Candidate gene associations with mood disorder, cognitive vulnerability, and fronto-limbic volumes

Abstract

Introduction

Understanding relationships between candidate genes and mood disorder is crucial for advancing toward molecular-based treatment approaches. Several candidate genes have been identified for mood disorders (Kupfer et al. 2012; Sullivan et al. 2012), with the strongest statistical signals for bipolar disorder (Lohoff et al. 2005; Baum et al. 2008; Ferreira et al. 2008). However, genome-wide association studies of common variants suggest that only a small proportion of the disease is accounted for by accumulation of these variants (Cichon et al. 2009). The modest fraction of phenotypic variance explained is likely a function of the heterogeneity of mood disorders, even within specific categories (Kupfer et al. 2012).

An important intermediate step is evaluation of relationships between candidate genes and structural brain changes or cognitive processes implicated in mood disorders (Gottesman and Gould 2003; Drevets et al. 2008). Structural and functional fronto-limbic brain abnormalities have been implicated in mood disorders (Drevets et al. 2008), most prominently bipolar disorder (Price and Drevets 2010). Additionally, a broad range of cognitive deficits have been observed in mood disorder. Numerous studies have identified memory impairments and impulsivity as likely trait characteristics in bipolar disorder (Robinson et al. 2006; Joseph et al. 2008; Bora et al. 2009; Kurtz and Gerraty 2009) and, to a lesser extent, major depression (Snyder 2013). These endophenotypes may be more closely related to genetic variation (Gottesman and Gould 2003) and provide a window into specific vulnerabilities that increase the probability of mood disorder evolution.

A growing number of studies have evaluated the relationships between candidate genes and cognitive processes or brain volumes in healthy and mood disordered samples (Bigos et al. 2010; Krug et al. 2010; Thimm et al. 2011; Frodl et al. 2012; Radua et al. 2012). For example, Frodl et al. (2012) identified associations of genes important for glucocorticoid and immune function with hippocampal volume in patients with major depressive disorder. Previous data from our group has found altered anterior cingulate volumes in individuals with bipolar disorder who carried the BDNF minor allele (Matsuo et al. 2009). A single nucleotide polymorphism (SNP) rs1006737 in CACNA1C, implicated in bipolar disorder (Ferreira et al. 2008) and other neuropsychiatric disorders (Gargus 2006, 2009), has been found to increase brain volumes, particularly grey matter volumes (Kempton et al. 2009), and impair appropriate functioning of fronto-temporal circuits (Wang et al. 2011) important for emotional processing (Radua et al. 2012). Similarly, variations in ANK3 and DGKH, also implicated in bipolar disorder (Baum et al. 2008) and other neuropsychiatric disorders (Weber et al. 2011), have been associated with altered brain structure and function (Hatzimanolis et al. 2012; Linke et al. 2012; Whalley et al. 2012). Taken together, these data support a model where genes important for ongoing neural plasticity and immune system functioning influence cognitive and structural brain endophenotypes representing key nodes of mood disorder vulnerability.

The present study evaluated four strong candidate polymorphisms from genes important for neurotransmission and plasticity - ANK3 (rs10994336), BDNF (rs6265), CACNA1C (rs1006737), and DGKH (rs1170191). The functionality of the intronic variants in ANK3, CACNA1C, and DGKH have not been demonstrated. However, it was reported that CACNA1C rs1006737 AA genotype subjects have greater mRNA expression in the dorsolateral prefrontal cortex than subjects with GG or GA genotypes (Bigos et al. 2010). Additionally, each of these candidate polymorphisms is believed to have functional relevance to neuropsychiatric disorders, including mood disorder: ANK3 is thought to influence the function of voltage-gated sodium channels, BDNF regulates neuronal growth and participates in plasticity of neurons throughout the lifespan, CACNA1C is the alpha 1C subunit of the L-type voltage-gated calcium channel, and DGKH participates in the lithium-sensitive phosphatidyl inositol pathway. Therefore, the primary aim of this study was to explore associations between these candidate genes, cognitive processes associated with mood disorder, and mood symptoms and diagnoses. Associations were investigated in a well-characterized sample of adults with bipolar disorder, major depressive disorder, or healthy controls. Candidate polymorphisms were expected to show only modest associations with mood disorder symptoms and diagnoses. However, genetic variation was expected to be significantly associated with individual differences in cognitive processing (global ability, impulsivity, memory) and fronto-limbic volumes. Fronto-limbic volumes were expected to mediate the relationships between genetic variation and cognitive vulnerability to mood disorder.

Materials and Methods

Results

Genetic correlates of cognitive processes and fronto-limbic volumes

Table ​Table33 also presents relationships between the candidate SNPs and cognitive processes related to mood disorder vulnerability. CACA1C minor allele carriers (GA and AA genotypes) were associated with higher overall, verbal, and nonverbal IQ scores. ANK3, BDNF, and DGKH genotype groups were not significantly associated with IQ. These relationships remained significant after covarying for race/ethnicity. DGKH minor allele carriers (AC and CC genotypes) had reduced verbal learning and memory. However, this did not survive false discovery rate correction. None of the remaining SNPs evaluated were found to be significantly associated with memory or attention/impulsivity measures.

Table ​Table44 presents relationships between candidate SNPs, total brain, and fronto-limbic volumes. CACNA1C minor allele carriers (GA/AA) showed significant positive relationships with total brain and total white matter volumes and fronto-limbic volumes in five of 17 regions examined, implying that the presence of minor alleles is related to greater brain volumes. Of these five significant fronto-limbic regions, only the caudate region remained significant following false discovery rate correction. Of the remaining 12 of 17 nonsignificant associations, 11 were in the positive direction, suggesting a general trend toward greater brain volumes with patients carrying at least one CACNA1C minor allele (sign test P < 0.0001). Figure ​Figure11 presents Q-Q plots of P-values from correlations between candidate SNPs and brain volumes. These plots indicate that relationships between CACNA1C minor allele carriers and brain volumes consistently deviated from the null hypothesis, suggesting a statistically reliable positive association with increased total brain and fronto-limbic volumes. Furthermore, the effect appears to be graded across homozygous minor, heterozygous, and major allele genotypes (Fig. ​(Fig.2).2). Covarying for race/ethnicity did not alter the pattern of CACNA1C clinical correlates. ANK3, BDNF, and DGKH genotypes were not significantly associated with total brain or fronto-limbic volumes in any region, with the exception of a negative relationship reflecting smaller anterior cingulate volume in DGKH minor allele carriers that did not survive false discovery correction. The pattern of relationships between genotypes, cognitive measures, and brain volumes remained the same after controlling for current psychotropic medication use and past history of substance use disorder.

Table 4

Relationships between candidate SNPs, total brain, and fronto-limbic volumes. Minor allele genotypes in CACNA1C were significantly associated with increased total brain volume and with volume of several fronto-limbic regions

	ANK3 rs10994336 N = 258	BDNF rs6265 N = 281	CACNA1C rs1006737 N = 259	DGKH rs1170191 N = 239
	r	r	r	r
Total brain volume	−0.06	0.00	0.13*	0.03
Total cortical grey matter	−0.01	0.03	0.08	−0.08
Total cortical white matter	−0.06	−0.04	0.13*	0.03
Frontal pole	−0.08	0.00	0.12*	−0.01
Superior frontal	−0.01	0.02	0.09	−0.07
Middle frontal	−0.04	0.00	0.07	−0.10
Lateral orbital frontal	0.01	−0.03	0.02	−0.09
Medial orbital frontal	−0.06	−0.05	0.08	−0.06
Anterior cingulate	−0.01	0.03	0.03	−0.13*
Nucleus accumbens	0.02	0.01	0.11	−0.07
Caudate	0.02	−0.02	0.21**	−0.05
Putamen	0.06	0.05	0.10	0.04
Globus pallidus	0.04	0.05	0.12*	0.04
Hippocampus	−0.04	0.04	0.04	0.01
Amygdala	0.01	−0.02	0.03	0.01
Insula	−0.03	0.09	0.14*	−0.09
Pars opercularis	0.03	−0.07	−0.08	−0.03
Pars triangularis	−0.03	0.00	0.08	−0.09
Pars orbitalis	−0.02	0.03	0.07	−0.02
Ventral diencephalon	0.01	0.03	0.15*	0.05

Open in a separate window

^*P < 0.05,

^**P < 0.001. Correlations were computed in all three study groups. The significant relationship between CACNA1C genotype groups and caudate volume survived false discovery rate correction. The relationship between CACNA1C genotype groups and caudate volume is significantly larger than the relationship between CACNA1C genotype groups and cortical grey matter volume [t(256) = 2.18, P = 0.030].

Open in a separate window

Figure 1

Q-Q plots of observed P-values by P-values expected under the null hypothesis for relationships between candidate genes and brain volumes. For CACNA1C, observed P-values are consistently more significant than those expected under the null hypothesis.

Open in a separate window

Figure 2

Total, cortical grey matter, cortical white matter, and caudate volumes by CACNA1C genotypes (GG n = 123, GA n = 103, AA = 33). Significant differences were observed between major allele homozygotes and minor allele carriers for total brain volume and total cortical white matter. For caudate volumes, minor allele genotype carriers differed significantly from the major allele homozygotes. *P < 0.05, **P < 0.001.

Mediational models

The significant relationship between increased IQ and brain volumes with the patients carrying at least one CACNA1C minor allele raised the possibility that increased brain volumes may mediate increased IQ in individuals with these genotypes. Figure ​Figure33 presents mediational modeling results for the relationships between CACNA1C genotype group, brain volumes, and full scale IQ. CACNA1C minor allele genotypes increased full scale IQ scores independently of increases in total and fronto-limbic brain volumes. The same pattern of results was observed for verbal and nonverbal IQ scores. The nominally significant relationship between DGKH minor allele carriers and reduced verbal memory was not mediated by reduced anterior cingulate volumes.

Open in a separate window

Figure 3

Mediational modeling results. CACNA1C minor allele genotype carriers had higher full scale IQ scores and this relationship was independent of increases in total and fronto-limbic brain volumes. Direct effects are given in blue (all P < 0.05) and listed in order from total brain to ventral diencephalon volume. Indirect effects are given in red (all P > 0.10).

Discussion

The present data highlight the complexity of relationships between candidate genes, structural neural and cognitive endophenotypes, and mood disorder phenotypes.

None of the four genotypes tested showed significant association with categorical diagnoses (bipolar disorder, major depressive disorder, or any mood disorder), which is perhaps to be expected given the small sample size with regard to case/control association studies. Although not significant, it is interesting to note that the SNPs in the BDNF gene showed the strongest evidence of association with any mood disorder diagnosis, in comparison to specific diagnoses (bipolar disorder or major depressive disorder). This reinforces the notion that candidate polymorphisms may predispose to broader neural system dysfunction rather than to specific neural abnormalities that map to precise mood dysregulation patterns. Instead, the combination of numerous alleles may increase neural system vulnerability to mood dysregulation, and this vulnerability may then be further shaped by environmental influences and mood episode triggers.

The effects observed for CACNA1C further underscore the need to better understand the influence of candidate polymorphisms on neural system functioning and neuropsychiatric phenotypes. Previous large sample genetic association studies have supported a role of the CACNA1C minor allele in bipolar disorder and schizophrenia. Smaller sample studies of healthy and psychiatric disorder populations have found that the CACNA1C risk allele increases mRNA transcript (Bigos et al. 2010) and alters Akt pathway activation (Balog et al. 2010). These molecular changes result in increased brain volumes (Kempton et al. 2009; Wang et al. 2011), reductions in emotional and cognitive processing (Bigos et al. 2010; Krug et al. 2010), increased brain activation signals during cognitive processing (Bigos et al. 2010; Krug et al. 2010), and decreased regional connectivity (Wang et al. 2011). While the present study did not replicate associations between the CACNA1C polymorphism and mood disorder, this is likely due to the need for very large samples to detect weak associations between candidate polymorphisms and neuropsychiatric diagnoses (Ferreira et al. 2008; Sullivan et al. 2012). However, the present work does reinforce previous observations regarding stronger effects for the CACNA1C risk allele on cognitive and neuroimaging endophenotypes (Bigos et al. 2010) and clarifies the nature of these downstream phenotypic effects. CACNA1C minor allele carriers had increased global brain volume, with larger effects for specific fronto-limbic regions - especially the caudate. In contrast with previous literature (Kempton et al. 2009; Wang et al. 2011), total cortical white matter increases were a prominent driver of increased brain volume. Total cortical grey matter increases were present, but more modest and not statistically significant. The CACNA1C risk allele also appears to increase global cognitive ability. This is opposite of most findings suggesting reductions in specific aspects of cognitive and emotional processing (Bigos et al. 2010; Krug et al. 2010; Soeiro-de-Souza et al. 2012), but congruent with a recent study suggesting that CACNA1C minor allele carriers with bipolar disorder may have improved working memory (Zhang et al. 2012). Taken together, these data imply that CACNA1C risk allele effects may be valuable for some aspects of cognition, but harmful for others; an interpretation that fits with an evolutionary view of mood disorder (Akiskal and Akiskal 2005) and disease-associated genes as having both adaptive and nonadaptive value depending on context (Crespi et al. 2007; Tennessen and Akey 2011). Co-incident measurement of mood, global, and specific cognitive processes; brain structure and function; and downstream molecular mechanisms will be needed to more accurately characterize effects of this polymorphism.

The observation of stronger effects of the CACNA1C polymorphism on caudate volumes than on global brain volumes is intriguing. CACNA1C minor alleles have been associated with both bipolar disorder and schizophrenia and nonmotor caudate subregions have been found to be important for integrating cognitive and emotional processing. Thus, the CACNA1C polymorphism may preferentially predispose individuals toward mood disorder or schizoaffective phenotypes. An important next step will be to explore specificity in the patterns of emotion or cognitive dysfunction (developmental course, severity, episodes, etc.) in individuals with this genotype.

The present study is also the first to our knowledge to show that the strong positive relationship between CACNA1C minor allele genotypes and global cognitive ability was not mediated by total brain or fronto-limbic volume increases. The molecular mechanisms by which CACNA1C minor allele genotypes increase brain volumes, particularly fronto-limbic volumes, and independently increase IQ deserve further exploration. It is possible that increases in IQ are mediated by other brain regions not specifically investigated in the present study or that distinct molecular mechanisms influence neural system and cognitive vulnerability. Alternatively, CACNA1C influences on brain structure may be completely separate from influences on function. In this scenario, there may be distinct downstream molecular and cellular effects of the CACNA1C polymorphism that deserve elaboration.

ANK3, BDNF, and DGKH genotypes did not show significant bivariate relationships with imaging volumes or cognitive processes. However, there was a nominally significant relationship between DGKH minor allele genotypes and smaller anterior cingulate volumes and, independently, with reduced verbal memory. These observations are consistent with literature identifying a role of DGKH in mTOR and ERK1/2 signaling influencing cell growth (Merida et al. 2008), including regulation of neural morphology (dendritic branching and spine formation) in fronto-limbic regions and hippocampal long-term potentiation influencing memory (Shirai et al. 2010; Tu-Sekine and Raben 2011).

The major study limitations are the modest sample size for evaluating genetic associations with phenotypic characteristics (particularly within diagnostic groupings), restriction to fronto-limbic volumes, missing data on genotypes, lack of information on smoking status, and lack of ancestry-informative markers to provide more detailed evaluation of population stratification. In spite of the modest sample size for genetic associations with complex neuropsychiatric disease, this study is one of the largest cross-level (polymorphisms-structural imaging-cognition-diagnosis) studies completed to date. Multiple comparison corrections were applied at each stage of analysis to ensure that identified associations are not simply a result of multiple testing. Additionally, covarying for race/ethnicity did not alter the pattern of findings. However, future work should use ancestry-informative markers to more carefully examine population stratification issues.

The presence of missing data on genotypes may have influenced some of the relationships with phenotypic measures, although analyses suggest that any influence of missing data is likely minimal for the analyses presented. The focus on fronto-limbic volumes is both a strength and a weakness because it decreases concerns of Type I error that might arise from evaluation of all brain regions, but also may have resulting in missing important functional relationships. Additionally, lack of information on smoking status is unfortunate because smoking may influence brain volumes in psychiatric groups (Schneider et al. 2014). Future research examining genotype-phenotype relationships should examine both smoking and any other current or past substance abuse as these may be important moderating or confounding factors. Finally, the cross-sectional nature of the study is also a limitation, as longitudinal relationships between levels of the genotype-phenotype pathway and temporal sequencing of possible mediation could not be investigated. Future cross-level studies involving candidate genes would be wise to include multiple assessment points to clarify whether changes in molecular and neural systems measurements influence mood disorder progression. These studies can further elucidate whether molecular or neural system changes are influenced by genotype, as has been shown for BDNF and changes in brain volumes during recovery from drinking (Mon et al. 2013).

The present study greatly extends knowledge of the role of CACNA1C variation in brain structure, function, and vulnerability to mood dysregulation. It also provides a model, as well as a cautionary tale, that informs future cross-level studies evaluating the role of common genetic variation in complex neuropsychiatric diseases. Future large-scale multimodal studies will be needed to clarify relationships between candidate genes, structural and functional brain characteristics, and cognitive processes related to mood disorder vulnerability. These types of studies, if well-powered, have the potential to identify specific molecular-neural systems relationships involved in mood disorder. In doing so, translational studies may identify sensitive neural treatment targets for genetically informed therapeutics, enhancing the speed of development and efficacy evaluation of new medicines.

Acknowledgments

References