, linoleic, eicosapentaenoic, and docosahexaenoic acids, known substrates of Cyp2c epoxygenases (20,21), were upregulated in Cyp2c44^−/− livers (Supplementary Fig. 1D). Consistent with increased levels of AA, total hepatic EETs were significantly lower in Cyp2c44^−/− mice versus WT mice (Supplementary Fig. 1E), and livers and plasma of Cyp2c44^−/− SCD-fed mice showed significantly higher levels of total unsaturated FA versus WT mice (Fig. 1D and E).

Hepatic lipid influx was measured by C1-BODIPY uptake by hepatocytes and found to be increased in Cyp2c44^−/− versus WT hepatocytes (Fig. 1F). As hepatic lipid uptake leads to the production/accumulation of DAG (22), we quantified hepatic DAG levels. Initially, enzymatic reactions were used to quantify the molar concentration of the glycerol backbone, then we measured the mass quantity of FA species in isolated DAG (GC-MS). DAG levels were significantly higher in livers of Cyp2c44^−/− mice independent of the technique used (Fig. 1G and H). The major FA species in DAGs were also identified using GC-MS (Supplementary Fig. 1F). The sum of FA species in each group was equal to the total mass quantity of DAG. GS-MS showed significantly higher levels of arachidonic, palmitic, oleic, linoleic, eicosapentaenoic, and docosahexaenoic acids in Cyp2c44^−/− livers (Supplementary Fig. 1G).

To determine the effect of HFD on hepatic lipid accumulation, WT and Cyp2c44^−/− mice were fed an HFD for 6 weeks. H&E and Oil Red O staining revealed a significant increase in hepatic lipid content in WT and Cyp2c44^−/− HFD-fed mice, which was more pronounced in Cyp2c44^−/− livers (Fig. 1A–C). Unsaturated FA content in livers and plasma of HFD-fed Cyp2c44^−/− mice was also significantly higher than that of HFD-fed WT mice (Fig. 1D and E). C1-BODIPY assay showed the highest uptake in hepatocytes of Cyp2c44^−/− HFD-fed mice (Fig. 1F), which correlated with the highest hepatic DAG content (Fig. 1G and H).

No differences were detected in body weight, liver weight, and percentage of body fat between WT and Cyp2c44^−/− SCD-fed mice (Supplementary Fig. 2). HFD significantly increased all parameters, although no differences were detected between the two genotypes (Supplementary Fig. 2).

CD36 (23) and FATP2 (24) regulate intracellular transport of long-chain FA (LCFA) into hepatocytes. Liver and hepatocyte plasma membrane–rich fractions revealed low CD36 levels, with no changes between WT and Cyp2c44^−/− mice on either diet (Supplementary Fig. 3). We detected a significant increase in plasma membrane–associated FATP2 in liver and hepatocytes of SCD-fed Cyp2c44^−/− mice versus WT mice (Fig. 2A, B, E, and F). FATP2 localization further increased in livers and hepatocytes of HFD-fed WT and Cyp2c44^−/− mice, which was most pronounced in HFD-fed Cyp2c44^−/− mice (Fig. 2A, B, E, and F). To determine the role of FATP2 in FA uptake, C1-BODIPY assay was performed in hepatocytes from SCD-fed WT and Cyp2c44^−/− mice treated with the FATP2 inhibitor Grassofermata. This inhibitor significantly decreased lipid uptake in Cyp2c44^−/− hepatocytes (Supplementary Fig. 4). Intracellular LCFA binds to cytosolic FA–binding proteins, with FABP1 highly expressed in the liver (25). We detected higher levels of FABP1 in the liver and hepatocytes of SCD-fed Cyp2c44^−/− mice compared with WT mice (Fig. 2C, D, G, and H). HFD significantly increased FABP1 levels in liver and hepatocytes of WT and Cyp2c44^−/− mice, with a more pronounced effect in Cyp2c44^−/− mice (Fig. 2C, D, G, and H).

We assessed insulin sensitivity as glucose response during IPITT. Compared with WT mice, glucose was increased in SCD-fed Cyp2c44^−/− mice and decreased in parallel after insulin administration in both genotypes (Fig. 3A). The AUC glucose below baseline glucose response during SCD was slightly greater in Cyp2c44^−/− mice (Fig. 3B). Insulin sensitivity was reduced in HFD-fed Cyp2c44^−/− compared with HFD-fed WT mice (Fig. 3B).

We analyzed glucose tolerance in mice on both SCD and HFD by IPGTT. Blood glucose during SCD was significantly higher in Cyp2c44^−/− mice versus WT mice (Fig. 3C). In HFD-fed WT mice, glucose tolerance worsened compared with SCD-fed mice (Fig. 3C). Glucose excursion increased slightly in HFD- versus SCD-fed Cyp2c44^−/− mice but to a lesser extent than in HFD-fed WT mice, such that glucose tolerance was similar in HFD-fed WT and HFD-fed Cyp2c44^−/− mice (Fig. 3C). Glucose AUC confirmed that HFD significantly worsened glucose tolerance in WT, but not in Cyp2c44^−/− mice (Fig. 3D).

Treatment of SCD-fed Cyp2c44^−/− mice with EET-A ameliorates glucose intolerance by restoring hepatic insulin signaling (5). We analyzed hepatic lipid content and glucose tolerance in HFD-fed WT and Cyp2c44^−/− mice before and after 4 weeks of treatment with vehicle or EET-A. EET-A significantly reduced body weight in Cyp2c44^−/− HFD-fed mice and liver weight in HFD-fed WT and Cyp2c44^−/− mice, without changing the overall body weight and percentage of body fat (Supplementary Fig. 2). Compared with vehicle-treated group, livers of EET-A-treated WT and Cyp2c44^−/− mice had significantly less lipid accumulation (Fig. 4A and B) and lower levels of total unsaturated FA (Fig. 4C and D). These effects were more pronounced in EET-A–treated Cyp2c44^−/− mice. EET-A increased insulin sensitivity in Cyp2c44^−/− mice similarly to vehicle-treated or EET-A–treated HFD-fed WT mice (Fig. 3A and B). EET-A significantly reduced fasting glucose and improved glucose tolerance in HFD-fed WT and Cyp2c44^−/− mice (Fig. 3C and D).

Although Cyp2c44^−/− SCD-fed mice have high levels of hepatic DAG levels, which are further increased by HFD (Fig. 1G and H), SCD-fed and HFD-fed Cyp2c44^−/− mice have comparable glucose tolerance. DAG promotes plasma membrane localization and activation of PKCδ and PKCε, which inhibit insulin signaling by phosphorylating IRS-1 on Ser307 (1,26,27). Thus, Cyp2c44^−/− HFD-fed mice should have increased glucose intolerance compared with Cyp2c44^−/− SCD-fed mice. To determine why this was not the case, we analyzed the plasma membrane localization of PKCδ and PKCε in livers and hepatocytes. Low hepatic plasma membrane–associated PKCε levels were observed between WT and Cyp2c44^−/− mice on either diet (Supplementary Fig. 3). PKCδ plasma membrane localization was detectable in the liver and hepatocytes of WT and Cyp2c44^−/− SCD-fed mice, with higher levels in the latter group (Fig. 5A–D). Livers and hepatocytes of WT HFD-fed mice showed significantly higher plasma membrane levels of PKCδ compared with WT SCD-fed mice; however, no overall differences were observed between SCD-fed versus HFD-fed Cyp2c44^−/− mice (Fig. 5A–D). Thus, hepatic PKCδ in the Cyp2c44^−/− mice cannot be further regulated by HFD-mediated increased DAG content.

Next, we analyzed insulin-activated signaling in hepatocytes. Insulin induced IR-β activation in WT but not Cyp2c44^−/− hepatocytes (Fig. 5E and F), as described previously (5). Insulin/IR-β axis leads to tyrosine phosphorylation of IRS-1, thus promoting insulin signaling. Consistent with reduced insulin-stimulated IR-β activation, IRS-1 p-Tyr608 was reduced in Cyp2c44^−/− hepatocytes (Supplementary Fig. 5). Consistent with increased plasma membrane levels of PKCδ, we observed IRS-1 p-Ser307 only in insulin-treated Cyp2c44^−/− hepatocytes (Fig. 5E and G). As IRS-1 p-Ser307 cannot activate PI3K-dependent pathways (2,26), insulin-mediated phosphorylation of AKT (Ser473) and glycogen synthase kinase-3β (GSK3β; Ser9) was significantly attenuated in insulin-treated Cyp2c44^−/− hepatocytes (Fig. 5E, H, and I).

We next defined the role of the DAG/PKCδ axis in hepatocytes by treating WT hepatocytes with the DAG kinase inhibitor D5919, which prevents the conversion of DAG to phosphatidic acid (Supplementary Fig. 6A). This caused a significant increase in DAG content starting at 5 μmol/L with a maximal effect at 10 μmol/L (Supplementary Fig. 6B). D5919 (10 μmol/L) significantly increased plasma membrane levels of PKCδ in WT hepatocytes (Fig. 6A and B) and IRS-1 p-Ser307, with concomitant decrease in insulin-mediated pAKT and pGSK3β (Fig. 6C–F). We also inhibited PKCδ in Cyp2c44^−/− hepatocytes with delcasertib (2 μmol/L), which resulted in a significant decrease in IRS-1 p-Ser307 (Fig. 6G and H) and increase in insulin-mediated pAKT and pGSK3β (Fig. 6G, I, and J).

Cyp2c44^−/− SCD-fed mice have increased levels of palmitate (Supplementary Fig. 1D, F, and G), which, in high doses, promotes the expression and membrane localization of FATP2 in rat hepatocytes (28). To determine whether hepatic lipid levels affect FATP2 membrane localization, we fed Cyp2c44^−/− mice with an LFD or SCD. After 3 weeks, LFD did not change body weight of Cyp2c44^−/− mice compared with SCD-fed Cyp2c44^−/− or WT mice (Fig. 7A). Cyp2c44^−/− LFD-fed mice showed significant decreased liver and plasma lipid content versus Cyp2c44^−/− SCD-fed mice (Fig. 7B–E), with FFA levels like those of WT SCD-fed mice (Fig. 7D and E). No changes were observed in plasma membrane localization of FATP2 between SCD- or LFD-fed Cyp2c44^−/− mice, which were significantly higher than those of WT SCD-fed mice (Fig. 7F and G). Thus, FFA levels do not control hepatic FATP2 membrane localization in Cyp2c44^−/− mice.

Next, we analyzed whether Cyp2c44 requires EETs to control FATP2 membrane localization. Treatment with EET-A reduced FATP2 plasma membrane localization in WT and Cyp2c44^−/− HFD-fed mice, although it reached significance in the latter group only (Fig. 8A and B). We also detected significantly lower levels of hepatic PKCδ in plasma membranes of EET-A–treated Cyp2c44^−/− HFD-fed mice (Fig. 8C and D).

Next, we determined whether EET-A–mediated decreased plasma membrane localization of FATP2 and PKCδ restored hepatic insulin signaling in Cyp2c44^−/− HFD-fed mice. No IR-β activation was detected in vehicle-treated WT or Cyp2c44^−/− HFD-fed mice (Supplementary Fig. 7A). Insulin activated IR-β in WT but not Cyp2c44^−/− HFD-fed mice (Supplementary Fig. 7A), although IR-β activation in WT HFD-fed mice was significantly lower than that of insulin-treated WT SCD-fed mice (Supplementary Fig. 7B). Consistent with PKCδ plasma membrane localization, increased IRS-1 p-S307 and decreased p-AKT and p-GSK3β were detected in the livers of Cyp2c44^−/− compared with WT HFD-fed mice (Fig. 8E–I) or vehicle-treated mice (Supplementary Fig. 7A and C–E). EET-A significantly enhanced hepatic insulin–mediated signaling in Cyp2c44^−/− HFD-fed mice, with decreased IRS-1 p-S307 and increased p-AKT and p-GSK3β (Fig. 8E–I). Thus, EET-A improves insulin signaling in Cyp2c44^−/− HFD-fed mice by decreasing plasma membrane levels of FATP2 and improving insulin signaling.

We show that, in male mice, Cyp2c44 epoxygenase acts as a regulator of lipid metabolism by linking it to insulin signaling. Male mice were used because estrogens are linked to insulin sensitivity (29), and female mice require ovariectomy to manifest insulin resistance (30). In addition, the metabolism of EETs is affected by estrogens (31), introducing additional complexity to our study.

SCD- and HFD-fed Cyp2c44^−/− mice have increased hepatic lipid content and impaired glucose tolerance, supporting studies in mice showing a correlation between Cyp epoxygenase expression and fatty liver disease. Diet-induced nonalcoholic fatty liver disease resulted in suppression of Cyp enzymes and increased inflammation (30,31), and CYP2J2 overexpression decreased HFD-induced hepatic lipid accumulation and inflammation (32). Thus, Cyp2c/j epoxygenases could protect the liver from lipid-induced injury by producing the anti-inflammatory EETs. In addition, Cyp2c44 negatively regulates hepatic lipid uptake, thus preserving insulin signaling by regulating plasma membrane levels of FATP2 and activation of the DAG/PKCδ pathway.

Hepatic FATP2 plasma membrane levels are upregulated in Cyp2c44^−/− SCD-fed mice. FATP2 is expressed in liver and controls LCFA transport (24). Liver-specific FATP2 knockdown reduced LCFA uptake (24), which is consistent with increased FATP2 plasma membrane levels and hepatic lipid content in Cyp2c44^−/− mice.

In hepatocytes, high levels of palmitate increase FATP2 expression, plasma membrane localization, and insulin resistance (28). SCD-fed Cyp2c44^−/− mice have high levels of hepatic palmitate, and HFD further increased hepatic FATP2 localization as well as hepatic and plasma lipid levels in Cyp2c44^−/− mice. Interestingly, Cyp2c44 deletion alleviated an effect of diet on hepatic FATP2 membrane localization, despite decreased plasma and hepatic lipid content in LFD-fed mice. Thus, Cyp2c44 may control FATP2 localization in an FFA-independent manner. Lack of Cyp2c44 in SCD-fed mice prevents insulin-mediated phosphorylation of FOXO1 (5). Unphosphorylated FOXO1 translocates to the nucleus to regulate gene transcription. In adipocytes, nuclear FOXO1 suppresses the expression of the peroxisome proliferator activated receptor (PPAR)γ (32), a negative regulator of FATP2 expression (33). Thus, loss of Cyp2c44 could result in increased FOXO1 nuclear localization, decreased PPARγ gene transcription, and increased FATP2 synthesis and membrane localization.

Hepatic levels of FABP1 are increased in Cyp2c44^−/− SCD-fed mice, and they are further enhanced by HFD. FABP1 has a high affinity for AA (34,35) and overexpression of FABP1 enhances AA uptake in vitro (36). We detected increased hepatic levels of AA in Cyp2c44^−/− SCD-fed mice, which is consistent with increased levels of FABP1. AA induces FABP gene expression (37), thus creating a vicious cycle between FABP1 expression and AA levels.

We show no differences in glucose intolerance between Cyp2c44^−/− SCD- and HFD-fed mice, although Cyp2c44^−/− HFD-fed mice have DAG levels higher than Cyp2c44^−/− SCD-fed mice. Because DAG regulates the membrane localization of PKCδ (38), DAG levels should correlate with plasma membrane localization of this enzyme. Surprisingly, no differences in plasma membrane–associated PKCδ were detected in livers of SCD-fed or HFD-fed Cyp2c44^−/− mice. Thus, a plateau of plasma membrane–associated PKCδ is reached in Cyp2c44^−/− mice, which cannot be modulated by increased DAG levels. PKCδ could translocate to mitochondria in a DAG-independent but reactive oxygen species–dependent manner (39). Excessive hepatic lipid accumulation in Cyp2c44^−/− mice might lead to reactive oxygen species production with consequent increased mitochondrial PKCδ localization.

Although PKCɛ contributes to hepatic insulin resistance (40), its direct role in the liver is questionable. In rats, fatty liver caused increased hepatic PKCε activation and reduced hepatic insulin signaling (41) and in vivo downregulation of PKCε reduced fat-induced hepatic insulin resistance (42). We failed to detect plasma membrane–associated hepatic PKCɛ in 129/SVJ mice, suggesting that, in this strain, DAG does not regulate hepatic PKCɛ plasma membrane localization. As PKCε is also found associated with intracellular membranes, it is difficult to rule out PKCε activation in our study. However, mice lacking hepatic PKCɛ were not protected against HFD-induced glucose intolerance or hepatic insulin resistance (43), suggesting that the effect of PKCɛ in regulating insulin signaling is not exerted directly in the liver.

EET-A–treated Cyp2c44^−/− HFD-fed mice have decreased FATP2 and PKCδ plasma membrane localization and improved insulin signaling. Although it is not clear how EETs regulate FATP2 plasma membrane localization, EETs control the function of transmembrane ion channels and receptors. EETs regulate the phosphorylation of the β- and γ-subunits of the epithelial sodium channel ENaC (44), and EET-A promotes hepatic insulin–mediated signaling in SCD-fed Cyp2c44^−/− mice by enhancing the retention of activated IR-β in the plasma membrane (5).

Increased plasma membrane localization of PKCδ in Cyp2c44^−/− mice correlated with hepatic IRS-1 p-Ser307. p-Ser307 was detected only in Cyp2c44^−/− mice, although insulin promotes IRS-1 p-Ser307 in WT cells (45). We speculate that binding of insulin to IR-β in Cyp2c44^−/− hepatocytes combined with increased plasma membrane–associated PKCδ is sufficient to cause increased IRS-1 p-Ser307. In addition, reduced EET levels in Cyp2c44^−/− mice might lead to inhibition of serine phosphatases responsible for dephosphorylating IRS-1 Ser307. Consistent with this statement, EET-induced type 2A–like protein phosphatase activity is required for activation/function of the large-conductance, Ca²⁺-activated K⁺ (BK[Ca]) channel in vascular smooth muscle (46). Thus, in WT hepatocytes, EETs might stimulate insulin signaling by promoting phosphatase-mediated serine dephosphorylation of IRS-1. Finally, PKCδ could contribute to hepatic lipid accumulation through increased expression of lipogenic genes. Deletion of PKCδ protected the mice from HFD-induced glucose intolerance (47), while overexpression of PKCδ in the liver leads to hepatic insulin resistance characterized by enhanced expression of lipogenic genes (48).

In conclusion, Cyp2c44 epoxygenase contributes to lipid-mediated glucose intolerance by regulating the hepatic plasma membrane localization of FATP2 and, in turn, FFA uptake and DAG production. We provide a novel mechanism whereby an epoxygenase modulates insulin signaling via an FATP2/DAG/PKCδ pathway.

Funding. This work was supported, in part, by American Diabetes Association grant 1-19-IBS-282 (A.P.); Vanderbilt Equipment Grant 1S10OD023514 (Mass Spectrometry Core); National Institutes of Health (NIH) grants R01-DK119212 (A.P.), R01-DK117875 (J.M.L.), U24-DK135073 (D.H.W.), DK20593 (D.H.W.), R01-DK050277 (D.H.W.), and R01 DK069921 (R.Z.); Department of Veterans Affairs Merit Reviews 1I01BX002025 (A.P.), 1I01BX002196 (R.Z.); and JRF Ortho, Welch Foundation I-0011 (J.R.F.). A.P. is the recipient of a Department of Veterans Affairs Senior Research Career Scientist Award IK6 BX005240. The Metabolomics Workbench repository is supported by NIH grant U2C-DK119886.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. K.G. conducted experiments, acquired data, analyzed data, and wrote the first draft of the manuscript; S.B.P. and S.C. conducted experiments and analyzed data; J.R.F. provided reagents; J.M.L., R.Z., and D.H.W. contributed to discussion and reviewed/edited the manuscript; A.P. designed research studies and wrote the manuscript. A.P. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.