Altered gamma-aminobutyric acid type B receptor subunit 1 splicing in alcoholics.

Gene specific library constructionRibosomal depleted total RNA was used to prepare the gene specific library. Single stranded cDNA (sscDNA) was generated by reverse transcription using a gene specific primer. Double stranded (dscDNA) was synthesized by template switching, and then another gene specific primer was used for amplification. After gel extraction, size selected dscDNAs were sonicated, and adaptors were ligated. The gene specific library was sequenced using the ABI SOLiD™ system.

Comparison of gene specific and whole transcriptome librariesAll TopHat mapping output files, .wig and .sam, were loaded on the IGV and are shown at the GABAB1 gene in blue and grey, respectively. The .wig files (blue) contain the number of mapped reads at each nucleotide position on top. The .sam files (grey) contain individual mapped reads. For the .sam files, the thick lines represent mapped reads, and the thin lines represent splice junctions. Mapped read numbers were compared in log space using the .wig files. With the two files, the gene specific library and whole transcriptome library results are shown in A and B, respectively. Because GABAB1 is negatively oriented at chromosome 6, the 5’ and 3’ ends of GABAB1 are located on the right and left during IGV visualization, respectively. For the gene specific library, only the upper portion of the .sam file data is shown due to space limitations. The gene specific library results had more GABAB1 specific mapped reads and splice junctions than the whole transcriptome library. For this comparison, both libraries were prepared from the same commercial human prefrontal cortex RNAs. The black bar on top of Figure A represents the location of gene specific primers. Exon numbers are labeled below the gene model of A and B.

Data analysis to maximize splice junction detectionTo maximize the identification of splice junctions using TopHat, a splice junction search step was added to the data analysis pipeline. We used gene specific libraries (GSLs) and whole transcriptome libraries (WTLs) prepared from commercial and postmortem human prefrontal cortices. After filtering low quality reads and undesired sequence containing reads, the data were converted to .fq files. After TopHat mapping, splice junction information was collected for the remapping step. We used the whole transcriptome library data from 14 alcoholic and 15 control human brain samples for the remapping step. After recollecting splice junctions, we defined the minimum GABAB1 gene length.

GABAB1 splice junctions and gene boundariesAfter combining splice junctions from the gene specific and the whole transcriptome libraries, we found that there are at least 65 splice junctions associated with the GABAB1 gene. The 65 splice junctions were visualized with the UCSC genome browser using a .juncs file and hg18. From these results, we determined that the GABAB1 gene is at least 72,333bp (chr6:29,653,437-29,725,769) long, which is about 41,000bp larger than the expected size of 30,958bp (chr6:29,677,984-29,708,941) based on the RefSeq Genes model. From top to bottom, the tracts include User Supplied Track (.juncs file loaded track) and RefSeq Genes.

GABAB1 gene expression comparison in alcoholic and control brainsGABAB1 expression was compared in 14 alcoholic and 15 control brains. A. At the known GABAB1 gene boundary of the RefSeq Genes model, expression levels were compared between groups. After log2 transformation of RPGM values, unpaired t-tests were performed to define statistically significant differences. Y-axis represents log2 RPGM. From top to bottom, the horizontal lines of each boxplot represent the largest non-outlier, upper quartile, median, lower quartile, and the smallest non-outlier values, respectively. B. GABAB1 gene expression levels were compared based on the minimum GABAB1 gene boundary from RNA-seq data. C. Gene expression levels were assessed based on the minimum GABAB1 gene boundary obtained from the UCSC genome browser database.

GABAB1 exon and intron expressionsRPGM values were calculated for each GABAB1 exon and intron from15 alcoholic and 14 control samples. 5’ GABAB1 exons showed much lower expression and indicated the abundance of small splicing variants instead of splicing variants containing all 23 exons. Y-axis represents RPGM values. Data are expressed as the mean ± standard error of the mean (SEM).

GABAB1 exon, intron, and splice junction changes in alcoholic brainsExon and intron expression and splice junctions were compared between alcoholic and control brains. Based on the RefSeq Genes model, RPGM values were calculated for all exons and introns. For all splice junctions found from TopHat mapping, RPJMs were calculated. Among significant RPGM and RPJM changes, GABAB1 specific data were selected and summarized. The expression of 11 exons was changed in alcoholic brains. In alcoholic brains, exons 10 and 17 expression decreased while increases were observed for 9 exons (7, 8, 11, 12, 14, 15, 16, 19, and 22). Exons 7-8, exon 10, exons 11-12, exons 14-16, and exon 17 were grouped based on expression directions and levels in Figure 6. Two other splice junction changes did not alter neighboring exon expression and showed the complexity of GABAB1 splicing. Red and blue rectangles indicate significant expression increases and decreases in exons, respectively. ** and ## indicate p-values < 0.001; * and # represent p-values < 0.05. Red and blue bent lines represent splice junctions that significantly increased and decreased in alcoholic brains, respectively.