Methylation by Set9 modulates FoxO3 stability and transcriptional activity

Abstract

The FoxO family of transcription factors plays an important role in longevity and tumor suppression by regulating the expression of a wide range of target genes. FoxO3 has recently been found to be associated with extreme longevity in humans and to regulate the homeostasis of adult stem cell pools in mammals, which may contribute to longevity. The activity of FoxO3 is controlled by a variety of post-translational modifications that have been proposed to form a 'code' affecting FoxO3 subcellular localization, DNA binding ability, protein-protein interactions and protein stability. Lysine methylation is a crucial post-translational modification on histones that regulates chromatin accessibility and is a key part of the 'histone code'. However, whether lysine methylation plays a role in modulating FoxO3 activity has never been examined. Here we show that the methyltransferase Set9 directly methylates FoxO3 in vitro and in cells. Using a combination of tandem mass spectrometry and methyl-specific antibodies, we find that Set9 methylates FoxO3 at a single residue, lysine 271, a site previously known to be deacetylated by Sirt1. Methylation of FoxO3 by Set9 decreases FoxO3 protein stability, while moderately increasing FoxO3 transcriptional activity. The modulation of FoxO3 stability and activity by methylation may be critical for fine-tuning cellular responses to stress stimuli, which may in turn affect FoxO3's ability to promote tumor suppression and longevity.

Figure 1. FoxO3 is methylated by Set9 in vitro

(A) In vitro methylation of the N-terminal portion of FoxO3 (amino acids 1-300) by 8 different methyltransferases. ◁: FoxO3 methylated by Set9, *: Set9 auto-methylation. (B) In vitro methylation of the C-terminal portion of FoxO3 (amino acids 301-673) by 8 different methyltransferases. *: Set9 auto-methylation. (C) Methylation of the full-length FoxO3 protein by Set9. ◁: full-length FoxO3, ◀: FoxO3 degradation product, *: Set9 auto-methylation.

Figure 2. Set9 mono-methylates FoxO3 at K271 in vitro

(A) Tandem mass spectrometry on in vitro methylated full-length FoxO3. Peptides containing methylated lysines are shown and the number of the methylated lysine in the human FoxO3 amino acid sequence is indicated. Methylated lysines are followed by an *. The type of methylation (mono-, di-, or tri-) is indicated by 1, 2, or 3, respectively. (B) Deletion analysis to map the region of FoxO3 containing the methylation site. In vitro methylation of overlapping fragments spanning the N-terminal domain of FoxO3 by Set9. (C) Methylation of FoxO3 WT or mutants of specific lysine residues. Each mutant was made in the context of the GST-FoxO3 protein (amino acid 1-525). ◁: FoxO3, ◀: FoxO3 degradation product, *: Set9 auto-methylation. (D) Location of K271 compared to other domains and PTMs of FoxO3. Listed are Akt phosphorylation sites (T32, S253, and S315), DNA binding domain, and NLS and NES (nuclear export sequence). K271 is the final amino acid in the second part of the bipartite FoxO3 NLS. (E) Alignment of the region surrounding the residues methylated by Set9 in a series of known Set9 substrates.

Figure 3. Set9 preferentially methylates FoxO3 among FoxO family members

(A) In vitro methylation of different mammalian FoxO family members or FoxO orthologs. The FoxO1, FoxO3 and FoxO4 constructs were generated from the human sequence, whereas the FoxO6 construct was generated from the mouse sequence. ◀: FoxO6 degradation product, ←: full-length FoxO6, *: Set9 auto-methylation. (B) Alignment of the region of FoxO3 surrounding K271 to that of other mammalian FoxO family members and FoxO orthologs.

Figure 4. FoxO3 is methylated at K271 in cells

(A) Scheme for generation of K271me1 antibody. A branched mono-methylated 11 amino acid peptide surrounding FoxO3 K271 was used as the epitope. (B) The K271me1 antibody is specific to FoxO3 that has been methylated at K271 in in vitro methylation assays. In vitro methylation assays were conducted with cold SAM and analyzed by western-blot with the K271me1 antibody [33]. Coomassie staining was used to show equal levels of FoxO3 and Set9 were used in each condition (bottom). ◁: FoxO3, ◀: FoxO3 degradation product. (C) The K271me1 antibody is specific to FoxO3 K271 methylation in cells. Flag-FoxO3 and Flag-Set9 (WT or H297A methyltransferase-deficient mutant) were co-expressed in 293T cells, and FoxO3 methylation was analyzed by western-blot with the K271me1 antibody. (D) Endogenous Set9 methylates ectopically expressed FoxO3. The methylated form of FoxO3 was immunoprecipitated from a 293T cell line with a stable knock-down of Set9 using the K271me1 antibody and the IP eluates were analyzed by western-blot with the Flag antibody. (E) Endogenous FoxO3 is methylated by overexpressed Set9. 293T cells were transfected with Flag-Set9, and FoxO3 was immunoprecipitated with an antibody to total FoxO3. The IP eluates were analyzed by western-blot with the K271me1 antibody. (F) Endogenous FoxO3 is methylated by endogenous Set9. Methylated FoxO3 was immunoprecipitated from the stable Set9 shRNA cell line used in (D) using the K271me1 antibody. The IP eluate was analyzed by western-blot with an antibody to total FoxO3. ◁: FoxO3, *: non-specific band.

Figure 5. The methylated form of FoxO3 has similar localization properties to FoxO3

(A) Immunofluorescence in NIH 3T3 cells expressing FoxO3-GFP (green), using the K271me1 antibody (red). Nuclei were stained with DAPI (blue). White bar: 10 mm. (B) Quantification of FoxO3-GFP localization in U87 cells. Cells were co-transfected with FoxO3-GFP and empty vector (pcDNA), Set9 (WT or H297A), or a constitutively active version of Akt as a positive control (Akt CA). Localization was scored as ‘nuclear’, ‘cytoplasmic’ or ‘ubiquitous’. The data represent the mean and SEM from 3 independent experiments. In each experiment, at least 200 cells per condition were scored. *** p<0.001, one way ANOVA, Bonferroni post-test (C) Chromatin fractionation of 293T cells co-transfected with Flag-FoxO3 and Flag-Set9. Cells were grown in media with 10% FBS (Normal Serum) or in media without serum in the presence of 20 μM LY294002 (Serum Starvation+LY294002) and then fractionated. A subset of the fractions was treated with micrococcal nuclease to digest the chromatin. Presence of chromatin-bound proteins in fractions was assessed by western-blot with antibodies to the core nucleosome histone H2B (H2B). S2: detergent lysis, S3: hypotonic lysis fraction, Ch: chromatin fraction, Nuc: supernatant from nuclease digest of chromatin.

Figure 6. The methylated K271 peptide does not bind to kown methyl-lysine binding domains

Twenty-one amino acid peptides surrounding K271, either mono-methylated (left) or unmethylated (right) were incubated with potential methyl-lysine and acetyl-lysine binding domains on a protein array. GST-Alexa Fluor 555 is used as a positive control. A detailed description of the binding domains present on the protein array is provided in the lower panels.

Figure 7. Methylation of FoxO3 at lysine 271 promotes FoxO3 degradation, while increasing FoxO3 transcriptional activity

(A) Methylated FoxO3 is less stable than the unmethylated form. 293T cells were co-transfected with Flag-FoxO3 and Flag-Set9 (WT or H297A). The cells were then treated with MG132 (20 μM), ALLN (40 μM) or PSI1 (40μM) for 16 hours prior to lysis. Extracts were analyzed by western-blot with antibodies to K271me1, Flag, c-Jun, and b-actin. (B) Western-blots from (A) were analyzed by densitometry and the ratio of intensity in the presence or absence of proteasome inhibitor was compared after normalization to b-actin. * p< 0.05, **p<0.01, One way ANOVA, Bonferroni post test. (C) K271R is more stable than wild type FoxO3. 293T cells were transfected with Flag-FoxO3 (WT or K271R) together with Flag-Set9 (WT or H297A). Cells were treated for the indicated length of time with 20 μM cycloheximide (CHX) and FoxO3 expression was tested by western-blot with antibodies to Flag. (D) Western-blots from (C) were analyzed by densitometry at each time point of cycloheximide treatment. Results were normalized to the untreated well for each condition. (E) Set9 increases the transcriptional activity of the constitutively nuclear form of FoxO3. Flag-Set9 (WT or H297A) was co-expressed with Flag-FoxO3 (WT, K271R, TM [T32A/S253A/S315A], or TM K271R [T32A/S253A/K271R/S315A]) in 293T cells together with a 6xDBE-luciferase reporter construct, and luciferase activity was measured. The data represents the mean and SEM from 5 independent experiments, conducted in triplicate. All samples were normalized to the Flag-FoxO3 WT + Flag-Set9 WT condition. * p< 0.05, ** p< 0.01, *** p< 0.001, ns: p> 0.05, One way ANOVA, Bonferroni post test.