Association of a Novel Mutation in the Retinol Dehydrogenase 12 (RDH12) Gene With Autosomal Dominant Retinitis Pigmentosa

Objective  To identify the gene causing retinitis pigmentosa (RP) in an autosomal dominant pedigree.

Methods  Family members with RP were studied with linkage analysis using single-nucleotide polymorphism and short tandem repeat polymorphic markers. Candidate genes in the linked region were evaluated with DNA sequencing.

Results  Nineteen family members had a mild form of RP. Multipoint linkage analysis of single-nucleotide polymorphism genotypes yielded a maximum nonparametric linkage score of 19.97 with markers located on chromosome 14q. LOD scores higher than 3.0 were obtained with 20 short tandem repeat polymorphic markers, and recombinants defined a 21.7-centimorgan locus on chromosome 14q. The retinol dehydrogenase 12 (RDH12) gene lies within this locus and was evaluated as a candidate gene. A frameshift mutation (776delG) was detected in all affected family members and was not detected in 158 control subjects.

Conclusions  Heterozygous mutations in RDH12can cause autosomal dominant RP with a late onset and relatively mild severity. This phenotype is dramatically different from the other disease associated with mutation in this gene, autosomal recessive Leber congenital amaurosis.

Clinical Relevance  The demonstration that mutations in a gene previously associated with recessive Leber congenital amaurosis can also cause dominant RP illustrates the wide phenotypic variability of retinal degeneration genes.

Retinitis pigmentosa (RP) is a collection of inherited, progressive retinal degenerations of the photoreceptors with typical clinical features including attenuated retinal arterioles, intraretinal bone spiculelike pigmentation, and posterior subcapsular cataract. Retinitis pigmentosa is characterized by marked reduction of both rod and cone responses in the electroretinogram, peripheral visual field defects, and reduction of central vision later in the course of the disease. The prevalence of RP is approximately 1 in 4000 and more than 1 million individuals may be affected with RP worldwide.¹

Details of the genetic features of RP have been recently reviewed.^1-3Retinitis pigmentosa may have an autosomal dominant (30%-40%), autosomal recessive (50%-60%), or X-linked (5%-15%) inheritance pattern.¹At present, 16 genetic loci for autosomal dominant RP (ADRP) have been identified and the genes at 14 of the loci have been discovered (RetNet: http://www.sph.uth.tmc.edu/Retnet/). The 14 known ADRP genes are CA4,^4,5CRX,^6,7FSCN2,⁸GUCA1B,⁹IMPDH1,^10-12NR2E3,¹³NRL,¹⁴PRPF3,^15,16PRPF8,^17,18PRPF31,^19,20RDS,^21,22RHO,^23-26ROM1,²⁷RP1,^28-31RP9,^32,33and SEMA4A.³⁴These genes have a range of functions, including phototransduction (RHO); RNA splicing (PRPF3, PRPF8, PRPF9, and PRPF31); signaling (SEMA4A); and retinal structure (RDS/peripherin, FSCN2, and RP1). Mutations associated with ADRP are most commonly detected in rhodopsin (RHO), RDS/peripherin, and PPRF31, which account for approximately 25%, 10%, and 8% of ADRP, respectively. The other genes are associated with smaller fractions of disease.³⁵Overall, mutations in these known disease-causing genes can be detected in nearly half of all ADRP cases, which suggests that many more ADRP genes remain to be identified.

There is considerable overlap between the inheritance patterns and the specific types of retinal dystrophies that are associated with mutations in a particular gene. For example, mutations in RHO,NRL, and RP1were initially associated with dominantly inherited RP,^{14,23-26,28-31}while different sets of mutations in these same genes were later shown to cause autosomal recessive RP.^36-38Similarly, some of the same genes that cause ADRP (CRX, IMPDH1, RDS, RHO, and SEMA4A) have also been associated with a number of other retinal phenotypes, including pattern dystrophy, Leber congenital amaurosis, cone dystrophy, and congenital stationary night blindness.^39-42Consequently, genes known to cause one retinal dystrophy are excellent candidates for causing others.

In this study, we report the genetic analysis of a 6-generation family from North Carolina with ADRP. The gene that causes ADRP in this family was mapped to chromosome 14q with linkage studies and recombination analysis. Family members were tested for disease-causing mutations in candidate genes contained within this new ADRP locus. A novel mutation in the retinol dehydrogenase 12 (RDH12) gene was detected that cosegregates with ADRP in this large pedigree. Mutations in RDH12have been previously associated with recessively inherited retinal dystrophies described clinically as early-onset retinal degeneration or Leber congenital amaurosis.^43-45However, this study presents the first case, to our knowledge, of ADRP associated with mutations in RDH12.

The research study was approved by the internal review board of the University of Iowa and informed consent was obtained from study participants.

Thirty-five family members had complete eye examinations and 19 were judged to have ADRP. Visual fields were assessed with Goldmann perimetry and ISCEV standard electroretinograms were obtained from a subset of family members. Patients were judged to be affected if they had classic signs of RP, including bone spiculelike pigmentation of the retina, attenuation of retinal arterioles, waxy pallor of the optic nerve, characteristic ring scotomas, and attenuated electroretinograms.

All patients (n = 273) and healthy control subjects (n = 158) were ascertained from the same outpatient ophthalmology clinic population at the University of Iowa. Subjects underwent complete eye examinations and were judged to be affected if they exhibited signs of a primary photoreceptor degeneration, including bone spiculelike pigmentation of the retina, reduced ISCEV standard electroretinogram amplitudes, and characteristic visual field defects. Control subjects had no clinical signs or family history of a retinal degeneration. Blood samples were obtained from study participants and DNA was prepared using a nonorganic method.⁴⁶

Pedigree members were first genotyped with short tandem repeat polymorphism (STRP) genetic markers flanking previously identified ADRP genes including rhodopsin (OMIM 180380), RDS/peripherin (OMIM 179605), and RP1(OMIM 603937). Genotyping with STRP genetic markers was conducted using standard methods as previously described.⁴⁷A genome-wide scan was next performed with Affymetrix microarrays (Sty1array of the GeneChip Human Mapping 500K Array Set, Affymetrix, Santa Clara, California), which interrogated 238 000 single-nucleotide polymorphisms (SNPs). Sample processing and labeling were performed using the manufacturer's instructions. The arrays were hybridized, washed, and scanned in the University of Iowa DNA core facility. Array images were processed with GeneChip DNA Analysis software.

Microarray data were analyzed and multipoint nonparametric linkage scores were calculated using the Genespring GT software package (Agilent Technologies, Palo Alto, California). Pairwise linkage analysis using STRP markers was performed with the MLINK and LODSCORE programs as implemented in the FASTLINK (v2.3) version^48,49of the LINKAGE software package.⁵⁰Penetrance and disease gene frequency were set to 99% and 0.1%, respectively. For each STRP marker, the allele frequencies were assumed to be equal. True allele frequencies could not be reliably estimated from the small number of spouses in the pedigree. To show that the assumption of the equal allele frequencies would not significantly affect our linkage results, we recalculated the LOD scores using allele frequencies for the “affected” allele of the most tightly linked marker (D14S587) ranging from 0.01 to 0.5. The Zmax for D14S587 was 4.5 when the affected allele frequency was arbitrarily set to 50%. In the 10 spouses who were studied, the actual frequencies of the affected alleles of D14S587 were much lower than 50%. In this small sample, the frequency of the affected allele of D14S587 was 10%, which provides additional evidence that our use of equal allele frequencies for D14S587 (11%) was reasonable.

Candidate genes were selected from among the genes in the chromosome 14q–linked interval based on their function, expression pattern, and prior association with retinal disease. DNA samples from 2 affected family members and from 2 healthy control subjects were tested for mutations in candidate genes using bidirectional sequencing of polymerase chain reaction products that encompassed the entire coding sequence. The first, and only, candidate to be evaluated was retinol dehydrogenase 12 (RDH12, OMIM 608830). Sequencing was performed using dye-terminator chemistry on an ABI 3730 DNA sequencer (Applied Biosystems, Foster City, California). Polymerase chain reaction amplification was performed with a standard protocol⁵¹using primer sequences that are available on request. Potential mutations were identified by comparing the DNA sequence of the affected family members and healthy control subjects. Similarly, the DNA sequences of the affected family members were compared with National Center for Biotechnology Information reference sequences (RDH12, NM_152443). Identified sequence variations were evaluated as potential disease-causing mutations using standard criteria.⁵²A single-strand conformation polymorphism (SSCP) assay was developed to detect the 776delG mutation in the control population (n = 158) with a standard protocol⁵¹using primer sequences that are available on request. The cohort of 273 patients with primary photoreceptor degenerations and 90 of the 158 healthy control subjects were tested for mutations in the entire RDH12gene using a combination of SSCP analysis and bidirectional sequencing using standard protocols.⁵¹

The frequency of RDH12variations detected in our cohort of patients with photoreceptor degeneration and our cohort of healthy control subjects was compared using the Fisher exact test for rare variants and χ²analysis for common variations. A 2-tailed Pvalue <.05 was considered statistically significant.

Members of a 6-generation family (pedigree 041D) received complete eye examinations and 19 family members were found to be clinically affected with RP. The disease in this family demonstrated an autosomal dominant mode of transmission through several generations (Figure 1). Family members had retinal findings typical of RP, including intraretinal bone spiculelike pigmentation and attenuation of retinal arterioles (Figure 2). Clinical information about the onset of disease was available from 4 of the affected family members. The average age at diagnosis in these family members was 28.5 years (range, 12-43 years). Some affected family members have maintained excellent central visual acuity (ie, 20/25 OU) and driving privileges into their eighth decade of life.

DNA samples from the family, including 19 affected members, were subsequently studied with linkage analysis using a stepwise approach. After linkage to several loci containing genes already associated with ADRP was excluded (data not shown), a genome-wide scan for linkage was conducted by genotyping DNA samples from 8 of the affected family members with microarrays of SNPs. Analysis of the SNP data identified a region of chromosome 14q with a maximum nonparametric multipoint linkage score of 19.97. All 8 affected pedigree members were found to share an allele of each of the 1350 consecutive SNPs that span 15.2 mega base pairs (Mbp) between rs4901408 and rs4902610.

The chromosome 14q linkage was confirmed by genotyping all 19 affected pedigree members with 26 STRP markers in this region (Figure 3). Two-point parametric LOD scores higher than 3.0 were obtained from 20 STRP markers and a maximum LOD score of 6.81 (θ = 0) was obtained with marker D14S587. The analysis of patients with recombination events near the linked interval is also shown in Figure 3. These recombination events indicate that the disease-causing gene lies within the 21.7-centimorgan (18.6-Mbp) interval between markers D14S1018 (telomeric) and D14S251(telomeric). This chromosome 14q locus contains 173 known genes.

One of the top candidate genes in the chromosome 14q locus is retinol dehydrogenase 12 (RDH12). RDH12is predominantly expressed in the photoreceptor cells of the retina⁵³and has an important role in the visual cycle.^44,45Mutations in this gene have been previously associated with autosomal recessive retinal dystrophies including Leber congenital amaurosis⁴³and early-onset retinal dystrophy,^44,45which share some clinical features of RP. Consequently, RDH12was the first candidate gene to be evaluated.

Two family members with RP were tested for disease-causing mutations in the coding sequence of RDH12using a DNA sequencing-based assay. A total of 3 DNA sequence variations were detected. Two variations are located within intervening sequences and are benign polymorphisms, while one variation causes a change in the predicted protein sequence encoded by RDH12. A heterozygous deletion was detected at position 2 of codon 259 (776delG), which causes a frameshift mutation and a premature termination at codon 277. The 776delG mutation was subsequently detected in all affected family members and was absent from 158 control subjects.

The conservation of the RDH12 protein sequence was examined to provide support for the pathogenicity of the 776delG mutation. Comparison with other homologous genes suggests that amino acids 37 to 240 are responsible for the dehydrogenase activity of RDH12.⁵⁴The 776delG mutation does not directly alter the dehydrogenase functional domain; however, it alters or eliminates the terminal 57 amino acids of RDH12, which are highly conserved (Figure 4).

To assess the role of RDH12in the pathogenesis of RP, we screened a panel of 273 patients with primary photoreceptor degenerations and 90 ethnically matched controls for disease-causing mutations using a combination of SSCP analysis and DNA sequencing. Patients were unselected for family history or for inheritance pattern of disease; however, patients with a diagnosis of Leber congenital amaurosis were excluded from this cohort. A total of 7 different RDH12variations were detected. Two variations were detected within intervening sequences, 1 variation was a synonymous codon change, and 4 variations were nonsynonymous codon changes (Table).

Multiple analyses were used to assess whether the 4 nonsynonymous coding sequence variations detected in RDH12were likely to be pathogenic. Analysis of coinheritance with disease was possible for the Glu260Asp variation. The patient harboring the Glu260Asp variation had affected family members available for study. However, the Glu260Asp variation was not coinherited with disease in this family (data not shown), suggesting that this variation is a benign polymorphism. Analysis of conservation of protein sequence was conducted using the blosum62 matrix.^52,55Some amino acid substitutions are tolerated without harm to protein function better than others. The blosum62 matrix was used to estimate the effects of the 4 nonsynonymous variations on the function of RDH12.Three of the 4 variations (Leu144Val, Arg161Gln, and Glu260Asp) cause changes in the amino acid sequence predicted by RDH12that are well tolerated by evolution, which is not indicative of disease-causing mutations. One variation (Ala126Glu) causes an amino acid substitution that is mildly supported by the blosum62 matrix as a disease-causing mutation. Finally, statistical analysis of these variations either individually or as a group failed to detect an association between the variations and disease (Table). One commonly detected variation (Arg161Gln) was observed at the same frequency in patients and controls (Pvalue = .91). This variation was similarly reported as a benign polymorphism in prior studies of RDH12.⁵⁶The other 3 nonsynonymous variations (Leu144Val, Ala126Glu, and Glu260Asp) were each detected only once in the cohort of patients and were not statistically associated with disease (Pvalue = >.99).

Fourteen genes associated with ADRP have been discovered, and in this article, we report the identification of another disease-causing gene using positional cloning and candidate gene screening techniques. Linkage studies of a large multiplex pedigree revealed a novel ADRP locus on chromosome 14q, which contains 173 known genes, including RDH12.

RDH12was considered the top candidate gene for causing ADRP in the chromosome 14q locus because of its function, expression pattern, and prior association with other retinal dystrophies. RDH12is predominantly expressed in the neurosensory retina⁵⁷and has an essential role in the conversion of all-transretinal to all-transretinol,⁴⁴which is an essential step in the visual cycle. Autosomal recessive mutations in RDH12have been associated with profound photoreceptor dysfunction and reduced visual function that is diagnosed at birth or in the first decade of life.^43,44Consequently, RDH12was the first gene we evaluated as the cause of RP in our pedigree.

Testing the family members for RDH12variations revealed a frameshift mutation (776delG) that causes premature termination of the translation of the RDH12transcript. Several lines of evidence suggest that this mutation causes ADRP in our family. First, the 776delG mutation cosegregates with disease in the family. Second, this mutation was not detected among 158 control subjects. Third, the 776delG mutation causes a truncation of the encoded RDH12 protein eliminating 57 amino acids from the conserved carboxy terminus. This mutation significantly alters the structure of the RDH12 protein and is likely to impair its function. Finally, mutations in RDH12have been previously associated with retinal degenerations. Taken together, these data strongly suggest that the 776delG mutation in RDH12causes ADRP in our pedigree. Functional studies of the 776delG mutation would be helpful to further establish its mechanism of action.

We additionally tested a large cohort of patients with primary retinal degenerations for disease-causing mutations in RDH12. No additional instances of the del776C mutation were detected; however, 3 other RDH12variations (Leu144Val, Ala126Glu, and Glu260Asp) were each detected once in our cohort of patients. These variations were analyzed for coinheritance with disease, alteration of conserved RDH12 protein sequence, and statistical evidence to support an association with disease. While it is possible that any of these 3 RDH12variations are rare causes of retinal degeneration, our study was unable to provide evidence for their pathogenicity.

The biologic events that lead to visual perception begin when light is absorbed by photoreceptors in the retina and triggers the isomerization of 11-cisretinal to all-transretinal. This reaction initiates transmission of visual information as a chemical signal. For continued signal transduction, all-transretinal must be recycled to 11-cisretinal (the visual cycle). RDH12plays an important role in this cycle by catalyzing the conversion of all-transretinal to all-transretinol. Both missense and truncating mutations in RDH12have been associated with early-onset autosomal recessive retinal dystrophies. Functional assays have shown that some of these RDH12mutations significantly reduce the enzymatic activity of the encoded protein,^44,56which implies that autosomal recessive retinal dystrophies and severe visual impairment are caused by loss of function mutations in RDH12. The 776delG mutation identified in our ADRP family is likely to cause disease via a mechanism that is different than that previously reported for RDH12mutations. This heterozygous mutation likely causes milder disease via a gain of function or dominant negative mechanism rather than loss of function. Some RDH12mutations cause severe and early-onset retinal dystrophy when 2 alleles are inherited, while a single 776delG allele is capable of causing a mild, late-onset form of disease. Truncating mutations, similar to 776delG, have been detected in each of RDH12's 7 exons. However, only the 776delG mutation has been associated with RP in the heterozygous state. Thus, the different behavior of these mutations does not appear to be due to their gross position within the RDH12gene. Further study of the mechanism by which the 776delG mutation causes disease may clarify the basis of RDH12genotype-phenotype correlations as well as provide valuable insight into the biology of the visual cycle and vision.

Correspondence:Edwin M. Stone, MD, PhD, Department of Ophthalmology, The University of Iowa Carver College of Medicine, Iowa City, IA 52242 (edwin-stone@uiowa.edu).

Submitted for Publication:August 13, 2007; final revision received January 24, 2008; accepted January 29, 2008.

Financial Disclosure:None reported.

Funding/Support:This work was supported by the Foundation Fighting Blindness, Research to Prevent Blindness, and the Grousbeck Family Foundation. Dr Fingert is supported by a Research to Prevent Blindness Career Development Award.