Model for a transcript map of human chromosome 21: isolation of new coding sequences from exon and enriched cDNA libraries.

The construction of a transcriptional map for human chromosome 21 requires the generation of a specific catalogue of genes, together with corresponding mapping information. Towards this goal, we conducted a pilot study on a pool of random chromosome 21 cosmids representing 2 Mb of non-contiguous DNA. Exon-amplification and cDNA selection methods were used in combination to extract the coding content from these cosmids, and to derive expressed sequences libraries. These libraries and the source cosmid library were arrayed at high density for hybridisation screening. A strategy was used which related data obtained by multiple hybridisations of clones originating from one library, screened against the other libraries. In this way, it was possible to integrate the information with the physical map and to compare the gene recovery rate of each technique. cDNAs and exons were grouped into bins delineated by EcoRI cosmid fragments, and a subset of 91 cDNAs and 29 exons have been sequenced. These sequences defined 79 non-overlapping potential coding segments distributed in 24 transcriptional units, which were mapped along 21q. Northern blot analysis performed for a subset of cDNAs indicated the existence of a cognate transcript. Comparison to databases indicated three segments matching to known chromosome 21 genes: PFKL, COL6A1 and S100B and six segments matching to unmapped anonymous expressed sequence tags (ESTs). At the translated nucleotide level, strong homologies to known proteins were found with ATP-binding transporters of the ABC family and the dihydroorotase domain of pyrimidine synthetases. These data strongly suggest that bona fide partial genes have been isolated. Several of the newly isolated transcriptional units map to clinically important regions, in particular those involved in Down's syndrome, progressive myoclonus epilepsia and auto-immune polyglandular disease. The study presented here illustrates the complementarity of exon-amplification and cDNA selection techniques for generating a large resource of new expressed landmarks, which contribute to the construction of a chromosome 21 transcript map.

Putative X-linked adrenoleukodystrophy gene shares unexpected homology with ABC transporters.

Adrenoleukodystrophy (ALD) is an X-linked disease affecting 1/20,000 males either as cerebral ALD in childhood or as adrenomyeloneuropathy (AMN) in adults. Childhood ALD is the more severe form, with onset of neurological symptoms between 5-12 years of age. Central nervous system demyelination progresses rapidly and death occurs within a few years. AMN is a milder form of the disease with onset at 15-30 years of age and a more progressive course. Adrenal insufficiency (Addison's disease) may remain the only clinical manifestation of ALD. The principal biochemical abnormality of ALD is the accumulation of very-long-chain fatty acids (VLCFA) because of impaired beta-oxidation in peroxisomes. The normal oxidation of VLCFA-CoA in patients' fibroblasts suggested that the gene coding for the VLCFA-CoA synthetase could be a candidate gene for ALD. Here we use positional cloning to identify a gene partially deleted in 6 of 85 independent patients with ALD. In familial cases, the deletions segregated with the disease. An identical deletion was detected in two brothers presenting with different clinical ALD phenotypes. Candidate exons were identified by computer analysis of genomic sequences and used to isolate complementary DNAs by exon connection and screening of cDNA libraries. The deduced protein sequence shows significant sequence identity to a peroxisomal membrane protein of M(r) 70K that is involved in peroxisome biogenesis and belongs to the 'ATP-binding cassette' superfamily of transporters.

Large-scale mapping and chromosome jumping in the q27 region of the human X chromosome.

We have used pulsed field gel electrophoresis for further physical mapping studies in the q27 region of the human X chromosome. We show that the DXS 102 locus and the F9 gene are separated by only 300 kb despite a genetic distance of 1.4 cM; this linkage orients our large-scale map and shows that the mcf.2 transforming sequence is telomeric to F9. A BssHII complete-digest jumping library was used to jump toward the DXS 105 locus; a 130-kb jump was achieved and the corresponding "linking clone" was obtained.

A long-range restriction map encompassing the cystic fibrosis locus and its closely linked genetic markers.

The cystic fibrosis (CF) locus has been localized to the long arm of chromosome 7 by linkage analysis, and the genetic relationship between CF and the probes J3.11, met, and 7C22 has been extensively studied. To extend this genetic analysis to higher resolution, to provide information on physical distances underlying the genetic relationships, and to set limits to the position of the cystic fibrosis mutation, we have constructed a partial restriction map covering approximately 5 Mb that defines the physical relationship between these and the more recently isolated markers CS.7, XV-2c, Lcn2, and C2/5. Allelic association indicates that CS.7 and XV-2c are close to the CF locus, and an expressed sequence from this region has been described as a candidate gene for this mutation (X. Estivill et al., 1987, Nature (London) 326: 840-845). Using pulsed-field gel electrophoresis we have determined the physical order of these markers to be cen-7C22-Lcn2-met-C2/5-XV-2c-CS.7-J3.11-tel and have localized the CF mutation to an interval of less than 1500 kb. A (not unexpected) disproportionality was observed between the currently best estimates of genetic and physical distances, with the interval J3.11-met showing an approximately fourfold higher frequency of recombination than the met-7C22 interval.