A SNP was considered valid if the frequency of minor allele was larger than 0.1 and genotyping missing rate was smaller than 30%. Forty-seven out of 120 SNPs met the validity criteria. The 47 SNPs span across 1591 kb around D1S251 marker (Table I) and cover 11 known functional genes of COG2, AGT, CAPN9, FLJ14525, FLJ 22584, ARV1, GNPAT, DKFZP547NO43, EGLN1, TSNAX, and DISC1. Except five SNPs (SNP 495, 506, 513, 527, 581), the other 42 SNP markers are compatible with the Hardy-Weinberg equilibrium.
Construction of SNP block
In order to perform haplotype analysis haplotype block was evaluated by inter-marker linkage disequilibrium coefficient. . Two SNP blocks were identified by using 2 criteria: 1) inter-marker association test based on chi-squared test was significant and 2) Coefficient D’ was higher than 0.8. The locations of these two SNP blocks are shown Figure 1. The first block covers the SNP markers of 482 (intron 2), 485 (intron 2), 479 (intron 5), 488 (intron 11) and 489 (intron 13) in the GNPAT gene region; the second block covers the markers of 517 (intron 4) and 518 (intron 5) in the DISC1 gene region as shown in Figure 2.
Single-locus association analysis
The preliminary analysis investigates the effect of single locus under phenotype models and provides information for the later multi-loci analysis. From the result of single-locus association analyses using FBAT program version 1.4.1 (Horvath et al 2001; Laird et al 2000), we found some SNP variants exhibited significant consequences in the broad and narrow models of schizophrenia phenotypes, respectively. In the broad model, the significant SNP marker on GNPAT gene is 485 (rs508908) (p=0.038), and on DISC1 gene are 517 (rs2793092) (p=0.005) and SNP518 (rs2793091) (p=0.008). In the narrow model, SNP marker 485 (rs508908) on GNPAT gene shows a modest effect p=0.070, and the highly significant results on DISC1 gene are 517 (rs2793092) (p=0.001) and 518 (rs2793091) (p=0.007). The details are summarized in Table 2. The results based on TRANSMIT program version 2.5.4 (Clayton 1999) yielded more significant results with similar pattern, especially for SNPs on GNPAT gene under narrow model, 485 (rs508908) (p=0.019) and 479 (rs538643) (p=0.049).
Haplotype-based association analysis
In order to evaluate the relationship between the haplotypes that were constructed and schizophrenia, haplotype-based association analysis was applied. The results from haplotype FBAT program version 1.4.1 (Horvath et al 2004) showed that GATTT haplotype in the GNPAT gene SNP block (block 1) is only slightly significant for the broad model of schizophrenia phenotype (p=0.0461). The AG and the GA haplotypes in the DISC1 gene SNP block (block 2) reveal significant association for both of the narrow model and the broad model of schizophrenia. It seems that haplotype GA is a risk haplotype and AG possesses protection effect. The corresponding p-values are p=0.0091 and p=0.0008 for the narrow model, and p=0.03 and p=0.002 for the broad model. The results are summarized in Table 2. Analyses carried out by TRANSMIT
60
program version 2.5.4 (Clayton 1999) yielded similar pattern and the results are not shown here. The interaction of two haplotype blocks was investigated by using the GEE method (Liang and Zeger 1986). However, no significance was found.
Quantitative TDT for phenotypic indicators
When correlating the SNP genotypes with the pathological symptoms of schizophrenia, we analyzed the symptomatic variables of age of onset of the initial symptom, global score of negative symptom, global score of positive symptoms, adjusted Z value of undegraded and degraded CPT, adjusted lβvalue of undegraded and degraded CPT, and WCST indicators of total errors, perseverative errors, perseverative response, categories achieved, learn to learn response, and failure to maintain set.
The analysis of heritability using QTDT version 2.4.3 (Abecasis et al 2000a,b) was conducted to screen the important quantitative phenotypes and the results are shown in Table III. Table III lists the variance components due to environment (
σ ) and
E2 genetic effect (σ
G2 ). Heritabilityh
2 is defined as the explained proportion due to genetic component. Two highly heritable phenotypes, d’ indices of undegraded CPT and degraded CPT, were identified by likelihood ratio tests. The former yields a very significant p-value 8*10-7 and heritabilityh
2 is 0.831, the latter yields p-value 2*10-7 andh
2 is 0.751.Quantitative haplotype analysis based on variance component analysis was investigated using these two highly heritable traits. Based on the Akaike information criterion (AIC) of model selection, the optimum variance-component models for two traits and haplotypes in two blocks were selected respectively. Indices YLN and MYLN are regarded as the adjusted covariates in the model fitting. The results show that significant association between the first haplotype block and degraded CPT index. The second haplotype block is associated to undegraded CPT index.
DISCUSSION
DISC1 gene has been known involved in major psychiatric illness, including schizophrenia (50-52). In our previous study, we have found that the D1S251 marker upstream the DISC1 gene has the maximum NPL score of 1.73 (p=0.03) in narrow phenotype model, and the highest peak NPL score of 2.18 (p=0.01) in the broad phenotype model (32). In exactly the same marker, the populations of British and Iceland have shown LOD scores of 2.5 (p=0.002) (31). These results suggest that there are positional candidate genes in this region susceptible to schizophrenia.
In further fine mapping study, the Finnish schizophrenia samples have shown a two SNP haplotype located between the intron 1 and exon 2 of the DISC1 gene significantly under-transmitted to affected individuals (p=0.0031) with gender differences (51). In this study, the result has suggested a two SNP haplotype between intron 4 and intron 5 of DISC1 gene significantly associated with schizophrenia (p<0.01). The exon1 and 2 encode the putative globular domains, and the exon3-13 encode the putative helical tail (coiled-coil motif) of DISC1 gene (30). The DISC1 181-357 domain (within exon 2) binds strongly with the cytoplasmic microtubules of α-tubulin, and the C-terminal coiled-coil motif is essential for interaction with neurodevelopmental protein Nudel (53). Therefore, we suspect that in different ethnic groups may have different
61
pathogenic mechanisms associated with DISC1 gene. There is no difference in our haplotype genetic transmission between the male and the female. Consistently in both studies agreed that the DISC1 gene is not associated with the onset of schizophrenia.
Since we have rule out the possible existence of the balanced translocation in our schizophrenia DNA samples (submitted manuscript), we would confirm that the participation of DISC1 gene in our schizophrenia patients may be exerted through other mechanisms besides genetic disruption in the pathological process of schizophrenia.
Most documents have been noticed the 1q42.1 regions are important for the pathological cause of schizophrenia (51, 54-57). Here is the first report showing another gene located within this region may have also involved in the pathological process of schizophrenia, the dihydroxyacetone phosphate acyltransferase (DHAPAT or GNPAT). This enzyme has been shown deficiency of varying degrees than normal in the congenital peroxisomal disorders (58). In recessive peroxisomal disorder of cerebrohepatorenal syndrome (ZS), the patients do not survive beyond the first year, and their DHAPAT activity is found only 10% of normal control (59). In mental status of phenotype, a case report has shown severe mental retardation, developmental delay and growth failure in a DHAPAT deficient girl (60). Our result suggests that the GNPAT may not be a schizophrenia specific etiological factor, but may play a role in the broad definition of schizophrenia. Further works in our future study, we will examine the SNP markers located within the exons of GNPAT gene and analyze their association degree to schizophrenia.
The continuous performance test (CPT) has been used as a measure of sustained attention and demonstrated highly sensitive to brain damage or dysfunction (61). These brain dysfunction areas measured by CPT have demonstrated involving in the interaction of cortical (frontal, temporal, parietal), subcortical (limbic, basal ganglia), and functional systems including the pathways between the basal ganglia, thalamus, right hemisphere and frontal lobes (61). These brain area matches the DISC1 gene expression regions found in primates (62). In our previous study, we have found a heritability of performance on the CPT ranged from 0.48 to 0.62 in the schizophrenic patients of Taiwan (63). We have further identified the CPT as a marker of genetic susceptibility to schizophrenia due to reasons of the existence of CPT deficit in probands’ CPT performance, and not amenable to neuroleptic treatment (64). In this
study, we further demonstrate that the DISC1 gene haplotype associated with undegraded CPT and the GNPAT gene haplotype associated with degraded CPT. This suggests that the two genes linking to schizophrenia may have different pathological process of attention deficit. This result along with our another finding in the promoter
region of DISC1 gene which significantly associated with attention deficit (submitted manuscript) suggest that the 1q42.1 region may be important in regulating the attention process in the pathophysiological dysfunction of schizophrenia.The Wisconsin Card Sorting Test (WCST) is a test paradigm to detect working memory as an indicator for the function of human frontal lobe in schizophrenia (65, 66). In this study, the candidate genes of DISC1 and GNPAT have no significant association with the pathological process of working memory as tested by WCST. This result is different
from the Finnish twin study which demonstrated significant association between the 1q
D1S2833 marker and the spatial working memory performance measured by the Visual
Span subtest (67). The D1S2833 (deCODE map position at 233.97 cM) and the
D1S251 (deCODE map position at 235.23 cM) have a chromosomal distance over 1.26
62