The free radical theory of aging is currently the leading hypothesis for a fundamental mechanism of aging. It potentially unifies aging mechanisms across an array of species [1–5]. The theory postulates that reactive oxygen species (ROS) such as superoxide and hydroxyl radicals, mainly generated as toxic by-products of normal cellular metabolism during aerobic respiration in mitochondria, give rise to damage to wide variety of macromolecules, including DNA. Animals possess enzymes that scavenge ROS. Among the most important of such antioxidant defenses are the superoxide dismutases (SOD), which facilitate the dismutation of superoxide free radical (O₂ ^-.) [6]. There are three types of SODs in higher eukaryotes, a copper/zinc containing cytoplasmic enzyme, SOD1; a manganese containing mitochondrial enzyme, SOD2; and a copper/zinc containing SOD3 that is prevalent in extracellular spaces [6, 7].

SOD2 is constitutively expressed in most cells [8]. SOD2 gene dosage correlates with longevity in fruit flies and in mice [9]. Mice that are heterozygous for an SOD2 knockout accumulate excessive damage to mitochondrial DNA as they age [10]. In these mouse lines, there were no compensatory increases of other antioxidant enzymes such as SOD1, catalase or glutathione peroxidase [11]. There is little understanding of the role of SODs in the modulation of longevity or health status in aging humans. Human disorders due to the SOD2 mutations have yet to be identified. However, a number of studies describe associations of SOD2 polymorphisms (particularly Ala16Val) with age-related disorders. One such study of elderly people with an Ashkenazi ethnic background suggested a possible association between Val at codon 16 of SOD2 and longevity [12]. Individuals who were homozygous for the valine allele at position 16 (also described as the -9 position of the mitochondrial targeting sequence) were shown to have an increased risk of non-small cell lung cancer [13, 14]. The Val allele may also be a predisposing factor for diabetic polyneuropathy (DPN) in Russians with type 1 diabetes mellitus [15]. Finally, homozygosity for the Val allele was also associated with non-familial idiopathic dilated cardiomyopathy among Japanese [16]. This was thought to be due to a decreased efficiency of SOD2 transport into mitochondria [16].

In contrast to the above reports, the alanine allele has been associated with high risk of breast cancer among U.S. and Finnish Caucasians [17, 18] and with an increased risk of developing psoriatic arthritis in the Taiwanese population [19].

As is typical of such association studies, most have not been confirmed in independent populations. Linkage studies are therefore of interest. A particularly cogent example is a report of a genome scan of microsatelite markers among hypertensive African-Americans; linkage was described at 6q24 and 21q21 [20], with Z scores, respectively, of 4.14 and 4.34. Markers associated with hypertension in African-Americans are between 112 cM and 166 cM of 6q. SOD2 is located at 160 cM.

The National Long-Term Care Survey (NLTCS) reports on the functional status of cross-sectional and longitudinal population-based samples from U.S. Medicare recipients ≥ age 65 [21]. During the 1999 wave of that survey, biological samples were obtained. This provided an opportunity to examine SOD2 polymorphisms among different ethnic groups.

The T9/T10 SOD2 polymorphism was newly identified in our laboratory in the course of screening of the 1999 NLTCS population for SOD2 polymorphisms. Our analysis indicated that the T9 allele is highly prevalent in elderly African-Americans as compared to elderly Caucasian-Americans (p < 0.0001). The location of T9/T10 at an intron/exon boundary region suggested a role in pre-mRNA processing. An exon trap analysis was quite consistent with this interpretation. There are a number of examples in the human genetics literature in which intron polymorphisms appear to have altered the ratio of wild type and alternative splicing patterns [22, 23]. In some cases, altered splicing was shown be responsible for the pathogenetic mechanism leading to disease. Interesting examples are the sequence variations at the 3' end of intron 8 of cystic fibrosis transmembrane conductance regulator (CFTR). These are thought to affect splicing of exon 9, resulting in the reduction of functional CFTR [24]. For the case of our SOD2 T9/T10 polymorphism, however, it was not possible to confirm the splicing bias using RNA from intact cells (lymphoblastoid cell lines). Similarly, we were unable to determine what fractions of the transcript lacking exon 4 were derived from T10 and T9 alleles in the heterozygote.

Previous work from our laboratory showed that intronic polymorphisms generated cryptic splicing products by exon trapping or by assays utilizing a mini gene and RT-PCR, but there was little evidence that this occurred in vivo [25]. The most likely interpretation of this discordance in the results of test tube-based versus cell-based methods is that the former exclude important cis-acting domains residing in more distal domains of DNA. In addition, trans-acting factors may differ among different cell types and species. These elements are likely to be capable of modulating splicing.

Although we have so far failed to provide molecular evidence for a functional distinction between the polymorphic allele of interest, given the striking difference in prevalence among African-Americans as compared to Caucasian-Americans, it will be important to carry out appropriate epidemiological studies of the impact of the two alleles (and, eventually, contrasting haplotypes), upon health and survival in the two ethnic groups. The NLTCS should provide valuable material for such an investigation, as there is linkage to Medicare data and data on functional status and survival. A definitive study, however, must await the availability of biological material from future cohorts, as the 1999 wave of the study could not provide biological specimens from many institutionalized subjects.

Perhaps the most immediate question would be the potential impact of this allelic variation of SOD2 upon susceptibility to hypertension among African-Americans. First, hypertension is pervasive in this population. Secondly, it has deleterious effects upon multiple organs. Third, while susceptibility to hypertension is clearly modulated by multiple genetic loci, a region of chromosome 6 that encompasses the SOD2 locus is among those that have been implicated [26]. Fourth, African-Americans provide an excellent opportunity to carry out admixture mapping (reviewed by [27, 28]. It has been estimated that ~75% of the alleles of African-Americans are of African ancestry whilst ~25% are of European ancestry [29]. Gene scans on the parental and admixed populations were successfully employed to identify hypertension loci by Zhu et al. (2005). The most robust finding was a region on the long arm of chromosome 6 that included the SOD2 locus. While the region contains many genes, potentially including more than one susceptibility genes, our observations suggest that one important candidate gene is SOD2.

We demonstrated that the 10T/9T polymorphism in the SOD2 intron 3 clearly results in alternative splicing of exon 4 by an in vitro exon trap assay using minigenes. This difference was masked when endogenous genes were expressed within cells. Over-representation of 9T alleles in African-Americans nevertheless raises the possibility that this intron 3 polymorphism or polymorphisms in linkage disequilibrium may be involved in the altered susceptibilities to a range of disorders that are more common in African-American populations.