Potential CpG-rich islands clustering around single-minded gene in Down syndrome chromosomal region.

Abstract

Down syndrome (DS), the most frequent birth defect, is caused by trisomy of human Chromosome (Chr) 21 and is associated with numerous phenotypic features, including mental retardation. Recently, the major features have been correlated with rare DS patients carrying partial trisomies that allow the definition of a 3.6Mb chromosomal segment around locus D21S55 in the cytogenetic G-band of 21q22.2 to be designated as Down syndrome chromosomal region or DCR (Delabar et al. 1993; Rahmani et al. 1989; Dufresne-Zacharia et al. 1994). This raises questions whether there exist any candidate genes transcribed from the inert G-band and, if present, how they can induce multiple phenotypes. To identify genes involved in the DS phenotypes, we have constructed an ordered clone library with Mb-sized cosmid/PAC contigs spanning the DCR (Fig. 1; Osoegawa in press), and sought potential CpGrich islands within the contigs, using the rare cutting enzymes that have the recognition sequences containing one or two CpG dinucleotides (Lindsay and Bird 1987). We found that potential CpG-rich islands were located at high density in a 330-kb region around the hSIM (Chen et al. 1995), which was characterized as a mini R-band. Interestingly, most of the CpG islands clustered in the vicinity of the 5' upmost region of the hSIM. For the location of the rare cutting sites, 138 cosmid and 8 PAC DNAs covering the DCR were digested doubly with EcoRI plus one of the seven rare cutting enzymes described below, and analyzed by agarose gel electrophoresis. A large number of the cutting sites were precisely mapped. We found at least 50 sites for BssHII, 84 for EagI, 58 for SaclI, 23 for MluI, 31 for NruI, 13 for Nod, and 53 for SalI. The distribution of the cutting sites was neither random nor uniform throughout the DCR, and some of these sites appeared to be clustered in a region smaller than 1 kb. Assuming a potential CpG-rich island to be a close clustering of three, or more, of the cutting sites within 1 kb DNA, we could identify at least 20 CpG islands. Of these, 12 CpG islands were concentrated in a 330-kb region between loci LA230E8 and RA152F7 located at 760 kb upstream from D21S55 (Fig. 1). In addition to the high island density, a striking structural feature was observed in the region telomeric to LA230E8 in which at least 6 CpG islands, or 34 cutting sites, clustered within 20 kb, had distinctly different distances from each other than the other islands. This has revealed that there exist two classes of CpG-rich islands, a cluster and single CpG islands. Single CpG islands are accepted as defining individual genes. Cluster, to our knowledge, is quite novel in the human genome and may be characterized to be in a T-band, subset of R-bands (Crig and Bickmore 1994). More recently, the human homolog to Drosophila sim gene (Crews et al. 1988) has been isolated, and subsequently mapped as the hSIM (Chen et al. 1995, Lucente et al. 1995). By analogy to the