FOXO3 longevity interactome on chromosome 6

Abstract

FOXO3 has been implicated in longevity in multiple populations. By DNA sequencing in long-lived individuals, we identified all single nucleotide polymorphisms (SNPs) in FOXO3 and showed 41 were associated with longevity. Thirteen of these had predicted alterations in transcription factor binding sites. Those SNPs appeared to be in physical contact, via RNA polymerase II binding chromatin looping, with sites in the FOXO3 promoter, and likely function together as a cis-regulatory unit. The SNPs exhibited a high degree of LD in the Asian population, in which they define a specific longevity haplotype that is relatively common. The haplotype was less frequent in whites and virtually nonexistent in Africans. We identified distant contact points between FOXO3 and 46 neighboring genes, through long-range physical contacts via CCCTC-binding factor zinc finger protein (CTCF) binding sites, over a 7.3 Mb distance on chromosome 6q21. When activated by cellular stress, we visualized movement of FOXO3 toward neighboring genes. FOXO3 resides at the center of this early-replicating and highly conserved syntenic region of chromosome 6. Thus, in addition to its role as a transcription factor regulating gene expression genomewide, FOXO3 may function at the genomic level to help regulate neighboring genes by virtue of its central location in chromatin conformation via topologically associated domains. We believe that the FOXO3 'interactome' on chromosome 6 is a chromatin domain that defines an aging hub. A more thorough understanding of the functions of these neighboring genes may help elucidate the mechanisms through which FOXO3 variants promote longevity and healthy aging.

Keywords: FOXO3; fluorescence in situ hybridization; gene-gene interactions; single nucleotide polymorphisms; transcription factor binding.

Figure 1

FOXO3 functional variant identification strategy. Our original case–control study involving 530 participants identified three SNPs associated with longevity. We then sequenced DNA from 95 participants who were ≥95 years old and identified 110 SNPs that were used to screen the RegulomeDB and HaploReg databases for overlap with functional features. By limiting our search to only those SNPs that were present in our study population at a minor allele frequency (MAF) of ≥0.05, we were able to reduce the number of candidates from 1753 to 110. We used the sequencing data to identify proxies for SNPs that were not genotyped in an effort to prioritize SNPs based on significance (P‐values) and effect (OR). We also performed a longitudinal study to examine healthy aging phenotypes using seven SNPs – rs2253310, rs2802288, rs9384683, rs2802292, rs2764264, rs3800230, and rs1935952. Abbreviations: MAF, minor allele frequency; −log P, −log10 P‐value; OR, odds ratio for longevity; TFBS, transcription factor binding site.

Figure 2

Recognition logos for transcription factor binding sites in FOXO3. There were 13 SNPs that modify 18 transcription factor binding sites (i.e., canonical sequences). Their logos (colored) are shown, with the height of the nucleotide designating the strength of requirement for efficient binding. The genomic sequence (black) is shown below the logo. The SNP rs768023 modifies both FOXA and HDAC2 sites where the major allele is an A and minor allele a G, which is predicted to eliminate binding at both sites. The SNP rs1536057 modifies NRF1, E2F, and POU5F1 binding sites, in which the major (C) allele is replaced by a T. The minor allele is predicted to abolish NRF1 and E2F binding but create a POU5F1 site. The minor allele (C) of rs2253310 would create a TFCP2L1 site. The minor allele (G) of rs12212067 would create an MZF1 site. The minor allele (C) of rs1935952 would abolish an MZF1 site. The minor allele (T) of rs1536057 would abolish NRF1 and E2F sites and create a POU5F1 site. The minor allele (C) of rs2764264 would abolish an NKX3 site. The minor allele (C) of rs2253310 would create a TFCP2L1 site. The minor allele (G) of rs17069665 would abolish a TFE site. The minor allele (G) of rs3800230 would abolish a FOXP1 site. The minor allele (C) of rs9398171 would create NR2F1 and HNF4 sites, but abolish an HNF6 site. The minor allele (G) of rs12202234 would abolish an HNF3 site. The canonical recognition motifs are shown in color, where the height of the letter indicates the strength of the requirement for that nucleotide. The red box denotes the location of the SNP in the recognition site, and the sequence below the motif is the genomic FOXO3 sequence. R = ‘A’ or ‘G’; Y = ‘C’ or ‘T’; S = ‘G’ or ‘C’; W = ‘A’ or ‘T’; and K = ‘T’ or ‘G’.

Figure 3

Effects of SNPs on chromatin contacts in FOXO3 and neighborhood. (a) Aggregation of transcriptional components in FOXO3. Proposed mechanism whereby the pro‐longevity haplotype influences the aggregation of transcription factors (blue diamonds) with the FOXO3 promoter (green star). The aggregation promotes FOXO3 to migrate toward the transcriptional apparatus (gray sphere) and to promote aggregation of neighboring genes. For simplicity, only eight transcription factors/TFBSs are shown. (b) Aggregation of FOXO3 neighboring genes. The coalescence of all the FOXO3 transcriptional components would further integrate chromatin contacts (i.e., looping via topologically‐associated domains [TADs]) between neighboring genes on chromosome 6q21 to enable transcriptional control (transcription apparatus, gray sphere) in specific tissues. Colored circles denote FOXO3 and, for simplicity, only four of its 46 neighboring genes on chromosome 6q21.

Figure 4

Effect of stress on FOXO3 interaction with neighborhood genes. FISH was used to localize FOXO3 and the two most distant genes, HACE1 and LAMA4, in the FOXO3 neigborhood in lymphoblastoid cell lines in response to cellular stress caused by 200 μm H₂O₂ + serum deprivation. (a) Shows the order of the three genes and color of fluorescent dye. (b) Shows the measurements taken. (c) Is an example of FISH results for one of ten lymphoblastoid cell lines from different subjects. (d) Depicts a model in which neighboring genes aggregate to form a stronger transcription complex in response to stress (only the three genes indicated, of 46 in the region, are shown).