4.1.3 Limitations
On top of limitation addressed in our publication, which include the retrospective assessment of phenotypes, as well as a lack of information on healthy controls, it is important to mention that we are fully aware of the fact that candidate approaches are nowadays outdated. However, as so far GWAS on subphenotypes are scarce, the specific subphenotypes reported in the original studies have not been studied at all using GWAS and the original articles are still being cited, we decided it was important to publish this data.
4.1.4 Personal comments
This study was one of the first studies performed during this PhD thesis and was meant to build up onto a larger study on the role of AKT/GSKβ signalling, especially in light of lithium response. However, as the first step, the replication of the studies by Karege et al. was unsuccessful, and upcoming GWAS studies on BD did not give any indications on an involvement of AKT1 in BD, we decided to stop the project. The publication process of this data proved to be difficult. This study exemplifies all projects with negative results, published or unpublished, that I have encountered during my doctoral studies.
This study is reprinted under the Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0) licence.
with BD, for example the NMDA-receptor subunit GRIN2A44 or the SYNE1/CPG2 gene247. Further support comes from imaging studies, in particular 1H-MRS studies (cf. 3.3). As discussed in section 1.2.3.2, BD has been associated with elevated Glx and glutamate levels in ACC and prefrontal regions53,54. In contrast to these findings, meta-analyses on major depressive disorder found lower glutamate and glutamine levels in ACC248, and the PFC249.
Few studies so far have analyzed the effects of genetics on glutamate levels in the brain, mostly in the context of schizophrenia spectrum disorders250,251. A recent twin study analyzed the heritability of several metabolite levels in the thalamus and the ACC in the context of schizophrenia spectrum disorders. The authors report heritability estimates of about 0.3 for both glutamate and Glx in the ACC252. In bipolar disorder, a genetic variation in bcl-2 was associated with changed Glx/Creatine in ACC253, while the val66met polymorphism in the BDNF gene was associated with changes of Glx/Creatine in the left hippocampus254.
4.2.1 Results
In this study, we aimed at exploring the effects of genes of the glutamate circle on Glx and glutamate levels in the ACC and the left dorsolateral PFC in a mixed cohort of depressed patients. We included SNPs in essential regulatory elements and coding sequences of GLUL, SLC1A3, SLC1A2, and SLC1A7 encoding for glutamine synthetase, excitatory amino acid transporter (EAAT) 1, EAAT2 and EAAT5, respectively.
While no difference was found between patient with MDD or BD in any of the regions, the minor alleles of rs3812778 and rs3829280, two SNPs in perfect linkage disequilibrium (LD) in the 3’-untranslated region (UTR) of SLC1A2, were associated with elevated 2D JPRESS mean ACC glutamate levels. No associations were observed for the combined glutamate/glutamine levels.
In silico analysis using the BrainCloud eQTL browser revealed an association between the minor allele of rs3829280 and higher levels of CD44, a gene situated downstream of SLC1A2 on chromosome 1. These results were corrobated by data from the UK Brain Expression Consortium, where associations in the same directions were found between the full-length transcript and rs3812778 in the cerebellar cortex, the putamen, the substantia nigra, as well as in the average of all brain regions.
CD44 codes for a receptor for hyaluron and plays important roles in cell-matrix binding, signaling, cell migration255, inflammation256 and has been shown to play an important role in brain development257. Having also been identified as a marker for astrocyte precursor cells258, we assessed the correlation with several astrocytic markers and markers of the glutamate-glutamine cycle and found very high correlation of CD44 with markers expressed in astrocytes (AQP4, GFAP, S100B, GLUL, SLC1A3, SLC1A2, SLC38A3 ), but not with
Figure 4.1: Pproposed Mechanism for the Impact of Genetic Variations in SLC1A2 on Glutamatergic Neurotransmission and Rapid Cycling. Figure from Veldic, Millischer et al., 2019, Translational Psychiatry.
those expressed in neurons (GLS, SLC1A1, SLC1A6, SLC38A1, SLC17A7, SLC17A6, SLC17A8, and SLC1A7 ).
Finally, to explore potential disease relevance, we followed up the finding in a mixed cohort composed of both MDD and BD. While no significant difference in the percentage of minor allele carriers was found between patients with MDD and patients with BD, patients with rapid cycling BD had a significantly higher percentage of minor allele carriers (26.9% [95% Confidence Interval (CI): 23.5–30.5]) in comparison to non-rapid cycling (21.8% [95% CI: 17.9–25.8]) or patients with MDD (21.9% [95% CI: 19.1–24.9]).
4.2.2 Summary
Based on our findings, we propose a mechanism presented in Fig. 4.1. We hypothesize that the minor alleles of rs3812778/rs3829280 are associated with increased levels of CD44, possibly a sign of higher numbers of astrocytes. This increase of EAAT-expressing cells could lead to increased glutamate recycling, dysregulation of glutamatergic neurotransmission, and increased risk for rapid-cycling. While some of these association have newly been reported by this study (blue arrows), other associations in this hypothesis (green arrows) have been reported previously259,260.
4.2.3 Limitations
There are several limitations to our study. The main one being the small number of patients included in the MRS study. Furthermore, not all genes that could have an impact on the glutamate levels were analysed. A replication in a bigger cohort, including a more complete list of genes (and their genetic variation) is therefore necessary. In order to address the power problematic, such a study would most probably have to be performed in a collaborative manner.
Furthermore, 1H-MRS measurements of glutamate have an intrinsic limitation, as this method captures the total concentration of glutamate in the area of interest, making it difficult to separate (1) intra- and extracellular glutamate, as well as (2) glutamate used for metabolic functions from glutamate used as neurotransmitter53.
Finally, there is some heterogeneity in the diagnoses of BD and MDD. While BD diagnosis was based on medical interviews, medical records and questionnaires performed in clinical settings, MDD cases where selected from a population cohort using the Major Depression Inventory.
This study is reprinted under the Attribution 4.0 International (CC BY 4.0) licence.