Association study of polymorphisms in the autosomal mitochondrial complex I subunit gene, NADH dehydrogenase (ubiquinone) flavoprotein 2 (NDUFV2), and bipolar disorder

Summary

The mitochondrial dysfunction hypothesis of bipolar disorder (BPD) was proposed based on a maternal inheritance pattern in family-based studies (McMahon et al., 1995). For the BPD linkage signal around human chromosome 18p11.2, Gershon et al. (1996) found a parent-of-origin effect only when maternal and paternal inheritance was considered. An autosomally encoded mitochondrial protein imported into mitochondria to perform important functions might explain this finding. NDUFV2, encoding the 24 kDa subunit of mitochondrial complex I, is one such gene found within the aforementioned BPD linkage region.

Single nucleotide polymorphisms (SNPs) in NDUFV2 showed nominally significant associations with BPD in various ethnic populations (Washizuka et al., 2003; Xu et al., 2008; Zhang et al., 2009). To replicate or extend these findings, we studied three SNPs in NDUFV2 (Chr18:9,102,675-9,134,336; build GRCh37:Feb2009:hg19): −3188C>T (rs2377961, Chr18:9,099,554) and −602G>A (rs1156044, Chr18:9,102,140) in the promoter, and +86C>T (rs906807, Chr18:9,117,867), a missense (A29V) SNP in exon 2, using the National Institute of Mental Health (NIMH) Caucasian control (no psychiatric or chronic neurological disease history) and BPDI (ascertained by DSM-IV criteria) populations. Informed consent was obtained from all individuals according to Institutional Review Board (IRB) requirements. 741 control and 569 BPDI individuals were genotyped using the Taqman® Genotyping Assay system (Applied Biosystems, Inc., Foster City, CA). Individuals with “undetermined” genotypes at any locus, representing 1.5% of control and 6.0% of BPDI samples, were removed from further analysis. Thus, 730 control and 535 bipolar individuals were included in our data analyses.

All SNPs were in Hardy-Weinberg equilibrium (rs2377961, BPDI p=0.140, Controls p=0.281; rs1156044, BPDI p=0.494, Controls p=0.794; rs906807, BPDI p=0.408, Controls p=0.889). rs2377961 was in strong LD with rs1156044 (D’=0.843, r²=0.288) and rs906807 (D’=0.832, r²=0.325), and rs1156044 was in strongest LD with rs906807 (D’=0.914, r²=0.721). There were no statistically significant associations between genotypes and BPDI at rs2377961 (X²=3.873; p=0.144), rs906807 (X²=4.711; p=0.095) or rs1156044 (X²=5.798; p=0.055). Neither allele at rs2377961 (X²=0.572; p=0.449) showed association with BPDI; however, the “A” allele at rs1156044 (X²=5.362; p=0.021) and “C” allele at rs906807 (X²=4.173; p=0.041) were nominally associated with BPDI. Notably, association of the “A” allele at rs1156044 agrees with a study of a smaller Caucasian population (Xu et al., 2008), but is opposite the trend observed for BPDII in a Japanese population (Washizuka et al., 2003). While our findings may bolster those of Xu et al. (2008), statistical significances of the allelic associations in the present study were not upheld after Bonferroni correction (α=0.0167 for 3 tests).

Washizuka et al. (2003) found association of a “CTGT” promoter haplotype in NDUFV2 with BPD, where “G” is rs1156044. Haplotype (rs1156044-rs906807) analysis of our data revealed no statistically significant association of the “GT” (X²=2.765; p=0.096) or “AT” (X²=1.259; p=0.262) haplotypes, but the “AC” haplotype showed statistical significance (X²=7.033; p=0.008) after Bonferroni correction (α=0.0125 for 4 haplotype comparisons). However, after 10,000 permutations of our data, no associations remained statistically significant (lowest permutation adjusted p-value=0.064 for the “AC” haplotype). Thus, we cautiously conclude that the “AC” haplotype in NDUFV2 may be associated with BPDI in this Caucasian population.