Support for schizophrenia susceptibility locus on chromosome 2q detected in a Swedish isolate using a dense map of microsatellites and SNPs.

Extended pedigrees are not only very useful to identify disease genes for rare Mendelian conditions, but they may also help unravel the genetics of complex diseases such as schizophrenia. In this study we performed genome-wide multipoint non-parametric linkage (NPL) score calculations using 825 microsatellites and 5,366 single nucleotide polymorphisms (SNPs), respectively, and searched for haplotypes shared by affected individuals, in three multiplex families including 29 genotyped affected individuals which in total contains 49 relative pairs useful for linkage studies. The most consistent results for microsatellites and SNPs were observed on 2q12.3-q14.1 (NPL scores 2.0, empirical P-value 0.009). However, the overall highest NPL score was observed on chromosome 2q33.3 using SNPs (NPL score 2.2, empirical P-value 0.007). Other chromosomal regions were detected on 5q15-q22.1, with microsatellites (NPL scores 1.7, empirical P-value 0.021) and with SNPs (NPL scores 2.0, empirical P-value 0.010) and on 5q23.1 (NPL score 1.9, empirical P-value 0.012) and 8q24.1-q24.2 (NPL score 2.1, empirical P-value 0.009) when using SNPs. The analysis of extended pedigrees allowed the search for haplotypes inherited identical by decent (IBD) by affected individuals. In all regions with NPL score >1.9 we found haplotypes inherited IBD by multiple cases. However, no common haplotypes were found for affected individuals in all families. In conclusion our NPL results support earlier findings suggesting that 2q and possibly 5q and 8q contain susceptibility loci for schizophrenia. Haplotype sharing in families helped to delimit the detected regions that potentially are susceptibility loci for schizophrenia.

Inflammation-related genes up-regulated in schizophrenia brains.

BACKGROUND: Multiple studies have shown that brain gene expression is disturbed in subjects suffering from schizophrenia. However, disentangling disease effects from alterations caused by medication is a challenging task. The main goal of this study is to find transcriptional alterations in schizophrenia that are independent of neuroleptic treatment. METHODS: We compared the transcriptional profiles in brain autopsy samples from 55 control individuals with that from 55 schizophrenic subjects, subdivided according to the type of antipsychotic medication received. RESULTS: Using global and high-resolution mRNA quantification techniques, we show that genes involved in immune response (GO:0006955) are up regulated in all groups of patients, including those not treated at the time of death. In particular, IFITM2, IFITM3, SERPINA3, and GBP1 showed increased mRNA levels in schizophrenia (p-values from qPCR < or = 0.01). These four genes were co-expressed in both schizophrenic subjects and controls. In-vitro experiments suggest that these genes are expressed in both oligodendrocyte and endothelial cells, where transcription is inducible by the inflammatory cytokines TNF-alpha, IFN-alpha and IFN-gamma. CONCLUSION: Although the modified genes are not classical indicators of chronic or acute inflammation, our results indicate alterations of inflammation-related pathways in schizophrenia. In addition, the observation in oligodendrocyte cells suggests that alterations in inflammatory-related genes may have consequences for myelination. Our findings encourage future research to explore whether anti-inflammatory agents can be used in combination with traditional antipsychotics for a more efficient treatment of schizophrenia.

Reconstruction of ancestral haplotypes in a 12-generation schizophrenia pedigree.

We searched for candidate chromosomal regions inherited identical by descent in 19 patients suffering from schizophrenia or schizoaffective disorder that are related 12 generations back, to an ancestral couple born in the middle of the seventeenth century. To accomplish this goal, we constructed complete chromosomal haplotypes for each patient using genotype data from 450 markers. In total, 12 haplotype regions (with sizes ranging from 0.6 to 10.9 cM) constituted by three markers each were identical in three or more of the affected individuals. The largest genomic segment was located on 6q25, a region previously shown to be significantly more frequent in patients than controls, and proposed to contain a schizophrenia susceptibility locus. For the remaining 11 candidate haplotypes, we estimated haplotype frequencies from all the 43 affected members collected from the same family and 46 unrelated control individuals. This analysis indicated that at least four of the 11 candidate haplotypes are ancestral, since the frequencies were significantly higher in patients than in controls. Five additional haplotypes showed higher estimated frequencies in the patients but the differences were not significant. Interestingly, five of these 11 genomic regions are located in, or close to, candidate regions previously suggested to contain susceptibility genes for schizophrenia. The regions are 5q21-23, 8p21-22, 10p13-15, 13q12-13 and 22q12-13. Several of these haplotypes are probably ancestral linkage disequilibrium blocks inherited from the original couple. There exists, however, the possibility that one or more of these regions harbour schizophrenia susceptibility loci that may have epistatic interactions among them.

Human QKI, a new candidate gene for schizophrenia involved in myelination.

We have previously shown that chromosome 6q25-6q27 includes a susceptibility locus for schizophrenia in a large pedigree from northern Sweden. In this study, we fine-mapped a 10.7 Mb region, included in this locus, using 42 microsatellites or SNP markers. We found a 0.5 Mb haplotype, likely to be inherited identical by decent, within the large family that is shared among the majority of the patients (69%). A gamete competition test of this haplotype in 176 unrelated nuclear families from the same geographical area as the large family showed association to schizophrenia (empirical P-value 0.041). The only gene located in the region, the quaking homolog, KH domain RNA binding (mouse) (QKI), was investigated in human brain autopsies from 55 cases and 55 controls using a high-resolution mRNA expression analysis. Relative mRNA expression levels of two QKI splice variants were clearly downregulated in schizophrenic patients (P-value 0.0004 and 0.03, respectively). The function of QKI has not been studied in humans, but the mouse homolog is involved in neural development and myelination. In conclusion, we present evidence from three unrelated sample-sets that propose the involvement of the QKI gene in schizophrenia. The two family based studies suggest that there may be functional variants of the QKI gene that increase the susceptibility of schizophrenia in northern Sweden, whereas the case-control study suggest that splicing of the gene may be disturbed in schizophrenic patients from other geographical origins. Taken together, we propose QKI as a possible target for functional studies related to the role of myelination in schizophrenia.

Comparative informativeness for linkage of multiple SNPs and single microsatellites.

Single nucleotide polymorphisms (SNPs), or biallelic markers, are popular in genetic linkage studies due to their abundance in the genome, stability, and ease of scoring. We determined the 'information ratio' (IR) of closely spaced SNPs in simulated nuclear families and affected sib pairs (ASPs). (The IR is the ratio of actual average maximum lod score to the maximum lod score attainable if the marker were fully informative.) The nuclear families included parental information, whereas the ASPs did not. We analyzed these SNPs in two ways: (1) using multipoint analysis, and (2) treating the SNPs as 'composite markers' (i.e., haplotypes, as assigned by GENEHUNTER). (3) We also calculated the IR of a single microsatellite marker with multiple alleles and compared with the IR from the SNPs. For each set of input conditions, we simulated 1000 nuclear families, of 2, 3, 4, or 5 children each, as well as 1000 ASPs. We generated SNP marker data for strings of k = 1, 2, 3, 5, 7, and 10 SNP loci, with no recombination (theta = 0) and no linkage disequilibrium among the SNPs. The MAF (minor allele frequency) was either 0.5 or 0.25, and allele frequencies were the same for all k loci in any analysis. We also generated marker data for one single-locus microsatellite marker, with m = 3, 4, 5, 6, 7, and 9 equally frequent alleles. In all simulations, the disease was fully penetrant dominant, and there was no recombination or linkage disequilibrium among markers or between marker and disease. When multipoint analysis was used, we found that 5-7 closely spaced SNPs were usually enough to yield an IR of approximately 100%, for nuclear families of any size. However, for the ASPs, even 7-10 SNPs yielded an IR of only 70-80%. A microsatellite with 9 equally frequent alleles yielded about the same IR (86-88%) as a string of 4-5 SNPs, in nuclear families. SNPs analyzed as 'composite markers' analyses performed worse, due to the inherent ambiguity of SNP haplotyping.

The reliability of haplotyping inference in nuclear families: misassignment rates for SNPs and microsatellites.

Single nucleotide polymorphisms (SNPs) are widely used when investigators try to map complex disease genes. Although biallelic SNP markers are less informative than microsatellite markers, one can increase their information content by using haplotypes. However, assigning haplotypes (i.e., assigning phase) correctly can be problematic in the presence of SNP heterozygosity. For example, a doubly heterozygous individual, with genotype 12, 12, could have haplotypes 1-1/2-2 or 1-2/2-1 with equal probability; in the absence of additional information, there is no way to determine which haplotype is correct. Thus an algorithm that assigns haplotypes to such an individual will assign the wrong one 50% of the time. We have studied the frequency of haplotype misassignments, i.e., haplotypes that are misassigned solely because of inherent marker ambiguity (not because of errors in genotyping or calculation). We examined both SNPs and microsatellite markers. We used the computer programs GENEHUNTER and SIMWALK to assign the haplotypes. We simulated (a) families with 1-5 children, (b) haplotypes involving different numbers of marker loci (3, 5, 7 and 10 loci, all in linkage equilibrium), and (c) different allele frequencies. Misassignment rates are highest (a) in small families, (b) with many SNP loci, and (c) for loci with the greatest heterozygosity (i.e., where both alleles have frequency 0.5). For example, for triads (i.e., one-child families with both parents genotyped), misassignment rates for SNPs can reach almost 50%. Family sizes of 4-5 children are required in order to ensure a misassignment frequency of < or = 5% for ten-SNP haplotypes with allele frequencies of 0.25-0.5. For microsatellites, a family size of at least 2-3 children is necessary to keep haplotyping misassignments < or = 5%. Finally, we point out that it is misleading for a computer program to yield haplotype assignments without indicating that they may have been misassigned, and we discuss the implications of these misassignments for association and linkage analysis.

Genome scan meta-analysis of schizophrenia and bipolar disorder, part II: Schizophrenia.

Schizophrenia is a common disorder with high heritability and a 10-fold increase in risk to siblings of probands. Replication has been inconsistent for reports of significant genetic linkage. To assess evidence for linkage across studies, rank-based genome scan meta-analysis (GSMA) was applied to data from 20 schizophrenia genome scans. Each marker for each scan was assigned to 1 of 120 30-cM bins, with the bins ranked by linkage scores (1 = most significant) and the ranks averaged across studies (R(avg)) and then weighted for sample size (N(sqrt)[affected casess]). A permutation test was used to compute the probability of observing, by chance, each bin's average rank (P(AvgRnk)) or of observing it for a bin with the same place (first, second, etc.) in the order of average ranks in each permutation (P(ord)). The GSMA produced significant genomewide evidence for linkage on chromosome 2q (PAvgRnk<.000417). Two aggregate criteria for linkage were also met (clusters of nominally significant P values that did not occur in 1,000 replicates of the entire data set with no linkage present): 12 consecutive bins with both P(AvgRnk) and P(ord)<.05, including regions of chromosomes 5q, 3p, 11q, 6p, 1q, 22q, 8p, 20q, and 14p, and 19 consecutive bins with P(ord)<.05, additionally including regions of chromosomes 16q, 18q, 10p, 15q, 6q, and 17q. There is greater consistency of linkage results across studies than has been previously recognized. The results suggest that some or all of these regions contain loci that increase susceptibility to schizophrenia in diverse populations.