Introduction

The forkhead box O3A gene (FOXO3A) is a key regulator of the insulin receptor/insulin-like growth factor-I signaling pathway and participates in the modulation of lifespan in Caenorhabditis elegans and Drosophila.1, 2, 3, 4, 5 In humans, genetic variation in FOXO3A has been shown consistently in different populations to be associated with survival into old age.6, 7, 8, 9, 10, 11 The underlying molecular mechanisms, however, still remain to be elucidated because most of the longevity-associated single-nucleotide polymorphisms (SNPs) analyzed are located in intronic regions.6, 7, 8, 9, 10, 11 In order to identify the causative allelic variants that modulate gene functionality, studies are therefore needed that explore the full genetic variation of FOXO3A by in-depth resequencing of exons, exon–intron boundaries and the promoter.

Before embarking on such investigations, it has to be taken into consideration that a pseudogene, called FOXO3B, is located on chromosome 17p11. FOXO3B is 99% identical to FOXO3A over a stretch of approximately 7.1 kb of exonic sequence. The pseudogene comprises all the three exons of FOXO3A NM_001455 and lacks the two introns (Figure 1).12 Because exon 2 and 3 of FOXO3A are comparatively large (1.4 kb and 4.9 kb, respectively), the FOXO3A-specific intronic regions do not provide sufficient primer targets for the analysis of these two exons. For example, to generate high-quality reads with Sanger sequencing, PCR amplicons should not exceed 600 bases. However, because exon 2 and 3 are each larger than 1400 bases, their sequencing primers need to be placed within the exonic regions, despite their strong homology with FOXO3B. Nevertheless, for some sequencing or genotyping assays, it may be possible to place primers particularly on FOXO3A/FOXO3B sequence differences (which comprise only 1% of the critical regions) (Supplementary Figure S1). For the genetic investigation of the complete exonic region (7.1 kb), however, a conventional primer design, using the homologous exonic regions for primer placement, would most likely result in unspecific genotyping or mixed FOXO3A/FOXO3B sequencing amplicons. Unfortunately, the outcome cannot be predicted reliably by, for instance, in silico test PCR prior to the actual wet lab experiment.

Figure 1
figure 1

Illustration of the FOXO3A gene model showing the two isoforms NM_201559 (four exons) and NM_001455 (three exons) on chromosome 6. The exons are indicated by boxes and are presented in comparison with the FOXO3B pseudogene on chromosome 17. FOXO3B is 99% identical to FOXO3A over approximately 7.1 kb of related sequence comprising all the three exons of FOXO3A NM_001455 while lacking the two introns in-between. The physical positions (in kilobases) of the transcripts refer to the UCSC Genome Browser on the Human March 2006 Assembly.

Full size image

Here, we present an experimental set-up that allows FOXO3A-specific genetic investigation of exonic regions without FOXO3B ‘contamination’: with the generation of long-range products that span around 8000 bases, exon 2 and 3 can be amplified in a FOXO3A-specific manner using the non-homologous intronic regions for primer placement. Genotyping and sequencing experiments are described exemplarily for exon 3: first, exon 3 was amplified by long-range PCR with FOXO3A-specific primer placement in the nearby unique intronic regions (Supplementary Table S3) to ensure targeted amplification of the FOXO3A exonic region (ie, devoid of FOXO3B sequence). Next, the long-range amplicons were used as templates for genotyping and sequencing reactions. We tested the suitability of this experimental set-up for two SNPs located in exon 3 of FOXO3A, namely rs4945816 and rs4946936. These two SNPs were selected because they are the only ones in the FOXO3A exonic region that are listed on the dbSNP page as common SNPs in CEU-samples (http://www.ncbi.nlm.nih.gov/projects/SNP/).

Results and Discussion

Sequenom test set-up: differences in FOXO3A specificity

Genotyping with Sequenom technology (Sequenom, Hamburg, Germany) was carried out for both SNPs on 23 individuals employing three different templates for comparison: FOXO3A long-range products of exon 3 (5.523 kb), FOXO3B long-range products (8.540 kb) and genomic DNA (gDNA). Table 1 shows exemplary data obtained for nine individuals. Data on the remaining 14 individuals are given in Supplementary Table S1. Sequenom PCR primers for both SNPs were placed on FOXO3A/FOXO3B sequence differences (Supplementary Figure S1). Despite this particular primer placement, comparison of genotyping results for SNP rs4945816 obtained from the three different templates shows that the seemingly heterozygous genotypes derived from the gDNA template (samples 1, 3, 4, 5 and 8) are due to unspecific typing and represent mixtures of FOXO3A and FOXO3B genotypes (Table 1a, Supplementary Table S1a). Fictitious heterozygosity becomes evident only if FOXO3A and FOXO3B genotypes differ from each other, as it was the case for samples 1, 3, 4, 5 and 8 (Table 1a). On the other hand, the second SNP assay (rs4946936) demonstrated FOXO3A specificity by presenting the same genotyping results for the FOXO3A long-range product and the gDNA template, both of which differed from FOXO3B (Table 1a). The differences in FOXO3A specificity for the two Sequenom assays clearly indicate that FOXO3A-assay specificity cannot be predicted reliably prior to the actual wet lab experiment. Only the comparison of our genotyping results for the gDNA template with those derived from the FOXO3A and FOXO3B long-range products demonstrated that the assay for rs4946936 was FOXO3A-specific, whereas the assay for rs4945816 was not (Table 1 and Supplementary Table S1). Consequently, to obtain FOXO3A-specific results for rs4945816 and rs4946936 with the applied Sequenom assays, the typing of rs4945816 would need to be performed on FOXO3A long-range products, whereas rs4946936 could be investigated on gDNA.

Table 1 Genotyping and sequencing data for rs4945816 and rs4946936 on nine exemplary individuals
Full size table

FOXO3A-specific sequencing

We further showed that our experimental set-up is also suitable for FOXO3A-specific sequencing of the problematic homologous exonic regions. To this end, we sequenced exon 3 of FOXO3A in the same 23 individuals using FOXO3A long-range products as targets for sequencing primers instead of gDNA. The subsequent SNP calling for rs4945816 and rs4946936 yielded the same FOXO3A-specific genotypes (Table 1b, Supplementary Table S1b) as before (Table 1a, Supplementary Table S1a).

FOXO3A-specific genotyping

For large-scale case–control association studies, it might be too time- and money-consuming to produce FOXO3A-specific long-range PCR products for thousands of samples to obtain the FOXO3A-specific genotyping results. Hence, in order to obviate the need for such products, our study design can also be used to test commercially available assays (eg TaqMan; Applied Biosystems, Foster City, CA, USA) for primer specificity: First, the assays of interest are typed in a small number of samples using the gDNA, FOXO3A and FOXO3B templates. The genotypes derived from the three different templates are then compared to evaluate assay specificity. Next, if the assays tested on gDNA show the same genotypes as those from FOXO3A long-range products, the custom assays can also be used for genotyping larger samples using gDNA as a template. We applied this set-up to evaluate FOXO3A specificity for one custom TaqMan assay (rs4945816; hCV27392797) and one self-designed TaqMan assay (rs4946936; rs4946936_SDA). Both TaqMan assays demonstrated FOXO3A specificity by yielding the same genotyping results as obtained from the FOXO3A long-range product (Table 1c, Supplementary Table S1c). As these two assays proved to generate reliable genotypes from normal gDNA, we decided to investigate them subsequently in our whole-study sample comprising 747 centenarians and 1102 controls. Both SNPs showed a highly significant association with longevity in an allele-based case-control analysis (rs4945816, P=0.0008; rs4946936, P=0.0008).

In silico investigation of functional relevance

Because both SNPs are located in the 3′ untranslated region of FOXO3A, they could be of functional relevance. Applying different prediction tools for in silico analysis, we tested the two SNPs rs4945816 and rs4946936 for potentially functional importance. There were no functional or structural effects, microRNA-binding sites or splice sites predicted (Supplementary Table S2). Only those tools that analyzed splicing regulation elements (SREs) detected differences between the reference and mutated sequences. However, these results have to be seen with caution because SREs are known to be less conserved, and consequently, this kind of tool is unlikely to generate robust and accurate results.13 The two SNPs rs4945816 and rs4946936 have been investigated exemplarily to test our set-up. Future studies that aim to identify new SNPs of functional relevance in the FOXO3A exonic region should consider the aspects presented here.

Conclusion

Taken together, our results clearly show that, for reliable genetic investigation of FOXO3A exonic regions, a special study design is mandatory accounting for the FOXO3B sequence homology. This finding is of particular importance because some studies have already used unspecific primers for the amplification of FOXO3A exonic regions. For example, Mikse et al14 applied various primer sets to determine FOXO3A deletions in tumor cells by qPCR. When we tested these pairs with the UCSC in silico Test-PCR (http://genome.ucsc.edu/cgi-bin/hgPcr?command=start), the output showed additional binding to FOXO3B for two of them (see Supplementary Materials and Methods).

The 99% FOXO3A/FOXO3B homology does not affect the former FOXO3A longevity association findings because, as has been mentioned above, most of the associated and validated SNPs are located in FOXO3A-specific intronic regions. Nevertheless, the pseudogene has to be taken into consideration in future FOXO3A genetic studies. Furthermore, the presented strategy could also be applied to other gene pairs or pseudogenes with high sequence homologies that render the design of PCR primers or reliable genotyping assays difficult, for instance the cytochrome P450 genes CYP2D6, CYP2D7, CYP2D8 or the RH blood-group genes (RH).15, 16