Do nuclear-encoded core subunits of mitochondrial complex I confer genetic susceptibility to schizophrenia in Han Chinese populations?

Schizophrenia is one of the most prevalent psychiatric disorders with complex genetic etiology. Accumulating evidence suggests that energy metabolism and oxidative stress play important roles in the pathophysiology of schizophrenia. Dysfunction of mitochondrial respiratory chain and altered expression of complex I subunits were frequently reported in schizophrenia. To investigate whether nuclear-encoded core subunit genes of mitochondrial complex I are associated with schizophrenia, we performed a genetic association study in Han Chinese. In total, 46 tag single nucleotide polymorphisms (SNPs) from 7 nuclear-encoded core genes of mitochondrial complex I were genotyped in 918 schizophrenia patients and 1042 healthy controls. We also analyzed these SNPs in a large sample mainly composed of Europeans through using the available GWAS datasets from the Psychiatric Genomics Consortium (PGC). No significant associations were detected between these SNPs and schizophrenia in Han Chinese and the PGC data set. However, we observed nominal significant associations of 2 SNPs in the NDUFS1 gene and 4 SNPs in the NDUFS2 gene with early onset schizophrenia (EOS), but none of these associations survived the Bonferroni correction. Taken together, our results suggested that common SNPs in the nuclear-encoded core subunit genes of mitochondrial complex I may not confer genetic susceptibility to schizophrenia.

function of brain and mitochondrial dysfunction has been frequently reported in patients with brain diseases, including neurodegenerative diseases and psychiatric disorders 2,4 .
Several lines of evidence support the dysfunction of mitochondria in schizophrenia. First, defect of mitochondrial oxidative phosphorylation and altered expression of mitochondria-related genes were reported in brains from patients with schizophrenia [5][6][7] . Second, data from transcriptomics, proteomics and metabolomics revealed aberrant brain metabolism and oxidative stress in schizophrenia patients 8 . Third, perturbations in mitochondrial network dynamics and in complex I dependent cellular respiration were also reported in schizophrenia 9 . Besides, genetic variations of mitochondrial DNA (mtDNA) were also reported to be susceptible to schizophrenia and other neurological disorders 2,10,11 , although with controversies [12][13][14][15] . Despite the significant progress in recent years, the role of mitochondria in the pathophysiology of schizophrenia remains elusive, and further studies using more rigorous methodological and statistical standards are necessary to avoid false positive results.
Among the mitochondrial complexes, NADH ubiquinone oxidoreductase (complex I) may have a role in schizophrenia. Human mitochondrial complex I, which contains 45 subunits, is the largest and most complicated component of the respiratory chain and plays a central role in electron transportation 16 . The 14 "core" subunits of complex I are conserved from bacteria to human and are sufficient for catalysis 17,18 , suggesting their importance in energy metabolism. Seven of the 14 "core" subunits were nuclear-encoded which ligate the flavin mononucleotide and the iron-sulfur clusters 19 . Due to its pivotal role in energy metabolism, it has been speculated that mitochondrial complex I may be involved in the pathophysiology of schizophrenia. Consistent with this speculation, previous studies showed aberrant expression of mitochondrial complex I subunits or altered complex I activity in schizophrenia 6,[20][21][22] . In addition, several "core" genes of mitochondrial complex I have been reported to be functionally related to schizophrenia [23][24][25] .
To further explore the potential association between mitochondrial complex I and schizophrenia, we conducted a comprehensive genetic association study by genotyping 46 tag SNPs from seven nuclear-encoded genes of mitochondrial complex I in Han Chinese with and without schizophrenia and reanalysis of the Psychiatric Genetics Consortium (PGC) dataset 26 . In addition, we also performed stratified analyses to test if genetic variants of the seven nuclear-encoded genes of mitochondrial complex I were associated with early onset schizophrenia (EOS) in Han Chinese.

Results
In total, 46 tag SNPs were successfully genotyped in our samples. Considering an average population minor allele frequency (MAF) of 0.1, the power to detect an odds ratio as low as 1.5 for a risk allele/ genotype/haplotype was above 90% for the comparison between schizophrenia patients and controls in our sample set (Supplementary Figure 1).
None of the 46 SNPs was deviated from HWE in both case and control samples (Supplementary Table 1). The linkage disequilibrium (LD) structures of the 46 SNPs showed high degrees of similarity between cases and controls ( Fig. 1). Allelic and genotypic association analyses revealed no significant association between these 46 tag SNPs and schizophrenia (Table 1). In addition, haplotype-based analysis also showed no significant association between schizophrenia cases and healthy controls (Supplementary Table 2). Further stratified association analysis revealed that two SNPs (rs4147713, P = 0.019; and rs1044120, P = 0.049) in the NDUFS1 gene and 4 SNPs (rs3924264, P = 0.004; rs1136224, P = 0.013; rs2070902, P = 0.006; and rs4233368, P = 0.038) in the NDUFS2 gene were nominally associated with EOS (onset age ≤ 18 years) ( Table 2). Nevertheless, none of the six SNPs survived for multiple testing corrections.
We further examined the genetic association between these 46 SNPs and schizophrenia in the PGC data 26 . Overall, 35,476 schizophrenia cases and 46,839 controls were included for association analysis. We observed marginally significant associations with schizophrenia of rs10908826, rs4656994, rs5085 and rs2307424 in the NDUFS2 gene and rs12457810, rs12964485 and rs2377961 in the NDUFV2 gene. However, none of these SNPs showed significant association after correcting for multiple testing (Table 1).
We further performed interaction analysis to test whether there is an interaction between SNPs in both case-control samples and the EOS-control samples. Our results showed that although there were many marginally significant interactions, none of them could survive the Bonferroni correction (Supplementary Tables 3-4).

Discussion
Mitochondrial dysfunction has been frequently reported in schizophrenia 8,27 . Previous studies have suggested that genes related to energy metabolism and oxidative stress may be responsible for mitochondrial dysfunction in schizophrenia 8 . As the largest component of the three membrane-bound enzymes, mitochondrial complex I plays a vital role in energy metabolism and altered activity of mitochondrial complex I was repeatedly reported in schizophrenia 20,22,28,29 . Though multiple studies have reported that mitochondrial dysfunction may be involved in schizophrenia pathogenesis, most of the conclusions were based on gene expression 6,21 . We previously analyzed the NDUFS7 gene in Han Chinese with and without schizophrenia, with an intention to discern the effect of the complex I genes on this disorder 30 . In this study, we analyzed other complex I core subunit genes including NDUFS1, NDUFS2, NDUFS3, NDUFS8, NDUFV1 and NDUFV2, as well as four more common SNPs of the NDUFS7 gene in a larger sample set. We systematically investigated the potential association between these nuclear-encoded mitochondrial Scientific RepoRts | 5:11076 | DOi: 10.1038/srep11076 complex I genes and schizophrenia in a Chinese case-control sample. Our results revealed no significant association between genetic variants from the seven selected genes and schizophrenia, suggesting that these genes are unlikely to confer risk of schizophrenia in Han Chinese population. Consistent with the finding in Han Chinese, we also found no robust association between these SNPs and schizophrenia in the PGC data 26 . It should be mentioned that if we did not consider the effect of multiple corrections, SNPs of the NDUFS2 gene would show (marginally) significant association with schizophrenia in both Han Chinese and the PGC data 26 .
It is interesting that we observed nominal significant association between genetic variants in the NDUFS1 and NDUFS2 genes and EOS in Han Chinese. The nominal significance would not survive from adjustment and we cannot rule out the possibility of false positive association caused by relatively small sample size of EOS used in the analyses. It should be mentioned that genetic variant in the NDUFS1 gene was reported to be associated with schizophrenia and negative symptoms in Han Chinese from Eastern China, albeit the samples size was also relatively small 31 . We also noticed that comparing with whole samples (odds ratio [OR] > 1), 2 of the 6 nominal significant SNPs (rs4147713 and rs3924264) were associated with EOS in reverse style (OR < 1). This observation indicated population stratification may exist in the EOS-subpopulation.
Considering that mitochondrial dysfunction was repeatedly reported in schizophrenia, it is amazing that none of the selected SNPs showed significant association with schizophrenia. One of the possible explanations is that other genes but not the seven selected core genes of mitochondrial complex I contribute to schizophrenia susceptibility. Moreover, possibility of rare variant(s) in these genes contributing to schizophrenia susceptibility needs to be studied as well. In addition, given that schizophrenia has a strong genetic heterogeneity, it is also possible that these genes would have a strong effect in other populations but not in Han Chinese 23,24 .
There are several limitations in this study. First, the sample size is relatively modest in this study. As a result, it may be difficult to detect a robust significant association. Second, we only analyzed seven nuclear-encoded genes in the mitochondrial complex I in this study, we could not exclude the possibility that other genes of complex I are associated with schizophrenia.  In summary, we detected no significant association between the genetic polymorphisms of the nuclear-encoded core subunit genes of mitochondrial complex I and schizophrenia by using a rigorous statistical standard so as to avoid false positive results. Further work is needed to test if the expression and rare variants, but not common variants of these genes, contribute to schizophrenia.

Subjects.
A total of 1960 subjects, including 918 unrelated patients with schizophrenia (561 males: mean age ± SD, 38.5 ± 13.6 years; 357 females: 38.4 ± 16.8 years) and 1042 healthy controls (631 males: mean age ± SD, 38.5 ± 14.2 years; 411 females: 46.0 ± 8.2 years) were recruited. All of the individuals are of Han Chinese origin from Hunan Province of South Central China. Among the 918 patients, 189 were EOS (first age of onset ≤ 18 years old). The patients were clinically diagnosed according to Diagnostic and Statistical Manual of Mental Disorders IV (DSM-IV) and had at least two-year history of schizophrenia. Diagnosis and review of schizophrenia cases were independently checked and verified by two senior psychiatrists prior to blood sample collection. The healthy controls were collected from local hospitals and assessed by experienced psychiatrists. Individuals with psychiatric history, alcohol dependence, drug abuse, or family history of psychiatric disorders were excluded. All of the schizophrenia patients and healthy controls have been reported in our previous studies 11,32 . Written informed consents conforming to the tenets of the Declaration of Helsinki were obtained from all participants or the appointed guardians of the patients (for those who were unable to provide informed consent at the time of blood collection) prior to this study. The experimental methods were carried out in accordance with the approved guidelines. All experimental protocols of this study were approved by the institutional review board / Ethics Committee of Kunming Institute of Zoology, Chinese Academy of Sciences. SNP selection and genotyping. Genomic DNA of all participants was extracted from peripheral blood using the AxyPrep TM Blood Genomic DNA Miniprep Kit (Axygen, USA) according to the manufacturer's instruction. Seven nuclear-encoded genes of mitochondrial complex I (NDUFS1, NDUFS2, NDUFS3, NDUFS7, NDUFS8, NDUFV1 and NDUFV2) were chosen in this study. To select the tag SNPs, we retrieved genotypic data of Han Chinese (CHB) from the HapMap database (http://hapmap.ncbi.nlm. nih.gov/) and defined LD blocks using the Haploview 4.2 33 . The gene region and potential regulatory sequences (20 kb of both upstream and downstream regions) were taken into consideration during the selection of tag SNPs. In total, 51 tag SNPs were selected based on the following criteria: minor allele frequency (MAF) ≥ 0.1 and r 2 ≥ 0.8 (Supplementary Figures 2-5). Three tag SNPs (rs2074896, rs2074897 and rs2074898) in the NDUFS7 gene were analyzed in our previous study 30 , therefore were not included in this study. The remaining 48 tag SNPs were divided into four panels (12 SNPs for each panel) according to their compatibility in multiplex PCR. Genotyping of each panel was conducted by SNaPshot assays reported in our previous studies 34,35 . The GeneMarker software was utilized to read the genotyping results 36 .
Power calculation and statistical analysis. Among the 48 tag SNPs, two (rs12798346 and rs3751084) were failed to be genotyped in our samples. Therefore, these two SNPs were excluded from our statistical analysis. The genotyping call rate of each SNP was above 99.0% in 1960 individuals. LD plot of the genotyped SNPs of each gene was constructed using Haploview 4.2 program (version 4.2). We tested deviation from the Hardy-Weinberg equilibrium (HWE), individual SNP association, haplotype comparison and SNP-SNP interaction by using PLINK 37 . Quanto software 38 was used for power analysis under the gene only hypothesis and log additive model and following parameters: risk allele frequency from 0.1 to 0.5 in increments of 0.1; overall disease risk in the general population = 0.01; sample size = 918 cases vs. 1042 controls; range of OR from 1.0 to 2.0 in increments of 0.1; two-sided type I error rate = 0.05.  Table 2. Association results of the 46 SNPs in 189 early onset schizophrenia patients and 1042 controls. a A 12 , minor allele and major allele. b P-value < 0.05 was marked in bold. OR -odds ratio; 95% CI -95% confidence interval