A 3p26-3p25 genetic linkage finding for DSM-IV major depression in heavy smoking families

Abstract

Objective

The authors tested for genetic linkage of DSM-IV-diagnosed major depressive disorder in families that were ascertained for cigarette smoking.

Method

Within a study that targeted families characterized by a history of smoking, analyses derived a subset of 91 Australian families with two or more offspring with a history of DSM-IV major depressive disorder (affected sibling pairs, N=187) and 25 Finnish families (affected sibling pairs, N=33). Within this affected sibling pair design, the authors conducted nonparametric linkage analysis.

Results

In the Australian heavy smoking families, the authors found a genome-wide significant multipoint LOD score of 4.14 for major depressive disorder on chromosome 3 at 24.9 cM (3p26-3p25).

Conclusions

Genome-wide significant linkage was detected for major depressive disorder on chromosome 3p in a sample ascertained for smoking. A linkage peak at this location was also observed in an independent study of major depressive disorder.

Introduction

Genetic linkage studies of major depressive disorder have found suggestive evidence across multiple genomic regions with little convergence of findings ¹. Genetic association, including genomewide association studies (GWAS), findings have been similarly disparate ^2–7. These inconsistencies raise the question of whether major depressive disorder encompasses a number of poorly understood subtypes (e.g., depression in smokers). Among smokers seeking cessation treatment, lifetime rates of major depression have been estimated at over 60%⁸. Cigarette smokers with a history of depression tend to report more severe nicotine withdrawal symptoms ^9–11, are more likely to relapse to smoking after a quit attempt⁸, and may be at increased risk for a recurrent episode of depression after smoking cessation¹². Models describing the relationship between symptoms of depression and smoking include examples of shared genetic risk¹³. We carried out genetic linkage analyses of DSM-IV diagnosed major depressive disorder in two samples that are part of the Nicotine Addiction Genetics project^14;15, an international consortium focused on tobacco dependence. We used an affected sibling pair design, in which at least two adult offspring per family reported a history of DSM-IV major depressive disorder, and tested for linkage. Results appear to confirm a genome-wide significant linkage signal at 3p26-3p25, also reported in an independent linkage study of major depressive disorder¹⁶.

Method

Samples

The Nicotine Addiction Genetics linkage project enrolled participants at the Queensland Institute of Medical Research in Australia and the University of Helsinki in Finland. Both sites utilized twin registries and targeted families of index cases who previously reported a history of cigarette smoking in interview or questionnaire surveys; the Australian site using a higher threshold measure when recruiting index cases with a history of heavy smoking¹⁴. These original linkage samples included microsatellite marker data and telephone diagnostic interviews in 289 families from the Australian site (offspring, N=917; parents, N=392) and 161 families from the Finnish site (offspring, N=522; parents, N=19). More than 90% of the participants from the Australian site were of Anglo-Celtic or Northern European ancestry, and all of the participants from the Finnish site were of Finnish ancestry. The assessment included a diagnostic telephone interview, adapted from the Semi-Structured Assessment for the Genetics of Alcoholism^17;18, which obtained a comprehensive assessment of lifetime DSM-IV¹⁹ major depressive disorder as well as a tobacco use and dependence assessment derived from the Composite International Diagnostic Interview²⁰. Additional details on the original samples and assessments have been described elsewhere ^14;15;21. For the present study, we conducted genetic linkage analyses using the 91 Australian families who had one or more affected sibling pairs concordant for history of major depressive disorder (N=187). In the Finnish sample, only 25 families had one or more affected sibling pairs (N=33), and thus the primary analyses focused on the Australian sample. A summary of the two major depressive disorder linkage samples is presented in Table 1 of the data supplement.

Analyses

For both the Australian and Finnish samples, 381 autosomal microsatellite markers were genotyped and spaced at approximately 10 cM across the genome, positioned using the deCODE genetic map²². Details of genotyping, including quality control procedures are described elsewhere^14,15. Single- and multi-point affected sibling pair nonparametric linkage (the latter using a 2 cM grid) was conducted in MERLIN (Multipoint Engine for Rapid Likelihood Inference)²³, which generated LOD scores ^{24, 25}. Upon detection of LOD scores > 3, genome-wide corrected p values were calculated from 1,000 replicates simulated in MERLIN^26;27.

Follow-up fine mapping included the addition of 290 single nucleotide polymorphism (SNPs) on chromosome 3 that were available for a portion of the Australian major depressive disorder linkage sample (84 out of 91 families), from other Australian projects with overlapping samples that have obtained GWAS data (see reference ²⁸). Through the use of Snagger software ²⁹, these SNPs were selected to be in low linkage disequilibrium (maximum pairwise r²₌ 0.2), to have a minor allele frequency > 40%, and to have a minimum distance between two tags of 450 kb in order to optimize information content.

Results

For the Australian linkage sample, a multipoint LOD of 4.14 for major depressive disorder was found on chromosome 3 (at 24.9 cM). The highest single-point linkage signal emerged at microsatellite marker D3S1304, with a LOD score of 3.7. The LOD of 4.14 at 3p in this sample met genomewide significance (p = 0.004). A complete report of these single- and multipoint findings, along with marker positions by chromosome, is provided in Table 2 of the data supplement. The addition of 290 SNPs on chromosome 3 narrowed the linkage locus and increased the LOD score to 4.55 (at 25.3 cM) (see Figure 1). For the Finnish major depressive disorder sample, a multipoint LOD score of 2.10 was found on chromosome 20 (at 90.9 cM).

Figure 1.

Chromosome 3: DSM-IV Major Depression NAG Australian sample; the LOD of 4.14 meets genome-wide significance at p = 0.004 with 1000 simulations in MERLIN;

Given that no linkage signals over 1.5 overlapped across both samples, we did not conduct a combined analysis. However, if we were to correct conservatively for testing in both samples by doubling the p value generated from the 1,000 simulations, we would still maintain a genomewide significant p value of 0.008.

Discussion

Overall, major depressive disorder was associated with a significant genetic linkage peak (a LOD score of 4.14) on chromosome 3 in the Nicotine Addiction Genetics Australian sample, replicating the linkage reported at the same location by Breen et al ¹⁶. These converging findings suggest that the genomic region spanning 3p26-3p25 is an important area for further investigation in genetic research on major depressive disorder. Given the small number of Finnish affected sibling pairs, the lack of confirmation in this particular sample is not unexpected. The genetic variants accounting for this linkage signal have not yet been convincingly identified. Although our highest single-point microsatellite marker (D3S1304 [LOD score =3.7]) lies within the metabotropic glutamate receptor 7 gene (GRM7), subsidiary association analyses within a one-LOD support interval, using GWAS data available in the Australian sample (²⁸), found only nominal association for two SNPs within GRM7 (p < .05). Even our strongest association effect, with was a p value of 0.00014 for rs6765537 (a nonsynonymous SNP within C3or20 at 33.87 cM), did not replicate in the Finnish families, nor did it replicate in the Genetic Association Information Network major depressive disorder sample from the Netherlands^{30, 2}. Thus, while others have found suggestive association between SNPs in GRM7 and Major Depressive Disorder (^7;31), genome-wide significant effects have not been reported. Further, because linkage implicates very broad regions, GRM7 is among many genes that might be hypothesized to explain our signal.

In terms of other linkage findings, one for quantity smoked in samples ascertained for depression has been reported near our chromosome 3 finding for major depressive disorder ³², raising the possibility of common genetic influences across major depressive disorder and smoking-related behavior or of gene-by-environment (i.e., smoking) interaction effects on major depressive disorder.

There are important limitations associated with our results. Our sample of 91 families is small by standards of modern genomic efforts. Thus, we cannot exclude the possibility that our finding is a false positive that coincidentally appears to replicate Breen et al.¹⁶, whose sample is much larger (> 800 families). Additionally, our finding at 3p26-3p25 (highest peak at 24.9 cm) does not align with the meta-analyses results reported by McMahon et al.³³, who suggest that variants located at 3p21.1 (at approximately 70–72 cM) are associated with mood disorders. Our future efforts to localize the genetic variants influencing major depressive disorder will entail confirmatory analyses in other samples and additional genotyping in this region on 3p.