Extreme longevity variants at the FOXO3 locus may moderate FOXO3 isoform levels

The rs2802292, rs2764264 and rs13217795 variants of FOXO3 have been associated with extreme longevity in multiple human populations, but the mechanisms underpinning this remain unclear. We aimed to characterise potential effects of longevity-associated variation on the expression and mRNA processing of the FOXO3 gene. We performed a comprehensive assessment of FOXO3 isoform usage across a wide variety of human tissues and carried out a bioinformatic analysis of the potential for longevity-associated variants to disrupt regulatory regions involved in isoform choice. We then related the expression of full length and 5′ truncated FOXO3 isoforms to rs13217795 genotype in peripheral blood and skeletal muscle from individuals of different rs13217795 genotypes. FOXO3 isoforms displayed considerable tissue specificity. We determined that rs13231195 and its tightly aligned proxy variant rs9400239 may lie in regulatory regions involved in isoform choice. The longevity allele at rs13217795 was associated with increased levels of full length FOXO3 isoforms in peripheral blood and a decrease in truncated FOXO3 isoforms in skeletal muscle RNA. We suggest that the longevity effect of FOXO3 SNPs may in part derive from a shift in isoform usage in skeletal muscle away from the production of 5′ truncated FOXO3 isoforms lacking a complete forkhead DNA binding domain, which may have compromised functionality. Supplementary Information The online version contains supplementary material available at 10.1007/s11357-021-00431-0.


Introduction
Human ageing is highly heterogeneous. Some people reach old age with good maintained health and quality of life, whilst others succumb to age-related diseases whilst still relatively young. The molecular basis behind these differences is still unclear. Long life often goes hand-in-hand with avoidance of agerelated disease, in a phenomenon termed compression of morbidity [12]. This suggests that the factors that govern lifespan also govern 'health span'. Although environment and lifestyle contribute the largest component to successful ageing, genetic influences are also apparent. All-cause mortality was seen to decrease by 19% and 14% for the children of longlived mothers and long-lived fathers respectively, compared to their cohabiting spouses [8]. Offspring of long-lived parents demonstrated lower incidence of cancer, diabetes, heart disease and stroke [8] and were also demonstrated to demonstrate lower rates of age-related cognitive decline [9]. The heritability of longevity is reported to be ~ 16% [19]. These observations suggest that some people are genetically predisposed to a long life.
Several genetic variants associated with longevity have been reported. A large meta-analyses of associations with parental lifespan carried out in 1 million participants reported associations with several loci including those containing the CDKN2B-AS1, ATXN2/BRAP, FURIN/FES, ZW10, PSORS1C3, ABO, ZC3HC1, IGF2R EBF1 and FOXO3 genes [30]. FOXO3 (Forkhead Box O3) is one of four FOXO human homologues of Daf16, the first gene to be associated with lifespan in the nematode C. elegans [20]. It is a multifunctional transcriptional regulator with roles in multiple processes such as apoptosis, insulin signalling, metabolism, immunity and inflammation, proteostasis and autophagy, response to cellular stress and control of cellular differentiation [26]. Associations between genetic variation within or near the FOXO3 gene and longevity were first reported in a population of male American individuals of Japanese ancestry, which provided evidence for an association between rs2802292, rs2764264 and rs13217795 and extreme longevity [31]. These findings have now been replicated in 11 different studies in populations of diverse ancestries [26]. The strongest association reported is with variant rs2802292, located in intron 2 of the FOXO3 gene, which confers an odds ratio for extreme longevity of 2.75 and is also linked with biological markers of successful ageing and lower incidence of age-related disease [31]. This variant has also been linked with preservation of telomere length in the peripheral blood of individuals carrying longevity alleles at these loci [4].
Genetic variation can affect gene expression or activity by many mechanisms. Coding variants can change the amino acid sequence of the peptides produced, but the vast majority of variation associated with longevity, like most other traits, maps to non-coding regions of the genome [14]. Non-coding regions such as the 5′ and 3′ untranslated regions (UTRs), along with introns and intergenic regions often contain regulatory elements such as enhancers or sites of posttranscriptional modification. The FOXO3 longevity-associated variants are primarily located in or near intron 2 and do not alter the coding sequences of the gene. The underlying mechanism by which these variants lead to extreme longevity remains to be elucidated, although a search for linked functional variants indicated 13 putative regulatory SNPs that significantly modified 18 transcription factor/enhancer binding sites in or near FOXO3 [7]. Intronic variation is also commonly associated with alterations in alternative isoform usage, by virtue of disruption of splicing regulatory elements such as intron splicing enhancer or silencer elements [1]. ENSEMBL indicates that human FOXO3 gene encodes three protein-coding alternatively expressed isoforms with differences in upstream exon usage (ENST00000343882.10, ENST00000406360.2 and ENST00000540898.1), although these are poorly characterised and little studied.
Here, we aimed to (i) characterise an expression profile for alternatively-expressed FOXO3 isoforms in a panel of human tissues, (ii) to map the rs2802292, 1 3 rs2764264 and rs13217795 variants of FOXO3 (and their proxies) to potential splicing or expression regulatory elements that may control FOXO3 isoform expression and (iii) to relate the expression of variants located in potential isoform regulatory elements to FOXO3 genotype in human peripheral blood and skeletal muscle. We confirmed the presence of two classes of FOXO3 isoforms: long (FOXO3-FL) and short (FOXO3-TR), and determined that whilst FOXO3-FL isoforms are present in all most FOXO3expressing tissues, the profile of FOXO3-TR is more restricted. We determined that variant rs13217795 lies in a region of the genome with characteristics of an alternative promoter for FOXO3 with potential to regulate FOXO3-TR, and that genotype at this variant is associated with increased FOXO3-FL expression in peripheral blood and with decreased FOXO3-TR expression in human skeletal muscle. FOXO3-TR is predicted to encode a truncated FOXO3 protein, which lacks part of the forkhead domain and several important points of posttranslational modification. We propose that the longevity effect conferred by rs13217795 may arise from a change in the balance of FOXO3-FL and FOXO3-TR isoforms towards a decrease in the production of the FOXO3-TR, which is predicted to lack DNA binding and to have lower activity.

Design of qRT-PCR probes to long and short FOXO3 isoforms
The sequences of FOXO3 transcripts corresponding to long and short FOXO3 isoforms were accessed from the ENSEMBL genome browser (https:// www. ensem bl. org), and assays specific to the two full length isoforms (FOXO3-FL; ENST00000343882.10 and ENST00000406360.2) or the truncated FOXO3 isoform (FOXO3-TR; ENST00000540898.1) were designed. The FOXO3-FL assay was targeted over the exon 2/exon 3 boundary shared by both full length isoforms and thus did not differentiate between them. The FOXO3-TR assay was targeted over the exon 1a/exon 3 boundary and was specific to ENST00000540898.1 (Fig. 1). Probes and primers were as follows:

Validation of FOXO3 isoform-specific assays
The performance of FOXO3-FL and FOXO3-TR assays was assessed using standard curve analysis. Fig. 1 The structure of FOXO3 isoforms. The structure of the exon structure and potential regulatory regions of the FOXO3 gene are given here. Exons are given by black or grey boxes, and introns by a black line. The position of the FOXO3-FL and FOXO3-TR assays that span the exon 2-3 and exon 1a-3 splice junctions respectively are indicated. The potential pro-moter is indicated by the red line and the presence of a layered histone modification peak indicated in blue, as observed using https:// www. genome. ucsc. edu/. The positions of the extreme longevity index SNPs rs2802292, rs2764264 and rs13217795 are given by a black star, and the rs9400239 proxy variant of rs13217795 is given by a grey star are indicated 1 3 One hundred nanograms of skeletal muscle total RNA was reverse transcribed for each sample in a total volume of 20µl using the Evoscript cDNA synthesis kit (Roche, Burgess Hill, UK) according to the manufacturer's instructions. cDNA then underwent seven 1:2 serial dilutions to establish accuracy, sensitivity and linear range. Reactions contained 2.5 µl TaqMan Universal Mastermix II (Thermo Fisher, Waltham, USA) 0.9µM each primer, 0.25µM probe (Thermo Fisher, Waltham, USA), 1.25µl H 2 O and 1µl cDNA in a total volume of 5µl. Cycling conditions were 95 O C for 10 min prior to cycling, then 95 O C for 15 s and 60 O C for a minute for 40 cycles.
Characterisation of the expression profile of FOXO3-FL and FOXO3-TR in human tissues Samples from different human tissues (skeletal muscle, pancreas, lung, pons, ovary, oesophagus, adipose, bladder, thyroid, breast, testes, adrenal gland, small intestine, thymus, trachea, cervix, prostate, medulla oblongata, colon, stomach, postcentral gyrus, kidney, hippocampus, spleen, salivary gland, parietal lobe, liver, cerebellum, heart, cerebral cortex, placenta, temporal lobe, uterus, total brain, pituitary, bone marrow, pooled islets and blood) were obtained from commercially-sourced, tissue-specific RNA panels (Becton, Dickinson & Co., Franklin Lakes, NJ, USA: BioChain Institute, Newark, CA, USA: Ambion®, Austin, TX, USA). One hundred nanograms of total RNA was reverse transcribed for each sample in a total volume of 20µl using the Evoscript cDNA synthesis kit (Roche, Burgess Hill, UK) according to the manufacturer's instructions. PCR reactions contained 2.5µl TaqMan Universal Mastermix (no AMPerase) (Applied Biosystems, Foster City, USA), 0.9µM each primer, 0.25µM probe, 1.25µl H 2 O and 1µl cDNA reverse transcribed as above in a total volume of 5µl. PCR conditions were a single cycle of 95ºC for 10 min followed by 40 cycles of 95 °C for 15 s and 60 °C for 1 min. The amount of FOXO3-FL and FOXO3-TR isoforms in each tissue sample was then quantified by isoform-specific qRT-PCR by the Comparative Ct method relative to the geometric mean of Beta 2 Microglobulin (B2M), Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) and beta glucoronidase (GUSB) endogenous control gene expression, with adjustment for assay efficiencies. Expression levels were then normalised to the average level of FOXO3-FL isoforms over the entire tissue panel.

Bioinformatic analysis of potential FOXO3 regulatory regions
To determine for rs2802292, rs2764264 and rs13217795 to interfere with splicing or expression of the FOXO3 gene, we firstly determined all the proxy SNPS that were in complete linkage disequilibrium (D' = > 0.95, r 2 > 0.80) with these index SNPs using SNPsnap (https:// data. broad insti tute. org/ mpg/ snpsn ap/ about. html). Index SNPs and proxies were then mapped to the known regulatory regions of the FOXO3 gene with reference to the UCSC genome browser (https:// www. genome. ucsc. edu/). SNPs were designated for follow-up if they were located within a predicted promoter element or within 50 bp of a splice site.

Profiling of FOXO3 isoforms in peripheral blood samples
We next profiled FOXO3 isoform levels in RNA extracted from 51 previously genotyped peripheral blood samples from the Exeter 10,000/Peninsula Research Bank study (https:// exete rcrfn ihr. org/ about/ exeter-10000-prb/) selected on the basis of rs13217795 genotype. This collection is a cross sectional population study consisting of samples collected from volunteer individuals living in the South West of England and recruited since 2010. Whole blood samples were collected in 2011/2012 using the PAXgene system [5] and extracted using the PAXgene Blood RNA kit (Qiagen, Paisley, UK). Access to these anonymised samples was approved by the ethically approved Peninsula Research Bank Steering Committee (14/SW/1089). Our sample set consisted of 18 individuals homozygous for the major allele of rs13217795 (TT), 17 samples heterozygous for rs13217795 (CT) and 17 samples homozygous for the longevity allele of rs13217795 (CC). Participant characteristics are given in Table 1. Levels of FOXO3-FL and FOXO3-TR isoforms were then quantified relative to the geometric mean of the B2M and GAPDH genes by qRT-PCR as described above, with adjustment for assay efficiencies. Data were then normalised to the mean level of the full-length isoforms in the rs13217795 heterozygous samples. Statistical In total, 64 human skeletal muscle samples were obtained from previously published studies of human muscle metabolism [6,18,21,29]. Four samples were omitted due to unsuccessful genotyping. All volunteers abstained from alcohol and strenuous exercise for at least 48 h prior to the sample obtained. Samples were obtained from the vastus lateralis muscle and were dissected free of visible connective tissue, fat and blood immediately prior to snap freezing. Patient characteristics are given in Table 1. RNA was extracted from 20 mg homogenised muscle sample using TRIzol reagent (Thermo Fisher, Waltham, USA). RNA concentration was quantified by ND-2000 Nanodrop Spectrophotometer (NanoDrop,Thermo Fisher, Waltham City, USA). The genotype of each sample was determined by qPCR from the presence of heterogeneous RNA (hnRNA) in the RNA samples using 2.5 µl TaqMan Universal Mastermix II (Thermo Fisher, Waltham City, USA), 0.9µM of each primer and 0.2µM of each probe (genotyping assay ID = C_9174543_10) and 1 µl RNA in a total volume of 5 µl. Cycling conditions were 60 °C for 30 s, followed by 95 °C for 10 min prior to cycling. Samples were then cycled between 95 °C for 15 s and 60 °C for a minute for 40 cycles. Following this, samples were held at 60 °C for 30 s. FOXO3 isoform levels were quantified by qRT-PCR as described above. Levels of FOXO3-FL and FOXO3-TR isoforms were then quantified relative to the geometric mean of the B2M and GAPDH genes by qRT-PCR as described above, with adjustment for assay efficiencies. GUSB was omitted from this analysis as it proved unstable baseline in our sample set. Data were then normalised to the mean level of the full length isoforms in the rs13217795 heterozygous samples. Statistical significance was then determined by ANOVA comparing age, sex and body mass index (BMI) covariates by genotype. Significant covariates were taken into account within a univariate linear model using the IBM SPSS Statistics 26 program (IBM SPSS version 26 release 26.0.0.0, Armonk, NY).

Assessment of FOXO3-FL isoform expression by age in human peripheral blood
To determine whether FOXO3 isoform expression alters with age, we carried out an assessment of the association between age and FOXO3-FL isoform expression exclusively in the 16 individuals homozygous for the major (T/T) allele, to negate the effect of genotype by linear regression analysis. We were unfortunately unable to assess the effect of age on the truncated FOXO3-TR isoform since this isoform was only expressed in skeletal muscle, and the age range of the donors of our muscle samples was very limited (18 to 31 years). as well as in salivary and pituitary glands and whole blood. Levels of FOXO3-TR were highest in skeletal muscle, followed by many tissues of the endocrine, digestive or respiratory systems (pancreas, ovary, oesophagus, adrenal gland, adipose, testes, small intestine, thymus, trachea and lung) (Fig. 2).

Bioinformatic evaluation of FOXO3 SNPs for potential regulatory effects
We next assessed the potential for FOXO3 rs2802292, rs2764264 and rs13217795 SNPs and their tightly linked proxies to disrupt predicted regulatory regions. Variant rs2802292 had seven proxies in almost complete linkage disequilibrium (LD) (D' of 0.99 and r 2 from 0.89 to 0.99; Table 2). None of these variants, or rs2802292 itself, was located within 50 bp of a splice site or within a predicted promoter sequence ( Table 2). We identified no proxies in complete LD with rs2764264 and rs2764264 did not map close to any splice site or predicted regulatory region (Table 2). Conversely, rs13217795 demonstrated seven proxy SNPs in tight LD (D' = 0.96 to 0.99; r 2 = 0.84-0.95). Of these, although none mapped in proximity to a splice site, rs9400239 mapped to the alternative noncoding exon 1a of FOXO3-TR, and rs13217795 itself mapped to a predicted regulatory region for an alternative promoter with potential to regulate the expression of FOXO3-TR (Table 2; Fig. 1). As rs9400239 and rs13217795 are in almost complete LD, we elected to take forward rs13217795 for analysis on the basis that it will capture both SNPs.
The longevity-associated C allele of rs13217795 is associated with elevated expression of FOXO3-FL isoforms in peripheral blood We measured levels of FOXO3-FL and FOXO3-TR isoforms in RNA from human peripheral blood from the Exeter 10 K study, and related these to genotype at rs13217795. We identified that increased expression of the FOXO3-FL isoforms was associated with the extreme longevity 'C' allele, and that effects were apparent in both heterozygotes and longevity allele homozygotes, with effects demonstrating a dominant pattern, with no difference between individuals carrying either one or two C alleles (Fig. 3). In heterozygous individuals ('C/T'), the logged mean expression of FOXO3-FL isoforms was 1.05 (SD = 0.33) compared with − 0.73 (SD = 0.19) in individuals homozygous for the 'T' allele (p = 0.004). In individuals carrying two copies of the longevity 'C' allele, logged mean expression was 1.1 (SD = 0.40) compared with 0.73 (SD = 0.249) in individuals homozygous for the major 'T' allele (p = 0.003). We detected no expression of FOXO3-TR isoforms in peripheral 1 3 blood, consistent with our findings in the tissue panel (Fig. 2).
The longevity-associated C allele of rs13217795 is associated with decreased expression of FOXO3-TR isoforms in skeletal muscle We measured levels of FOXO3-FL and FOXO3-TR isoforms in RNA from human skeletal muscle samples, and related these to genotype at rs13217795. In contrast to our findings in in peripheral blood, we detected no association of FOXO3-FL isoforms with genotype at rs13217795 in skeletal muscle (Fig. 4). FOXO3-TR isoforms were expressed in muscle, consistent with our findings in the tissue panel (Fig. 2).
Furthermore, we determined that in the skeletal muscle of individuals carrying two copies of the longevity 'C' allele, the logged mean expression of the FOXO3-TR isoform was 0.07 (SD = 0.07) compared with 0.12 (SD = 0.06) in individuals heterozygous for rs13217795 (p = 0.013). Levels of FOXO3-TR isoforms also demonstrated a trend towards reduced expression in C/C individuals compared with individuals homozygous for the 'T' allele, although this was not statistically significant, probably due to the large spread of data in the 'T/T' individuals. Comparison of FOXO3-TR levels in individuals heterozygous for the C allele did not display a similar reduction relative to TT homozygotes; however, levels are if anything slightly elevated in heterozygotes. This may suggest that 2 copies of the C allele may be necessary to observe a reduction in level.
The full length FOXO3-FL isoforms display an association with age in human peripheral blood To determine whether the FOXO3-FL isoforms displayed any evidence of differential expression with age, we correlated FOXO3-FL expression in peripheral blood mRNA with participant age in individuals homozygous for the major (T/T) allele of rs13217795 by linear regression. We identified that as expected, FOXO3-FL isoform expression declined with age (beta coefficient = − 0.64; SE = 0.003, p = 0.012; supplementary figure S2).

Discussion
The FOXO3 gene is one of 4 human homologues of the C. elegans Daf-16 gene, which was amongst the first genes associated with lifespan in animal models [20]. FOXO3 belongs to a gene family which encodes 4 transcription factors (FOXO1, FOXO3, FOXO4 and FOXO6) which are conserved from nematodes to mammals [24]. The unique characteristics of FOXO's place them perfectly to act as cellular sensors, to allow cells to adapt and respond to their internal and external environment. Accordingly, FOXO proteins have important roles in multiple fundamental processes in cells, such as in control of metabolism, cell division and differentiation status and response to cellular stress [17] and unsurprisingly, have particular importance in regulation of ageing and longevity [25]. Genetic variation in the FOXO3 gene has been linked with extreme longevity in multiple human populations [26,31], although the molecular basis for this remains to be elucidated. Here, we have characterised the expression profile alternatively expressed isoforms of the FOXO3 gene across a panel of human tissues, mapped SNPs associated with extreme longevity in human populations to the regulatory regions of the FOXO3 gene and identified that individuals carrying extreme longevity alleles at rs13217795 demonstrate elevated expression of full length FOXO3 isoforms in peripheral blood and decreased expression of the truncated FOXO3 isoform in skeletal muscle. FOXO proteins share a Forkhead DNA binding domain, which is approximately 100 bp long, and consists of three alpha helices and three beta pleated sheets flanked by looped and winged sections [13].
FOXO proteins bind as monomers to their downstream targets, and have been reported to be capable of both transactivation through direct binding to targets [10], or to gene repression, through DNAindependent indirect mechanisms [3,11]. Transactivation or inhibitory activity can also be moderated by binding with cofactors such as SIRT1, CBP and p300 [27]. FOXOs are themselves regulated by posttranslational modifications such as phosphorylation, acetylation, methylation and ubiquitination [32]. The FOXO3-FL and FOXO3-TR isoforms demonstrate marked differences in their structure (Fig. 5). The full length isoforms contain the full length forkhead domain, whereas the truncated isoform contains only part of the S2 spacer, the S3 spacer and the 5th Helix H5, but is missing Helices H1 to H4 and spacer S1. The lack of a complete forkhead domain makes it unlikely that the truncated isoform would be able to efficiently bind DNA and regulate its downstream targets. The truncated FOXO3 isoform is also lacking some points of posttranslational modification including phosphorylation sites at T32 and T179, and S207. T32 is one of three FOXO3 sites modified by AKT and SGK, that has been shown to create binding sites for the chaperone protein 14-3-3, which binds FOXO Fig. 3 Relationship between FOXO3 genotype and FOXO3 isoform expression in peripheral blood RNA samples. The relationship between FOXO3 full length and truncated isoform levels and genotype at rs13217795 is given for peripheral blood samples originating from the Exeter 10 K study. Isoform identity and genotype are given on the X axis, and median relative expression in arbitrary units (AU) is given on the Y axis. Statistical significance of logged data is indicated by stars; *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001. Circles indicate major allele (T) homozygotes, squares indicate heterozygous samples (C/T) and triangles indicate minor allele (C) homozygotes. Error bars represent interquartile range. No expression of FOXO3-TR isoform was detected in peripheral blood Fig. 4 Relationship between FOXO3 genotype and FOXO3 isoform expression in skeletal muscle RNA samples. The relationship between FOXO3 full length and truncated isoform levels and genotype at rs13217795 is given for muscle samples originating from the SPADE, BIER, TIME and INDIA studies. Isoform identity and genotype are given on the X axis, and median relative expression in arbitrary units (AU) is given on the Y axis. Statistical significance of logged data is indicated by stars; *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001. Circles indicate major allele (T) homozygotes, squares indicate heterozygous samples (C/T) and triangles indicate minor allele (C) homozygotes. Error bars represent interquartile range proteins in the nucleus and allows their active export [2]. This may mean that the truncated isoform also has altered ability to translocate between nucleus and cytoplasm, which has been shown to be a key factor in stress responses and determination of lifespan in nematodes [22,28].
There may also be differences in miRNA regulation between isoforms, since the 3′ UTR of FOXO3-TR is truncated. The FOXO3-FL isoforms contain additional binding sites for isoforms contain additional binding sites for hsa-miR-29-3p, hsa-miR-155-5p, hsa-miR-9-5p, hsa-miR-27-3p, hsa-miR-122-5p, hsa-miR-142-3p.2 and hsa-miR-217, which are not present in FOXO3-TR. Although not all of these predicted binding sites have been functionally validated, inhibition of hsa-miR-155-5p has been shown to rejuvenate aged mesenchymal stem cells and enhances cardioprotection following infarction [16]. The FOXO3-TR isoform may therefore compete with the full-length transcripts for miRNAs that modulate (downregulate) FOXO3 expression and, hence, the amount of protein. The longevity (minor) allele of rs13217795 and/or its proxies may functionally increase the amount of protein by serving as a miRNA sponge. There are also interactions with circ-FOXO3, a circular RNA, which we have previously described to be associated with cellular senescence and parental longevity in humans [15]. CircFOXO3 has a role in cellular proliferation via sponging of miR-9-5p [23].
Our study is the first to characterise the alternatively expressed FOXO3 isoforms in humans, and the first to report a shift in FOXO3 isoform usage in human samples from different tissues in response to genotype at loci associated with extreme longevity. The tissue-specific effects we have noted also raise the possibility that different tissues may have different responses to genotype at such loci, particularly where the truncated isoform is predominant. Our work also has caveats; we have demonstrated these effects only at the level of the mRNA transcript, but others have documented the presence of a similar 5′ truncated FOXO3 protein isoform by Western blot in mouse [33]. We are also unable to precisely pinpoint the causal variant in or near FOXO3 that underpins the effect we note, because of the tight LD between variants, but the most likely candidate is rs13217795 since it lies in a region of the FOXO3 gene with characteristics of an alternative promoter. We cannot rule out an effect of its tightly linked proxy variant rs9400239, which lies in the alternatively expressed exon 1a of the FOXO3 gene. Exon 1a is a non-coding exon, so will not alter the coding sequence of the truncated isoform, but it is possible that it confers some alteration in regulatory potential. Previous work has suggested that some of the longevity-associated SNPs at the FOXO3 locus may act as a cis-regulatory unit that influences the expression level of neighbouring genes [7]. It is possible therefore that the longevity effect is due to both the potential alteration in overall FOXO3 activity brought by the change in expression of FOXO3 isoforms, but also the changes in the expression of other genes that are influenced by FOXO3.
We propose here a potential explanation for the association reported in multiple human populations between genetic variation in intron 2 of the FOXO3 gene and extreme longevity. We suggest Fig. 5 The functional domains of the FOXO3 protein in relation to isoform structure. The structure of the forkhead domain and position of functional posttranslational modifications within the FOXO3 gene relative to its isoform structure is given. The position of the forkhead domain is indicated by the red arrow. The five helices are given by blue boxes, and the spacer regions by orange boxes. The exon structure is given by black and grey boxes. Points of posttranslational modification at T32 and T179, and S207 are given by blue stars that rs13217795, or one of its tightly-linked proxies, is capable of bringing about an upregulation of fulllength FOXO3 isoforms in some tissues, with downregulation of truncated forms of FOXO3 being associated with inheritance of two copies of the longevity allele in others, most notably skeletal muscle. Some tissues may exhibit both phenomena, since we have observed differences between tissues in response. The truncated isoform of FOXO3 is likely to have compromised DNA binding potential due to its interrupted forkhead domain, and may also have altered subcellular trafficking because of disrupted posttranslational modification. We propose that in individuals carrying one or more alleles of the rs13217795 variant (or a tightly linked proxy SNP), the shift towards full-length FOXO3 isoforms and away from truncated transcripts brings about an enhancement of FOXO3 activity, with consequent effects on cellular processes involved in determination of lifespan.
Author contribution RF carried out the experiments, analysed the data and wrote the initial draft of the manuscript. TT contributed to data analysis and reviewed the manuscript. TSOJ, FBS and BTW contributed to study design, provided data and skeletal muscle samples and reviewed the manuscript. BAK contributed to study design, sought ethical approval, provided peripheral blood samples and reviewed the manuscript. TAD, RCA, DCW and BJW contributed to study design and reviewed the manuscript. LWH designed and managed the study, interpreted the data and reviewed the manuscript. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.