Introduction

Major depressive disorder (MDD) and panic disorder (PD) are prevalent and serious psychiatric disorders commonly manifesting in young adult age and affecting females more often than males.1, 2 The data from twin and family studies suggest a substantial involvement of genetic factors in the familial transmission of MDD and PD, with heritability estimates for MDD in the range of 31−42%,3 or as high as 70%,4 and for PD near 40%.5 It has been established that MDD and PD are highly comorbid and that shared genetic risk factors largely account for their comorbidity.6 However, the specific genes or their variants responsible for heritability or expression of these disorders remain undetermined.

The limbic system-associated membrane protein (LSAMP) is a 64- to 68-kDa glycoprotein that is found on the somata and dendrites of neurons of cortical and subcortical regions comprising the adult mammalian limbic system, which is involved in the mediation of emotional behavior, learning, memory and central autonomic regulation.7, 8 LSAMP is a member of immunoglobulin superfamily, with three immunoglobulin domains and a glycosyl–phosphatidylinositol anchor.9 It has been suggested that LSAMP acts as a necessary molecular component for the formation of proper connections between the limbic system regions.7 In humans, the LSAMP gene is located in the 3q13.2-q21 chromosomal region, which consists 7 exons and shows 94% identity at the nucleotide level, and 99% sequence identity at the encoded 338-amino-acid polypeptides level with rat LSAMP cDNA.10 Human studies have shown that LSAMP may have a role in the neurobiology of suicidal behavior.11 Proteomic assessment has strongly implicated LSAMP in schizophrenia and bipolar disorder,12 demonstrating its increased level in the dorsolateral prefrontal cortex. Furthermore, stronger or weaker linkages have been shown in the long arm of chromosome 3 with recurrent early-onset MDD or anxiety,13 agoraphobia,14 schizophrenia,15, 16, 17, 18 bipolar disorder19, 20 and autism-spectrum disorders.21 A number of genome-wide association studies of depression have mainly given results that did not meet a genome-wide threshold for significance, but there were associations between depression and LSAMP or adjacent gene loci in top findings of all publications.22, 23, 24, 25, 26, 27, 28, 29

In animal studies, it has been shown that the anxious rats display 1.6-fold higher expression of Lsamp gene compared to the non-anxious rats in the periaqueductal gray.30 Similar results have been reported for the amygdala, demonstrating that the anxious rats display 2.4-fold higher expression of the Lsamp gene compared with the non-anxious rats.31 Upregulated Lsamp expression was also detected in the raphe, the hippocampus and the frontal cortex of the highly anxious rats.32 The studies in Lsamp-deficient mice demonstrated the reduced anxiety and increased sensitivity to anxiolytic action of diazepam in these animals.33 The findings above suggest the importance of LSAMP in emotional regulation and make the LSAMP gene a good candidate for genetic research on mood and anxiety disorders. The purpose of the present study was to detect associations of genetic polymorphisms in the LSAMP gene with MDD and PD.

Materials and methods

Subjects and psychiatric assessment

Unrelated patients with the diagnoses of MDD or PD (n=591) were enrolled in the study along with healthy control individuals (n=384). The patients were recruited from consecutive outpatients and inpatients at the Clinic of Psychiatry of the Tartu University Hospital, and healthy subjects were recruited by newspaper advertisement in Tartu, Estonia. The diagnostic status of the patients was substantiated by a psychiatric interview and verified by the Mini International Neuropsychiatric Interview (MINI 5.0.0) based on DSM-IV criteria.34 Controls were evaluated by the MINI and family history interview, to exclude those with psychiatric morbidity and a history of major psychiatric disorders in first-degree relatives. All subjects were of white European ancestry and living in Estonia. The patients were divided into MDD (n=395) and PD (n=196) subgroups according to the primary diagnosis. MDD or PD was considered primary if it was the principal diagnosis at the time of the investigation and/or had an earlier onset in the course of illness. Patients in MDD group were either pure MDD or MDD comorbid with anxiety disorders. Patients in PD group were either pure PD, or comorbid with mood or other anxiety disorders. The control group was the same for MDD and PD, with an exception of reducing the number of male subjects in control group for PD to reflect the respective male/female ratio in the PD group. There were no significant differences in age or sex between the patients and healthy controls (χ2 and P-value for gender of MDD and PD groups are 2.74 and P=0.098, and 3.35 and P=0.067, respectively). Demographic characteristics of the study subjects are presented in Table 1. The study protocol was approved by the Ethics Review Committee on Human Research of the University of Tartu. Each subject provided a written informed consent before participation.

Table 1 Demographic characteristics of the subjects
Full size table

SNP selection, DNA preparation and genotyping

Tag single-nucleotide polymorphisms (SNPs) were selected from the LSAMP gene and flanking regions (150 kb fragment from 116931222–117081732; and 242 kb fragment from 117428563–117670401) using the Tagger algorithm (r2=0.8, minor allele frequency >0.05) implemented in Haploview 3.32.11 Standard high-salt extraction method was used to isolate genomic DNA from 9 ml venous blood samples. SNP detection was performed by SNPlex Genotyping System (Applied Biosystems, Foster City, CA, USA). This system is based on the oligonucleotide ligation/PCR assay with a universal ZipChute (Applied Biosystems) probe detection for high throughput SNP genotyping. Fluorescent-labeled ZipChute probes are hybridized to complementary ZipCode (Applied Biosystems) sequences that are part of genotype-specific amplicons. These ZipChute probes are eluted and detected by electrophoretic separation on Applied Biosystems 3730 DNA Analyzers. The system analysis software collects and manages raw data and provides automated allele calling and quality metrics.35

Statistical analysis

Genotypes were determined by GeneMapper 3.7 (Applied Biosystems). Allelic association, linkage disequilibrium and haplotype analyses between groups of patients and controls, and Hardy–Weinberg equilibrium calculations in the control group were performed using Haploview Version 4.0 software.36 The significance level for all statistical tests was 0.05. The algorithm for defining haplotype blocks was an internally developed method Solid Spine of linkage disequilibrium (D′>0.75). Ten thousand permutations were used to correct P-values for multiple testing. Closer analysis of haplotype block nr 5 was performed with Statistica software (Information Technology Office, University of Tartu, Tartu, Estonia).

Results

We genotyped a total of 22 SNPs from the LSAMP gene in 591 unrelated psychiatric patients and 384 healthy controls. One SNP (rs4831129) was excluded from the further analysis because of a bias from Hardy–Weinberg equilibrium in the control group (P-value cut-off 0.01).

Allelic association analysis

The results of allelic association analysis of 21 SNPs are presented in Table 2. Four SNPs showed significant associations in MDD group—rs16824691, rs1461131, rs4831089 and rs9874470 (P-values 0.004, <0.001, <0.001 and <0.001, respectively). For the last three of these SNPs, the P-values remained statistically significant after 10 000 permutations (odds ratios, 95% confidence intervals and permutation P-values 1.43, 1.17–1.76 and 0.0228; 1.53, 1.24–1.89 and 0.0015; and 0.63, 0.5–0.78 and 0.0005, respectively). Two SNPs showed significant associations in PD group—rs1461131 and rs4831089 (P-values 0.033 and 0.018, respectively). These P-values did not remain statistically significant after correction for multiple testing.

Table 2 Results of allelic association analysis
Full size table

A separate analysis of allelic association by gender revealed significant associations with the same four SNPs as whole-group analysis. The SNPs rs1461131, rs4831089 and rs9874470 (P-values 0.005, 0.005 and 0.007, respectively) differed significantly between females of the MDD group and their respective controls. No significant association was detected in the female PD group. The SNPs rs1682469, rs4831089 and rs9874470 (P-values 0.01, 0.004 and <0.001, respectively) differed significantly between males of the MDD group and their controls. The SNPs rs1461131, rs4831089 and rs9874470 (P-values 0.003, 0.005 and 0.03, respectively) differed significantly between males of PD group and their controls. These P-values did not remain statistically significant after correction for multiple testing, except rs9874470 in the male MDD group (odds ratio (95% confidence intervals)=0.49 (0.33–0.73); permutation P=0.01).

Haplotype analysis

Haplotype analysis revealed seven haplotype blocks in both patient groups (see Figure 1 for haplotype blocks in MDD group). Haplotypes of block nr 5 (rs9874470, rs4831089, rs16824691 and rs1461131) in MDD and PD groups differed statistically significantly between patient groups and healthy control groups (Tables 3 and 4). Haplotype TATA was a risk haplotype in MDD group (P<0.001) and in PD group (P=0.020). P-value remained statistically significant after 10 000 permutations in MDD group (odds ratio (95% confidence intervals)=1.55 (1.26–1.91); permutation P=0.0008), but did not survive the correction in the PD group. Two haplotypes were protective in the MDD group—CGAG and CGTG, with respective P=0.003 and P=0.020. These P-values did not remain statistically significant after correction for multiple testing. ATT-risk haplotype from block nr 7 (rs4831140, rs6787168 and rs16824996) differed statistically significantly in the MDD group (P=0.042), but this P-value also did not survive the correction. Other haplotypes did not significantly differ between the patient and control groups (data not shown). On a closer analysis of frequencies of minor alleles of haplotype block nr 5, the risk haplotype TATA appeared more frequently than protective haplotype CGAG in the MDD group (41 vs 17%) as compared with their respective control group (31 vs 23%; χ2-test P<0.001). The same trend was observed in PD group-risk haplotype TATA was more frequent than protective haplotype CGAG (38 vs 20%) as compared with their respective control group (31 vs 24%; χ2-test P=0.003).

Figure 1
figure 1

Haplotype blocks of the limbic system-associated membrane protein (LSAMP ) gene in major depressive disorder (MDD) group.

Full size image
Table 3 Haplotypes of block 5 in MDD group
Full size table
Table 4 Haplotypes of block 5 in PD group
Full size table

Discussion

Our findings indicate that the LSAMP gene is possibly related to MDD and PD. We detected statistically significant allelic associations between four LSAMP SNPs and MDD. Two SNPs out of these four contributed to the increased risk (rs1461131 and rs4831089) and two were protective (rs16824691 and rs9874470). Only two risk SNPs above were related to PD. Haplotype block nr 5 was formed from the forenamed SNPs giving three statistically significant haplotypes in the MDD group: TATA being risk haplotype and CGAG and CGTG being a protective haplotype. Risk haplotype TATA was also statistically significant in the PD group. Closer analysis of minor alleles of haploblock 5 indicated that the proportion of risk SNPs was higher in patient groups as compared with control groups. Sub-analysis by gender suggested that LSAMP gene polymorphisms are evidently more related to males than to females, as no statistically significant association was detected in the female PD group. Plausibly, LSAMP gene polymorphisms contribute to shared genetic factors of MDD and PD, as associations were detected across both groups.

The LSAMP gene and protein show high similarity between the humans and rodents, suggesting strong phylogenetic conservation of the protein structure and associated functional properties.10 The LSAMP gene has two alternative first exons (1a and 1b) separated by a 1.5-Mb-long intron, which give rise to two different mRNA transcripts.37 A preliminary in situ RNA hybridization analysis has demonstrated substantially different expression patterns of Lsamp 1a and 1b transcripts in mouse brain, with the 1b predominantly expressed in sensory nuclei of the brainstem, in the thalamus and in sensory areas of the cortex (Philips et al., unpublished). It has been suggested that large 5′-introns contain regulatory elements involved in silencing and spatial and temporal patterns of gene expression.37 Interestingly, all associated SNPs in our study reside in the first intron (1b) and may affect regulation of gene expression. Two different Lsamp gene knockout mouse lines have been created.33, 38 Overlapping between the phenotype of these lines is remarkable; both knockout mice are vital and fertile with no gross abnormalities.33 The first line, in which the second exon of Lsamp is deleted, exhibits hyperactivity in novel situations, altered synaptic transmission and impaired plasticity in the hippocampus, inducing deficits in spatial memory acquisition and retention.38, 39 The other line, in which the 1b exon of Lsamp is deleted, exhibits slower swim velocity, hyperactivity in novel situations, reduced anxiety, reduced aggressiveness and decrease of barbering behavior in male mice; however, their spatial memory and learning curve were similar to that of wild-type mice.33, 40 Altogether, these results suggest that altered expression of LSAMP leads to differences in neuronal connectivity, causing only subtle disturbance in signaling but measurable changes in functional output.38 Furthermore, the lack of LSAMP protein seems to result in an inability to adapt or react to novel environments or stressful environmental manipulations.40 Thus, a properly regulated level of LSAMP expression appears to be required for normal neuronal functioning. A former study has demonstrated that intraventricular administration of anti-LSAMP to postnatal rats resulted in aberrant growth of the mossy fiber projection in the hippocampus.9 However, more recent study showed no major alterations in the brain organization and gross connectivity in the Lsamp-deficient mouse.38 To our knowledge, the direct impact of LSAMP overexpression on neuronal circuitry has not yet been investigated, but the animal studies indicate a relationship between heightened expression of LSAMP and anxiety. Our speculation is that risk SNPs in 1b intron may cause an increased expression of the LSAMP gene, which results in elevated anxiety and a negative effect.

Several limitations of our study should be recognized. The sample sizes were rather small, especially the PD male group. The patients in both MDD and PD groups had comorbid diagnoses. Not all associations survived correction for multiple testing, and we cannot exclude false-positive results. Also, our study lacks functional analysis of the associated SNPs. Therefore, replication association studies in independent samples and imaging genetic studies are needed to confirm and extend our findings. Further functional research is required to determine the relevance of LSAMP function to the molecular mechanisms of psychiatric disorders.