Exploring the transcriptome of the brain

University dissertation from Stockholm : Karolinska Institutet, Dept of Neuroscience

Abstract: Our knowledge of the transcriptome has become much more complex since the days of the central dogma of molecular biology. We now know that splicing takes place to create potentially thousands of isoforms from a single gene, and we know that RNA does not always faithfully recapitulate DNA if RNA editing occurs. Collectively, these observations show that the transcriptome is amazingly rich with intricate regulatory mechanisms for overall gene expression, splicing, and RNA editing. Genetic variability can play a role in controlling gene expression, which can be identified by examining expression quantitative trait loci (eQTLs). eQTLs are genomic regions where genetic variants, including single nucleotide polymorphisms (SNPs) show a statistical association with expression of mRNA transcripts. In humans, many SNPs are also associated with disease, and have been identified using genome wide association studies (GWAS) but the biological effects of those SNPs are usually not known. If SNPs found in GWAS are also found in eQTLs, then one could hypothesize that expression levels may contribute to disease risk. Performing eQTL analysis with GWAS SNPs in both blood and brain, specifically the frontal cortex and the cerebellum, we found both shared and tissue unique eQTLS. The identification of tissue-unique eQTLs supports the argument that choice of tissue type is important in eQTL studies (Paper I). Aging is a complex process with the mechanisms underlying aging still being poorly defined. There is evidence that the transcriptome changes with age, and hence we used the brain dataset from our first paper as a discovery set, with an additional replication dataset, to investigate any aging-gene expression associations. We found evidence that many genes were associated with aging. We further found that there were more statically significant expression changes in the frontal cortex versus the cerebellum, indicating that brain regions may age at different rates. As the brain is a heterogeneous tissue including both neurons and non-neuronal cells, we used LCM to capture Purkinje cells as a representative neuronal type and repeated the age analysis. Looking at the discovery, replication and Purkinje cell datasets we found five genes with strong, replicated evidence of age-expression associations (Paper II). Being able to capture and quantify the depth of the transcriptome has been a lengthy process starting with methods that could only measure a single gene to genome-wide techniques such as microarray. A recently developed technology, RNA-Seq, shows promise in its ability to capture expression, splicing, and editing and with its broad dynamic range quantification is accurate and reliable. RNA-Seq is, however, data intensive and a great deal of computational expertise is required to fully utilize the strengths of this method. We aimed to create a small, well-controlled, experiment in order to test the performance of this relatively new technology in the brain. We chose embryonic versus adult cerebral cortex, as mice are genetically homogenous and there are many known differences in gene expression related to brain development that we could use as benchmarks for analysis testing. We found a large number of differences in total gene expression between embryonic and adult brain. Rigorous technical and biological validation illustrated the accuracy and dynamic range of RNA-Seq. We were also able to interrogate differences in exon usage in the same dataset. Finally we were able to identify and quantify both well-known and novel A-to-I edit sites. Overall this project helped us develop the tools needed to build usable pipelines for RNA-Seq data processing (Paper III). Our studies in the developing brain (Paper III) illustrated that RNA-Seq was a useful unbiased method for investigating RNA editing. To extend this further, we utilized a genetically modified mouse model to study the transcriptomic role of the RNA editing enzyme ADAR2. We found that ADAR2 was important for editing of the coding region of mRNA as a large proportion of RNA editing sites in coding regions had a statistically significant decrease in editing percentages in Adar2 -/-Gria2 R/R mice versus controls. However, despite indications in the literature that ADAR2 may also be involved in splicing and expression regulatory machinery we found no changes in gene expression or exon utilization in Adar2 -/-Gria2 R/R mice as compared to their littermate controls (Paper IV). In our final study, based on the methods developed in Papers III and IV, we revisited the idea of age related gene expression associations from Paper II. We used a subset of human frontal cortices for RNA sequencing. Interestingly we found more gene expression changes with aging compared to the previous data using microarrays in Paper II. When the significant gene lists were analysed for gene ontology enrichment, we found that there was a large number of downregulated genes involved in synaptic function while those that were upregulated had enrichment in immune function. This dataset illustrates that the aging brain may be predisposed to the processes found in neurodegenerative diseases (Paper V).

  This dissertation MIGHT be available in PDF-format. Check this page to see if it is available for download.