General discussion 119
General discussion
The candidate gene approach
In the first step of this thesis, candidate genes for canine congenital sensorineural deafness (CCSD) in Dalmatian dogs were analyzed by means of microsatellite markers or alternatively by single nucleotide polymorphisms (SNPs).
The candidate genes, for which a set of in total 43 microsatellites was available, included the following 24 genes: CDH23, CLDN14, COCH, COL11A2, DFNA5, DIAPH1, EDN3, EDNRB, EYA4, GJA1, GJB2, GJB6, MITF, MYH9, MYO6, MYO7A, MYO15A, OTOF, PAX3, POU4F3, SLC26A4, SOX10, TECTA, and TMPRSS3 (Rak 2003). These genes are known to be involved either in human non-syndromic deafness or in the human Waardenburg syndrome. The Waardenburg syndrome (WS) manifests with sensorineural deafness and pigmentation defects in iris, hair and skin. WS is classified into four types, depending on the presence or absence of additional symptoms, which are caused by mutations in the five genes EDN3, EDNRB, MITF, PAX3 and SOX, respectively.
For another eight recently identified genes responsible for different forms of human non-syndromic deafness including TMC1, TMIE, USH1C, MYH14, MYO3A, PRES, WHRN, and ESPN, linkage and association analyses were performed using newly developed SNPs.
In the last years most projects have exploited canine traits for which either direct candidate genes could be proposed and evaluated, or for which large informative pedigrees were available to enable linkage mapping to identify candidate regions. A major component of such research efforts comprised the cloning, sequencing and mapping of individual canine homologs of genes either proposed as candidate genes, or expected to be located in candidate regions. This was necessary to identify new informative polymorphisms (e.g. SNPs, microsatellites) for high resolution mapping of candidate regions, and to examine each exon and exon/intron boundary for positional candidates. Availability of the second version of the dog genome assembly (build 2.1) of the NCBI database shortcuts this effort and increases the investigators efficiency. Now, either additional candidate genes for canine congenital sensorineural deafness can be found directly by its gene symbol in the 2.1 of the
General discussion 120
NCBI's genome annotation or, if a candidate gene is not yet annotated, a BLAST (Basic Local Alignment Search Tool) search against the canine whole genome shotgun (wgs) sequence resource can be used to obtain the sequence of the canine genomic contigs containing the human homologous gene and thus, intragenic markers can be developed for subsequent linkage and association analyses.
Over the past decade it has become increasingly clear how far structural and functional homologies at the gene level extend across even distantly related species.
Much is known about deafness-causing gene mutations in humans and mice, including the fact that the clinical and histopathological findings are often very similar to those of deafness in Dalmatian dogs. Thus, genes responsible for non-syndromic congenital hereditary deafness in humans seem to be appropriate candidate genes for CCSD (Rak and Distl, 2005). In this thesis we first concentrated on the candidate gene approach combined with linkage analysis method using affected pedigree members. Once a significant linkage was found, only the linked genes with the required low error probability values were used for further molecular genetic analysis.
The method of candidate gene approach using either gene-associated microsatellite or alternatively SNP markers was applied for in total 32 candidate genes, which comprise nearly all of the identified mutated genes causing non-syndromic hereditary hearing impairment in humans.
Linkage and association analysis
In principle, linkage and association are totally different phenomena. Linkage is a relation between loci and association is a relation between alleles.
Linkage means that a haplotype characterised by microsatellites or SNPs is significantly more often present in family members with the phenotype under study than expected by random assortment. For construction of haplotypes, sets of closely linked genetic markers on the same chromosome are needed, which tend to be inherited together as they are not likely to be separated by recombination. Linkage creates associations within families, but not among unrelated induviduals.
Association is a statistical statement about the co-occurrence of alleles or phenotypes. Association analysis can be carried out as a method of genetic analysis, that compares the frequency of alleles between affected and unaffected individuals across all families. Thus, for association family structures are not necessary. A given allele is considered to be associated with the disease if the presence of that allele
General discussion 121 explains a significant proportion of the phenotypic trait variation. For association studies the developing of a marker set consisting of SNPs rather than microsatellites is needed.
In this thesis a total of 32 candidate genes for canine congenital deafness were analyzed, which showed an appropriate clinical and histological disease pattern in comparison to deafness in Dalmatian dogs. Rak (2003) developed a set of 43 microsatellites for in total 24 candidate genes, among them the CLDN14 gene on canine chromosome (CFA) 31 and the MYH9 gene on CFA10. The GJA1 on CFA1 was also considered as a candidate gene for CCSD (Rak 2003), and therefore two gene-associated microsatellites have been developed. Recently it turned out, that GJA1 is not responsible for human sensorineural non-syndromic deafness, but for a human syndromic disorder that can be related with conductive deafness.
By the use of a non-parametric linkage analysis using the existing set of 43 microsatellites associated to 24 candidate genes, we found linkage to markers associated to CLDN14 on CFA31, MYH9 on CFA10 and GJA1 on CFA1.
For another another eight candidate genes it was possible to develop SNPs.
Performing linkage analyses as well as association and haplotype studies it was possible to exclude these eight candidate genes from being responsible for the CCSD phenotype.
Over the past ten years, significant progress has been made in the identification of deafness gene localisations. Up to now, approximately 120 loci have been reported for both autosomal dominant and recessive forms of non-syndromic hereditary deafness in humans and only for one third the responsible gene mutation could be detected. Thus, it can be expected that additional potential human candidates for CCSD in Dalmatian dogs will become available in future (Van Camp and Smith, 2003).
The extreme heterogeneity of human deafness often hampered genetic studies because many different genetic forms of hearing loss give rise to similar clinical phenotypes, and, conversely, mutations in the same gene can result in a variety of clinical phenotypes. In man, genes, that transport ions across membranes to maintain appropriate solute concentration and pH as well as regulatory genes, mainly transcription factors, and genes, that play a part in structural integrity, are essential for the hearing process.
General discussion 122
The results of this thesis indicate that the inheritance of hearing loss in Dalmatian dogs is probably as heterogenic in origin as it is in humans. Genetic heterogeneity means, that different mutations cause a similar phenotype; the different mutations can either be found at the same locus (allelic heterogeneity) or even at different loci (non-allelic heterogeneity). As linkage was found for different candidate genes in different families, subsequently only the families indicating linkage were chosen for further molecular analyses.
GJA1 on CFA1, MYH9 on CFA10 and CLDN14 on CFA31 and their flanking regions are further analyzed with a combined approach using microsatellite and SNP markers.
CFA1
By the use of GJA1-associated microsatellites, one large French Dalmatian dog family reached a Zmean value of 2.95 (p<0.002) for GJA1_MS1 and a Zmean value of 2.86 (p<0.002) for GJA1_MS2, indicating linkage. Subsequently, a sequence analysis of the GJA1 gene using the above mentioned French Dalmatian dog family was performed. None of the observed polymorphism did alter the predicted amino acid sequence of GJA1 nor showed the identified haplotypes an association with the CCSD phenotype. Thus, evaluation of the SNP markers debilitates the linkage to CCSD for the French half-sib family. It is unlikely that the GJA1 gene is involved in the pathogenesis of CCSD in Dalmatian dogs. To see whether significant test statistics for other genomic regions on CFA1 and for more families can be shown, a non-parametric linkage analysis was performed with 27 microsatellite markers covering CFA1 completely. In total 176 animals were genotyped. We could not find linkage to any microsatellite in the analyzed families. Furthermore, it was revealed that GJA1 is not a candidate gene for sensorineural non-syndromic deafness in humans (W.A. Paznekas cited a personal communication from the senior author (W.
E. Nance) of the paper by Liu et al., 2001). GJA1 is participating in a human syndrome called oculodentodigital dysplasia (ODDD) that can be accompanied with hearing impairment (Paznekas et al., 2003). But the type of deafness in human ODDD is conductive rather than sensorineural. As deafness in dogs, especially in Dalmatians, is almost exclusively caused by sensorineural non-syndromic forms, also known as cochleosaccular degeneration, the GJA1 gene should not be considered as a candidate gene for CCSD anymore.
General discussion 123 CFA31
By the use of microsatellites as well as SNPs we found significant linkage with CCSD for CLDN14-associated microsatellites in four full-sib and one half-sib Dalmatian dog familiy with a Zmean value of 3.83 (p<0.00007). A mutation analysis was performed for exon three, as this is the only translated one in man. None of the observed polymorphisms did alter the predicted amino acid sequence. However, to clarify the importance of the CLDN14 gene and its flanking regions in the CCSD phenotype, more SNPs have to be developed within the CLDN14 gene as well as in its flanking regions with the aim to find significant linkage disequilibrium of SNP markers.
CFA10
A significant co-segregation of markers alleles and the phenotypic expression of deafness in a large sample of German Dalmatian dog families was determined for one marker (Z-mean of 1.56 [p= 0.06] and a Lod score of 0.58 [p= 0.05]) associated to the MYH9 gene. Subsequently, we evaluated whether MYH9 gene mutations are responsible for CCSD in these Dalmatian dog families. An initial priority in defining gene structure is to obtain a full-length cDNA sequence and identify translational initiation and termination sites, and polyadenylation site(s). Exon/intron structure can then be determined by referencing the cDNA sequence against sequences of cognate genomic DNA. One popular method of obtaining full-length cDNA sequences is the RACE (rapid amplification of cDNA ends) technique. RACE-PCR is an anchor PCR modification of RT-PCR. The rationale is to amplify sequences between a single previously characterised region in the mRNA (cDNA) and an anchor sequence that is coupled to the 5 or the 3 end. A primer is designed from the known internal sequence and the second primer is selected from the relevant anchor sequence.
To provide the genomic organization and the complete sequence of the canine MYH9 gene, the isolation of full length cDNAs was achieved with the help of a modified rapid amplification of cDNA ends (RACE) protocol. A mutation analysis was performed to identify single nucleotide polymorphisms (SNPs) in this gene. We analyzed the association of the MYH9 haplotypes with the CCSD phenotype in three families of Dalmatian dogs with frequent occurrence of CCSD and significant linkage to gene-associated microsatellites. Using the RT-PCR analyses for cDNA-genomic sequence comparisons we detected that the canine MYH9 gene is bigger compared to the human sequence due to the untranlated first exon in the 5’-UTR.
General discussion 124
The canine MYH9 gene consists of 41 exons with an untranslated exon 1 and exon/intron boundaries that conform perfectly to the GT/AG rule.
None of the observed polymorphisms did alter the predicted amino acid sequence of MYH9 nor showed the identified haplotypes an association with the CCSD phenotype.
Thus, these silent point mutations found in affected and unaffected Dalmatian dogs do not seem to be responsible for the CCSD phenotype in these three families.
To clarify if other regions on CFA10 are responsible for the CCSD phenotype, we added in a second step 27 microsatellite markers derived from the NCBI database to cover CFA10 with a high density in the flanking region of the MYH9 gene. A linkage analysis was carried out for 23 Dalmatian dog families consisting of 176 animals that were genotyped with the marker set of 27 microsatellite markers.
We found significant linkage to microsatellites in the region spanning 36 Mb to 48 Mb.
Consequently, we screened this 12 Mb spanning region for SNPs. Out of the 23 analyzed Dalmatian dog families, five full-sib families were chosen to screen for SNPs because of their obviously significant values at the above mentioned region.
Twenty-six new SNP markers covering the region of 36 Mb to 48 Mb equally were developed and added to the linkage analysis. The significant Zmeans on CFA10 was confirmed after adding the SNP markers. Furthermore, with the use of SNPs, the apparently linked region spanning 36 Mb to 48 Mb could be narrowed down to 5 Mb spanning from 39 Mb to 44 Mb. For this reason, the use of SNPs in addition to informative microsatellites appears highly recommendable.
In further studies more SNPs have to be developed within the identified CCSD region on CFA10 to localize the deafness causing gene or to find unambiguously associsted SNP markers, which could be used for a population-wide genetic test for CCSD.
Chapter 10
Summary
Summary 127
Summary
Molecular genetic analysis of canine congenital sensorineural deafness in Dalmatian dogs
Katharina Mieskes (2006)
The objective of the present study was to localize the gene or genomic region that is involved in the development of canine congenital sensorineural deafness (CCSD) in Dalmatian dogs.
In man as in different dog breeds deafness is an often diagnosed disorder with the Dalmatian dog showing the highest incidence. Many genetic disorders in humans and domestic dogs (Canis familiaris) demonstrate a high level of clinical and molecular similarity.
Altogether 39 genes have already been found causative for sensorineural non-syndromic hearing impairment in humans. Out of this 39 deafness causing genes a total of 32 candidate genes were selected for canine congenital deafness, which showed an appropriate clinical and histological disease pattern in comparison to deafness in Dalmatians dogs.
On the one hand an existing set of 43 microsatllite markers for in total 24 candidate genes were used for a non-parametric linkage analysis, among them the claudin-14 (CLDN14) gene on canine chromosome (CFA) 31 and the myosin, heavy polypeptide 9 (MYH9) gene on CFA10. The gap junction protein, alpha 1 (GJA1) gene on CFA1 was also considered as a candidate gene for CCSD and thus GJA1-associated microsatellites were part of the existing set. Recently it turned out, that GJA1 is not responsible for human sensorineural non-syndromic deafness, but for a human syndromic disorder that can be related with conductive deafness. In the last few years more human deafness genes have been identified, among them eight genes that were considered as appropriate candidates for CCSD. For these eight genes a total of 21 SNPs were newly developed and used for non-parametric linkage and association analyses.
Summary 128
The used microsatellite and SNP markers derived either from a partial sequence analysis of BAC clones each containing a different candidate gene, or from sequences deposited in the current dog genome assembly (Boxer genome assembly 2.1) of the NCBI GenBank. We found significant linkage to markers associated to CLDN14 on CFA31, MYH9 on CFA10 and GJA1 on CFA1. To substantiate the linkage we searched for sequence variations within these three genes. SNPs found in intronic sequences of either gene were included in the linkage analyses. Sequence analysis neither revealed a causative mutation nor significant linkage disequilibrium of SNP markers with CCSD. Subsequently, CFA1, 10 and 31 were scanned completely with microsatellite markers derived from the NCBI database with the purpose to see if other regions on this three chromosomes harbour a gene that is involved in the development of CCSD.
The analyses of SNPs and more microsatellite markers on CFA1 revealed no significant linkage to CCSD. Thus, it is unlikely that the canine chromosome 1 and the GJA1 gene are responsible for the CCSD phenoptype. As deafness in dogs, especially in Dalmatians, is almost exclusively caused by sensorineural non-syndromic forms, the GJA1 gene should not be considered as a candidate gene for CCSD anymore.
On CFA10 we could exclude MYH9 for being causal for deafness, but by adding more microsatellites covering CFA10 completely we found significant linkage to the CCSD phenotype in the region distal of MYH9. Hence, new SNP-markers for fine mapping the region spanning 36 to 48 Mb were developed by sequence analyses of different Dalmatian dogs. The search for SNPs was carried out on genomic sequences of genes located in the significantly linked region. The sequences of these genomic sequences were derived from the NCBI GenBank. The SNPs confirmed the linkage and narrowed the region harbouring a causative CCSD gene down to 5 Mb spanning from 39 to 44 Mb.
After scanning CFA31 we could not exclude CLDN14 for being responsible for the CCSD phenotype as microsatellite markers associated to CLDN14 indicated linkage.
However, to clarify the importance of CLDN14 in the CCSD phenotype, more SNPs have to be developed within the CLDN14 gene as well as in its flanking regions with the aim to find linkage disequilibrium for SNP markers.