Characterisation of genetic variation in ST8SIA2 and its interaction region in NCAM1 in patients with bipolar disorder

Abstract

Alpha-2,8-sialyltransferase 2 (ST8SIA2) is an enzyme responsible for the transfer of polysialic acid (PSA) to glycoproteins, principally the neuronal cell adhesion molecule (NCAM1), and is involved in neuronal plasticity. Variants within ST8SIA2 have previously shown association with bipolar disorder, schizophrenia and autism. In addition, altered PSA-NCAM expression in brains of patients with schizophrenia or bipolar disorder indicates a functional dysregulation of glycosylation in mental illness. To explore the role of sequence variation affecting PSA-NCAM formation, we conducted a targeted re-sequencing study of a ∼ 100 kb region--including the entire ST8SIA2 gene and its region of interaction with NCAM1--in 48 Caucasian cases with bipolar disorder using the Roche 454 platform. We identified over 400 DNA variants, including 47 putative novel variants not described in dbSNP. Validation of a subset of variants via Sequenom showed high reliability of Roche 454 genotype calls (97% genotype concordance, with 80% of novel variants independently verified). We did not observe major loss-of-function mutations that would affect PSA-NCAM formation, either by ablating ST8SIA2 function or by affecting the ability of NCAM1 to be glycosylated. However, we identified 13 SNPs in the UTRs of ST8SIA2, a synonymous coding SNP in exon 5 (rs2305561, P207P) and many additional non-coding variants that may influence splicing or regulation of ST8SIA2 expression. We calculated nucleotide diversity within ST8SIA2 on specific haplotypes, finding that the diversity on the specific "risk" and "protective" haplotypes was lower than other non-disease-associated haplotypes, suggesting that putative functional variation may have arisen on a spectrum of haplotypes. We have identified common and novel variants (rs11074064, rs722645, 15:92961050) that exist on a spectrum of haplotypes, yet are plausible candidates for conferring the effect of risk and protective haplotypes via multiple enhancer elements. A Galaxy workflow/pipeline for sequence analysis used herein is available at: https://main.g2.bx.psu.edu/u/a-shaw-neura/p/next-generation-resources.

Figure 1. Coverage of ST8SIA2 gene locus by 454 and Sanger sequencing.

The 96∼20 kb flanking sequence (chr15∶92,919,255–93,013,920 bp; UCSC Genome Browser hg19 http://genome.ucsc.edu) was divided into eight long-range PCR amplicons (454 Amplicons 1–8, in pink) for library preparation prior to 454 sequencing. The mean depth of coverage across all 48 samples for each long-range amplicon is shown in burgundy. A GC-rich region surrounding the promoter and exon 1 was sequenced via Sanger sequencing using 8 short overlapping amplicons (Sanger amplicons 1–8; inset). Pink dots on the GC percent or mean depth per sample tracks indicate loci where the level of GC richness, or read depth is above the range indicated. Regions which failed to amplify, or which contained simple repeats, long homopolymers (length > = 10), or short interspersed elements (SINEs) tended to have lower depth of coverage, and were inaccessible to next-generation sequencing in phase I of the 1000 genomes project (1000 G Accs strict).

Figure 2. Region of NCAM1 gene covered via Roche 454 sequencing.

The sequenced region spanned 6.6(chr11∶113,100,972–113,107,571 bp; hg19) and included exons 8–12 (red boxes) and less frequently used exons 8a and mini-a (grey boxes) . The locations of the 5^th and 6^th glycosylation sites and acidic patch required for glycosylation are in exons 9, 10 and 11 and are indicated by green, blue and white lines, respectively. The locations of SNPs identified in this study are indicated with arrows. Names of known SNPs are in black text, and novel SNPs are listed as NV (novel base substitutions) or InDel (novel insertion-deletion change) followed by their base position (hg19). Two coding SNPs located in exons 8a and 11 are marked with closed boxes, with their effect on amino acid sequence shown.

Figure 3. Heatmap of mean depth of coverage per amplicon and per sample.

Amplicons are listed along the x-axis, and samples (x1-48) are listed along the y-axis. Depth of coverage is indicated according to the colour key, with white indicating <2 reads coverage and red indicating >35 reads coverage. The plot was created using R .

Figure 4. Comparison of detected SNP variants with all SNPs detected in 1 kG data.

The minor allele frequency (MAF) of all SNPs identified in the sequenced region from 174 Caucasian Europeans from phase I of the 1 kG project (n = 519) and 48 bipolar disorder cases (n = 412) were compared. SNPs found in bipolar individuals only (yellow) and 1 kG individuals only (light blue) were predominantly rare alleles whereas those found in both sets (green) were predominantly common.

Figure 5. Haplotype phasing of all detected nucleotide variants in ST8SIA2 gene region.

The gene structure is shown above, with the minor allele frequency of each SNP detected shown by the histogram immediately below the gene (red represents frequency of alternate allele). The red, green and yellow bars on the left indicate the haplotype on which the detected variation resides (risk  =  red; protective  =  green; other haplotypes  =  yellow). Detected variation with respect to the specific haplotype on which it resides is shown to the right of haplotype bars, where non-reference alleles are coloured in light blue (homozygous) and dark blue (heterozygous). The locations of the six SNP markers that were used to define the haplotypes are shown below the variation. The location of the PreSTIGE enhancer elements within the 454 sequenced region from human skeletal muscle myoblasts (HSMM) are indicated in orange, and from neuronal precursor cells (NPC) are indicated in purple. PreSTIGE enhancer locations were obtained from http://genetics.case.edu/prestige/).