The desmocollins (DSC) and the desmogleins (DSG) are members o f the cadherin family of adhesive proteins present in the desmosomal type o f cell-cell junction. All o f the known desmocollin and desmoglein isoforms, which have differing tissue and developmental expression patterns, are coded by very closely linked genes at human chromosome 18ql2.1. The six genes have previously been isolated on three overlapping YAC clones (Simrak et al, 1995; Cowley et al., 1997), obtained fi-om two different libraries to form a complete YAC contig of the region. The whole o f the
DSC/DSG locus occupies no more than about 700 kb, and the genes are arranged in the order cen-3’-DSC3-DSC2-DSCl-5’-5’-DSGl-DSG3-DSG2-3’-tel so that the two gene clusters are transcribed outwards from the interlocus region.
Within the human epidermis, the type 1 isoforms are expressed more suprabasally than the type 3 isoforms (Amemann et al., 1993; King et al., 1995,1996). The stratification-related expression o f the desmosomal cadherins is reminiscent of the differential expression o f the type II keratin genes on chromosome 12ql3, where more closely related genes are also physically closer together (Yoon et al., 1994; for review see Fuchs, 1993). In addition, the spatial order in which the DSC and DSG messages are expressed during morphogenesis of stratified epithelia correlates with the physical order of the genes at the desmosomal locus (King et al., 1997). The close linkage suggests an evolutionary selective pressure keeping the two gene clusters together which may also indicate a region in the locus that is involved in coordinating the expression o f the DSC and DSG genes. Recent experiments performed by Marcozzi et al (1998) have also shown that expression o f both a desmocollin and a desmoglein, together with the plaque protein plakoglobin, is required for strong intercellular adhesion in non-adhesive L-cells. The data suggests that the desmosome may obligatorily contain two cadherins and is consistent with a model in which desmocollins and desmogleins may form side by side heterodimers in contrast to the classical cadherins that are homodimeric.
The isolation and characterisation of bacterial clones covering the desmosomal locus provides an extremely valuable resource for studying the regulatory elements that govern and coordinate the expression o f the desmosomal cadherins. PAC and cosmid clones have been shown to be less prone to rearrangements and/or deletions than YACs, and these clones have been chosen in order to sequence the whole o f the human genome. The PAC and cosmid libraries were constructed from two independent DNA sources, thereby allowing the markers obtained from one library to be compared directly with the second library, thus confirming the integrity o f each
clone. Through the use o f standard molecular biology techniques such as Southern analysis and PCR, the positions o f the desmosomal genes were accurately determined, relative to each other. The data also revealed that the desmocollins were equally spaced apart within their cluster whereas the desmoglein genes were not. This was confirmed using automated sequencing techniques to isolate the ends of the PAC clones. Investigations at the HGMP subclone and fingerprint large DNA clones before sequencing begins (D. Bentley, personal communication). Direct sequencing is not the preferred method for sequencing large clones, but in this case, only sequence fi*om the ends o f this ‘ready to sequence’ contig was required. The protocols were optimised and accurate sequence data was obtained, which eliminated the time involved with cloning in order to obtain sequence. The clones would however, need to be fingerprinted to ensure the integrity o f the individual clones (work in progress).
The PACs covering both clusters were obtained easily since cDNA markers were readily available. The distance between the DSC genes was approximately 30-40 kb. The PAC clones were also used to isolate introns that were not easily cloned in order to determine the exon-intron structure o f DSC2 (Greenwood et al., 1997). From this data, the size o f the DSC2 gene was estimated to be approximately 32 kb.
In contrast, the distance between DSGl and DSG3 was approximately 90 kb. The distance between DSG3 and DSG2 was approximately 30-40 kb. A PAC clone, 147K3, which covered part of the desmoglein cluster, was also used to verify the absence of intron 14 in the DSG2 gene (M. Kruger, personal communication). It is possible that the area between DSGl and DSG3 contains another gene o f a size that would be consistent the sizes o f the other desmosomal genes. Previous attempts to look for other unidentified desmosomal genes on the locus involved using PCR primers directed towards conserved sequences o f the desmosomal genes, but this failed to show the existence of another isoform (Cowley et al., 1997). Exon trapping studies on PAC clone 87014, covering the region between DSGl and DSG3, also failed to reveal other genes (V.K. Sahota, unpublished data). It is important to note however, that exon trapping only traps around 40% of the exons present within a
given clone (D. Campbell, personal communication) and this technique should be used in conjunction with other ‘gene hunting’ techniques such as cDNA library screening or cDNA selection. The use of large clones to isolate intron or exon sequence also assists in the identification o f inherited mutations associated with the desmosomal cadherins. DNA from patients with diseases such as the autosomal dominant striated form o f palmoplantar keratoderma which has strong linkage to the desmosomal cadherin locus, may be compared with the intron or exon sequences obtained from these large clones, using techniques such as heteroduplex analysis or single strand conformational polymorphism (SSCP).
The interlocus region presented a more difficult problem as the size o f this region could only be estimated from one YAC clone and no markers had been isolated for this area. Chromosome walking techniques were employed to tackle this region, thereby generating new STS markers en route. Isolating the ends of the clones already identified provided such markers. These markers were used to screen clones originally isolated from the original library screen with the YAC clones 9G-C3, 14E-B5 and 38A-H2 as probes. O f the 247 clones picked, no new PAC clones were identified when using the ends o f PAC 159P13 as a PCR probe. Using the end o f a YAC clone F2/8R, which was isolated by vectorette PCR and which covered the desmoglein cluster (Simrak et al., 1995), five additional clones were obtained that covered the interlocus region, joining the desmocollin and desmoglein clusters together. O f the five additional PAC clones isolated, only one contained the 5’ end o f DSC 1 which was confirmed by direct sequencing of the clone ends and also by PCR. All five clones contained the F2/8R marker as well as the end of PAC clone 159P13 as shown by PCR. One clone extended further towards the DSG cluster, and contained the marker DT3, which was originally isolated from the end o f the cosmid clone D211 (D.M. Hunt, unpublished data). Two of the five clones were digested using rare cutting enzymes which showed common bands between the two clones, indicating the level o f overlap of each clone. In addition, the clone containing the very 5’ end of DSCl (PAC clone 100C18) was shown by Southern analysis to link the desmocollin
cluster to the desmoglein cluster, using the end o f PAC clone 159? 13 as a probe. From this data, the size of the interlocus region was estimated at 150-200 kb, completely covered by two PAC clones. Interestingly, further analysis o f the markers present on each clone indicated that one of the markers (the end o f PAC 159P13) was absent from YAC F2/8.
It was previously thought that the desmosomal locus spanned a distance o f no more than 700 kb. By sizing each clone using PFGE and FIGE, and comparing STS markers generated from both PACs and cosmids, the desmosomal locus appeared to cover no more than 600 kb. The gene order of both clusters may simply be as a result o f duplications that occurred throughout evolution. The fact that the direction of transcription of each gene cluster extends away from the interlocus region does raise the possibility that regulatory control elements exist within this region. The close proximity o f the desmosomal cadherin genes to each other suggests that evolutionary pressures kept the two clusters physically close to each other on the chromosome. This would be consistent with the hypothesis that regulatory elements common to both clusters may direct the tissue-specific expression o f the desmosomal cadherins.
The sequence date obtained from the ends of the PAG clones covering the desmosomal locus were compared against the DNA sequence databases using the BLAST program analysis package available through the HGMP internet website. Sequence analysis revealed that the SP6 end o f PAG clone 315L21 showed high homology to a LINE-1 repetitive element, whilst the T7 end also showed high homology to an Alu repeat region (data not shown). The ends o f PAG clone 159P13 showed no homology to anything in the databases. For sequence data obtained from PAG clones 87014, 147K3, 133E14 and 182N17 refer to Appendix.