FOXO3A-short is a novel regulator of non-oxidative glucose metabolism associated with human longevity

Abstract

Intronic single-nucleotide polymorphisms (SNPs) in FOXO3A are associated with human longevity. Currently, it is unclear how these SNPs alter FOXO3A functionality and human physiology, thereby influencing lifespan. Here, we identify a primate-specific FOXO3A transcriptional isoform, FOXO3A-Short (FOXO3A-S), encoding a major longevity-associated SNP, rs9400239 (C or T), within its 5' untranslated region. The FOXO3A-S mRNA is highly expressed in the skeletal muscle and has very limited expression in other tissues. We find that the rs9400239 variant influences the stability and functionality of the primarily nuclear protein(s) encoded by the FOXO3A-S mRNA. Assessment of the relationship between the FOXO3A-S polymorphism and peripheral glucose clearance during insulin infusion (Rd clamp) in a cohort of Danish twins revealed that longevity T-allele carriers have markedly faster peripheral glucose clearance rates than normal lifespan C-allele carriers. In vitro experiments in human myotube cultures utilizing overexpression of each allele showed that the C-allele represses glycolysis independently of PI3K signaling, while overexpression of the T-allele represses glycolysis only in a PI3K-inactive background. Supporting this finding inducible knockdown of the FOXO3A-S C-allele in cultured myotubes increases the glycolytic rate. We conclude that the rs9400239 polymorphism acts as a molecular switch which changes the identity of the FOXO3A-S-derived protein(s), which in turn alters the relationship between FOXO3A-S and insulin/PI3K signaling and glycolytic flux in the skeletal muscle. This critical difference endows carriers of the FOXO3A-S T-allele with consistently higher insulin-stimulated peripheral glucose clearance rates, which may contribute to their longer and healthier lifespans.

Keywords: FOXO; FOXO3A; PI3K; SNP; aging; glycolysis; insulin; skeletal muscle.

FIGURE 1

The FOXO3A locus encodes a novel transcript which contains a longevity SNP. (a) A genomic view of the FOXO3A locus rendered by the UCSC Genome Browser with FOXO3A transcripts and lead longevity SNPs annotated (Hg38). (b) A non‐scaled cartoon representation of the known FOXO3A coding exons and splice junction in blue, UTRs in red and the novel exon and splice junction that comprise FOXO3A‐short in green. Start codons (ATG) for each isoform are indicated. (c) Linkage disequilibrium (LD) analysis of the FOXO3A longevity SNPs using data supplied from the 1000 Genomes Project (Phase 3 Version 5) and represented as correlation coefficients (r ²) generated using the LDmatrix tool (https://ldlink.nci.nih.gov/?tab=ldmatrix). Linkages for Europeans are highlighted in blue, East Asians are highlighted in green.

FIGURE 2

FOXO3A‐S is predominantly expressed in the skeletal muscle. (a) GTEx version 8 (https://www.gtexportal.org) human tissue mRNA expression data for both FOXO3A and FOXO3A‐S depicted in TPM (transcripts per million). Data labels above each bar are the exact TPM value. (b) Representative light microscopy images from a differentiation time series of the human myoblast line LHCNM2. Day 0 is under normal myoblast growth conditions and Day 6 is under differentiation conditions. (c) RT‐qPCR measurement of FOXO3A, FOXO3A‐S, and MYH8 expression from the LHCNM2 differentiation time course imaged in (b) represented as ∆Ct. Error bars are SD. (d) RT‐PCR of pooled skeletal muscle mRNA from multiple individuals in multiple species for FOXO3A‐S transcript using primers designed to conserved genomic sequence. 18S rRNA served as a loading control.

FIGURE 3

The FOXO3A‐S transcript encodes different FOXO3A‐related proteins depending on the genotype of the rs9400239 SNP. (a) The complete sequence of exon 2A of FOXO3A‐S and partial sequence of the common FOXO3A/FOXO3A‐S exon 3. Highlighted in red is the rs9400239 SNP (C/T), in green the putative ATG start codon for FOXO3A‐S and in blue the splice junction between exon 2A and exon 3. In bold/underline are stop codons in‐frame with the putative ATG start/canonical FOXO3A reading frame. (b) Annotation of the FOXO3A‐S protein sequence relative to FOXO3A. The forkhead domain is colored blue, nuclear localization signals are red, nuclear export signals are green, the transactivation domain is yellow, and the three Akt phosphorylation sites are purple. (c) Western blot of the products of in vitro transcription/translation reactions performed with FOXO3A‐S clones. Two of the clones are full‐length FOXO3A‐S transcripts containing either rs9400239 variant (3A‐S(C) and 3A‐S(T)) ‐ the rs9400239 variant being the only difference between them. The third clone is one starting exactly from the first ATG (3A‐S(ATG)) which truncates the rs9400239 position and remaining 3’ UTR. (d) Western blot from HEK293T transfections of the same clones used in (c). (e) Western blot of LHCNM2 myotube lysates (Day 7) derived from either control (NT) or a FOXO3A/FOXO3A‐S‐targeting spJCRISPR construct single‐cell clone that was further transduced with Dox‐inducible empty vector (EV), 3A‐S(C) or 3A‐S(T) clones and then treated or not with 0.1 μg/ml Dox on Day 5. Nuclear Lamin A/C is shown as a loading control. (f) Western blot of immunoprecipitations of endogenous FOXO3A and FOXO3A‐S from LHCNM2 myotube lysates (Day 7) collected from spJCRISPR single‐cell clones. All myotube cultures were treated with 200 nM of the proteasome inhibitor Bortezomib for 18 h (starting Day 6) prior to harvest. (g) Western blots of lysates from cycloheximide (CHX) pulse‐chase experiments performed with the LHCNM2 lines from (e). Myoblasts were treated with 0.1 μg/ml Dox for 48 h post‐seeding, pulsed with 300 μM CHX then harvested at indicated time points (T = 0 no CHX). Western blots are representative of three independent experiments. (h) Plot of the band densitometry quantification of Western blots from (g). FOXO3A‐S protein levels were first normalized to corresponding Lamin A/C controls which were then normalized to T = 0 within each series. Plots are average of three replicate experiments with error bars being SD. (i) FOXO3A‐S protein half life calculated by averaging the computed half lives from each time point within (h). Error bars are SD.

FIGURE 4

The FOXO3A‐S T‐allele is positively associated with peripheral glucose clearance in vivo. (a) Association of the rs9400239 genotype with the Rd clamp glucose disposal parameter. The association was calculated using a mixed‐ANOVA regression analysis correcting for age, sex, BMI, twin pair, and zygosity. (b) RT‐qPCR for FOXO3A‐S mRNA from muscle biopsy pairs harvested before (Basal) and during insulin infusion (insulin) in 139 healthy individuals. Expression was calculated relative to 18S rRNA using the ∆Ct method. Error bars are expressed as SEM, and the significance was calculated using the two‐tailed paired Student's t‐test. (c) Allele‐specific RT‐qPCR of FOXO3A‐S mRNA under basal and insulin‐stimulated conditions correlated with Rd clamp. The expression of each allele was corrected to 18S rRNA using the ∆Ct method. A mixed ANOVA regression was used to assess the correlation between allele‐specific values and Rd clamp correcting for age, sex, and BMI. Error bars are 95% CI.

FIGURE 5

FOXO3A‐S suppresses glycolysis in myotubes with PI3K‐dependency altered by the rs9400239 genotype. (a) Western blot of the inducible overexpression of 3A‐S(C) or 3A‐S(T) constructs in LHCNM2 myotubes. Dox (0.1 μg/ml) was applied on Day 5 of differentiation, and on Day 7, the myotube lysates were harvested. (b) Seahorse assays performed following the same differentiation and Dox treatment as in (a) with the addition of DMSO or 5 μM GDC‐0941 (PI3Ki) for 1 h prior to the assay. All basal ECAR values were calculated as a percentage of the −Dox/DMSO control for each respective assay. The presented assay is representative of three independent replicates. Error bars are SEM. (c) Comparison of the maximal ECAR values upon oligomycin injection attained from the same overexpression assays performed in (b). The max ECAR was calculated as a percentage of the −Dox/DMSO basal ECAR within each assay. Error bars are SEM. (d) RT‐qPCR for inducible shRNA‐mediated knockdown of FOXO3A and FOXO3A‐S. Myoblasts were differentiated to myotubes and on Day 2 2 mM IPTG was applied with RNA harvested on Day 6. Both FOXO3A and FOXO3A‐S expression in the +IPTG condition was quantified relative to the −IPTG control. Error bars are SD. (e) ECAR measurement in myotubes where FOXO3A and FOXO3A‐S was inducibly knocked down (IPTG Day 2 measured Day 6). Basal ECAR values were calculated as a percentage of the −IPTG control. The presented assay is representative of three independent replicates. Error bars are SEM.

FIGURE 6

Model of FOXO3A‐S action in longevity and peripheral glucose clearance. The FOXO3A‐S proteoforms either constitutively or conditionally limit skeletal muscle‐mediated glucose clearance under insulin‐stimulated conditions. Longevity (T‐allele) carriers express a PI3K‐repressible form of the protein allowing for higher glucose clearance rates with insulin challenge while C‐allele carriers have more restricted insulin‐stimulated glucose clearance because the C‐allele proteoform is refractory to PI3K signaling. Increased peripheral glucose clearance confers health benefits to T‐allele carriers, which allows for longer and healthier lives.