MtDNA genes, in contrast to nuclear DNA (nDNA)-encoded genes, are in full linkage disequilibrium. The mutation rate of the mtDNA is ~10 times faster than that of the nDNA and thus it is the most variable coding region in the human genome. Since mtDNA is maternally inherited, it is prone to genetic drift, resulting in large differences in patterns of genetic variability among and within populations . Such genetic drift often leads to difficulties in replicating results of mtDNA association studies among populations. Hence, we hypothesized that, due to its high genetic divergence among populations, a subset of mtDNA alleles with functional consequences will differentiate among distinct populations. Here we have shown that recent bottleneck events within the three studied Jewish populations (Ash, Seph and NAF), underlie marked differences in mtDNA diversity in three ethnically-related Jewish populations, resulting in increased frequency of genotypes in some populations, some of which may act as susceptibility factors to T2DM complications. Such was the case in the haplogroup N1b1 that was significantly under-represented in certain complications of the Ashkenazi population and not present in the NAF population.
In contrast to haplogroup N1b1, the haplogroups identified as factors with risk trends to some T2DM complications (haplogroup J1 and haplogroup aggregate HV*, with marginal significant values considering a Bonferonni corrected α<0.017) were present in all three studied populations. Nevertheless haplogroup J1, showing association with increased risk for T2DM nephropathy or retinopathy in the Ashkenazi population does not have enough power to replicate in the Seph and NAF populations, i.e. ~250 subjects in each of the complications groups to replicate the significant results of the Ash population (power of 80%, α<0.05 (two tailed)) (Figure 1). The significant over-representation of haplogroup aggregate HV* in certain complications of the NAF population is harder to interpret, since although there was enough power to detect its effect in the Ashkenazi population, i.e. ~70 subjects in each of the complication groups in order to detect significance with a power of 80% and α<0.05 (two tailed), it did not show the same tendency as in the NAF population. Nevertheless one should take into account that HV* is a haplogroup aggregate; hence the different bottleneck events leading to the establishment of the Ashkenazi and North African Jewish populations could result in different compositions of lineages comprising the HV* haplogroup aggregate in the two populations. Testing for this possibility needs further genotyping of HV*, requiring increased sample sizes of the studied populations.
Our findings support association of mtDNA common genetic variants with sub-phenotypes of T2DM. Interestingly, the inconsistency of mtDNA genetic association with complications of T2DM as found here was described for other phenotypes as well: While haplogroup J has been associated with successful longevity in northern Italians and the Finnish , it was not associated with successful longevity in southern Italians  and subhaplogroup J1, but not J2, was associated with successful longevity in the Northern Irish . Therefore, both differences in mtDNA sub-groups and their differences in response to environment appear to affect the relationship between mtDNA genotypes and phenotypes. Since all of the functional SNPs in particular mtDNA lineages would have a collective effect on mitochondrial function, many mtDNA haplogroups and sub-haplogroups might interact with environmental variation differently. Furthermore, this difference can be further complicated by the interaction of mtDNA encoded subunits, harboring functional SNPs, with nuclear DNA encoded subunits, harboring their own genetic variation. This interpretation applies to our observation that N1b1 reduces the risk to T2DM complications only in Ashkenazi Jews. In addition, the tendency to develop complications is an interplay of environment and genetics, hence it is not solely dependent on a particular haplogroup, and thus the absence of the N1b1 haplogroup in the NAF cohort is not expected to change significantly the overall risk do develop complications.
Since our study observed disease-association with a population specific haplogroup, the Ashkenazi specific N1b1, it was of importance to assess the functional potential of this haplogroup defining mutations. During our sequence analysis of haplogroup N1b we noticed that the Ash-specific sub-haplogroup N1b1 harbors an amino acid substitution in mtDNA position 4917 (Figure 2), which also defines haplogroup T that was previously associated with reduced sperm motility . Interestingly, our permutation test suggests with borderline significance that haplogroup T might be in association (p = 0.059) with some T2DM complications in the Seph population. Previously [30, 36] we showed that this mutation alters a highly conserved amino acid in the ND2 gene, hence suggesting a functional potential. Since haplogroups N1b1 and T stem from very different branches in the human mtDNA phylogeny it can be concluded that the 4917 mutation was established at least twice during human evolution. All these evidence imply that the change at position 4917 contributes to the protective effect against certain T2DM complications.
Along with haplogroup J1 association with other multi-factorial phenotypes [30, 37], and its effect on the penetrance of mutations causing the eye disorder LHON  our results support the premise that mutations defining this haplogroup affect OXPHOS. Similar to haplogroup N1b1, some mutations defining haplogroup J1 alter amino-acids with high conservation degree: (1) a transition in position 10398 which is a Thr114Ala replacement in the ND3 subunit of complex I, shown to alter mitochondrial matrix pH in cell-culture experiments ; and (2) a transition in position 13708, causing a Ala458Thr replacement in the ND5 complex I subunit. In addition, haplogroup J1 harbors the G3010A substitution located within the 12SrRNA gene. Although this mutation has been found in several haplogroups , it is possible that the combination of this mutation with the mutations that generally define haplogroup J underlies the phenotypic effect of J1 in the Ash population.
Interestingly, the non-synonymous changes in haplogroups J1 and N1b1 altered highly conserved amino-acid positions in OXPHOS complex I, implying that they possibly affect the activity of this complex . Accordingly, anti-diabetic agents (metformin and thiazolidinediones) act specifically through the inhibition of complex I activity , which suggests a role for complex I functional alteration in the etiology of T2DM.