Novel PRPF31 and PRPH2 Mutations and Co-occurrence of PRPF31 and RHO Mutations in Chinese Patients With Retinitis PigmentosaNovel PRPF31 and PRPH2 Mutations in RP FREE

Retinitis pigmentosa (RP) is a group of heterogeneous inherited eye diseases characterized by progressive degeneration of photoreceptors that eventually leads to blindness. Most patients experience night blindness first, followed by loss of peripheral visual field and central vision at a later stage. It can be inherited in autosomal dominant (AD), autosomal recessive, or X-linked modes, whereas the simplex form lacks a family history.¹ It is the most common inherited eye disease, with an overall prevalence of approximately 1:4000.¹ More than 40 genes are associated with RP, and most of them have been cloned.^{1 - 2} Herein, we report the mutational analysis of 3 RP genes (PRPF31, RHO,and PRPH2) performed across several years in a cohort of Chinese patients with RP who had never been screened for mutations in any gene implicated in RP.

The PRP31 pre–messenger RNA processing factor 31 homologue (Saccharomyces cerevisiae) gene (PRPF31; OMIM 606419 and HGNC15446) has 14 exons and encodes a protein of 499 amino acids. The 61-kDa protein is required for formation of the U4/U6·U5 tri-small nuclear ribonucleoprotein spliceosome complex, which serves to ensure accurate and efficient splicing of pre–messenger RNA.³ Mutations in PRPF31that cause RP, first reported in 2001,⁴ result in only a retina-specific phenotype, although the gene is ubiquitously expressed.⁴

The rhodopsin (RHO) gene is a highly specialized G protein–coupled receptor that composes approximately 70% of the total protein in the outer segment of rod photoreceptors.⁵ It has 348 amino acids, consists of 7 transmembrane domains that form a compact bundle to hold the 11-cis retinal chromophore via Schiff base linkage, and is responsible for the activation of phototransduction.⁶ The first RHOmutation that caused RP was reported in 1990.⁷

The peripherin 2 gene (PRPH2; 179605 and HGNC9942), formerly known as the retinal degeneration slow gene (RDS), consists of 3 exons and encodes a 39-kDa integral membrane glycoprotein with 346 amino acids. The protein includes 4 transmembrane domains (M1-M4) and a large intradiscal domain (known as the D₂loop) and is located at the outer segment discs of the rod and cone photoreceptors.^{8 - 9} It forms a homo-oligomeric structure with itself and a mixture of homotetrameric and heterotetrameric complexes with another membrane protein (ROM1).¹⁰ This structure is important in disc morphogenesis and stabilization. Mutations in PRPH2that cause RP were first described in 1991.^{11 - 12}

Chinese patients with RP were recruited on a voluntary basis from Retina Hong Kong (http://www.retina.org.hk/) as part of the Hong Kong Patients' Register of Retinal Degeneration, and the procedures and examinations involved have been reported previously.¹³ The diagnosis of RP was based on the presence of night blindness, typical fundus findings (characteristic retinal pigmentation, vessel attenuation, optic atrophy, and severe loss of peripheral visual field), and abnormal electroretinogram findings (usually the complete absence of an electroretinographic response, ie, no a and b waves) in both eyes. The study adhered to the tenets of the Declaration of Helsinki, and written informed consent was obtained from all the participants.

Blood samples were taken from patients with RP and their available family members for mutational analysis. Seventy-five anonymous blood samples from healthy Chinese donors were obtained from the Hong Kong Red Cross Blood Transfusion Service and served as normal samples for establishing the allele frequencies of sequence variations in the general population. A modified salting-out method was used for DNA extraction.¹³

Primers were designed to amplify the exons of the PRPF31, RHO,and PRPH2genes and their flanking regions by means of polymerase chain reaction (PCR). The PCR was performed in a 25-μL mixture containing 1 × PCR buffer II, 50 ng of human genomic DNA, 0.2mM concentration of each deoxyribonucleoside triphosphate, and 0.5 U of AmpliTaq Gold polymerase (Applied Biosystems, Foster City, California). The concentrations of primers (0.1μM, 0.3μM, or 0.5μM) and magnesium chloride (1.5mM or 2.5mM) were specific to each fragment. Amplification was performed in the 96-well GeneAmp PCR System 9700 (Applied Biosystems) and using the same single PCR condition for all fragments. The PCR was initiated by heating at 95°C for 10 minutes, followed by 38 cycles of 20 seconds at 95°C, 25 seconds at 60°C, and 25 seconds at 72°C, plus a final extension of 10 minutes at 72°C.

Two microliters each of a pair of PCR fragments (1 approximately 200 base pair [bp] and 1 approximately 300 bp) were mixed with 5 μL of single-stranded conformation polymorphism (SSCP) loading buffer (95% formamide, 20mM disodium salt of EDTA, 0.05% bromophenol blue, and 0.05% xylene cyanol FF) and 1 μL of water. Two PCR fragments were loaded per lane to double the throughput. The SSCP analysis was performed as described previously, with every single PCR fragment being analyzed under 4 different combinations of conditions to ensure complete detection of sequence variations: gel without glycerol/4°C, gel with glycerol/4°C, gel without glycerol/20°C, and gel with glycerol/20°C.¹⁴ One modification was that electrophoresis was run at 400 V for 3 hours. The SSCP was also used to genotype normal DNA samples to establish the allele frequencies of sequence variations in the general population and was performed using 1 of the 4 conditions that gave the most distinctive SSCP pattern.

Representative samples showing variant SSCP banding patterns under any 1 condition mentioned in the previous subsection were sequenced. DNA was amplified using the same primers and conditions as above. The PCR products were purified by means of enzyme digestion and were sequenced using the ABI PRISM BigDye Terminators v2.0 Cycle Sequencing Ready Mix Kit (Applied Biosystems), as described by Yip et al.¹⁵

The PCR fragment containing the complex base changes in exon 1 of the PRPH2gene was cloned into the pCR2.1-TOPO vector (Invitrogen Corporation; Carlsbad, California) according to the instructions of the manufacturer. Plasmid DNA from recombinant clones was extracted using the NucleoSpin Plasmid DNA extraction kit (Macherey-Nagel GmbH & Co, KG; Düren, Germany). DNA cycle sequencing was performed using the M13 forward primer.

Seventy-six patients with RP from 54 families were recruited into this study. The inheritance modes were AD in 9 families, autosomal recessive in 8 families, and simplex in 31 families but were uncertain in 6 families. All DNA sequence changes were numbered with reference to the first translated base (+1) of the genes and were described according to standard nomenclature and its subsequent online revisions.^{16 - 17} In this study, 3 pathogenic mutations were identified together with 32 other nonpathogenic sequence changes (Table 1). Of these, 2 mutations and 13 nonpathogenic sequence changes were novel.

In a family (C8289) that showed a typical AD inheritance pattern (Figure 1), 8 women were diagnosed as having RP on eye examination (Table 2), and 3 other patients (I:1, I:2, and II:1), not available for study, were reported by other family members to have RP. In the second generation, the youngest brother (II:14) died 1 week after birth; thus, his disease status was unknown. Of the 5 affected females with DNA available for study, mutational screening showed that 4 (II:3, II:8, II:11, and III:3) were heterozygous for the novel PRPF31mutation c.785delT (p.Phe262SerfsX59), and 2 (II:10 and II:11) were heterozygous for the RHOmutation c.1040C>T (p.Pro347Leu) (Figure 1and Figure 2). The p.Pro347Leu mutation was first reported in white patients.¹⁸ Note that patient II:11 harbored both mutations. The PRPF31c.785delT mutation created a frameshift leading to a premature stop codon, which is 59 codons downstream (counting from the first altered codon), and was, thus, expected to produce a truncated protein of 319 amino acids, if any.

A novel complex PRPH2mutation (c.232G>C;c.232_233insT) was found in 1 male patient with simplex RP but not in control subjects (Table 1). Base position 232 is the first base of codon 78. On sequencing of cloned inserts, these 2 base changes were found to be on the same chromosome (Figure 3) and, thus, changed codon 78 from GCT (alanine) to CTC (leucine) and created a frameshift (p.Ala78LeufsX99). The frameshift was predicted to create a premature stop of translation 99 codons downstream and to produce a truncated protein of 175 amino acids. In other words, the patient with RP was heterozygous for this complex PRPH2mutation. He started to experience night blindness at age 25 years and was first diagnosed as having RP at age 32 years. He was recruited into our register at age 64 years.

Of the 32 nonpathogenic sequence changes identified in this study (Table 1), 19 were already documented in the dbSNP database (http://www.ncbi.nlm.nih.gov/SNP/) or were reported previously in the literature.^{18 - 26} The minor allele frequencies of these known sequence changes varied in the general Chinese population (from 0.00 to 0.42). In particular, the c.625G>A (p.Val209Met) sequence change of the RHOgene did not cosegregate with the RP phenotype in the family concerned and was not found in a cohort of 75 healthy individuals.

The remaining 13 sequence changes were novel (Table 1). Seven of them did not cosegregate with the RP phenotype although they were found at low frequencies in the cohort of RP probands and not in the general population: c.5C>G, c.697 + 29_30insC, c.1074-38A>G, c.1098G>C, and c.1216C>T of PRPF31and c.464C>G and c.946T>G of PRPH2. Two were present at similar, although low, frequencies in RP probands and the general population: c.1074-35C>T and c.*172C>T of PRPF31. One sequence change was a synonymous substitution (c.138T>C of PRPF31), and another (c.-201C>T of RHO) was found in the noncoding region, although both were found at low frequencies in RP probands but not in the general population. The remaining 2 sequence changes (c.855 + 40G>A of PRPF31and c.-300_-302delTTT of RHO) were found in the noncoding region only in the general population when another sequence change in the same PCR fragment was being genotyped in the cohort of healthy individuals to determine its frequency.

More than 10 frameshift mutations in the PRPF31gene have been reported and predicted to give rise to premature termination codons and to produce truncated proteins.² In the present study, the c.785delT (Figure 2) of PRPF31is a novel frameshift mutation that produces a predicted truncated protein of 319 amino acids, with the first changed amino acid serine at position 262 (p.Phe262SerfsX59). One reported mutation produced a frameshift at a nearby position (codon 253) and a predicted truncated protein of similar length (316 amino acids): c.758_767del (p.Gly253AlafsX65).²⁷ For PRPF31mutations that give rise to premature termination codons, it has been found that reduced mutant transcripts were produced as a result of nonsense-mediated messenger RNA decay,^{28 - 29} and no truncated proteins could be detected, probably because of nonsense-medicated translational repression.²⁹ Thus, the pathogenic effect of such PRPF31mutations is likely due to haploinsufficiency rather than to gaining of function.^{28 - 30} It is tempting to predict that the new mutation c.785delT also produces a reduced amount of the mutant transcript and probably none of the truncated protein. Even if any truncated protein is produced, it is likely that it cannot be translocated to the nucleus to perform its function because the truncated protein is predicted to be lacking the nuclear localization signal (RKKRGGRRYRKMKE), which is located between residues 351 and 364 of the wild-type protein.³⁰ Overall, the new mutation is predicted to behave like a null allele and to cause ADRP due to haploinsufficiency.

Note that this PRPF31mutation is amenable to detection by use of a PCR-based method followed by DNA sequencing. However, such PCR-based methods cannot detect gross genomic rearrangements, such as large deletions, which were found by the use of multiplex ligation–dependent probe amplification to account for 2.5% of ARDP families in a recent American study.³¹ Thus, the frequency (1 in 9 families) of PRPF31mutations in ADRP was likely to be an underestimate.

The p.Pro347Leu substitution of RHOis a known mutation that causes ADRP.¹⁸ Proline 347 is the penultimate amino acid located at the carboxyterminal tail of the intracellular domain of RHO. Mutations in the carboxyterminal tail are known to affect the RHOtrafficking from rod inner segment to rod outer segment and, hence, disc renewal.^{32 - 33} This induces the outer segment to shorten and triggers apoptosis. The rate of cell death correlates with the amount of accumulated mutant protein.³⁴ In general, carboxyterminal mutations produce a more severe disease phenotype than do RHOmutations at other locations.^{35 - 36} This study found 1 RHOmutation in 1 (11.1%; 95% confidence interval, 2.0%-43.5%) of 9 Chinese families showing AD inheritance, whereas RHOmutations are found in approximately 25% of white families showing an AD pattern.¹

Mutations in the PRPF31and RHOgenes can independently result in ADRP. Of 9 ADRP families studied, 1 (11.1%) was found to harbor mutations in both genes (Figures 1and 2and Table 2). The mutation in either gene alone did not seem to cosegregate with the RP phenotype in this particular family. Cosegregation was apparent only when mutations in both genes were considered together: 3 (II:3, II:8, and III:3) of the 5 affected members tested carried only the heterozygous PRPF31mutation, 1 (II:10) harbored only the heterozygous RHOmutation, and 1 (II:11) was heterozygous for both mutations (Table 2). Although the grandfather (I:1) was deceased and the grandmother (I:2) was not available for study, they were reported to have RP too. As such, 1 grandparent is expected to be heterozygous for the PRPF31mutation and the other heterozygous for the RHOmutation. Age at onset of night blindness was younger for the 3 affected members with only the PRPF31mutation than for the 1 with only the RHOmutation (≤10 vs 15 years) (Table 2).

Another interesting finding is the complex PRPH2base changes (+232G>C; +232_+233insT) identified in 1 male patient with simplex RP but not in any of the control samples (Table 1). The PCR cloning showed that these 2 base changes were on the same chromosome (Figure 3). Thus, the patient was heterozygous for the wild-type allele and the complex mutation. If +232G>C was the only missense substitution, the amino acid would change from alanine (GCT) to proline (CCT) at codon 78. However, the single T insertion following the missense substitution changed the amino acid to leucine (CTC) and led to a frameshift downstream. Although the effect of Ala78Leu and the following missense substitutions could not be simply predicted from this study, a truncated protein would be expected from the frameshift mutation. The truncated protein was predicted to have 175 amino acids and was, thus, only approximately half the size of the functional full-length protein (346 amino acids). The truncated protein would lack the C-terminus, the transmembrane domain M4, and most of the intradiscal D₂loop in addition to an altered amino acid sequence from residues 78 to 175.

More than 100 mutations in the PRPH2gene have been identified that cause different types of retinal dystrophies, including RP.^{2 ,37} Approximately 65% of these mutations are found in the D₂loop, which makes up only 41% of the full-length protein. The importance of the D₂loop in the folding and subunit assembly of the peripherin-2 protein is also supported by experimental studies.^{38 - 40} More than 25 truncating mutations have been reported in the coding regions of the PRPH2gene, including nonsense substitutions and frameshifts due to deletions, insertions, and indels.^{2 ,37} Thus, it is likely that the complex frameshift base change identified in this study is a pathogenic mutation and causes the RP phenotype in the male patient. Thus, PRPH2mutation accounted for 11.1% (1 in 9 families; 95% confidence interval, 2.0%-43.5%) in this cohort of Chinese families with ADRP. In comparison, the proportions of ADRP due to PRPH2mutations were 3% to 9% in populations with northern European ancestry but less in other populations (southern European and Japanese).³⁷

In this study, 9 nonpathogenic sequence changes identified are missense substitutions: 2 in PRPF31, 2 in RHO,and 5 in PRPH2(Table 1). Already reported as polymorphisms in previous studies,²⁴ 2 PRPH2changes (p.Gln304Glu and p.Gly338Asp) were also common in patients with RP and healthy individuals in this study. Two missense alterations (p.Ala299Ser in RHOand p.Gly170Ser in PRPH2) were not found in any patients with RP but were each found in a heterozygous state in 1 (of 75) healthy individuals in this study; p.Ala299Ser of RHOwas also reported as a rare polymorphism in another study,²¹ and p.Gly170Ser of PRPH2was first reported as a mutation that causes cone-rod dystrophy.²³ The p.Val209Met change of RHOwas first identified in a patient with simplex RP, with cosegregation remaining to be determined,¹⁹ whereas no cosegregation was demonstrated in the present study. The remaining 4 missense changes were novel and did not cosegregate with the RP phenotype in the families concerned: p.Ser2Cys and p.Arg406Cys in PRPF31and p.thr155Ser and p.Trp316Gly in PRPH2(Table 1). We also assessed the potential functional consequences of these 9 missense changes using the online bioinformatics tool PolyPhen (http://coot.embl.de/PolyPhen/).⁴¹ PolyPhen is used to predict the effects of amino acid substitutions on the basis of the sequence, phylogenetic, and structural information that characterize the substitution.⁴² All were predicted to be benign (most likely lacking any phenotypic effect) except 2. The 2 exceptions were p.Arg406Cys of PRPF31and p.Trp316Gly of PRPH2, which were predicted to be “probably damaging” (supposed with high confidence to affect protein function or structure), the same prediction for the known RHOmutation p.Pro347Leu. The discrepancy between the observation of no cosegregation and the prediction of damaging effect remains to be explained, and further experimental work remains to be performed to resolve the issue.

In conclusion, this study investigated 76 patients with RP from 54 Chinese families and identified 3 pathogenic mutations, 1 each in the PRPPF31, RHO,and PRPH2genes, in 2 of 9 families showing an AD inheritance pattern. Two of these mutations are novel. One family harbored mutations in PRPF31and RHO,with 1 affected member being heterozygous for both mutations and with a later age at onset of night blindness.

Correspondence: Shea Ping Yip, PhD, Department of Health Technology and Informatics, The Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong, China (shea.ping.yip@polyu.edu.hk).

Submitted for Publication: August 6, 2008; final revision received November 10, 2008; accepted November 16, 2008.

Author Contributions: Dr Yip had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Financial Disclosure: None reported.

Funding/Support: This study was supported by grant G-V938 from The Hong Kong Polytechnic University, and the Hong Kong RP Patients' Register of Retinal Degeneration was established with financial support (HCPF HF-Z40) from the Health Care and Promotion Fund of the Hong Kong SAR Government. Ms Cheung was also partly supported by a Dean Reserve Fund (Faculty of Health and Social Sciences) from The Hong Kong Polytechnic University.

Additional Contributions: We thank all the families participating in the study.