Genetic modulation of GABA levels in the anterior cingulate cortex by GAD1 and COMT

Abstract

Gamma-aminobutyric acid (GABA)-ergic transmission is critical for normal cortical function and is likely abnormal in a variety of neuropsychiatric disorders. We tested the in vivo effects of variations in two genes implicated in GABA function on GABA concentrations in prefrontal cortex of living subjects: glutamic acid decarboxylase 1 (GAD1), which encodes GAD67, and catechol-o-methyltransferase (COMT), which regulates synaptic dopamine in the cortex. We studied six single nucleotide polymorphisms (SNPs) in GAD1 previously associated with risk for schizophrenia or cognitive dysfunction and the val158met polymorphism in COMT in 116 healthy volunteers using proton magnetic resonance spectroscopy. Two of the GAD1 SNPs (rs1978340 (p=0.005) and rs769390 (p=0.004)) showed effects on GABA levels as did COMT val158met (p=0.04). We then tested three SNPs in GAD1 (rs1978340, rs11542313, and rs769390) for interaction with COMT val158met based on previous clinical results. In this model, rs11542313 and COMT val158met showed significant main effects (p=0.001 and 0.003, respectively) and a trend toward a significant interaction (p=0.05). Interestingly, GAD1 risk alleles for schizophrenia were associated with higher GABA/Cre, and Val-Val homozygotes had high GABA/Cre levels when on a GAD1 risk genotype background (N=6). These results support the importance of genetic variation in GAD1 and COMT in regulating prefrontal cortical GABA function. The directionality of the effects, however, is inconsistent with earlier evidence of decreased GABA activity in schizophrenia.

Figure 1

Magnetic resonance spectroscopy (MRS) voxel position. Sagittal, coronal, and axial images showing the position of the anterior cingulate voxel of interest, placed in the medial prefrontal cortex while trying to maximize gray matter.

Figure 2

Spectroscopy output. A typical difference spectrum is displayed, showing the GABA resonance at 3.0 p.p.m. and co-edited Glx at 3.8 p.p.m. The N-acetyl-aspartate (NAA) signal at 2 p.p.m. appears inverted due to the editing that suppresses frequencies around this area of the spectrum. The subtraction of the unedited from the edited spectrum will thus result in a negative peak for NAA. The dotted line represents the automated fit to the raw data (continuous line).

Figure 3

Linkage disequilibrium (r²) between the 10 genotyped single nucleotide polymorphisms (SNPs). Figure obtained with Haploview (http://www.broadinstitute.org/mpg/haploview).

Figure 4

Directionality of main effects of GAD1 single nucleotide polymorphisms (SNPs) on GABA/Cre. The alleles associated with increased risk for schizophrenia according to Straub et al (2007) were A for rs11978340 and C for rs769390. Error bars represent 95% confidence limits. The p-values for post hoc comparisons (Tukey's ‘honest significant difference' correction for multiple comparisons) are also shown, with lines connecting the bars that showed significant differences. All results were adjusted for age in an ANCOVA model.

Figure 5

Effects of age (a), COMT Val158Met (b), GAD1 rs11542313 (c), and the interaction of the two genotypes (d). Error bars represent 95% confidence intervals, as does the dotted line in (a). The significant post hoc comparisons are also shown, as in Figure 4. The number of individuals in each cell is specified in (c) and (d).