The Genom of Homo sapiens.pdf

GENOMIC VARIATION IN MULTIGENIC TRAITS: HSCR 375MATERIALS AND METHODSGenome scan. We genotyped 43 trios from the Old OrderMennonite community in Lancaster and Berks Counties,Pennsylvania. Ascertainment was conducted underprotocols approved by the Institutional Review Board ofJohns Hopkins University School of Medicine. A completedescription of the study population has been publishedelsewhere (Carrasquillo et al. 2002). Followingthe manufacturer’s protocol, we genotyped 4,363 SNPsusing the WGSA-EcoRI-p502 array (Affymetrix).Briefly, following EcoRI digest of genomic DNA and ligationwith adapters, samples were PCR-amplified as ameans of size-selecting fragments up to 1 kb. Details ofsample labeling and array hybridization have been publishedpreviously (Kennedy et al. 2003).Also included in the analysis are data from an additional569 microsatellites and 1,384 SNPs genotyped earfied.Furthermore, joint transmission tests of alleles fromEDNRB and RET strongly suggested that mutations atthese loci do not act independently. These data promptedus to assay for noncomplementation between these genesusing extant mutations in the mouse. We have successfullyestablished mouse strains, simultaneously mutant atboth Ret and Ednrb, which completely recapitulate theclinical and genetic features of HSCR and now provide aunique system with which to test hypotheses about themolecular interactions that result in disease (McCallion etal. 2003). However, in the absence of identifying thenovel 16q locus and examining the potential involvementof others, we could not explain disease transmission in allaffected individuals.As predicted, the disease-causing variants identified todate in the Mennonites are common (10–20%) in the population,suggesting that if other common variants exist atadditional loci, they may also be detected in associationscans of sufficient resolution. Here we report the completionof such a genome-wide screen that implicates three additionalloci in disease susceptibility (1p34, 4q31, and11p15) in this isolated population. However, despite detectionof additional loci, our genome scan lacks the resolutionnecessary to reveal the specific genes underlying disease.Often the ability to refine a genetic locus and narrowthe list of possible candidate genes is dependent on theavailability of biological and functional annotation for thecorresponding genes in the interval. Unfortunately, at thisstage, this information is incomplete. To reduce the list ofgene candidates in HSCR susceptibility regions, we hypothesizedthat the pertinent loci may comprise downstreamcomponents of RET and EDNRB signaling andthat disruption of RET and/or EDNRB signaling maythus influence the transcript levels of these genes. Onepowerful approach to assay such changes is through array-basedtranscript profiling technologies, permittingthe simultaneous analysis of thousands of transcripts. Weposited that RET- and EDNRB-responsive genes lyingwithin known HSCR susceptibility genomic intervalswould comprise ideal disease gene candidates. To testthis hypothesis, we examined the transcript profile in theintestinal tracts of our two-locus HSCR mouse strains,mice harboring Ret and Ednrb mutations independently,and in wild-type mice. Our data demonstrate that a groupof 9 genes, whose human orthologs localize to a knownHSCR susceptibility locus on 21q22 (Puffenberger et al.1994b), are differentially expressed in the colons of wildtypeand HSCR mutant mice. This report affirms thevalue of integrating genomic approaches in dissecting themechanisms underlying complex inherited disease. Theultimate goal, of understanding the molecular nature ofthe variants contributing to clinical expression, will alsobe facilitated by the integration of genomic informationfrom multiple organisms.The absence of a coding sequence change in RET in theMennonite population is not completely surprising. Despitethe identification of over 100 RET mutations inHSCR and related syndromes, the majority of RET-linkedfamilies lack a frank coding sequence mutation. We recentlycompleted a genome screen in L-HSCR families,demonstrating linkage at RET in 11/12 (92%) families butidentifying RET mutations in only 6/11 linked families(Bolk et al. 2000). Similarly, the majority (88%) of S-HSCR sib pairs, in a second study, demonstrated allelesharing at RET, but mutations were found in only 40% ofthese families (Gabriel et al. 2002). The lack of identifiablemutations in RET in the outbred population suggeststhat noncoding mutations in RET probably underlie mostforms of HSCR. This is consistent with the observation ofa role for a relatively common RET haplotype with nocoding sequence mutation in the Mennonites. We haveused these observations to posit that the same (or similar)common haplotype may be present at elevated frequencyin HSCR patients in the general population. To test thishypothesis, we used the transmission disequilibrium test(TDT) in samples drawn from a general, outbred populationof HSCR patients. We demonstrate here, using an 8-marker study, that the HSCR-susceptibility haplotype observedin the Mennonites is also found in association withdisease in the general population. Therefore, a noncodingmutation at RET likely underlies the genesis of most casesof HSCR.Perhaps the greatest challenge lies ahead as we seek tonarrow the variants in noncoding sequence to determinewhich may be causative. We posited that sequence conservationbetween orthologous sequences from distantlyrelated organisms is a reliable predictor of functional constraint(Thomas et al. 2003). Thus, we have aligned humanand mouse RET genomic sequences, resulting in theidentification of both coding and noncoding multispeciesconserved sequences (MCSs) within the RET locus. Byoverlapping the peak transmission bias from our TDTdata with MCSs in noncoding regions, we have discoveredthe specific functional element involved in HSCR.Hence, through an integrated approach to the study ofcomplex disease, using all available genomic resources inmouse and human, we have made significant progress inelucidation of the genetic basis of HSCR. We expect thatthe causative variants at the remaining linked loci will beidentified by similar studies and that our work will serveas a paradigm for the field of complex disease research.