The Genom of Homo sapiens.pdf

464 LIPOVICH AND KINGto-tail antisense to the LOC51306 gene. The KLHL3gene shared two exons in the sense orientation with38I10.TU1 (Fig. 1A). Although the multiple UGPs associatedwith 38I10.TU1 and other genes in the locus wereevident from manual curation of EST-to-genomic alignments,no mention of these UGPs, or of the existence of38I10.TU1, is seen in published analyses of the genomicstructures of genes which we show to be adjacent to oroverlapped by the TU (Godley et al. 1999; Guo et al.2000; Lai et al. 2001).Comparative annotation of the syntenic region in themouse genome was undertaken to determine whether38I10.TU1 has a direct homolog or a positional equivalentin the mouse. The mouse Ttid and Pkd2l2 genes werefound to be 10 Mb from each other due to a local inversionwithin a larger syntenic stretch. This sharply contrastswith the 2-kb distance separating the human orthologsof these genes and precludes the existence of amouse TU that would connect or span the two genes.Along with the major contribution of expressed primatespecificrepeats to the definition of splice junctions andhence exons of 38I10.TU1, these results indicate that theTU, along with its potential impact on expression regulationand structural modification of its multiple overlappingand adjacent coding genes, is likely primate-specific.If true, this would not be the only case of alineage-specific TU simultaneously involved in multipleUGPs (Ohinata et al. 2003).Example 2 was discovered during our annotation of the5q31 protocadherin (PCDH) gene cluster. PCDHs aremembers of the cadherin superfamily, known to representmajor structural and functional components of synapses.Cell-specific combinations of PCDHs are expressed atsynaptic junctions. PCDHs likely account for some of theneuronal combinatorial complexity in both developingand adult brain, affecting brain development and possiblymemory formation (Noonan et al. 2003). Open questionsregarding the mechanism of cell specificity and splicingregulation of PCDH expression remain (Wang et al.2002). Our annotation demonstrated EST support for 8TUs cis-antisense to human PCDH exons (Table 1). In allcases, ESTs comprising the TUs were erroneouslygrouped by UniGene together with the sense-strandPCDH ESTs from the appropriate locus, even thoughtheir antisense nature is clear from a distinct pattern ofEST-to-genomic alignments coupled with unambiguousassignment of GT-AG splice sites and/or canonicalpolyadenylation signals to the strand opposite that fromwhich the PCDH exons are transcribed.Two of the eight anti-PCDH TUs overlap newly pseudogenicPCDH exons inactivated during mammalianevolution, in a fashion reminiscent of that exhibited by38I10.TU1 in its overlap of TTID’s apparently silencedamino-terminal-coding exon. Specifically, the humanPCDHβΨ5 pseudogene, which has an antisense TU, andthe human PCDHβ15 gene are the putative products of alineage-specific duplication of the ancestral PCDHβVgene, which is single-copy in mouse (Vanhalst et al.2001). Similarly, the human PCDHγ variable exon Ψ3,which also has an antisense TU, is pseudogenic, but itsmouse ortholog, Pcdhγ variable exon b8, has an intactORF (Wu et al. 2001). This again suggests a role for cisantisensein gene structure evolution.Lineage specificity of such gene-structure evolution isaccentuated by the absence of equivalents of any humananti-PCDH TUs in the mouse, revealed by manual annotationof all mouse ESTs representing true orthologs ornearest homologs of human PCDH exons. It is highly intriguingthat not only are PCDHs directly relevant to neuronaland behavioral complexity (which is more extensivein mammals, and especially primates, than in anyother animals), but they may also be subject to an antisense-mediatedregulatory mechanism that arose after theprimate–rodent divergence. Such a mechanism would bein agreement with suggestions of species-specific evolutionarypressures on PCDH genes (Vanhalst et al. 2001).Major Properties of UGPs and TUs at 5q31The results of our 5q31 annotation can be summarizedin four major trends. First, the number of EST-supportednovel TUs approximately equals that of known genes.Second, certain genes and TUs are simultaneously involvedin multiple UGPs. Third, UGPs occur more frequentlyin specific genomic intervals and may be nonran-Table 1. Putative Novel Transcriptional Units (TUs) cis-Antisense to Protocadherin (PCDH) Genes at 5q31Human PCDH Mouse Expression Antisense ESTs Exons in humangene or exon Characteristics ortholog human/mouse human/mouse antisense transcriptPCDHα9 variable exon Pcdhα7 + + 10 0 1PCDHα11 variable exon Pcdhα11 + + 1 0 1PCDHα12 variable exon none + na 2 na 1PCDHαc1 variable exon Pcdhαc1 + + 1 0 1PCDHβ3 single-exon gene Pcdhβ3 + + 6 0 2PCDHβ16 single-exon gene Pcdhβ16 + + 4 0 1PCDHΨ5 single-exon unprocessed β-class none – na 16 na 3pseudogenePCDHΨ3 unprocessed γ-class pseudogenic Pcdhγb8 – + 2 0 1variable exonTable was derived from our SeqHelp-aided manual annotation of genomic clones AC005609, AC010223, AC025436, AC008727,AC005618, and AC005366.(na) Not applicable.