Introduction

Amelogenesis imperfecta (AI) is the collective term for an inherited group of dental diseases where the common clinical feature is an abnormality of tooth enamel. The enamel may be thin but otherwise normal (hypoplastic), or poorly mineralised (hypomineralised), or a combination of the two. AI can be inherited as an autosomal dominant, autosomal recessive or X-linked trait.1,2 Mutations have been identified in the amelogenin gene, AMGX (locus designated AIH1), on Xp22.1-22.3 in X-linked pedigrees.3 There is also a second locus on the X chromosome at Xq22-28 (AIH3).2,4 The only known locus for autosomal dominant AI to date (AIH2) is on chromosome 4q5 and the gene mutated in AI at this locus, enamelin, has recently been identified.6 Prior to this study, no locus had been reported for autosomal recessive AI, though pedigrees suggesting its existence have been documented.7,8 This mode of inheritance is particularly prevalent in Sweden, where autosomal recessive families are thought to account for 12% of cases in one study.7 Syndromic forms of AI can be associated with: epilepsy (MIM: 226750), platyspondyly (MIM: 601216), nephrocalcinosis (MIM: 204690) sensorineural hearing loss (MIM: 234580) cone-rod dystrophy (MIM: 217080) in the pedigree studied here and with keratin and bone abnormalities in tricho-dento-osseous syndrome (MIM: 190320).

Cone-rod dystrophies (CORD) are a group of retinal disorders characterised by degeneration of cone and subsequently rod photoreceptors. Clinically this is manifest as an initial loss of central vision, colour vision and photophobia with subsequent night blindness and restricted visual fields as rod involvement begins.9 Several genes have been implicated in autosomal dominant cone-rod dystrophy; namely peripherin/RDS,10 CRX,11 RETGC-112 and GUCA1A.13 A further locus exists; CORD7 on 6q11-q1514 for which the gene has yet to be identified. Loci for recessively inherited forms of cone-rod dystrophy are less well characterised. Four loci exist; these are CORD5 on 17p13-p12,15 CORD 8 on 1q12-q24,16 CORD9 on 8p12-q1117 and the ABCA4 gene on 1p21-p22.18

In this study we describe linkage analysis in a large consanguineous autosomal recessive Arab pedigree originating from the Gaza strip in which all 29 affected family members studied were found to suffer from both cone-rod dystrophy and amelogenesis imperfecta (Figure 1). Their phenotype is described in detail elsewhere19 and their visual complaints can be summarised as colour blindness, photophobia, nystagmus and poor visual acuity (beginning within the first 2 years of life). Older patients begin to show loss of the scotopic ERG, reduction in peripheral visual fields and optic disc pallor, characteristic of rod photoreceptor involvement. Fundus examination revealed an early bulls eye maculopathy with later chorioretinal atrophy at the macula producing a macular coloboma. AI was found to co-segregate perfectly with the retinal dystrophy; enamel was either absent or vestigial on erupted and unerupted teeth. Three teeth were examined microscopically and showed a generalised absence of enamel with scanty areas of full thickness enamel away from the occlusal surface.19

Figure 1
figure 1

The large inbred consanguineous recessive pedigree described in which recessive amelogenesis imperfecta and cone-rod dystrophy cosegregate. Affected individuals are represented by a filled symbol. Not all consanguineous loops are shown due to lack of information regarding inheritance, particularly via the female line. Haplotypes are shown from markers D2S417, D2S388, D2S2175, D2S2209, D2S2311, D2S2187, D2S1309, D2S373, D2S436 and D2S340, listed proximal to distal using the order given in the Marshfield integrated map.

Materials and Methods

Genomic DNA was extracted from peripheral blood leukocytes using standard techniques from 21 affected and 16 unaffected family members. Initial analysis was performed on pooled equimolar DNA solutions from affected individuals, and results were compared with those obtained in pooled carriers and unaffected siblings. These DNA mixes were genotyped by PCR using standard parameters and incorporating 32P-labelled dCTP. Fragments were resolved on 50 cm denaturing polyacrylamide gels, then visualised by autoradiography. After linkage had been established, individual fluorescently tagged polymorphic microsatellite markers from the region were amplified by PCR using genomic DNA from a cohort of 34 family members (Figure 1) and the products were resolved using an ABI 377 automated DNA sequencer (Applied Biosystems). Products were identified and sized using GENESCAN (version 2.0.2) and GENOTYPER (version 1.1.1) software. A Lod score was obtained using the CYRILLIC 2.1 data management package and the MLINK program from the LINKAGE package (VERSION 5.1) software.20 Penetrance was assumed to be 100%.

Exons of the CNGA3 gene were screened in this pedigree by direct sequencing of both DNA strands in two affected family members. Primers for sequencing were designed to include a minimum of 20bp of intronic sequence flanking each exon. Exons were PCR amplified, purified using the Qiaquick PCR purification kit (Qiagen) then sequenced directly using the ABI Prism BigDye Terminator cycle sequencing kit from Applied Biosystems.

Results

DNAs consisting of pooled affected individuals and carriers/normal siblings were screened for linkage using 121 polymorphic microsatellite markers closely linked to known retinal dystrophy loci and a further three markers at the known autosomal AI locus (AIH2). The retinal dystrophy loci tested were those reported up to 1996, and included seven dominant and seven recessive RP loci, eight central retinal dystrophy loci, five for Ushers syndrome, four for Bardet Biedl syndrome and eight other loci at which mutations cause an assortment of retinal phenotypes. Subsequently a locus on chromosome 2q11 was implicated in achromatopsia, another retinal disease.21 We noted the clinical similarities between the retinal phenotype in the Gaza pedigree studied herein and this disease. Achromatopsia (total colour blindness; MIM: 216900) is an autosomal recessive retinal disorder in which patients are unable to differentiate colours. They are photophobic and have a markedly reduced level of daytime visual acuity from early childhood together with nystagmus. All of these features are prominent in the phenotype studied here. In addition however, members of this pedigree show progressive retinal degeneration with involvement of both rods and cones. The marker D2S417 on chromosome 2q, known to be closely linked to the achromatopsia locus, was therefore tested on pooled DNAs from affected and normal individuals, and this analysis revealed excess allele sharing.

We genotyped further markers spanning the achromatopsia locus on 2q11 (ACHM2) in a panel of individual family members. Polymorphic microsatellite markers D2S417, D2S388, D2S2175, D2S2209, D2S2311, D2S2187, D2S1309, D2S373 D2S436 and D2S340 were amplified by PCR using genomic DNA from a cohort of 34 family members. A maximum Lod score of 7.03 was obtained for marker D2S2187. Genotyping of affected individuals revealed a region of homozygosity of approximately 2 cM (5 Mb) flanked by markers D2S2209 and D2S373, as shown in Figure 1. This result is the first report of a locus for autosomal recessive AI and is also confirmation of a further locus for autosomal recessive CORD.

Additional genotyping was performed for markers D2S2187 and D2S1309 on four other consanguineous recessive pedigrees in which AI was segregating in the absence of CORD. No homozygosity was seen for any pedigree with either marker. It is therefore unlikely that AI in any of these pedigrees results from mutations at this locus.

In view of the clinical similarities between achromatopsia and CORD in this pedigree, and in the light of a recent report of CNGA3 mutations in patients with progressive central retinal dystrophies22, CNGA3 appeared a strong candidate gene for a mutation causing the retinal phenotype in this pedigree. CNGA3 encodes the alpha subunit of a cyclic GMP-gated channel present in the plasma membrane of the cone-photoreceptor outer segment. This cyclic GMP-gated channel is responsible for the influx of sodium and calcium ions as a response to light and represents a key element of the phototransduction cascade in cone photoreceptors.23 The genomic sequence of CNGA3 is contained within the genomic clone AC027241 along with marker D2S2311 (Goldenpath), one of the markers contained in the region of homozygosity in our pedigree, providing good evidence for its consideration as a positional candidate. All seven exons of this gene, together with the two new exons (Exons 0 and 2b) described recently22 were screened by direct sequencing of both DNA strands in two affected family members. A polymorphism was found in the 5′UTR which appeared to segregate with disease (84014A>T, reference sequence AC092675). The existence of a SNP at this site had been previously described by B Wissinger et al (personal communication). In addition two new variants were found within the 5′UTR although neither segregated with disease.

In order to further ensure the accuracy of the sequencing data, the entire CNGA3 gene was then rescreened independently using direct sequencing in a second lab (Tuebingen). This analysis confirmed the absence of causal mutations anywhere within the coding sequence or within the 5′ region surrounding exon 0, including a putative promoter site, located approximately 24 kb upstream of the start codon.

Discussion

These data demonstrate the existence of a locus for autosomal recessive AI and CORD on chromosome 2q11. In addition this study reveals locus heterogeneity in autosomal recessive AI, since this locus is excluded in four smaller consanguineous AI pedigrees. However two questions remain. Firstly, could the CNGA3 gene still be involved in the retinal phenotype seen in this pedigree? Though mutations in the coding sequence, splice sites and putative promoter have been excluded, an intronic change or large-scale rearrangement within the gene or in nearby sequence remains a possibility. Also large-scale genomic changes up to 1 Mb away from the gene in question have been implicated in inherited human diseases, altering the level of transcription of the gene with a so-called position effect.24 Such an effect might alter the expression of several genes in the vicinity and could therefore account for both tooth and retinal phenotypes. An alternative hypothesis, that mutations in a second gene within this relatively small interval could result in such a similar retinal phenotype to that described for CNGA3 mutations seems unlikely. A similar hypothesis was proposed to explain the lack of mutations in the RPGR gene in RP3 linked, retinitis pigmentosa families.25 This was subsequently disproved with the discovery of a mutational hotspot in a previously unidentified RPGR exon.26

Secondly, it remains to be determined whether the two disparate phenotypes in this pedigree result from a defect in a single gene or from defects in two closely linked genes. In the light of this question it is interesting to note that expression of CNGA3 is not confined to cone photoreceptors, with evidence of expression in bovine kidney, spermatozoa, heart and colon.27 Identification of genes expressed during enamel formation is far from complete and so it is a matter of speculation as to whether expression of CNGA3 occurs. There is however evidence for transcellular active transport of calcium through ameloblasts, the cells involved in the mineralisation of enamel.28,29,30

A further 23 characterised or predicted genes map within the interval predicted to contain the mutation(s) in this pedigree (Goldenpath) defined by haplotype analysis. Of these, none are retina specific, four are hypothetical proteins with no known function and four others are ubiquitously expressed, including expression in brain and/or eye. No genes are documented as having expression in teeth. It remains possible that these could be involved in one or both of the phenotypes seen in this family.

Analysis of the contig available for this region using the human genome browser (UCSC: Goldenpath) shows a discrepancy for the position of the marker D2S2209. Haplotype reconstruction in this pedigree reveals a proximal crossover in this marker, which correlates well with its position on the Marshfield genetic map where it is placed proximal to D2S2175 but distal to D2S2311. Its position on the contig is however distal to D2S1309. This marker order depends upon the arrangement of clones bridging a gap in genomic sequence, which are linked together by the gene LAF4. Until the finished draft of the sequence is available it will be impossible to be certain of the true marker order for the region. However if the Goldenpath marker order were confirmed, the locus would be further refined to a 2Mb interval between markers D2S2209 and D2S373, and could potentially exclude CNGA3 from the disease interval.

In summary, we have identified a new locus on chromosome 2q11 at which recessive CORD and AI cosegregate. Further experiments will be required to determine whether this represents two distinct but closely linked phenotypes or a new syndromic locus where both CORD and AI result from mutations in the same gene. The CNGA3 gene remains a strong candidate for involvement in the retinal phenotype although it has been sequenced in its entirety in this pedigree and no mutations were identified.