Type 2 Diabetes Locus on 12q15: Further Mapping and Mutation Screening of Two Candidate Genes

It is becoming increasingly clear that the etiology of type 2 diabetes is much more heterogeneous than previously thought. Several attempts to identify type 2 diabetes genes by linkage studies have led to conflicting results, indicating that distinct genes are probably involved in different populations (1,2). Even within the same ethnic group, different genes may be involved in different families (2). The situation does not appear to be simpler for monogenic forms of diabetes, such as those transmitted with an autosomal-dominant pattern of inheritance. Six distinct genes have been described for the best-known form of this group—maturity-onset diabetes of the young (MODY) (3,4)—and several additional loci probably exist. In France and England, ∼25% of MODY families have diabetes unaccounted for by known MODY genes, but this proportion seems to be much higher among families with an older age of diabetes diagnosis than classical MODY (5,6,7). We recently described a novel locus for autosomal-dominant type 2 diabetes on chromosome 12q15, 50 cM from the previously described locus NIDDM2 (8). All of the evidence of linkage came from four families that had individual LOD scores >1.0. By analyzing obligate recombinants in the two families with highest LOD scores (2.35 and 2.1), we assigned this putative locus to the 6 cM between markers D12S375 and D12S1684. Here we report the further mapping of this diabetes locus and the results of the study of two candidate genes placed in the critical region.

To narrow the linked interval, the four families that had a LOD score >1.0 in the original analysis (families 8, 19, 24, and 32 [8]) were genotyped for 11 microsatellite markers located between D12S375 and D12S1684 (Fig. 1). Included were 27 affected and 25 nonaffected family members. The 11 markers were chosen from contig WC12.4 of the WI/MIT integrated map (9). Seven of the microsatellites were also in the Marshfield genetic map (Fig. 1). Within each family, all affected members shared at least one portion of this region identically by descent (Fig. 2A), although the shared haplotype differed among families. In family 8, which had a maximum LOD score (Z_max) of 2.35 in the original analysis, the shared haplotype went from D12S1693 on the centromeric side to D12S326 on the telomeric side (Fig. 2A). The LOD score for linkage at this location peaked to 3.1 (Fig. 2B). The shared haplotype in family 24, which had a Z_max of 2.15, further moved the telomeric boundary, narrowing the linked region to the interval between D12S1693 and D12S1347 (Fig. 2A and B). No additional information was provided by family 19, in which no recombinants were identified. The affected haplotype of family 32 did not overlap with the critical interval defined by families 8 and 24, but extended instead from D12S1693 toward the centromere, with the LOD score peaking between D12S1702 and D12S375 (Z_max = 1.5). A possible explanation for this finding is that there are two diabetes loci in this region: one operating in family 8, placed between D12S1693 and D12S326, and the other in family 32, centromeric to D12S1693, with diabetes in families 19 and 24 being attributable to either locus. Indeed, it is not unusual that two adjacent loci responsible for the same disorder are initially mapped as a single locus (10). Another possibility is that the recombinants defining either the centromeric boundary in family 8 or the telomeric boundary in family 32 are actually phenocopies. Finally, because family 32 has a Z_max of only 1.5, it is also possible that the linkage observed in this family is entirely due to chance.

One gene placed in the D12S1693-D12S326 interval that caught our attention is protein tyrosine phosphatase receptor type R (PTPRR, also known as NC-PTPCOM1). Tyrosine phosphorylation is a key component of the cascade of reactions linking the insulin receptor to insulin action in peripheral tissues (11). Furthermore, a tyrosine phosphatase of the same class as PTPRR has been shown to be associated with insulin-containing vesicles in β-cells (12). Another interesting aspect was that the major site for PTPRR expression was the brain, followed by pancreatic islets and lung (Fig. 3). This distribution pattern is remarkably similar to that of another gene (NEUROD1) that has been found to be mutated in MODY and type 2 diabetes (4). Thus, PTPRR was a plausible candidate gene for diabetes in these families. The PTPRR gene had been assigned to YAC clones 959-C-8 and 788-E-12, which also contain D12S1025, but its exon-intron structure had not been resolved. After isolating and sequencing three overlapping BAC clones containing the whole gene, we identified 14 exons ranging in size between 80 and 473 bp (Fig. 4A and Table A1, which can be found in an on-line appendix at www.diabetes.org/diabetes/appendix.htm ). Mutation screening of the coding sequence of probands from the four linked families and other 12 families in which linkage could not be excluded (LOD score > -2.0) revealed several polymorphisms, one of which changed the amino acid sequence (Arg³¹⁴→Lys [Table 1 and Table A2, which can be found in an on-line appendix at www.diabetes.org/diabetes/appendix.htm ]). However, no mutations segregating with diabetes were identified. Furthermore, the Arg³¹⁴→Lys polymorphism was similarly frequent among the 32 original family probands (0.266), 173 subjects with type 2 diabetes (0.266), and 181 nondiabetic control subjects (0.226), making PTPRR unlikely as a type 2 diabetes locus in these families.

Because there was evidence that the linked interval might extend centromeric to D12S1693, or that another diabetes gene might be located beyond this marker, we also analyzed a candidate gene in the vicinity of D12S375. In the National Center for Biology Information GeneMap 99 (13), the carboxypeptidase (CP) M gene was mapped to this region, and we confirmed this location by assigning it to CEPH YAC clones 916-C-11 and 883-H-12, which also contain D12S375. CPM is a membrane-bound enzyme belonging to the regulatory carboxypeptidase subfamily—a class of proteins that have been implicated in the processing and sorting of polypeptide hormones, either at the site of hormone production or in target tissues (14,15). Of note, a mutation in a member of this family (CPE) has been shown to be responsible for diabetes in the fat/fat mouse model (16). After isolating BAC clones containing the CPM gene, we identified nine exons ranging in size from 98 to 240 bp (Fig. 4B and Table A3, which can be found in an online appendix at www.diabetes.org/diabetes/appendix.asp ). In the same 16 families that were screened for PTPRR, we identified several silent polymorphisms together with a private Tyr⁹⁹→His mutation that showed incomplete segregation with diabetes in one family (Table 1 and Table A4, which can be found in an online appendix at www.diabetes.org/diabetes/appendix.asp ). No mutations, however, were identified in the four linked families, suggesting that CPM is unlikely to be the diabetes locus on 12q15.

In summary, we have narrowed the diabetes locus on 12q15 to the region between D12S1693 and D12S326. The LOD score for this location is 3.1 in one family, but it is still possible that the critical interval extends beyond these boundaries. It is unlikely that this locus corresponds to the PTPRR or CPM genes, although a role of mutations in regulatory regions cannot be excluded at this time. New clues are expected from the upcoming completion of the Human Genome Project, which will help us define the exact size of the linked interval, and will provide an inclusive list of the genes placed in this region.