3.14 L inkage o f D H R D to chrom osom e 2p21-16
Following a genome wide search utilising over 230 (CA)n microsatellite markers, pairwise linkage analysis enabled the DHRD causative gene to be localised to chromosome 2 p l6 . Haplotype analysis of recombinant individuals enabled the disease to be placed within a 5 cM interval between markers D2S1316 (centromeric) and D2S378 (telomeric). Three markers within the region provided significant positive lodscores above 3.0 including the most informative tetranucleotide marker D2S133.‘\ which is known to map within this interval, but has not been placed at a genetic distance relative to markers D2S2316 and D2S378. No double recombination events were observed across this region encompassing 55 meiotic events.
3.15 C andidate genes and other ocular diseases m apping w ithin the DHRD
region
Aside from Malattia leventinese which has already been described (see introduction to this chapter), another ocular condition, congenital glaucoma (GLC3A) has recently been assigned to 2p21 (Sarafarazi et al, 1995), this however is a disorder of the anterior segment o f the eye and is therefore unlikely to be allelic with DHRD. There are at present two genes mapping within the 2p21-16 interval which can be regarded as good candidates for DHRD. This includes the calmodulin gene {CALM2), which is one of a number of calmodulin genes known to be expressed in the retina (Pochet et al, 1991) and the nonerythroid form of (3 spectrin known as (3 fodrin (SPTBNl) which is known to be expressed in the RPE (Gunderson et al, 1991). The calmodulin protein is ubiquitously expressed and known to be involved in calcium binding. In the retina, its interaction with Ca^^ is believed to modulate the affinity of the voltage-gated ion channel for cGMP thus modulating the events of phototransduction (Hsu et al, 1993), a pathway in which disruptions have led to numerous retinal dystrophies (Daiger et al, 1995; Berson, 1996). It has not however been determined whether the retinal calmodulin is the product o f the CALMl, CALM2 or CALMS gene expression, as all three genes code for the same protein. Furthermore, "iRe structure of the human calmodulin gene promoter indicates this gene to belong to a class of house keeping genes (Roller et al, 1990) making it an unlikely candidate for the retinal specific disease phenotype observed inBHRD, however tissue specific regulatory mechanisms maybe in operation. (3 fodrin is a protein of the cell membrane cytoskeleton, displaying properties such as calmodulin binding and the ability to cross-link actin filaments, and may therefore play a role in cell structure and stability. It has been mapped by in situ hybridisation to chromosome 2p21, a region in which the DHRD northern flanking recombinants are seen and therefore lies in a position bordering the DHRD region, but is still worthy of investigation due to its functional properties and pattern o f tissue expression.
Genetic mapping o f D oyn e’s honeycomb retinal dystrophy
that lies between the genes coding for malate dehydrogenase {Mor2) and reticuioendotheliosis oncogene {Rel). Scrutiny of the Mouse Genome Database (1995) for genes mapping to this region however, provided no further candidates for DHRD.
3.16 T he d ifficulties encountered in m apping D H R D
(a)
The importance of correct disease assignment
The importance o f correct disease assignment must be emphatically stated from the insight gained from this study. Clinicians need to be cautious when diagnosing patients and their phenotypes. The late onset of phenotypic manifestations in this disease coupled with the extent o f phenotypic variability observed across this family proved a difficult task in DHRD evaluation. Recent detailed electrodiagnostic tests including autofluorescent imaging, reveals the properties o f the subretinal deposits found in this family to vary in different regions o f the retina (Evans et al. In Press). This is supported by the previous work o f Pauleikhoff and coworkers (1992) who demonstrated that the nature of drusen deposits in different regions o f Bruchs membrane varied, implying possible variation in their biochemical composition. Differing ratios o f deposits of various chemical composition may culminate in the varying clinical manifestations and extent of atrophy seen in this disease. All these factors contributed to the difficulty o f correct disease status allocation in DHRD and was one o f the reasons responsible for the several changes made to the pedigree structure used for linkage analysis in this study.
Discordance between the phenotype and underlying genotype can be regarded as a misclassification or misdiagnoses. Such incorrect diagnoses can be extremely damaging to a linkage study. Family size is a important factor that relates to the effect of a misdiagnosis on linkage analysis. It is o f critical importance in smaller pedigrees, whereas 1-2 such errors do not significantly affect the lodscore in a large multigenerational family. For example, in a fully informative sibship of 11 sibs and 2 living parents, a single misdiagnosis can decrease the lodscore from a value o f 3.0 to 1.3 if a single unaffected individual is classified as being affected. If the genetic marker being tested lies at a distance o f 10 cM from the disease, a single misdiagnosis can result in a maximal lodscore of only 0.5 which may preclude the region from further analysis (Stone and Sheffield, 1992). It is for this reason that penetrance values of disease are set at 0.99 to compensate for any discrepancies in diagnosis and not at a value o f 1.0 even for diseases clearly displaying complete penetrance. Although the DHRD family used in this study was a large kindred, the linkage panel was sparsely multigenerational and in this situation, correct diagnosis was also found to be crucial to achieve reasonable lodscores.
(b) Family structure
The structure of the family also meant that a number of markers recognised as being relatively informative with 6-8 different alleles randomly segregating in the general population were uninformative in this family where most parental haplotypes had to be inferred, this was n o t always unequivocally possible. Hindsight also reveals that for such family structures it is necessary to be closer to the disease before evidence o f linkage emerges, therefore implying the necessity to test markers spaced at closer genetic intervals (5 - 10 cM) for certainty o f detecting linkage on preliminary screening o f the genome. Penetrance was not however a problem in this family as all affected individuals displaying the phenotype had a single affected parent either displaying the disease characteristics or known to have been affected by the information gained from living descendants.
The ancestral consanguineous marriage of two affected individuals in generation IV (branch Al and A2, pedigree IV) which came to light during the course of the linkage study made aware the 50 % possibility of ‘double dominant’ individuals carrying two affected genes being present in the offspring o f these two individuals. This brings a degree o f complexity to the study as two different disease haplotypes could presumably be segregating in the family, a feature beyond the analytical capacity o f the LINKAGE program used for lodscore calculations. Therefore, during the visual scoring of alleles not one, but two alleles may have been segregating with the disease phenotype in the different branches of the family which may have been overlooked as recombination events and thus non-linkage. The ‘double dom inant’ genotype would have been far simpler to analyse had the two disease chromosomes been identical with the same alleles segregating throughout most of its length, however without the knowledge of an existing consanguinity, such homozygous alleles would have been regarded as being uninformative and plummeted the lodscore. Re-analyses o f genotypic data based on this possibility was carried out for regions in which positive lodscores were obtained but to no avail. Recent haplotype analysis on the linked pedigree by a colleague in the group has confirmed the presence of a single disease associated haplotype.
The difficulties encountered with mapping of this disease led us at one stage o f the project to divide the 4 branches of the main family (A l, A2, and C) into 3 separate smaller families from which individual lodscores could be assessed and the cumulative lodscore obtained. The reasoning behind this being that the 3 branches o f the family were far removed from each other, especially branch C and thus separated by many recombination events which could result in different alleles at marker loci segregating with the disease phenotype in the different branches. In regions where the insignificant but positive lodscores were obtained such re calculations were found to not greatly alter the existing lodscore from that calculated for the family as a whole and this approach was discarded.
Genetic mapping o f Doyne's honeycomb retinal dystrophy