Skip Navigation


Human Molecular Genetics Advance Access originally published online on March 21, 2007
Human Molecular Genetics 2007 16(6):704-715; doi:10.1093/hmg/ddm015
This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow Supplementary Data
Right arrow All Versions of this Article:
16/6/704    most recent
ddm015v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (5)
Right arrowRequest Permissions
Google Scholar
Right arrow Articles by Stone, J. L.
Right arrow Articles by Nelson, S. F.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Stone, J. L.
Right arrow Articles by Nelson, S. F.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

© The Author 2007. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oxfordjournals.org

High density SNP association study of a major autism linkage region on chromosome 17

Jennifer L. Stone1,{dagger}, Barry Merriman1, Rita M. Cantor1, Daniel H. Geschwind1,5 and Stanley F. Nelson1,4,5,*

1 Department of Human Genetics, 2 Department of Neurology, 3 Program in Neurobiology, 4 Center for Neurobehavioral Genetics and 5 Department of Psychiatry and Biobehavioral Sciences, University of California, Los Angeles, CA, USA

* To whom correspondence should be addressed at: 695 Charles Young Drive South, Gonda Building, Room 5554, Los Angeles, CA 90095, USA. Tel: +1 3107947981; Fax: +1 3107945446; Email: snelson{at}ucla.edu

Received December 4, 2006; Revised January 31, 2007; Accepted February 2, 2007


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 RESULTS
 MATERIALS AND METHODS
 SUPPLEMENTARY MATERIAL
 REFERENCES
 
A region on chromosome 17 has recently been highlighted as linked to autism (MIM[209850]) in multiple studies and evidence has accumulated suggesting that male-only families (those families that have produced only affected males) provide the major contribution to linkage at this locus. In an attempt to comprehensively test for association of common variants to autism within the region on chromosome 17 defined in Stone et al. (Stone, J.L., Merriman, B., Cantor, R.M., Yonan, A.L., Gilliam, T.C., Geschwind, D.H. and Nelson, S.F. (2004) Evidence for sex-specific risk alleles in autism spectrum disorder. Am. J. Hum. Genet., 75, 1117–1123), a dense panel of single nucleotide polymorphisms (SNPs) was selected across the linkage peak and analyzed in a trio-based study design. SNPs were genotyped in 219 independent trios at an average intermarker distance of 6.1 kb across the 13.7 Mb interval. This provided ~80% coverage of common HapMap variation present in Caucasians, testing exonic, intronic, promoter and intergenic regions, as knowledge of important functional regions within the genome is currently limited. In this comprehensive association study of a linkage region in autism, no single SNP or haplotype association was sufficient to account for the initial linkage signal. Nominally significant single SNP and/or haplotype-based association results were detected in 15 genes, of which, MYO1D, ACCN1 and LASP1 stand out as genes with autism risk alleles requiring further study, with potential GRRs in the range of 1.34–2.29.


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
RESULTS
 MATERIALS AND METHODS
 SUPPLEMENTARY MATERIAL
 REFERENCES
 
Autism (MIM[209850]) is a pervasive neurobehavioral disorder marked by impairment along three dimensions: language development, development of reciprocal social interactions and engagement in stereotyped and ritualistic activities. By definition, onset is prior to age 3, however, signs of clinical impairment are often detected earlier, during the period of language acquisition (1). The autistic phenotype is highly heterogeneous and children with impairment in the three dimensions listed above can have remarkably distinct behaviors along a continuum. Autism is considered the most severe form of a spectrum of disorders referred to as Pervasive Developmental Disorders (PDDs), which includes Asperger's syndrome (MIM[608638]), childhood disintegrative disorder and PDD-not otherwise specified (PDD-NOS). A blanket term often used to describe this continuum of clinical affection status is autism spectrum disorder (ASD).

There is substantial evidence of genetic contributions to autism risk. Concordance rates for monozygotic (MZ) twins of greater than 90%, and a substantial decrease in concordance for dizygotic (DZ) twins to ~10% (2,3), is indicative of a disorder with a strong, yet multifactorial, genetic component (4). Inheritance patterns do not fit a purely dominant or recessive model and likely are the result of many interacting loci (5).

The evidence for a substantial genetic component in the etiology of autism has energized a number of genome-wide linkage and candidate gene studies. Through affected sibling pair linkage analysis, a region has been highlighted on chromosome 17 at which Yonan et al. (6) reported suggestive linkage (MLS 2.83) in 345 multiplex families obtained from the Autism Genetic Resource Exchange (AGRE). Through re-analysis of these data, Stone et al. (7) further demonstrated that families which have produced only male affected children (MO families) were the major contributors to linkage at this locus. Analyzing the MO families separately revealed significantly increased evidence of linkage and produced overall genomewide significance (MLS 4.3). This suggests that stratifying families by sex-of-affecteds reduced genetic heterogeneity at the putative chromosome 17 susceptibility locus. This region of linkage was also shown to be suggestive in the independent IMGSAC sample (8) (MLS 2.34), though an increased linkage signal through stratification by sex was not detected (9). Replication of male-specific linkage to chromosome 17 in a second, independent, AGRE sample (10), confirmed the sex-specific nature of this locus. While the linkage region includes the serotonin transporter gene (SLC6A4), which has long been studied as a candidate gene for a number of psychiatric disorders, including ASD, reports of association to autism have been inconsistent (11,12). Thus, there is substantial evidence this region harbors an as-yet unidentified susceptibility gene or genes of particular relevance to autism in males.

Here we present a comprehensive survey of 13.7 Mb of DNA flanking the centromere on chromosome 17 using a highly dense panel of SNPs to test for association of common variants to ASD. In total, 2053 SNPs were genotyped, directly assaying 180 genes (NCBI MapViewer, build 36) and 44 regions without genic annotation, in a sample consisting of 219 trios selected from the AGRE resource. We have capitalized on the nature of the linkage peak to focus analyses on those trios from families most likely to harbor susceptibility alleles and to prioritize SNPs for subsequent replication studies or sequence analysis. These data constitute a broad exploration of the common variant hypothesis within a region of significant autism linkage and highlight several genes for additional study. Some portions of the physical interval of linkage, however, were not analyzed for common variant association due to insufficient SNP coverage. These holes in coverage of the interval represent areas where further SNP genotyping would be useful.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
MATERIALS AND METHODS
 SUPPLEMENTARY MATERIAL
 REFERENCES
 
Single SNP analysis
SNPs were tested individually for association to ASD in the MO families using the WHAP program as discussed in the methods section. To provide a global view of results, empirical significance of the transmission bias for each SNP is plotted as a function of position in Figure 1 (gray triangles). For ease of visualization, the y-axis represents the –log10(P-value)for each SNP so that those SNPs with more significant evidence for association will stand out from the noise floor. It is clear from Figure 1 that many SNPs display significant (empirical P-value < 0.05) evidence of association to autism, as expected given the large number of tests. To limit the number of false positives identified at a P-value < 0.05 and provide a tenable number of SNPs for primary replication efforts, a second analysis was employed. This hypergeometric analysis (see Materials and Methods section) identified those SNPs presenting a significant increase in transmission bias as the sample analyzed was focused from all trios (ALL) to MO trios. Those SNPs found to be significant by MO TDT analysis and also significant by the hypergeometric analysis are highlighted by black squares in Figure 1 and listed in Table 2. As is clear from Figure 1, including the hypergeometric analysis has served to logically focus replication efforts; otherwise, there would be a large number of SNPs, all at approximately the same significance level, worthy of further study.


Figure 1
View larger version (32K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Figure 1. The negative log of single-SNP allelic P-values are plotted as a function of base pair position (gray triangles). Empirical significance was established by randomly permuting which allele was transmitted to affected children 1000 times at each SNP in the MO trios. Those SNPs that were found to be significantly associated in MO trios at an empirical P-value < 0.05 and also by hypergeometric analysis (see Methods and Materials for further information) at P-value < 0.05 are highlighted by black squares. Highlighted SNPs located within a known gene region are labeled as such. The black ticks below the graph denote the position of genes across the region according to NCBI MapViewer build 36. Positions of haplotypes highlighted in Table 3 are marked by horizontal black lines below the graph. Two blocks located around 28.9 Mb and two blocks located around 32.3 Mb, respectively, are unable to be resolved and look like a single block at this scale. A nominal P-value of 0.05 for TDT results is marked by the dashed gray horizontal line.

 


View this table:
[in this window]
[in a new window]

  		Table 1. Total number of samples with racial distribution for ALL and MO analysis groups

 


View this table:
[in this window]
[in a new window]

  		Table 2. The 15 SNPs displaying significant transmission bias (empirical P-value < 0.05) and evidence of enhancement of transmission bias by hypergeometric distribution analysis ({chi}2 P-value < 0.05)

 


View this table:
[in this window]
[in a new window]

  		Table 3. The 11 blocks displaying significant empirical omnibus evidence (P-value < 0.05) and significant over-transmission of an individual allele

 
The 15 SNPs highlighted by both the TDT and hypergeometric analyses (P-value < 0.05 in both analyses, listed in Table 2) are located within 10 kb of the following genes: DHRS7B, DKFZp434O047, LOC440421, LOC440442, LOC646037, MYO1D, TMEM98, ACCN1, LOC646202 and LASP1. A number of intergenic regions were also highlighted and are marked by squares without annotations in Figure 1. The Maximum Likelihood Genotype Relative Risk (ML GRR) was calculated under both a multiplicative and additive model for each of the 15 highlighted SNPs and listed in Table 2. ML GRRs for the 15 highlighted SNPs ranged from 1.27 (rs17669584) to 3.47 (rs772111) under a multiplicative model, and 1.35 (rs17669584) to 14.5 (rs498583) under an additive model. We estimated the 95% confidence interval (CI) given the ratio of transmitted and non-transmitted alleles observed at each individual SNP (Table 2). From this analysis, there are several SNPs for which the lower bound of the 95% CI remains above 1.0 indicating a strong likelihood of a significant affect of several alleles within or adjacent to DHRS7B(rs12452369), LOC646037(rs7217921), MYO1D(rs2470207), TMEM98(rs4795725, rs8082539) and ACCN1(rs4795750, rs12449864) under multiplicative and additive models.

The nominally significant SNPs (empirical P-value < 0.05) within ACCN1 (both long and short forms analyzed independently) and MYO1D were tallied to assess deviation from the expected proportion of 1/20 or 5% (Supplementary Material, Table S3). Results highlight MYO1D as having significantly more nominally associated SNPs than would be expected given the number of SNPs tested in that gene (total SNPs tested n = 93; tags selected at r2 > 0.8, n = 68, number of nominally associated SNPs = 14; Z = 6.987, P < 0.05). The trend of over-representation of significant SNPs in MYO1D persists when the stringency of independent marker selection in increased (r2 > 0.3) though the number of counts became too low (tags selected at r2 > 0.3 = 14, number of nominally associated SNPs = 3) for meaningful significance estimation. Neither isoform of ACCN1 displayed significant deviation from the proportion of SNPs expected to be significant at a P-value <0.05.

Haplotype block analysis
Groups of two or more SNPs were defined as haplotype blocks by the Four Gamete Test (FGT) (13) as implemented in Haploview. Each block was tested for empirically significant global levels of transmission bias in MO trios using WHAP. Twenty-five of the 348 blocks across the interval displayed nominally significant evidence (empirical omnibus P-value < 0.05) of transmission bias. These 25 blocks were further examined to reveal the haplotypes inferred from the genotypic data and the observed T:NT ratios for each allele as output by Haploview (Supplementary Material, Table S4). Significant transmission biases can arise from over-transmission of an allele to affected children, or from under-transmission. The mode of under-transmission seems much less biologically plausible in the present context, though in principle this would be evidence of a ‘protective’ allele. Since the goal of the current study is the identification of susceptibility alleles, haplotype-based results presented here focus on those haplotypes displaying significant evidence of over-transmission to affected children, as would be expected from a disease causing mutation. In 11 of the 25 blocks that displayed significant omnibus transmission bias, at least one significantly over-transmitted haplotype allele was identified. Table 3 is a focused list of the significantly ({chi}2 allelic P-value < 0.05) over-transmitted haplotype alleles in each of the 11 blocks. Among all 11 over-transmitted haplotypes, likely GRRs range from 1.41 to 1.99 under a multiplicative model. Within these analyses, the lower 95% confidence bound remained above 1.0 for all 11 haplotypes, including those adjacent to or within DHRS7B, MGC33894, PIPOX, PSMD11, MYO1D, ACCN1 and LOC646157. The largest GRR at the lower bound of the 95% CI was 1.32 for a haplotype bounded by rs4794966 and rs17782997 within the longer form of ACCN1.

Discussion
Here we have presented a comprehensive high density SNP association study covering a region of male-specific linkage to ASD on chromosome 17, in order to search for common variants contributing to the risk of developing autism. The trio-based nature of the study allowed use of the TDT as a test of association, which is robust against false-positive associations from cryptic population stratification. The a priori knowledge of linkage to the region allowed the TDT to focus on trios from MO families as, based on the linkage results, these are most enriched for susceptibility alleles within the region. The present study was further designed to interrogate a region of linkage to autism in the most thorough way possible given a limited knowledge of genomic regions of functional importance (14–16) and the limitations of current SNP genotyping technologies. Rather than focusing on a few, repeatedly studied candidate genes, the present study has explored a region of linkage to chromosome 17 in an entirety which has not been attempted before in the search for alleles conferring susceptibility to ASD. This approach of thorough SNP typing over a given interval has been successfully employed in the search for genetic variants conferring susceptibility to late-onset Alzheimer's Disease (17). Although this involves multiple comparisons that cannot be fully corrected for in this analysis, we consider this a first stage in a multi-step process to identify, replicate and confirm associated variants within this region, and report the nominal findings here to permit follow-up in additional data sets.

Significance of association is based on the properties of the null hypothesis which are completely known by Mendelian principles and interpretation is straightforward. Estimation of a likely GRR, on the other hand, depends on assumptions of disease model and ascertainment that cannot be completely known and thus can be overestimated. We have applied a post-hoc GRR calculation of 95% CI of likely affect size in order to determine which of the allelic associations identified in our screen are likely to remain significant on repeat studies. It is notable that for each highlighted SNP and haplotype in the current study the 95% CI covers a smaller range under the multiplicative model and therefore can be considered to ‘fit’ the disease better than the additive model. This is expected of the true susceptibility allele(s) within this region given that the original detection of linkage is based primarily on the excess of affected male siblings sharing two alleles over this interval, which is suggestive of a ‘recessive-like’ mode of inheritance. The GRRs reported here are for the SNPs or haplotypes directly genotyped in the current study and most likely do not include the true susceptibility allele(s). Given that the average correlation to SNPs present in the interval but not genotyped in the current study is r2~0.8 by comparison to HapMap (see Materials and Methods), GRR of the true susceptibility allele(s) is likely ~20% greater than those reported for the proxies used in the current study.

Using relationships established by Risch and Merikangas (18), linkage results over the interval suggest a GRR of ~4, assuming a single susceptibility locus within the region. None of the individual SNP associations or haplotype associations are of sufficiently high GRR to support a model of a single susceptibility allele of such magnitude. Only rs772111 was in this range with a GRR of 3.47, but this estimate is suspect as only 10 informative transmissions were observed. Of the remainder, estimated GRRs for the highlighted SNPs and haplotypes ranged from 1.27 to 2.66 (multiplicative model) suggesting they individually are unlikely to account for the linkage signal detected in this interval. The upper bounds of the GRR of alleles with at least 100 informative transmissions range from 1.97 to 2.33. It is more plausible that either many rare alleles are contributing to disease susceptibility or there are multiple genes of modest affect within the region.

It follows, therefore, that four scenarios are possible from the sum total of evidence presented in the current study: (i) the true common susceptibility allele remains undetected because it resides within the portion of the linkage interval not assayed by the SNPs genotyped, (ii) there is more than one risk factor within the region, each with a more modest GRR, summing together to achieve the GRR suggested by linkage results, (iii) there is allelic heterogeneity such that using the common alleles is underpowered to detect the multiple rare variants occurring in the affected individuals, (iv) the original interval covered densely by SNPs in this study was insufficiently broad to include the affect locus. All of these or portions may be true. It seems most reasonable that there is no single common allele within the interval tested that can account for the linkage signal, and this suggests that multiple common susceptibility alleles or rare alleles are a plausible model. Thus, the search for rare variants in some of the genes would strengthen the evidence of association which may be on the basis of large spontaneous genomic deletions, duplications, or point mutations. On the basis of availability of current technology, the interval is being carefully surveyed at high resolution for genomic deletions and duplications.

The single-SNP and haplotype block analyses highlighted 15 genes and seven intergenic regions. While several SNP or haplotype associations were within or immediately adjacent to genes, the most notable genes based on current knowledge of biological plausibility were MYO1D, ACCN1 (Fig. 2) and LASP1. Myosin family members are key components of the molecular machinery responsible for cell motility and intracellular transport, and select members are involved in transcription (19). Recently, the Drosophila homologue of MYO1D, Myo31DF, was implicated in proper development of bi-lateral symmetry of the gut and genital organs (20,21). Myosin 1D is strongly expressed in brain tissue of mammals (22), and could play a role in establishing normal brain symmetry or asymmetry. This is particularly relevant to autism research given the recent accumulation of imaging studies revealing marked abnormal brain asymmetries in autistic and language delayed individuals (23–25). It is interesting to note that MYO1D is further highlighted by an over-representation of nominally significant SNPs after random selection of tags to account for LD between markers (Supplementary Material, Table S3). Given that multiple single-SNP and haplotype results consistently highlight this gene, yet no one SNP or haplotype has been identified that can independently account for the strength of linkage to the interval, it is plausible that more than one variant within this gene may be contributing to susceptibility. This hypothesis is consistent with the TDT results of this study that display an over-representation of nominally significant SNPs within the MYO1D genic interval.


Figure 2
View larger version (38K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Figure 2. Two regions highlighted through single SNP analysis: MYO1D/TMEM98 (A) and ACCN1 (B). Significance of each priority SNP (green data points) and all other SNPs (gray data points) tested within the region are plotted according to position and significance. Colored arrows below the graph mark gene regions and strand orientation, while intersecting black bars mark exons of the respective genes. SNP locations (gray ticks) and corresponding D' measures are plotted below graph (red, high LD, white, low LD). There is an additional long isoform of ACCN1 that is listed in the UCSC Genome Browser, but not in NCBI build 36; this longer isoform is denoted by the dashed red extension of the genomic arrow of ACCN1 in (B).

 
The multiple associations detected within intronic regions of ACCN1 are interesting as this gene is highly restricted to neuronal expression and was discovered based on similarity to genes known to cause hereditary neurodegeneration of C. elegans (26). In Drosophila, the homologue was localized to the dendritic arbor subtype of the Drosophila peripheral nervous system multiple dendritic sensory neurons. These neurons are thought to be involved in mechanotransduction of stretch and/or touch which is intriguing given the extreme touch sensitivity common in children with autism. It is interesting to note, however, that in contrast to MYO1D, neither isoform of ACCN1 presented with more nominally significant SNPs than expected by chance (P > 0.05) (Supplementary Material, Table S3). This observation is inconclusive, however, and does not preclude the existence of a single susceptibility allele of modest effect or the presence of multiple susceptibility alleles not effectively assayed within this genic interval.

Finally, LASP1 was initially discovered as being over expressed in some breast carcinomas (27). However, recently, LASP1 has been localized to the post-synaptic density (PSD) fraction of rat cortical neurons (28), specifically the cytoplasmic side of the PSD and additional unidentified synaptic compartments (29). Though the function of LASP1 in the central nervous system is unknown, siRNA-based silencing of LASP1 in breast cancer cells resulted in decreased cell proliferation and motility (30) suggesting a possible role for LASP1 in the CNS as being involved in neuronal migration and/or neuronal proliferation. Further study is required to elucidate the role of LASP1 in normal brain development which may prove to be relevant to autism research given observed neuronal patterning defects in the brains of autistic individuals, specifically cortical mini-column patterning (31,32). Thus, we consider Myo1D, ACCN1 and LASP1 as reasonable functional candidates for an autism risk gene highlighted by these analyses.

Discussion of the chromosome 17 locus requires comment on the serotonin transporter (SLC6A4), which is one of the most repeatedly studied functional candidate genes in autism and is located at 25.5 Mb on chromosome 17, within the region of linkage. Numerous groups have studied this gene with inconsistent results (11,12,33–42). In general, results from the current study using the AGRE sample are consistent with the majority of studies using other autistic samples which have failed to detect association to common variants within this gene. The AGRE sample has been used by other groups to study SLC6A4 (12, 34), though the current study finds no evidence for association within or near this gene. Multiple rare variants in SLC6A4 were reported by Sutcliffe et al. (12) which suggested association with autism risk and this gene, but no significant rare allele associations were detected in Yonan et al. (43) from the sequencing of 96 affected individuals. Thus, it is clear that while it is plausible, multiple rare variants contribute to autism risk, rare variants in SLC6A4 do not account for the linkage signal in male only families.

Lastly, the initial finding by Stone et al. (7) provided peak evidence of linkage within MO families centered at ~53 cM. Cantor et al. (10) replicated the evidence of linkage at this location, and provided evidence that linkage extends an additional 20 cM toward the q-telomere. In aggregate from both studies, the evidence from microsatellite based fine mapping broadens the region of interest with an MLS >2 to ~25 cM on chromosome 17. Given the interval size and overall strength of linkage within the AGRE family set, it may well be that two (or more) susceptibility loci reside at different locations within the broad interval. The data presented here attempt to identify association from within the boundaries of the linkage signal reported in Stone et al. (7). It is appropriate to extend the interval of coverage to include the interval reported in Cantor et al. (10) and to explore the current data for allele–allele relationships and to fill in gaps of coverage within the interval covered here.

Although no findings of the current study are sufficiently significant in the context of multiple testing, the present high-density SNP screen has provided a number of nominally significant associations that warrant further study. Replication will be essential in validating the results presented here and to reject false-positives derived from this screen. The present study was also not designed to assess the contribution of rare variants to the etiology of autism. Results presented here do not preclude their existence or importance in autism. It may be important to combine sequence analysis to detect multiple different rare variants with small effects from common variants to uncover the true susceptibility gene or genes in this interval. The public sharing of the data from the AGRE repository freely permits additional investigation which can build on the framework provided here to enhance our knowledge of susceptibility alleles in autism.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 MATERIALS AND METHODS
SUPPLEMENTARY MATERIAL
 REFERENCES
 
Genetic material and preparation
Samples were selected from AGRE, an organization which facilitates the collection of biomaterials from autistic children and their families, and coordinates distribution of samples to approved researchers. Additional details regarding criteria for inclusion into AGRE, standards of diagnosis and protocol of sample collection have been previously published (44) and current details are available at http://www.agre.org. The majority of genomic DNA samples available from AGRE are collected from families with more than one affected child (multiplex families) diagnosed with ASD. Here, ASD includes classical autism, PDD and Asperger syndrome as defined in Liu et al. (45). All aspects of the study presented were approved by the UCLA IRB board.

For the current study, the mother, father and one affected child per family were chosen for genotyping from 333 unrelated AGRE families. The affected child was selected using criteria designed to best enrich for a genetic link to the locus. Specifically, from within each multiplex family, the most severely affected child was selected (i.e. full autism chosen over broad spectrum diagnosis). In families where all affected children received the same diagnosis, the oldest male child was selected. Female children were selected when no male child was available. Diagnoses were based on the ADIR and supplied by AGRE with the pedigree information for each individual. Please see Supplementary Material, Table S5 for a complete list of samples included in analyses and their affection statuses. Families were excluded from the current study if a non-idiopathic autism flag (i.e. fragile-X, abnormal brain imaging results, dysmorphic features, birth trauma) was recorded for any member of the family at the time of family selection.

Genomic DNA samples were obtained from the NIMH cell repository (Rutgers, Piscataway, NJ) and concentrations were established using a Nanodrop (Willmington, DE) instrument. A total of 10 µg of each of 999 samples (333 trios) was sent in 96 well plates (Qiagen, Valencia, CA) on dry ice to Perlegen Sciences, Inc. (Mountain View, CA) for genotyping.

Definition of the region, SNP selection and genotyping procedure
The region selected for high density SNP genotyping represents the 1-LOD drop from the maximal MLS over the interval defined by linkage results published by Stone et al. (7) and chosen to approximately cover the interval between markers D17S1871 (48.07 cM) and D17S1818 (60.4 cM) (33). The region spans 13.7 Mb from physical positions 20.7 Mb to 34.4 Mb (NCBI build 36). This region contains the centromere of chromosome 17, therefore 13.5 Mb of euchromatic DNA was available for assay design.

SNPs were selected to cover the region from those determined to perform at ≥80% call rate and have a minor allele frequency (MAF) >0.1 in samples previously genotyped by Perlegen Sciences. A total of 4267 SNPs within the region fit these criteria. Perlegen haplotype tagging SNPs (46,47) were selected for inclusion and an adaptation of the Lee-Kang algorithm (48) was applied to select additional SNPs for a target average density of 1 SNP/5.1 kb. In total, 2609 SNPs were selected to cover both genic and intergenic regions, as knowledge of functional genomic elements is incomplete (14). Genic intervals were defined by NCBI build 34. However, all references reported here have been converted to build 36 positions. We define ‘genic intervals’ to mean exonic, intronic or intergenic but within 10 kb the 5' and 3'-UTR sequences.

SNP genotyping was performed at Perlegen Sciences, Inc. (Mountain View, CA) using methods similar to those described elsewhere (47). Briefly, long-range PCR (LR-PCR) was used to amplify ~10 kb sections of the region of interest. Arrays were custom designed by Perlegen Sciences to allow allele specific hybridization from all SNPs on a single array. All LR-PCR products were pooled, fragmented, labeled with biotin, and hybridized to high-density oligonucleotide arrays manufactured by Affymetrix. The arrays were washed and stained with streptavidin R-phycoerythrin (SAPE). Signal was amplified through a second staining step using SAPE, and the arrays were then read using a custom built confocal scanner.

Summary of returned data
In total, 219 trios were successfully genotyped at 2053 SNPs by Perlegen QC criteria. Of these, 46 SNPs were flagged (see in what follows for flag criteria) for removal resulting in a final average intermarker distance of 6.7 kb. The SNPs were selected to cover both genic and intergenic regions. Of the cleaned genotype data, 913 SNPs were defined as intergenic SNPs (according to build 36 NCBI MapViewer annotations) and 1094 SNPs were defined as genic. Four SNPs were found not to be polymorphic in our sample, while the MAF was <5% for 52 SNPs and 5–10% for 135 SNPs. The remaining SNPs were approximately equally distributed from 10–50% MAF. See Supplementary Material, Table S1 for a list of markers and MAF. In total, 180 genes were directly assayed by at least 1 SNP within 10 kb of the genic interval and 44 intergenic regions were also covered. See Supplementary Material, Table S2 for gene-based coverage.

HapMap data
Genotype data from CEPH families were downloaded from the HapMap Project, release #20, between 20679826-34377098 base pairs on chromosome 17. Haploview was used to calculate MAF, heterozygosity, and pair-wise LD for SNPs separated by ≤300 kb. SNPs with MAF <0.01 were removed from analysis.

Calculation of allele characteristics
The minor allele frequency (MAF), number of Mendelian errors (ME) and Hardy–Weinberg equilibrium (HWE) P-values were calculated for each SNP using the Haploview program (49). HWE P-values were calculated using only unrelated founder genotypes. ME were identified on a per family basis using Merlin (50) and were removed by zeroing the genotypes of all members of the trio at problematic loci prior to all other analyses. SNPs were flagged if MAF <0.01, >5 ME were detected or HWE P-value <0.001. If a SNP received a flag in any of the sub-divided analytical groups (MO, ALL; see in what follows), it was removed from all subsequent calculations. Haploview was also used to determine T:NT counts for single SNPs and haplotypes across the interval. Phasing of genotype data for haplotype analysis was completed in Haploview, using its implementation of an expectation-maximization (EM) algorithm similar to a previously published partition/ligation method (51). Haplotype allele counts reflect uncertainty in phasing by summing the fractional likelihood of each haplotype for each individual to determine T:NT counts. Minor allele, MAF, number of ME, HWE P-value and T:NT are listed in Supplementary Material, Table S1 along with other SNP-specific information. Owing to space limitations, haplotype details are listed only for those blocks of interest (criteria for interest discussed subsequently; also see Table 3).

Calculation of LD and identification of redundant SNPs
Pair-wise LD measures (D' and r2) were calculated for each pair of SNPs within 300 kb using standard definitions (52) as implemented in the Haploview program (49). There is much debate on the appropriate way to deal with non-independence between markers (53). Here, SNPs were binned using Tagger (54) as implemented in Haploview to identify representative SNPs over the interval. For the purposes of this study, one SNP was randomly selected from pair-wise bins established by both r2 ≥ 0.8 and r2 ≥ 0.3 to remove redundancy at two levels of stringency and obtain a more accurate assessment of the number of independent tests. Significant over-representation of significant SNPs was identified using a test of proportions within the genes with greater than 50 independent SNPs when binned at r2 > 0.8 (ACCN1, ACCN1—long form and MYO1D). The expected proportion was fixed at 1/20 or 5%.

Estimation of regional coverage and comparison to HapMap
Effective ascertainment by proxy is essential to LD-based association studies. Over 75% of SNPs genotyped in the current study were found to be correlated at r2 > 0.3, while ~50% were correlated at r2 > 0.8. Though the HapMap provides incomplete information, particularly with respect to rare alleles, it is the most complete catalogue of human variation available over the genome as a whole and provides a clear standard by which to gauge coverage of known variation over an interval of interest in a given study. Using those SNPs that are common to both our generated data and those genotyped by the HapMap Project (n = 1630) as tags, 96% of all SNPs in the HapMap (total number of HapMap SNPs in region = 12809) were covered by proxy at average r2 = 0.8. Thus, we conclude that the interval is well covered as this represents minimum coverage of the region, given that there are an additional 377 SNPs included in the current study, but not present in the HapMap and therefore not included in coverage assessment. In summary, using the common variation present in CEPH individuals as provided by HapMap (release #20) as a reference, we estimate that more than 80% of the interval is assayed by the current study.

This genomic interval contains the typically diverse LD patterns found throughout the human genome and is consistent with a largely Caucasian sample (46,55). Of the eight physical gaps in the current study greater than 100 kb in size, four are also present in the HapMap (release #20). In these common gap regions no SNP was available or had been successfully genotyped by the HapMap consortium. The additional four large physical gaps in the current study are located in regions of average to above average LD (data not shown), but represent regions where additional SNP genotyping may be useful. Data from the current study reveals haplotype blocks ranging in size from 74 to 192 kb as defined using the four-gamete test (FGT; described in what follows) implemented in Haploview. The largest ‘block’ defined by the FGT encompasses three genes, including SSH2, FLJ46247 and DKFZP434K1421, located from 25.29 to 25.48 Mb (NCBI build 36) and is formed by 11 SNPs.

Selection of families for TDT association analysis
The linkage to chromosome 17 published by Stone et al. (7) was a result of excess sharing of two alleles IBD when MO families were analyzed separately from those families with an affected female. Therefore, primary analyses focus on MO families (n = 133) to enrich for those families most likely to harbor a susceptibility allele at this locus. All MO families were included in primary analysis regardless of any available information on IBD status with affected siblings. Allelic T:NT counts were also obtained from the complete 219 trio data set (ALL trios) for use in prioritizing SNPs for replication using a hypergeometric distribution function.

Single SNP TDT analysis
The empirical significance of observed transmission results was calculated using the WHAP program (56). WHAP establishes an empirical P-value based on random permutations of the allele transmitted from informative meioses under the null hypothesis of no transmission bias. There are many other programs available, including Haploview, which provide means to calculate the TDT. However, most programs evaluate an observed transmission count versus an expected transmission count using a {chi}2 distribution. Empirical significance does not rely on the assumption of a continuous {chi}2 distribution and is based on the true properties of the SNP in the parental population, therefore WHAP was chosen for primary analyses. Permutations consisted of fixing the genotypes of the parents and randomly permuting the allele passed to affected children 1000 times at each SNP to establish to significance of the observed results. For single-SNP analysis, WHAP was run under conditional mode with disease prevalence set to 0.006 (57) using a sliding window of one SNP per window. Since the data set consisted of trios, the ‘within’ family model was specified in the command line.

Haplotype TDT analysis
Disease susceptibility alleles arise on a chromosomal background of other previous mutations. Therefore, it has been suggested that if the true disease susceptibility allele is not itself directly assayed, combining markers into haplotype blocks to identify the genomic background on which the mutation occurred can be more powerful to detect true association than any one marker alone. Consecutive markers were combined into haplotype blocks by the FGT (13) as implemented in Haploview (49) at default settings. The FGT haplotype boundary defining algorithm defines blocks based on evidence of historical recombination (the presence of a fourth gamete under an infinite-sites model of mutation) and is closely related to the concept of r2 and effective association detection by proxy. Data from contiguous SNPs designated as haplotype blocks using this algorithm were selected from the complete data set and run as a single block using the WHAP software (56). Command line parameters were the same as listed for single-SNP analyses; however, no sliding window analysis was employed since the goal of the haplotype analysis was to analyze all SNPs in a block together rather than one by one. The output from WHAP provides many statistical metrics; however, the one used in the current study is the empirical omnibus test of evidence for haplotypic association of one or more alleles within the block. This metric was selected because it estimates the significance of the observed deviations from the expected 50/50 transmission for all alleles (n = H) of a given haplotype block with H-1 degrees of freedom. As with the single-SNP analysis, an empirical P-value was estimated by randomly permuting the haplotype allele transmitted at each informative meioses for the entire sample, 1000 times.

Hypergeometric distribution analysis
Division of the AGRE sample based on sex of affected children produced a strong enhancement of linkage in the MO families (7). If this reflects the underlying genetics of the disorder it is reasonable to expect that associations would also exhibit enhancement under the same splitting strategy. The actual strength of such an effect will depend on the number, strength and frequency of risk alleles within the region, and the actual degree of enrichment in association could in principle be greater than or less than the enrichment observed in the linkage analysis. Further, observation of the affect will be dependent on the number and placement of genotyped SNPs/haplotypes and their ability to tag true susceptibility alleles. Because of these confounding factors and unknown variables, the true power of this approach is difficult to estimate. Nonetheless, here we use the hypothesis that association should be enriched by sex-splitting as a means to further filter the most promising SNPs for primary replication efforts. It would be incorrect to infer that transmission bias identified within ALL trios is evidence of a false-positive association. Rather, our hypothesis is that limiting analysis to just the MO trios should enhance evidence of association in a similar way that limiting to MO families increased evidence of linkage. It is possible that the susceptibility allele is present and displays significant association to ASD within ALL families, though we hypothesize that this association, or transmission bias, should be exaggerated in the MO trios beyond what would be expected in a random sampling of ‘T’ and ‘NT’ as defined by ALL. In other words, the null hypothesis is that for a given SNP, restricting the TDT ‘T’ and ‘NT’ counts to just the MO family trios will simply be random sampling from the ALL trio counts, whereas the (anticipated) alternative hypothesis is that there will be excess of transmissions over what would be expected from random sampling. The null distribution is modeled by the hypergeometric distribution, and we use this to assess the significance of deviations from the null. Specifically, at each SNP, given the counts of transmission, T, and non-transmission, NT, observed in ALL trios, and further given the corresponding counts, t and nt, observed within just the MO trios, the null hypothesis would be that there were t successes in t + nt trials, as drawn randomly from a population of T + NT with T successes (58). We use the hypergeometric distribution to compute the P-value for obtaining t or more successes for each individual SNP. Hypergeometric analysis of haplotype blocks is not presented, as significant blocks were identified by global evidence (based on transmission bias of all haplotype alleles of a given block) of transmission bias. Since more than one haplotypic allele can contribute to the observed omnibus transmission bias, the hypergeometric analysis as presented here is confounded. Hypergeometric P-value calculations were implemented in a custom Mathematica script.

Estimated power of TDT
Stratification of families based on the sex of their affected children revealed an excess of ASPs sharing two alleles IBD in the MO families (~40% of MO ASPs versus 25% as expected under the null hypothesis) (7). Following guidelines suggested by Risch and Merikangas (18) and sex-specific linkage sharing statistics of ~20/40/40% (Z0/Z1/Z2) as published in Stone et al. (7), the genotype relative risk (GRR) was calculated to have a value between 3 and 4, assuming a single affect locus on chromosome 17. Power to detect association using the TDT was calculated using the Genetic Power Calculator (59) under both additive and multiplicative models, and a range of marker (0.05, 0.1, 0.2, 0.3, 0.4) and disease (0.05, 0.1, 0.2) allele frequencies (see Supplemental Material, Fig. S1). For simplification, markers were assumed to be in linkage disequilibrium with the true disease locus at a level of D' = 0.8, disease prevalence was fixed at 0.006 (57) and {alpha} fixed at 0.05. Under all parameters used, maximum power to detect association is achieved when the disease allele approaches 20%. In the event that the true disease allele frequency is ~0.05 or lower, power to detect association at ≥80% is still achievable only for GRR of 4 or greater, given marker alleles of similar frequency and sample size of the current study.

Post-hoc estimation of GRR and 95% CI
To assess the increase in disease risk conferred by inheritance of 0, 1 or 2 copies of the putative risk allele (the over-transmitted allele), the GRR and 95% CI were calculated for each highlighted SNP and over-transmitted haplotype based on observed T:NT results. The method used to calculate the GRR and 95% CI is similar to methods described previously (60,61) and described in detail in Supplementary Material, GRR calculation.

Multiple testing correction and nominal levels of significance used for current study
There is much uncertainty about the most appropriate method of correction for multiple testing in SNP-based association studies. Therefore, given the high level of non-independence between markers in the current study and to avoid false negatives given the screening nature of this study, no correction for multiple testing was made on P-values reported. It is intended that by providing full disclosure of the number of tests, the reader will interpret and apply correction appropriate for their purposes. In the current study, single SNPs displaying empirical P-values < 0.05 in the MO TDT analysis that also were significant at P-value < 0.05 through the hypergeometric analysis were given highest priority for primary replication efforts. From the haplotype analysis, priority haplotypes were required to display an empirical omnibus P-value < 0.05 by WHAP analysis and, on closer inspection of T:NT counts, display significant over-transmission of at least one allele ({chi}2 P-value < 0.05).


   SUPPLEMENTARY MATERIAL
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 MATERIALS AND METHODS
 SUPPLEMENTARY MATERIAL
REFERENCES
 
Supplementary Material is available at HMG Online.


   ACKNOWLEDGEMENTS
 
We thank the AGRE families who participated in this study, the Cure Autism Now Foundation (CAN) for supporting the AGRE program, and the scientists who have provided oversight to the AGRE consortium (listed in what follows). We also thank Drs David Cox and Kelly Frazer of Perlegen Sciences for advice and technical assistance in planning the study. This work was supported by National Institutes of Health grants MH64547 (D.H.G, S.F.N., R.M.C.) and by a Graduate Assistance in Areas of National Need (GAANN) pre-doctoral training fellowship (J.L.S.).

Members of AGRE Consortium are Daniel H. Geschwind, University of California at Los Angeles, Los Angeles; Maja Bucan, University of Pennsylvania, Philadelphia; W. Ted Brown, New York State Institute for Basic Research in Developmental Disabilities, Staten Island; Rita Cantor, University of California at Los Angeles, Los Angeles; John Constantino, Washington University, Saint Lewis; T. Conrad Gilliam, University of Chicago, Chicago; Martha Herbert, Harvard University, Boston; David H. Ledbetter, Emory University, Atlanta; Stanley F. Nelson, University of California at Los Angeles, Los Angeles; Carol A. Samango-Sprouse, George Washington University, Washington, D.C.; Gerard D. Schellenberg, University of Washington and Veterans Affairs Medical Center, Seattle; Matthew State, Yale University, New Haven; and Rudolph E. Tanzi, Massachusetts General Hospital, Boston.

Conflict of Interest statement. None declared.


   FOOTNOTES
 
{dagger} Present address: Psychiatric and Neurodevelopmental Genetics Unit, Center for Human Genetic Research, Massachusetts General Hospital, Harvard Medical School, Boston, MA 02114, USA Back


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 MATERIALS AND METHODS
 SUPPLEMENTARY MATERIAL
 REFERENCES
 

  1. Volkmar F., Chawarska K., Klin A. (2005) Autism in infancy and early childhood. Annu. Rev. Psychol. 56:315–336.[CrossRef][Web of Science][Medline]

  2. Bailey A., Le Couteur A., Gottesman I., Bolton P., Simonoff E., Yuzda E., Rutter M. (1995) Autism as a strongly genetic disorder: evidence from a British twin study. Psychol. Med. 25:63–77.[Web of Science][Medline]

  3. Ritvo E.R., Freeman B.J., Mason-Brothers A., Mo A., Ritvo A.M. (1985) Concordance for the syndrome of autism in 40 pairs of afflicted twins. Am. J. Psychiatry 142:74–77.[Abstract/Free Full Text]

  4. Risch N. (1990) Linkage strategies for genetically complex traits. I. Multilocus models. Am. J. Hum. Genet. 46:222–228.[Web of Science][Medline]

  5. Risch N., Spiker D., Lotspeich L., Nouri N., Hinds D., Hallmayer J., Kalaydjieva L., McCague P., Dimiceli S., Pitts T., et al. (1999) A genomic screen of autism: evidence for a multilocus etiology. Am. J. Hum. Genet. 65:493–507.[CrossRef][Web of Science][Medline]

  6. Yonan A.L., Alarcon M., Cheng R., Magnusson P.K., Spence S.J., Palmer A.A., Grunn A., Juo S.H., Terwilliger J.D., Liu J., et al. (2003) A genomewide screen of 345 families for autism-susceptibility loci. Am. J. Hum. Genet. 73:886–897.[CrossRef][Web of Science][Medline]

  7. Stone J.L., Merriman B., Cantor R.M., Yonan A.L., Gilliam T.C., Geschwind D.H., Nelson S.F. (2004) Evidence for sex-specific risk alleles in autism spectrum disorder. Am. J. Hum. Genet. 75:1117–1123.[CrossRef][Web of Science][Medline]

  8. International Molecular Genetic Study of Autism Consortium (IMGSAC). (2001) A genomewide screen for autism: strong evidence for linkage to chromosomes 2q, 7q, and 16p. Am. J. Hum. Genet. 69:570–581.[CrossRef][Medline]

  9. Lamb J.A., Barnby G., Bonora E., Sykes N., Bacchelli E., Blasi F., Maestrini E., Broxholme J., Tzenova J., Weeks D., et al. (2005) Analysis of IMGSAC autism susceptibility loci: evidence for sex limited and parent of origin specific effects. J. Med. Genet. 42:132–137.[Abstract/Free Full Text]

  10. Cantor R.M., Kono N., Duvall J.A., Alvarez-Retuerto A., Stone J.L., Alarcon M., Nelson S.F., Geschwind D.H. (2005) Replication of autism linkage: fine-mapping peak at 17q21. Am. J. Hum. Genet. 76:1050–1056.[CrossRef][Web of Science][Medline]

  11. Devlin B., Cook E.H., Coon H., Dawson G., Grigorenko E.L., McMahon W., Minshew N., Pauls D., Smith M., Spence M.A., et al. (2005) Autism and the serotonin transporter: the long and short of it. Mol. Psychiatry 10:1110–1116.[CrossRef][Web of Science][Medline]

  12. Sutcliffe J.S., Delahanty R.J., Prasad H.C., McCauley J.L., Han Q., Jiang L., Li C., Folstein S.E., Blakely R.D. (2005) Allelic heterogeneity at the serotonin transporter locus (SLC6A4) confers susceptibility to autism and rigid-compulsive behaviors. Am. J. Hum. Genet. 77:265–279.[CrossRef][Web of Science][Medline]

  13. Wang N., Akey J.M., Zhang K., Chakraborty R., Jin L. (2002) Distribution of recombination crossovers and the origin of haplotype blocks: the interplay of population history, recombination, and mutation. Am. J. Hum. Genet. 71:1227–1234.[CrossRef][Web of Science][Medline]

  14. Cheng J., Kapranov P., Drenkow J., Dike S., Brubaker S., Patel S., Long J., Stern D., Tammana H., Helt G., et al. (2005) Transcriptional maps of 10 human chromosomes at 5-nucleotide resolution. Science 308:1149–1154.[Abstract/Free Full Text]

  15. Crowe M.L., Wang X.Q., Rothnagel J.A. (2006) Evidence for conservation and selection of upstream open reading frames suggests probable encoding of bioactive peptides. BMC. Genomics 7:16.[CrossRef][Medline]

  16. Keightley P.D., Kryukov G.V., Sunyaev S., Halligan D.L., Gaffney D.J. (2005) Evolutionary constraints in conserved nongenic sequences of mammals. Genome Res. 15:1373–1378.[Abstract/Free Full Text]

  17. Grupe A., Li Y., Rowland C., Nowotny P., Hinrichs A.L., Smemo S., Kauwe J.S., Maxwell T.J., Cherny S., Doil L., et al. (2006) A scan of chromosome 10 identifies a novel locus showing strong association with late-onset Alzheimer disease. Am. J. Hum. Genet. 78:78–88.[CrossRef][Web of Science][Medline]

  18. Risch N. and Merikangas K. (1996) The future of genetic studies of complex human diseases. Science 273:1516–1517.[Abstract/Free Full Text]

  19. Grummt I. (2006) Actin and myosin as transcription factors. Curr. Opin. Genet. Dev. 16:191–196.[CrossRef][Web of Science][Medline]

  20. Speder P., Adam G., Noselli S. (2006) Type ID unconventional myosin controls left–right asymmetry in Drosophila. Nature 440:803–807.[CrossRef][Medline]

  21. Hozumi S., Maeda R., Taniguchi K., Kanai M., Shirakabe S., Sasamura T., Speder P., Noselli S., Aigaki T., Murakami R., et al. (2006) An unconventional myosin in Drosophila reverses the default handedness in visceral organs. Nature 440:798–802.[CrossRef][Medline]

  22. Bahler M., Kroschewski R., Stoffler H.E., Behrmann T. (1994) Rat myr 4 defines a novel subclass of myosin I: identification, distribution, localization, and mapping of calmodulin-binding sites with differential calcium sensitivity. J. Cell Biol. 126:375–389.[Abstract/Free Full Text]

  23. Herbert M.R., Harris G.J., Adrien K.T., Ziegler D.A., Makris N., Kennedy D.N., Lange N.T., Chabris C.F., Bakardjiev A., Hodgson J., et al. (2002) Abnormal asymmetry in language association cortex in autism. Ann. Neurol. 52:588–596.[CrossRef][Web of Science][Medline]

  24. Herbert M.R., Ziegler D.A., Deutsch C.K., O'Brien L.M., Kennedy D.N., Filipek P.A., Bakardjiev A.I., Hodgson J., Takeoka M., Makris N., et al. (2005) Brain asymmetries in autism and developmental language disorder: a nested whole-brain analysis. Brain 128:213–226.[Abstract/Free Full Text]

  25. De Fosse L., Hodge S.M., Makris N., Kennedy D.N., Caviness V.S., Jr V.S., McGrath L., Steele S., Ziegler D.A., Herbert M.R., Frazier J.A., et al. (2004) Language-association cortex asymmetry in autism and specific language impairment. Ann. Neurol. 56:757–766.[CrossRef][Web of Science][Medline]

  26. Waldmann R., Champigny G., Voilley N., Lauritzen I., Lazdunski M. (1996) The mammalian degenerin MDEG, an amiloride-sensitive cation channel activated by mutations causing neurodegeneration in Caenorhabditis elegans. J. Biol. Chem. 271:10433–10436.[Abstract/Free Full Text]

  27. Tomasetto C., Moog-Lutz C., Regnier C.H., Schreiber V., Basset P., Rio M.C. (1995) Lasp-1 (MLN 50) defines a new LIM protein subfamily characterized by the association of LIM and SH3 domains. FEBS Lett. 373:245–249.[CrossRef][Web of Science][Medline]

  28. Phillips G.R., Anderson T.R., Florens L., Gudas C., Magda G., Yates J.R. III, Colman D.R. (2004) Actin-binding proteins in a postsynaptic preparation: Lasp-1 is a component of central nervous system synapses and dendritic spines. J. Neurosci. Res. 78:38–48.[CrossRef][Web of Science][Medline]

  29. Li K., Hornshaw M.P., van Minnen J., Smalla K.H., Gundelfinger E.D., Smit A.B. (2005) Organelle proteomics of rat synaptic proteins: correlation-profiling by isotope-coded affinity tagging in conjunction with liquid chromatography-tandem mass spectrometry to reveal post-synaptic density specific proteins. J. Proteome Res. 4:725–733.[CrossRef][Web of Science][Medline]

  30. Grunewald T.G., Kammerer U., Schulze E., Schindler D., Honig A., Zimmer M., Butt E. (2006) Silencing of LASP-1 influences zyxin localization, inhibits proliferation and reduces migration in breast cancer cells. Exp. Cell Res. 312:974–982.[CrossRef][Web of Science][Medline]

  31. Casanova M.F., Buxhoeveden D., Gomez J. (2003) Disruption in the inhibitory architecture of the cell minicolumn: implications for autism. Neuroscientist 9:496–507.[Abstract/Free Full Text]

  32. Casanova M.F., van Kooten I.A., Switala A.E., van Engeland H., Heinsen H., Steinbusch H.W., Hof P.R., Trippe J., Stone J., Schmitz C. (2006) Minicolumnar abnormalities in autism. Acta. Neuropathol. (Berl.) 112:287–303.[CrossRef][Medline]

  33. Broman K.W., Murray J.C., Sheffield V.C., White R.L., Weber J.L. (1998) Comprehensive human genetic maps: individual and sex-specific variation in recombination. Am. J. Hum. Genet. 63:861–869.[CrossRef][Web of Science][Medline]

  34. McCauley J.L., Olson L.M., Dowd M., Amin T., Steele A., Blakely R.D., Folstein S.E., Haines J.L., Sutcliffe J.S. (2004) Linkage and association analysis at the serotonin transporter (SLC6A4) locus in a rigid-compulsive subset of autism. Am. J. Med. Genet. B Neuropsychiatr. Genet. 127:104–112.[Medline]

  35. Ramoz N., Reichert J.G., Corwin T.E., Smith C.J., Silverman J.M., Hollander E., Buxbaum J.D. (2006) Lack of evidence for association of the serotonin transporter gene SLC6A4 with autism. Biol. Psychiatry 60:186–191.[Web of Science][Medline]

  36. Persico A.M., Pascucci T., Puglisi-Allegra S., Militerni R., Bravaccio C., Schneider C., Melmed R., Trillo S., Montecchi F., Palermo M., et al. (2002) Serotonin transporter gene promoter variants do not explain the hyperserotoninemia in autistic children. Mol. Psychiatry 7:795–800.[CrossRef][Web of Science][Medline]

  37. Anderson G.M., Gutknecht L., Cohen D.J., Brailly-Tabard S., Cohen J.H., Ferrari P., Roubertoux P.L., Tordjman S. (2002) Serotonin transporter promoter variants in autism: functional effects and relationship to platelet hyperserotonemia. Mol. Psychiatry 7:831–836.[CrossRef][Web of Science][Medline]

  38. Conroy J., Meally E., Kearney G., Fitzgerald M., Gill M., Gallagher L. (2004) Serotonin transporter gene and autism: a haplotype analysis in an Irish autistic population. Mol. Psychiatry 9:587–593.[CrossRef][Web of Science][Medline]

  39. Coutinho A.M., Oliveira G., Morgadinho T., Fesel C., Macedo T.R., Bento C., Marques C., Ataide A., Miguel T., Borges L., et al. (2004) Variants of the serotonin transporter gene (SLC6A4) significantly contribute to hyperserotonemia in autism. Mol. Psychiatry 9:264–271.[CrossRef][Web of Science][Medline]

  40. Kim S.J., Cox N., Courchesne R., Lord C., Corsello C., Akshoomoff N., Guter S., Leventhal B.L., Courchesne E., Cook E.H., Jr E.H. (2002) Transmission disequilibrium mapping at the serotonin transporter gene (SLC6A4) region in autistic disorder. Mol. Psychiatry 7:278–288.[CrossRef][Web of Science][Medline]

  41. Mulder E.J., Anderson G.M., Kema I.P., Brugman A.M., Ketelaars C.E., de Bildt A., van Lang N.D., den Boer J.A., Minderaa R.B. (2005) Serotonin transporter intron 2 polymorphism associated with rigid-compulsive behaviors in Dutch individuals with pervasive developmental disorder. Am. J. Med. Genet. B Neuropsychiatr. Genet. 133:93–96.[Medline]

  42. Wu S., Guo Y., Jia M., Ruan Y., Shuang M., Liu J., Gong X., Zhang Y., Yang J., Yang X., et al. (2005) Lack of evidence for association between the serotonin transporter gene (SLC6A4) polymorphisms and autism in the Chinese trios. Neurosci. Lett. 381:1–5.[CrossRef][Web of Science][Medline]

  43. Yonan A.L., Palmer A.A., Gilliam T.C. (2006) Hardy–Weinberg disequilibrium identified genotyping error of the serotonin transporter (SLC6A4) promoter polymorphism. Psychiatr. Genet. 16:31–34.[CrossRef][Web of Science][Medline]

  44. Geschwind D.H., Sowinski J., Lord C., Iversen P., Shestack J., Jones P., Ducat L., Spence S.J. (2001) The autism genetic resource exchange: a resource for the study of autism and related neuropsychiatric conditions. Am. J. Hum. Genet. 69:463–466.[CrossRef][Web of Science][Medline]

  45. Liu J., Nyholt D.R., Magnussen P., Parano E., Pavone P., Geschwind D., Lord C., Iversen P., Hoh J., Ott J., et al. (2001) A genomewide screen for autism susceptibility loci. Am. J. Hum. Genet. 69:327–340.[CrossRef][Web of Science][Medline]

  46. Hinds D.A., Stuve L.L., Nilsen G.B., Halperin E., Eskin E., Ballinger D.G., Frazer K.A., Cox D.R. (2005) Whole-genome patterns of common DNA variation in three human populations. Science 307:1072–1079.[Abstract/Free Full Text]

  47. Patil N., Berno A.J., Hinds D.A., Barrett W.A., Doshi J.M., Hacker C.R., Kautzer C.R., Lee D.H., Marjoribanks C., McDonough D.P., et al. (2001) Blocks of limited haplotype diversity revealed by high-resolution scanning of human chromosome 21. Science 294:1719–1723.[Abstract/Free Full Text]

  48. Lee S. and Kang C. (2004) CHOISS for selection of single nucleotide polymorphism markers on interval regularity. Bioinformatics 20:581–582.[Abstract/Free Full Text]

  49. Barrett J.C., Fry B., Maller J., Daly M.J. (2005) Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics 21:263–265.[Abstract/Free Full Text]

  50. Abecasis G.R., Cherny S.S., Cookson W.O., Cardon L.R. (2002) Merlin—rapid analysis of dense genetic maps using sparse gene flow trees. Nat. Genet. 30:97–101.[CrossRef][Web of Science][Medline]

  51. Qin Z.S., Niu T., Liu J.S. (2002) Partition-ligation-expectation-maximization algorithm for haplotype inference with single-nucleotide polymorphisms. Am. J. Hum. Genet. 71:1242–1247.[CrossRef][Web of Science][Medline]

  52. Lewontin R.C. (1964) The interaction of selection and linkage. Ii. Optimum models. Genetics 50:757–782.[Free Full Text]

  53. van den Oord E.J. and Neale B.M. (2004) Will haplotype maps be useful for finding genes? Mol. Psychiatry 9:227–236.[CrossRef][Web of Science][Medline]

  54. de Bakker P.I., Yelensky R., Pe'er I., Gabriel S.B., Daly M.J., Altshuler D. (2005) Efficiency and power in genetic association studies. Nat. Genet. 37:1217–1223.[CrossRef][Web of Science][Medline]

  55. Altshuler D., Brooks L.D., Chakravarti A., Collins F.S., Daly M.J., Donnelly P. (2005) A haplotype map of the human genome. Nature 437:1299–1320.[CrossRef][Medline]

  56. Purcell S., Daly M.J., Sham P.C. (2007) WHAP: haplotype-based association analysis. Bioinformatics 23:255–256.[Abstract/Free Full Text]

  57. Fombonne E. (2005) Epidemiology of autistic disorder and other pervasive developmental disorders. J. Clin. Psychiatry 66:Suppl. 10, 3–8.

  58. Belle G.V. (2004) Biostatistics: a Methodology for the Health Sciences(John Wiley & Sons, Hoboken, NJ).

  59. Purcell S., Cherny S.S., Sham P.C. (2003) Genetic Power Calculator: design of linkage and association genetic mapping studies of complex traits. Bioinformatics 19:149–150.[Abstract/Free Full Text]

  60. Buyske S., Williams T.A., Mars A.E., Stenroos E.S., Ming S.X., Wang R., Sreenath M., Factura M.F., Reddy C., Lambert G.H., et al. (2006) Analysis of case–parent trios at a locus with a deletion allele: association of GSTM1 with autism. BMC Genet. 7:8.[CrossRef][Medline]

  61. Scherag A., Dempfle A., Hinney A., Hebebrand J., Schafer H. (2002) Confidence intervals for genotype relative risks and allele frequencies from the case parent trio design for candidate-gene studies. Hum. Hered. 54:210–217.[CrossRef][Web of Science][Medline]


Add to CiteULike CiteULike   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us    What's this?


This article has been cited by other articles:


Home page
 J. Neurosci.Home page
X.-m. Zha, V. Costa, A. M. S. Harding, L. Reznikov, C. J. Benson, and M. J. Welsh
ASIC2 Subunits Target Acid-Sensing Ion Channels to the Synapse via an Association with PSD-95
J. Neurosci., July 1, 2009; 29(26): 8438 - 8446.
[Abstract] [Full Text] [PDF]

Home page
 BloodHome page
J. A. Knight, A. D. Skol, A. Shinde, D. Hastings, R. A. Walgren, J. Shao, T. R. Tennant, M. Banerjee, J. M. Allan, M. M. Le Beau, et al.
Genome-wide association study to identify novel loci associated with therapy-related myeloid leukemia susceptibility
Blood, May 28, 2009; 113(22): 5575 - 5582.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow Supplementary Data
Right arrow All Versions of this Article:
16/6/704    most recent
ddm015v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (5)
Right arrowRequest Permissions
Google Scholar
Right arrow Articles by Stone, J. L.
Right arrow Articles by Nelson, S. F.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Stone, J. L.
Right arrow Articles by Nelson, S. F.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?