Elovl5 regulates the mTORC2-Akt-FOXO1 pathway by controlling hepatic cis-vaccenic acid synthesis in diet-induced obese mice.

Elevated hepatic expression of fatty acid elongase-5 (Elovl5) induces FoxO1 phosphorylation, lowers FoxO1 nuclear content, and suppresses expression of genes involved in gluconeogenesis (GNG). In this report, we define the molecular and metabolic basis of Elovl5 control of FoxO1 phosphorylation. Adenoviral-mediated (Ad-Elovl5) induction of hepatic Elovl5 in diet-induced obese, glucose-intolerant mice and HepG2 cells increased the phosphorylation of Akt2-S(473) [mammalian target of rapamycin complex-2 (mTORC2) site], but not Akt2-T(308) (PDK1 site). The Akt2 inhibitor Akti1/2 blocked Elovl5 induction of FoxO1-S(256) phosphorylation in HepG2 cells. Elevated Elovl5 activity in liver and HepG2 cells induced rictor mRNA, rictor protein, and rictor-mTOR interaction, whereas rictor knockdown (siRNA) attenuated Elovl5 induction of Akt2-S(473) and FoxO1-S(256) phosphorylation in HepG2 cells. FA analysis revealed that the abundance of cis-vaccenic acid (18:1,n-7) was increased in livers of obese mice and HepG2 cells following Ad-Elovl5 infection. Treating HepG2 cells with Elovl5 substrates established that palmitoleic acid (16:1,n-7), but not γ-linolenic acid (18:3,n-6), induced rictor protein, Akt-S(473), and FoxO1-S(256) phosphorylation. Inhibition of FA elongation blocked 16:1,n-7 but not 18:1,n-7 induction of rictor protein and Akt-S(473) and FoxO1-S(256) phosphorylation. These results establish a novel link between Elovl5-mediated synthesis of 18:1,n-7 and GNG through the control of the mTORC2-Akt-FoxO1 pathway.


Recombinant adenovirus
The source, construction, purifi cation, titration, and use of the recombinant adenovirus expressing luciferase (Ad-Luc) and Elovl5 (Ad-Elovl5) were described previously ( 2,4 ). Adenovirus expressing a form of FoxO1 that is resistant to phosphorylation control (Ad-ADA-FoxO1) was a generous gift from Dr. D. Accili (Columbia University Medical center, NY) ( 29 ).

Mouse liver extracts
The mouse hepatic protein extracts from whole liver, nuclei, or cytosol used in this study were obtained from the lean and obese-glucose intolerant C57BL/6J mice described previously ( 2 ). Briefl y, male C57BL/6J mice were fed a low-fat diet (10% calories as fat; Research Diets, D12450B) or a high-fat diet (60% calories as fat; Research Diets, D12492) for 12 weeks. After 12 weeks on these diets, mice fed the low-fat diet were lean and euglycemic, whereas mice fed the high-fat diet were obese, hyperglycemic, and insulin resistant. Five days prior to termination of the experiment, mice were infected with Ad-Luc or Ad-Elovl5. Five days later, mice were fasted overnight; half of the fasted mice were refed their diets for 4 h. Fasted and refed mice were euthanized at 8 AM and 12 noon for the recovery of blood and liver. Methods for liver recovery and preparation of protein extracts were described in our earlier report ( 2 ).

HepG2 Cells
Human hepatoma (HepG2) cells were obtained from American Type Culture Collection (Manassas, VA) and grown in DMEM with 25 mM glucose; media was supplemented with 10% fetal calf serum. All experiments were carried out in cells grown on 6-well or 12-well plates (Corning Life Sciences; Corning, NY) in a humidifi ed incubator at 37°C and 5% CO 2 . HepG2 cells were infected with recombinant adenovirus expressing either a control virus (Ad-Luc) or test viruses (Ad-Elovl5, Ad-ADA-FoxO1) at 20 infectious units (IU) per cell. In all HepG2 cell experiments, proteins were extracted as wholecell extract.

Immunoblotting
Proteins were extracted from mouse liver and HepG2 cells as described previously ( 1,2 ) in the presence of protease (Roche Applied Science) and phosphatase inhibitors (1 mM ␤ -glycerol phosphate, 2.5 mM Na-pyrophosphate, 1 mM Na 3 VO4). Cytosolic (post-nuclear), nuclear protein fractions and whole-cell protein extracts were separated electrophoretically by SDS-PAGE (NuPAGE, 4-12% polyacrylamide Bis-Tris; Invitrogen) and transferred to nitrocellulose (BA83) membranes. Blots were incubated with primary antibodies against various proteins overnight at 4°C. The following day, blots were washed and incubated with secondary antibody for 1 h at room temperature. Antigen-antibody reactions were detected and quantifi ed using LiCor Odyssey scanner and software ( 1,2 ).

RNA extraction and quantitative real time polymerase chain reaction
Total RNA was extracted from mouse liver of our earlier study ( 2 ) and HepG2 cells using Triazol (Invitrogen). Transcript levels were measured by qRT-PCR. Gene-specifi c primers are listed in Table 1 . Primer design and qRT-PCR methods were described previously ( 2 ). Cyclophilin was used as a control gene. Levels of target gene mRNA abundance were normalized to the abundance of cyclophilin mRNA.
G6Pc are important enzymes involved in gluconeogenesis (GNG) and hepatic glucose production.
Although our earlier studies suggested a possible role of Elovl5 in the control of Akt, FoxO1, and GNG, those studies did not establish the metabolic or molecular mechanisms for this Elovl5-mediated regulatory scheme ( 2 ). In this report, we extend our previous study by establishing the requirement for Akt and FoxO1 phosphorylation in the Elovl5 control of FoxO1 and its target genes. We also identify mTORC2 as a major target of Elovl5 control and establish that cis -vaccenic acid (18:1,n-7), a product of Elovl5 activity, is a mediator of the mTORC2-Akt-FoxO1 pathway.

ADA-FoxO1 overrides Elovl5 suppression of GNG genes in HepG2 cells
To determine whether Elovl5 effects on the expression of gluconeogenic enzymes (i.e., Pck1 and G6Pc) required FoxO1 phosphorylation, HepG2 cells were infected with Ad-Luc or Ad-Elovl5 in the absence and presence of Ad-ADA-FoxO1. Ad-ADA-FoxO1 expresses a form of FoxO1 that is insensitive to phosphorylation control ( 29 ). Infection of HepG2 cells with Ad-Elovl5 signifi cantly decreased Pck1 and G6Pc mRNA abundance by >70% and ‫ف‬ 50% ( P < 0.05), respectively ( Fig. 2 ). Including Ad-ADA-FoxO1 in the infection scheme induced Pck1 and G6Pc mRNA and completely abolished Elovl5 regulation of Pck1 and G6Pc mRNA abundance. Thus, Elovl5 control of Pck1 and G6Pc expression in HepG2 cells requires increased FoxO1 phosphorylation.

Rictor knockdown
Knockdown of rictor protein in HepG2 cells used an siRNA approach. HepG2 cells were grown in 12-well plates to ‫ف‬ 30-50% confl uence; cells were transfected with either siRNA rictor or control siRNA (scrambled RNA) at 100 pmol/well plus 2 l/well of the X-tremeGENE siRNA transfection reagent according to the manufacturer's recommendations. Cells were also infected with either Ad-Luc or Ad-Elovl5 (20 IU/cell). Cells were maintained in DMEM + FBS for 72 h before harvest for protein analysis.

Immunoprecipitation
Mouse liver extracts (1 mg protein/ml) were incubated with 10 l of antibody-conjugated sepharose beads. The antibodies conjugated to sepharose beads were against rictor or raptor; IgG served as a control. Extracts were incubated with antibody sepharose beads overnight at 4°C. The following day, the beads were centrifuged at 14,000 g for 30 s at 4°C. The immunoprecipitates were collected and washed with cell lysis buffer fi ve times. The beads were resuspended with protein denaturing buffer containing SDS, boiled, and centrifuged; the supernatants were applied to SDS-PAGE gels for electrophoresis. Specifi c proteins were detected by immunoblotting as described above.

Lipid extraction and FA analysis
Total lipids from liver and HepG2 cells were extracted, saponifi ed, methylated, and quantifi ed by gas chromatography as described previously ( 2,30 ).

Statistical analysis
The statistical analyses performed in this study included ANOVA plus post hoc Tukey test and Student's t -test by using the statistical software StatView. P < 0.05 was considered statistically different. Data are expressed as mean ± SD.

Elovl5 regulates Akt2 and FoxO1 phosphorylation in mouse liver and human hepatoma (HepG2) cells
Hepatic Elovl5 activity in obese glucose-intolerant mice is ‫ف‬ 60-75% below levels in livers of lean glucose-tolerant mice. Infection of obese mice with Ad-Elovl5 increases Elovl5 activity ‫ف‬ 3-fold; to a level ‫ف‬ 50% above the level expressed in livers of lean mice. Increased hepatic Elovl5 activity correlated with increased Akt2 and FoxO1 phosphorylation, decreased hepatic Pck1 expression, and restoration of euglycemia ( 2 ). When compared with Ad-Luc-infected mice, elevated Elovl5 activity (Ad-Elovl5 of mTOR-S 2448 in obese but not in lean mice ( Fig. 5A, B ). Phosphorylated mTOR-S 2448 is an active kinase ( 23,35 ). The mTORC2-associated regulatory protein, rictor, was signifi cantly induced by increased Elovl5 activity in both lean and obese mice ( Fig. 5C ), an effect that correlated with Ad-Elovl5-mediated increase in rictor mRNA ( Fig. 4 ). Whereas the high-fat diet signifi cantly suppressed hepatic raptor protein abundance, elevated Elovl5 activity had no further effect on hepatic raptor content ( Fig. 5A, D ). Neither diet nor changes in Elovl5 activity signifi cantly affected hepatic abundance of other mTOR-associated proteins, mSIN1 and G ␤ L ( Fig. 5A ).
mTOR is the catalytic (kinase) subunit for both mTORC1 and mTORC2 ( 36,37 ). Raptor and rictor play key roles in complex assembly and substrate selection for mTOR kinase ( 23,33 ). The observed increase in rictor protein abundance with Elovl5 overexpression may facilitate rictor-mTOR interaction. To test this possibility, immunoprecipitation and immunoblotting were used to examine the impact of Elovl5 on mTOR interaction with rictor We fi rst examined the impact of diet and Elovl5 activity on the mammalian target of rapamycin complex-1 (mTORC1) and the mammalian target of rapamycin complex-2 (mTORC2) components, namely raptor, rictor, and mTOR ( Fig. 4 ). Elevated Elovl5 activity increased rictor and mTOR mRNA in both lean and obese mice у 2-fold ( P < 0.05). Interestingly, raptor mRNA abundance was suppressed by ‫ف‬ 70% in livers of obese mice; elevated Elovl5 activity further decreased raptor mRNA. Similar effects were seen in HepG2 cells ( Fig. 4B ); increased Elovl5 expression induced rictor and mTOR, but suppressed raptor mRNA abundance.
To determine whether these changes in mRNA correlated with changes in protein and mTOR phosphorylation, hepatic extracts from lean and obese mice infected with either Ad-Luc or Ad-Elovl5 ( 2 ) were quantifi ed by immunoblotting ( Fig. 5 ). Elevated hepatic Elovl5 activity did not signifi cantly increase hepatic mTOR protein abundance ( Fig. 5A ), an effect that did not correlate with the changes in mTOR mRNA ( Fig. 4 ). Instead, elevated Elovl5 expression signifi cantly increased phosphorylation Mouse liver extracts from fasted obese glucose-intolerant mice infected with either Ad-Luc or Ad-Elovl5 were prepared and examined for total and phosphoprotein abundance as described previously ( 2 ). A: Representative immunoblots, n = three mice per treatment. B: Quantifi ed levels of protein phosphorylation for Akt2-S 473 , Akt2-T 308 , and FoxO1-S 256 . Results are from two separate studies and are expressed as mean ± SD, n = 6. * P р 0.05 versus Ad-Luc. Phosphorylation status is based on the ratio of phosphoprotein divided by total protein as quantifi ed by immunoblot and Licor Odyssey software. C, D: HepG2 cells infected with Ad-Luc or Ad-Elovl5 for 48 h were serum starved overnight. The next day, cells were harvested to prepare total cell extracts and quantify levels of total and phospho-Akt and FoxO1. C: Representative immunoblots for total and phospho-Akt2-S 473 , Akt2-T 308 , and FoxO1-S 256 . D: Phosphorylation status was quantifi ed as above. Results are representative of three separate studies; results are expressed as mean ± SD, n = 3. * P р 0.05 versus Ad-Luc.

Fig. 2.
Overexpressed ADA-FoxO1 abrogates Elovl5 suppression of Pck1 and G6Pc in HepG2 cells. HepG2 cells were infected with Ad-Luc (control virus) and Ad-Elovl5 in the absence and presence of Ad-ADA-FoxO1 for 48 h. ADA-FoxO1 is insensitive to phosphorylation control ( 29 ). As in Fig. 1 , cells were serum starved overnight and harvested the following day for RNA extraction and quantitation of Pck1 (A), G6Pc (B), and cyclophilin mRNA by qRT-PCR. Results are representative of three separate studies; results are expressed as mean ± SD, n = 3. * P р 0.05 versus Ad-Luc; # P р 0.05 versus no Ad-ADA-FoxO1 (None). electrophoresed, immunoblotted, and assayed for the presence of rictor, raptor, and mTOR. The rictor sepharose bead-conjugated antibody pulled down both rictor and mTOR. Elevated Elovl5 activity increased the immunoprecipitation of both mTOR and rictor. Elovl5 not only increased hepatic levels of rictor protein, Elovl5 promoted the physical interaction of rictor with mTOR. In contrast, raptor sepharose bead-conjugated antibody did not pull down raptor or mTOR. The low level of raptor in these hepatic extracts probably explains this result.
Further verifi cation of the involvement of rictor in mTORC2 in this regulatory scheme evaluated the impact of rapamycin (a selective chemical inhibitor for mTORC1) and PP242 (a dual chemical inhibitor for mTORC1 and mTORC2). Treatment of HepG2 with rapamycin lowered the phosphorylation of mTOR and p70S6K, a target of mTOR, but did not block the Ad-Elovl5-mediated increase in Akt2-S 473 or FoxO1-S 256 phosphorylation (see supplementary Fig. I A-D). In contrast, PP242 suppressed mTOR and and raptor ( Fig. 6 ). Extracts from diet-induced obese mice infected with Ad-Luc or Ad-Elovl5 were treated with antibodies against rictor (rictor-IP) or raptor (raptor-IP); IgG served as a control for nonspecifi c immunoprecipitation. The immunoprecipitates were collected, denatured,   Fig. 1A ) and HepG2 cells (B). mRNA abundance was measured by qRT-PCR using the primers for rictor, raptor, mTOR, and cyclophilin ( Table 1 ). Cyclophilin was used as a housekeeping gene. A: mRNA abundance in livers of lean and obese mice infected with either Ad-Luc or Ad-Elovl5. Results are expressed as fold change, relative to the abundance of the mRNA in liver of lean Ad-Luc-infected mice. Results are from two separate studies and are expressed as the mean ± SD, n = 6. B: mRNA abundance in HepG2 cells infected with either Ad-Luc or Ad-Elovl5. Results are expressed as fold change, relative to the abundance of the mRNA in cells infected with Ad-Luc. Results are representative of three separate studies; results are expressed as mean ± SD, n = 3. * P р 0.05 versus Ad-Luc. this pathway is the suppression of expression of genes involved in GNG.
To determine whether similar changes occur in HepG2 cells infected with Ad-Elovl5, we examined the FA profi le of HepG2 cells infected with Ad-Luc or Ad-Elovl5. The FA profi les of HepG2 cells and the medium used to grow cells are shown in supplementary Fig. IV . Although these cells contain 16:1,n-7 and 18:2,n-6, cellular levels of 18:3,n-6 are very low because of low 18:2,n-6 and low FADS2 expression ( 30 ). Elevated Elovl5 activity signifi cantly increased ( ‫ف‬ 40%, P < 0.05) the 18:1,n-7/16:1,n-7 ratio, but had no effect on other FA ratios ( Fig. 8D ). This effect of Ad-Elovl5 infection of HepG2 cells correlated with the increase in 18:1,n-7 production in liver of Ad-Elovl5-infected mice ( Fig. 8C ).
In contrast, hepatic expression of SCD1 and Elovl6 is elevated in mouse models of carbohydrate-induced obesity and leptin defi ciency ( 4,46,47 ). Ablation of these enzymes protects mice from diet-induced obesity and insulin resistance ( 46,47 ). Hepatic expression of SCD1, Elovl5, and Elovl6 is increased in response to elevated hepatic SREBP1 nuclear abundance and leptin defi ciency. Hepatic expression of SCD1 and Elovl6, but not Elovl5, is induced by high-carbohydrate diets and elevated presence of soraphen A (Sor A, 200 nM). Elevated Elovl5 activity increased rictor protein abundance and Akt and FoxO1 phosphorylation. Furthermore, adding 16:1,n-7 to the media augmented this response ( Fig. 12A-D ). Cells treated with soraphen A, however, blocked 16:1,n-7 and Elovl5-mediated induction of rictor protein. Changes in Akt-S 473 and FoxO1-S 256 phosphorylation paralleled changes in rictor protein ( Fig. 12A-D ).

DISCUSSION
Our previous study established that a high-fat lard diet (60% energy as fat) lowered hepatic Elovl5 expression and activity and also lowered hepatic content of several MUFA and PUFA products of Elovl5. This diet also increased hepatic nuclear abundance of FoxO1 and elevated expression of genes (Pck1 and G6Pc) involved in GNG. These changes in FoxO1 nuclear abundance and gene expression were associated with hyperglycemia, glucose intolerance, and insulin resistance. Adenoviral-mediated induction of hepatic Elovl5 activity in obese glucose-intolerant mice reversed these diet-induced effects on hepatic metabolism ( 2 ).
The fi ndings in this report extend our previous study by establishing that Elovl5 regulated 18:1,n-7 synthesis controls the mTORC2-Akt-FoxO1 pathway and the expression of genes (Pck1 and G6Pc) involved in GNG. We established that increasing hepatic Elovl5 activity increases hepatic rictor through a pretranslational mechanism. Increased rictor protein promotes rictor-mTOR interaction to form mTORC2 and the stimulation of Akt-S 473 but not Akt-T 308 phosphorylation in liver and HepG2 cells ( Figs. 1,(4)(5)(6). Based on studies that examined rictor-T 1135 and Gsk3 ␤ phosphorylation, Elovl5 controls rictor protein abundance but may not control its phosphorylation status (see supplementary Fig. III ). However, more studies are required to verify this mechanism. The selective effect of Elovl5 on Akt-S 473 phosphorylation, the requirement for active Akt for Elovl5 control of FoxO1, coupled with the knockdown of rictor ( Fig. 7 ) and inhibition of mTORC2 activity (see supplementary Figs. I, II ) support the role of rictor as a key mediator of Elovl5 regulation of FoxO1 and gluconeogenic gene expression. As far as we are aware, this represents the fi rst documented linkage between FA elongation and the mTORC2 pathway ( Fig. 13 ).
In summary, we have identifi ed cis -vaccenic acid (18:1,n-7) as a regulator of the mTORC2-Akt-FoxO1 pathway and a mediator of Elovl5 suppression of hepatic gluconeogenic gene expression. Decreased hepatic abundance of 18:1,n-7 is associated with increased expression of enzymes involved in GNG, glucose intolerance, and hyperglycemia. Increased hepatic production of 18:1,n-7 by elevated Elovl5 activity has the opposite effects on hepatic glucose metabolism and systemic control of blood glucose. These outcomes raise two key questions: 1 ) how does 18:1,n-7 regulate rictor expression, and 2 ) is dietary 18:1,n-7 alone suffi cient to regulate hepatic glucose production in mouse models of obesity and diabetes. If so, dietary 18:1,n-7 might be useful in the management of blood glucose in humans with hyperglycemia.
ChREBP/MLX nuclear content ( 4 ). Expression of SCD1 and Elovl6 directs saturated FAs originating from de novo lipogenesis to the formation of stearic (18:0) and oleic acid (18:1,n-9). In contrast, SCD1 and Elovl5 will direct palmitate to the formation of 18:1,n-7 (4)(5)(6). Based on the studies reported here, the relative abundance of hepatic SCD1, Elovl5, and Elovl6 and their n-7 and n-9 MUFA products are important determinants in the control of the mTORC2 pathway and GNG.
The notion that n-7 MUFAs play a role in metabolic regulation is not new; adipose tissue-derived palmitoleic acid (16:1,n-7) was reported to act as a lipokine and to improve hepatic insulin action ( 48 ). A recent report examined the effects of adipose tissue 16:1,n-7 and SCD1 activity on the prevalence of obesity in 1,926 adults in Costa Rica. The authors found no evidence to support a lipokine role for adipose tissue 16:1,n-7 in the reduction of obesity occurrence ( 49 ). Others have reported that 16:1,n-7 promoted fatty liver but suppressed hepatic infl ammation ( 50 ). Still another study suggests that 18:1,n-7 inhibits hepatosteatosis ( 51,52 ). Although our previous reports described effects of elevated Elovl5 activity on hepatosteatosis ( 2 ), we suspect that this effect is linked to the capacity of Elovl5 to regulate hepatic content of C 20-22 PUFA and hepatic SREBP1 nuclear abundance ( Fig. 13 ) (42)(43)(44)(45). In fact, addition of 18:3,n-6, but not 16:1,n-7, to the Elovl5infected HepG2 cells signifi cantly suppressed nuclear SREBP1 content (data not shown). As noted earlier, C [18][19][20][21][22] n-3 and n-6 PUFAs regulate hepatic SREBP1 abundance, and dietary C 20-22 n-3 attenuates hepatosteatosis in a mouse model of diet-induced fatty liver disease (42)(43)(44)(45)53 ). More recently, Wu et al. ( 54 ) reported that increased human plasma content of FAs derived from de novo lipogenesis, MUFA synthesis, and limited ␤ -oxidation, i.e., 16:1,n-7, 18:1,n-7, 16:1,n-9, was associated with increased risk of sudden cardiac arrest. These FAs are in low abundance or absent in rodent and human diets, but are generated by elongation and desaturation reaction in cells. These reports, coupled with the data presented in this study, make