The LXXLL motif of murine forkhead transcription factor FoxO1 mediates Sirt1-dependent transcriptional activity (2024)

The LXXLL motif is conserved among FoxO family members. We identified an LXXLL motif at amino acids 459–463 in murine FoxO1, which is highly conserved among FoxO family members in human, murine, and Xenopus laevis and weakly conserved in Drosophila melanogaster and Caenorhabditis elegans (Table 1). This finding suggests that the LXXLL motif may have an important role in the function of FoxO proteins.

Mutagenesis of the LXXLL motif of FoxO1 abolishes Igfbp-1 and G6Pase promoter activity. At first, in order to investigate roles of the LXXLL motif of FoxO1, we constructed an LXXAA mutant in which leucine 462 and 463 were substituted by alanine (Figure 1A) and performed Igfbp-1 and G6Pase promoter assays using this mutant FoxO1 in SV40-transformed hepatocytes (25). Both Igfbp-1 and G6Pase have already been identified as target genes of FoxO1 (16, 26). In the absence of serum, WT FoxO1 increased Igfbp-1 and G6Pase promoter activity by 2.5- and 3.2-fold compared with control, respectively (Figure 1, B and C). However, the LXXAA mutant increased Igfbp-1 and G6Pase promoter activity by only 1.3- and 1.9-fold, respectively (Figure 1, and C). These data indicate that disruption of the LXXAA motif of FoxO1 decreases its transcriptional activity on its target genes in the absence of serum.

Transcriptional activity of FoxO1 is largely dependent on its subcellular localization (1). Although serum deprivation increases nuclear localization of FoxO1, the amount of FoxO1 that is localized in the nucleus is about 60% and cannot reach 100% in this cell line (data not shown). In order to investigate effects of disruption of the LXXAA motif on transcriptional regulation without considering subcellular localization of FoxO1, we constructed 3A/LXXAA mutant in which all Akt phosphorylation sites and leucines 462 and 463 were also substituted by alanine (Figure 1A). Immunofluorescence studies demonstrated that subcellular localization of the 3A/LXXAA mutant was the same as in the 3A mutant, in which only 3 Akt phosphorylation sites were substituted by alanine (Figure 1A) under several conditions, i.e., it is constitutively nuclear (Figure 1, D–G). Using these mutant FoxO1, we investigated effects of disruption of the LXXLL motif of FoxO1 on transcriptional regulation of FoxO1 target genes without considering subcellular localization. We performed Igfbp-1 and G6Pase promoter assays to investigate the effects of mutagenesis of the LXXLL motif on target gene expression. We used adenovirus vectors encoding LacZ, WT, 3A, and 3A/LXXAA in order to fix FoxO1 expression level (Figure 2A). WT FoxO1 and 3A FoxO1 increased Igfbp-1 promoter activity by 2.3- and 5.4-fold compared with LacZ, respectively (Figure 2B). However, the 3A/LXXAA mutant increased Igfbp-1 promoter activity by only 2.8-fold (Figure 2B). Therefore, disruption of the LXXLL motif of 3A FoxO1 decreased Igfbp-1 promoter activity by 47%. Furthermore, WT FoxO1 and 3A FoxO1 increased G6Pase promoter activity by 5.0- and 11.6-fold, respectively (Figure 2C). However, the 3A/LXXAA mutant increased G6Pase promoter activity by only 5.7-fold (Figure 2C). Therefore, disruption of the LXXLL motif of 3A FoxO1 decreased G6Pase promoter activity by 50%. These data suggest that disruption of the LXXLL motif of FoxO1 abolishes Igfbp-1 and G6Pase promoter activity and that the LXXLL motif of FoxO1 plays an important role in mediating the transcriptional activity of FoxO1.

Mutagenesis of the LXXLL motif of FoxO1 inhibits endogenous Igfbp-1 and G6Pase gene expression. In order to investigate effects of disrupting the LXXLL motif of FoxO1 on endogenous gene expression, we transduced SV40-transformed hepatocytes with adenovirus vector encoding WT FoxO1, 3A FoxO1, or 3A/LXXAA FoxO1 (Figure 3A). In this cell line, Igfbp-1 and G6Pase are induced in a dexamethasone/8-Br-cAMP/IBMX–dependent manner (8-Br-cAMP, 8-bromoadenosine cAMP; IBMX, 3-isobutyl-1-methylxanthine). After serum deprivation, dexamethasone/8-Br-cAMP/IBMX increased endogenous Igfbp-1 gene expression by 43-fold compared with basal condition (data not shown). Furthermore, transduction with 3A FoxO1 increased dexamethasone/8-Br-cAMP/IBMX–induced Igfbp-1 gene expression by 23-fold compared with LacZ-transduced conditions (Figure 3B). However, transduction with 3A/LXXAA FoxO1 increased Igfbp-1 gene expression by only 5.2-fold compared with LacZ-transduced conditions (Figure 3B). Furthermore, we investigated the effects of overexpression of mutant FoxO1 protein on endogenous G6Pase gene expression in this cell line. Transduction with 3A FoxO1 increased dexamethasone/8-Br-cAMP/IBMX–induced endogenous G6Pase by 2.6-fold compared with LacZ-transduced condition. However, transduction with 3A/LXXAA FoxO1 increased by only 1.8-fold (Figure 3C). These data indicate that FoxO1 regulates endogenous Igfbp-1 and G6Pase gene expression in SV40-transformed hepatocytes in an LXXLL motif–dependent manner.

Because the C terminal region of FoxO1 is a transactivation domain, truncation of the terminus (Δ256) abolishes its transcriptional activity. In order to determine how much transcriptional activity the LXXLL motif contributes to full transcriptional activity of the terminus of FoxO1, we compared endogenous Igfbp-1 gene expression levels in SV40-transformed hepatocytes infected with adenovirus encoding 3A FoxO1, 3A/LXXAA FoxO1, or Δ256 FoxO1. In the absence of dexamethasone/8-Br-cAMP/ IBMX, 3A FoxO1, 3A/LXXAA FoxO1, or Δ256 FoxO1 induced endogenous Igfbp-1 gene expression by 9.9-, 4.3-, or 1.4-fold, respectively (Figure 3D). In the presence of dexamethasone/8-Br-cAMP/ IBMX, transduction with 3A FoxO1, 3A/LXXAA FoxO1, or Δ256 FoxO1 induced gene expression by 98-, 30-, or 6.6-fold, respectively (Figure 3D). These data suggest that 70% of transcriptional activity of FoxO1 depends on the LXXLL motif.

Disruption of the LXXLL motif in constitutively active mutant FoxO1 confers dominant-negative effect onto endogenous Igfbp-1 gene expression. In order to investigate whether disruption of the LXXLL motif of FoxO1 confers dominant-negative effect onto endogenous FoxO1 target gene expression, we transduced SV40-transformed hepatocytes with adenoviruses encoding hemagglutinin–T24A/S253D/S316A (HA-ADA), which has already been reported to be constitutively active FoxO1 (16), and Flag-3A/LXXAA FoxO1 (Figure 4A), and examined endogenous Igfbp-1 gene expression by real-time PCR. In this cell line, HA-ADA FoxO1 induced endogenous Igfbp-1 gene expression by 25-fold compared with nontransduced cells in the presence of dexamethasone/8-Br-cAMP/ IBMX (Figure 4B). However, overexpression of 3A/LXXAA FoxO1 inhibited endogenous Igfbp-1 gene expression induced by overexpression of HA-ADA by 65% and 70% in a dose-dependent manner (Figure 4B). These data suggest that disruption of the LXXLL motif of FoxO1 confers dominant-negative effect onto endogenous Igfbp-1 gene expression in SV40-transformed hepatocytes.

The LXXLL motif of FoxO1 is indispensable for binding to Sirt1 in vivo. From previous studies, the LXXLL motif of FoxO1 appears to be quite important for its transcriptional activity. Therefore, we sought to identify the mechanism by which the LXXLL motif of FoxO1 is involved in its transcriptional regulation. The nicotinamide adenine dinucleotide–dependent (NAD-dependent) deacetylase Sirt1 is abundantly expressed in SV40-transformed hepatocytes (Figure 5A). After transduction of SV40-transformed hepatocytes with adenovirus vector encoding LacZ, 3A FoxO1, or 3A/LXXAA FoxO1, we immunoprecipitated 3A FoxO1 or 3A/LXXAA FoxO1 using anti-Flag antibody and blotted with anti–silent information regulator 2 (anti-Sir2; mammalian Sirt1) antibody. We showed that 3A FoxO1 bound to endogenous Sirt1. However, 3A/LXXAA FoxO1 failed to do so (Figure 5A). Furthermore, immunoprecipitation with anti-Sir2 antibody revealed that endogenous Sirt1 interacted with transduced Flag 3A FoxO1 but not with Flag 3A/LXXAA FoxO1 (Figure 5A). These data suggest that the LXXLL motif of FoxO1 is important for binding to Sirt1.

In order to investigate the effects of disruption of the LXXLL motif of FoxO1 on its acetylation, we performed Western blotting with anti-acetylated lysine antibody using the same sample. Mutagenesis of the LXXLL motif in 3A FoxO1 showed increased acetylation compared with that of 3A FoxO1 (Figure 5, B and C). Furthermore, in order to investigate effects of disruption of the LXXLL motif on acetylation of FoxO1 in the presence of nicotinamide, an inhibitor of Sirt1, and/or trichostatin A (TSA), an inhibitor of class I and class II histone deacetylases (HDACs), we transfected HEK293 cells with pCMV5/cMyc/3A FoxO1 or 3A/LXXAA FoxO1, immunoprecipitated cell lysates with anti-cMyc antibody, and blotted with anti-acetylated lysine polyclonal antibody. Although acetylation of 3A/LXXAA FoxO1 was detected in the presence of nicotinamide and TSA (Figure 5D), 3A FoxO1 was not acetylated under the same conditions (Figure 5D). These data suggest that disruption of the LXXLL motif enhances acetylation of FoxO1.

Sirt1 inhibitor inhibits Igfbp-1 and G6Pase promoter activity in an LXXLL motif–dependent manner. Nicotinamide is a Sirt1-specific deacetylase inhibitor. The effects of Sirt1 on transcriptional activity of FoxO family members are controversial. These effects seem to be dependent on FoxO’s target genes (27–30). In order to investigate whether the LXXLL motif of FoxO1 mediates Sirt1-dependent promoter activity of FoxO1 target genes, promoter assay of Igfbp-1 or G6Pase was performed. Nicotinamide decreased Igfbp-1 promoter activity by 48% compared with promoter activity induced by both dexamethasone/8-Br-cAMP/IBMX and 3A FoxO1 (Figure 6A). In contrast, nicotinamide did not change Igfbp-1 promoter activity induced by both dexamethasone/8-Br-cAMP/IBMX and 3A/LXXAA FoxO1 (Figure 6A). Furthermore, nicotinamide decreased G6Pase promoter activity by 70% compared with promoter activity induced by both dexamethasone/8-Br-cAMP/IBMX and 3A FoxO1 (Figure 6B). However, nicotinamide did not change G6Pase promoter activity induced by both dexamethasone/8-Br-cAMP/IBMX and 3A/LXXAA FoxO1 (Figure 6B). Interestingly, Igfbp-1 and G6Pase promoter activity induced by both dexamethasone/8-Br-cAMP/IBMX and 3A/LXXAA FoxO1 was the same as the nicotinamide-treated promoter activity level of 3A FoxO1 (Figure 6, A and B). In order to confirm the effect of the Sirt1 inhibitor on transcriptional regulation of FoxO1, we used another Sirt1 inhibitor, BML-210 (27), and performed the same promoter assays. It has been reported that BML-210 was a less potent inhibitor of Sirt1 than nicotinamide (27). BML-210 significantly decreased Igfbp-1 promoter activity (by 25%) compared with promoter activity induced by both dexamethasone/8-Br-cAMP/IBMX and 3A FoxO1 (Figure 7A). In contrast, BML-210 did not change Igfbp-1 promoter activity induced by both dexamethasone/8-Br-cAMP/IBMX and 3A/LXXAA FoxO1 (Figure 7A). Furthermore, BML-210 significantly decreased G6Pase promoter activity (by 17%) compared with promoter activity induced by both dexamethasone/8-Br-cAMP/IBMX and 3A FoxO1 (Figure 7B). However, BML-210 did not change G6Pase promoter activity induced by both dexamethasone/8-Br-cAMP/IBMX and 3A/LXXAA FoxO1 (Figure 7B). These data demonstrate that nicotinamide can inhibit 3A FoxO1–induced promoter activity of Igfbp-1 and G6Pase to the same level as that induced by 3A/LXXAA FoxO1 and suggest that the LXXLL motif mediates the Sirt1-dependent transcriptional activity of FoxO1.

Resveratrol activates Igfbp-1 and G6Pase promoter activity in an LXXLL motif–dependent manner. Resveratrol was found to be a potent activator of Sirt1 (31). Recently, it has been reported that resveratrol increases glucose production by increasing FoxO1-dependent transcription of gluconeogenic genes in H4IIE rat hepatoma cells (32). In order to determine whether the LXXLL motif in FoxO1 is involved in Sirt1-dependent transcriptional regulation, we also assessed effects of resveratrol on Igfbp-1 or G6Pase promoter activity induced by both dexamethasone/8-Br-cAMP/IBMX and 3A FoxO1 or 3A/LXXAA FoxO1. Resveratrol increased Igfbp-1 or G6Pase promoter activity induced by both dexamethasone/8-Br-cAMP/IBMX and 3A FoxO1 by 30% and 37%, respectively (P < 0.005 or P < 0.001 by 1-way ANOVA, respectively) (Figure 8, A and B). In contrast, resveratrol had no effects on both Igfbp-1 and G6Pase promoter activity levels induced by both dexamethasone/8-Br-cAMP/IBMX and 3A/ LXXAA FoxO1 (Figure 8, A and B). These data suggest that resveratrol, a Sirt1 activator, activates promoter activity of FoxO1 target genes in an LXXLL motif–dependent manner.

Effects of acute hepatic overexpression of 3A/LXXAA FoxO1 on glucose metabolism in Lepr^db/db mice. From the previous data, it appears that 3A/LXXAA FoxO1 acts as a dominant-negative type FoxO1. In order to confirm the dominant-negative effect of 3A/LXXAA FoxO1 in vivo, we administered adenovirus (4 × 10⁸ pfu/g of body weight) encoding LacZ, 3A FoxO1, or 3A/LXXAA FoxO1 into leptin receptor–deficient Lepr^db/db mice (Figure 9, A and B). These mice show hyperphagia, obesity, and insulin resistance in their basal state, and it has already been reported that about 50% of total intracellular FoxO1 proteins were localized in nucleus in livers from Lepr^db/db mice (33). Therefore, it is appropriate to use Lepr^db/db mice for investigation of a dominant-negative effect of 3A/LXXAA FoxO1. Immunohistochemistry using anti-FoxO1 serum demonstrated that both exogenous FoxO1 mutants were localized in nucleus of liver (Figure 9B). Five days after injection, we performed intraperitoneal glucose tolerance tests. Fasting blood glucose levels of mice injected with 3A FoxO1 increased significantly compared with those of mice injected with adenovirus encoding LacZ (469.6 ± 30.7 versus 311 ± 27.5; P < 0.001 by 1-way ANOVA). In contrast, fasting blood glucose of mice injected with 3A/LXXAA FoxO1 decreased significantly compared with mice injected with LacZ or 3A FoxO1 (176 ± 30.7 versus 311 ± 27.5 or 469.9 ± 30.7; P < 0.001 by 1-way ANOVA) (Figure 9C). Furthermore, intraperitoneal glucose tolerance testing showed that mice injected with 3A/LXXAA FoxO1 were significantly more glucose tolerant than mice injected with LacZ or 3A FoxO1 (Figure 9D). These data suggest that overexpression of 3A/LXXAA FoxO1 in liver can alleviate insulin resistance of Lepr^db/db mice.

Overexpression of 3A/LXXAA FoxO1 in liver from Lepr^db/db mice inhibits glucose production through inhibition of G6Pase gene expression. Real-time PCR shows that G6Pase and Igfbp-1 gene expression levels in liver from fasted mice injected with 3A/ LXXAA FoxO1 are significantly decreased compared with those in mice injected with 3A FoxO1. In contrast, Pepck gene expression level was not different statistically among LacZ, 3A FoxO1, and 3A/LXXAA FoxO1–treated mice (Figure 10A). In order to confirm whether both 3A FoxO1 and 3A/LXXAA FoxO1 can bind to promoter regions of FoxO1 target genes, which include Igfbp-1, G6Pase, and Pepck, chromatin immunoprecipitation (ChIP) assay was performed using liver from Lepr^db/db mice injected with adenovirus. Both 3A FoxO1 and 3A/LXXAA FoxO1 bound to promoter regions of Igfbp-1 and G6Pase (Figure 10B). Interestingly, although FoxO1 didn’t affect Pepck gene expression in liver in the present study, both 3A FoxO1 and 3A/LXXAA FoxO1 can bind to Pepck promoter region (Figure 10B). Furthermore, hepatic glycogen content of mice injected with 3A/LXXAA FoxO1 increased significantly compared with mice injected with the LacZ or 3A FoxO1 (0.0162 ± 0.0014 versus 0.0071 ± 0.0008 or 0.0065 ± 0.0009 mg glycogen/mg liver, respectively, P < 0.01 by 1-way ANOVA) (Figure 10C). These data suggest that 3A/LXXAA FoxO1 can bind to the promoter region of FoxO1 target genes in liver and that acute over-expression of 3A/LXXAA FoxO1 inhibits G6Pase gene expression and suppresses glucose production in liver from Lepr^db/db mice.

Disruption of the LXXLL motif enhances acetylation of FoxO1 in liver. As described above, we demonstrated that disruption of the LXXLL motif inhibits binding of FoxO1 to Sirt1 and enhances acetylation of FoxO1 in SV40-transformed hepatocytes or HEK293 cells. In order to confirm effects of disruption of the LXXLL motif in FoxO1 on in vivo acetylation in liver, we performed immunoprecipitation of liver lysates injected with adenovirus encoding 3A FoxO1 or 3A/LXXAA FoxO1 using anti-Flag antibody and Western blotting using anti-acetylated lysine antibody. Although both 3A FoxO1 and 3A/LXXAA FoxO1 can be coimmunoprecipitated with endogenous Cbp and Pgc-1α, 3A/LXXAA FoxO1 binds to endogenous Sirt1 weakly compared with 3A FoxO1 (Figure 11). Consistent with data described above, acetylation of 3A/LXXAA FoxO1 is enhanced compared with that of 3A FoxO1 (Figure 11). These data suggest that disruption of the LXXLL motif in FoxO1 doesn’t affect binding to Cbp but inhibits interaction with Sirt1 and finally enhances acetylation of FoxO1 in vivo.