X-linked retinitis pigmentosa (XLRP) is a genetically heterogeneous group of progressive retinal degenerations. The disease process is initiated by premature apoptosis of rod photoreceptor cells in the retina, which leads to reduced visual acuity and, eventually, complete blindness. Mutations in the retinitis pigmen-tosa GTPase regulator (RPGR), a ubiquitously expressed gene at the RP3 locus in Xp21.1, accountfor ∼20% of all X-linked cases. We have analysed the expression of this gene by northern blot hybridization, cDNA library screening and RT-PCR in various organs from mouse and man. These studies revealed at least 12 alternatively spliced isoforms. Some of the transcripts are tissue specific and contain novel exons, which elongate or truncate the previously reported open reading frame of the mouse and human RPGR gene. One of the newly identified exons is expressed exclusively in the human retina and mouse eye and contains a premature stop codon. The deduced polypeptide lacks 169 amino acids from the C-terminus of the ubiquitously expressed variant, including an isoprenylation site. Moreover, this exon was found to be deleted in a family with XLRP. Our results indicate tissue-dependent regulation of alternative splicing of RPGR in mouse and man. The discovery of a retina-specific transcript may explain why phenotypic abberations in RP3 are confined to the eye.
Retinitis pigmentosa (RP) describes a clinically heterogeneous group of retinal dystrophies characterized by premature death of the photoreceptor cells in the human retina. Apoptotic cell death starts in the periphery and progresses towards the centre during the third to fourth decade of life. Rod photoreceptors are affected first but, with disease progression, cones degenerate as well. The ocular symptoms include early night blindness, progressive constriction of the visual field, tunnel vision, reduced visual acuity and, eventually, complete blindness. Upon fundoscopic examination, patients show characteristic bone spicule-like pigment deposits and attenuated blood vessels. RP occurs with a frequency of ∼1 in 4000 individuals, and molecular genetic studies have revealed a remarkable locus heterogeneity (http://www.sph.uth.tmc.edu/Ret-Net/disease.html).
About 10–25% of the familial cases are X-linked, and linkage studies have defined at least five distinct loci, four in Xp (RP2, RP3, RP6 and RP15) and one in Xq (RP24) (1–5). Most of the families showed linkage to the RP2 and RP3 intervals and, recently, both genes were isolated by positional cloning (6–9). The RP2 gene is located at Xp11.3. Mutations in this gene were found in 10–18% of the families with X-linked RP (9–11), which is in good agreement with linkage data.
The RP3 gene resides in Xp21.1, contains 19 exons and codes for a polypeptide of 815 amino acids. The N-terminal moiety of the deduced protein is homologous to the regulator of chromosome condensation (RCC1), a guanine nucleotide exchange factor of the small nuclear GTPase Ran (12). Therefore, the RP3 gene was designated RPGR (retinitis pigmentosa GTPase regulator) (6). In exon 19, at the very C-terminus of RPGR, an isoprenylation site is present. This motif is responsible for targeting the protein to the Golgi complex, as shown by transient expression of the mouse Rpgr cDNA in COS cells (13). Golgi localization was abolished when the isoprenylation site was deleted or mutated. The precise cellular and subcellular distribution of RPGR in the retina remains to be elucidated. The mutation spectrum in patients comprises missense, nonsense and splice site mutations, but also deletions removing one or several exons (6,8,14–21). Remarkably, the vast majority of mutations affect the RCC1-homologous domain (RHD) of RPGR. This RHD was shown in the yeast two-hybrid system to interact with the delta subunit of the rod cGMP phosphodiesterase (PDEδ) (22). RPGR molecules carrying RP-associated missense mutations in the RHD showed significantly reduced binding affinities in vitro (22). PDEδ is able to solubilize the membrane-bound form of the rod phosphodiesterase (23) and this process may be regulated by RPGR. So far, RPGR mutations had been detected in only 20% of all X-linked RP cases (6,8,16), which represents a much lower frequency than estimated by linkage analyses (60–80%). The paucity of mutations, even in families with linkage to the RP3 interval, may be explained by: (i) locus heterogeneity for RP3, i.e. the presence of another as yet unidentified RP gene in the vicinity of RPGR; (ii) mutations in regulatory elements which affect gene expression; and (iii) not yet identified RPGR exons, which are targets of mutations in a significant number of patients.
Here we describe a detailed analysis of RPGR transcription in mouse and human tissues. In both species, the gene is characterized by an extraordinary degree of alternative splicing. Several tissue-specific splice variants are reported, one of which contains a novel retina-specific exon. The corresponding transcript carries a premature stop codon upstream of the isoprenylation site. Moreover, this exon is deleted in a patient with typical symptoms of the disease, emphasizing the functional relevance of the retina-specific RPGR transcript.
Isolation and expression pattern of the mouse Rpgr gene (mRpgr)
Initially, a 330 bp fragment of the mouse Rpgr cDNA was generated by low stringent PCR on a newborn mouse brain cDNA library (Stratagene) with primers corresponding to exons 3 and 6 of the human RPGR gene. Subsequent hybridization of this fragment to 800 000 clones from a newborn mouse brain library identified 10 clones that contained different parts of mRpgr. Sequence comparison with its human counterpart revealed that the isolated mouse cDNA represents an alternative isoform where exon 13 was spliced to exon 16. The human and mouse sequences have an average base pair identity of 76.4%, which is higher in the 5′ portion than near the 3′ end of the cDNA (e.g. 90.3% for exon 3 and 65.3% for exon 16). Hybridization of a full-length clone to a multiple tissue northern blot detected a ubiquitous expression pattern of mRpgr (Fig. 1A). The main transcript was observed at ∼2.7 kb, except in testis, where the most prominent band appeared at 3.6 kb. Additionally, signals with higher molecular weights were present in all tissues. In the eye and brain, for example, novel transcripts of ∼5, 9.5 and 20 kb were detected. The strongest mRpgr expression is observed in testis, while transcription in the eye is comparatively low.
In summary, our RT-PCR and northern blot experiments have revealed tissue-specific isoforms of mRpgr. In order to characterize them in more detail, we have performed RT-PCR analysis of RNA from several mouse and human tissues.
Intron 14 of RPGR is used as exon in mouse tissues
The whole open reading frame (ORF) of mRpgr was amplified from mouse brain, testis and eye cDNAs by long-range PCR employing forward and reverse primers specific for exons 1 and 19, respectively. PCR products from eye and brain revealed a major band at ∼2.1 kb, which corresponds to the size of the mRpgr transcript isolated from the mouse brain library described above. In testis, an additional prominent band of ∼3 kb was observed (data not shown). Furthermore, weak smaller bands of 0.9 and 1.5–1.6 kb were present in testis and brain. These bands were enriched by excision from agarose gel and subsequent re-amplification. Cloning of the PCR products and sequencing of individual clones revealed 10 alternative isoforms of the mRpgr cDNA, summarized in Figure 2A. The small PCR products found in testis and brain correspond to splice variants that lack large parts of the cDNA (Fig. 2A, isoforms 7–10). The 2.1 kb band represents three different isoforms of similar size. They all share splicing of exons 13–16, but contain alternative stop codons (Fig. 2A, isoforms 4–6). Skipping of exon 18 or exon 19a leads to a shift in the reading frame and, thus, to the usage of premature stops. Exon 19a representsthe first 30 bp of exon 19 which can be used as a separate exon in the mouse.
Sequence analysis of the 3 kb PCR product cloned from testis cDNA identified a novel exon, located between exons 14 and 15 and designated 14a [Fig. 2A, isoforms 1 (EMBL accession no. AJ238396), 2 and 3]. This exon comprises 546 bp and contains an ORF. Sequence comparison with the human RPGR gene showed that mouse exon 14a is 73% homologous to human intron 14 at the nucleotide level. A database search revealed the same identity to a partial RPGR cDNA sequence of the dog (GenBank accession no. AF030561). The testis isoform with exons 14, 14a and 15 exists in three variants with different stop codons generated by skipping of either exon 18 or 19a. Hybridization of a mouse multiple tissue northern blot with exon 14a as probe gave rise to a 3.6 kb testis-specific signal, indicating that the isoform containing exon 14a is prevalent in mouse testis (Fig. 1B). However, by RT-PCR with primer combinations in exons 14a and 13 or 16, expression of 14a was also detectable in eye and brain (data not shown). In these tissues, exon 14a may be expressed at lower levels ins ufficient for detection by northern blot hybridization.
The finding of alternative splicing and of a novel exon of mRpgr led us to perform long-range PCR with primers in exons 3 and 19 on human testis and retina cDNA. In contrast to mouse, only the 2.1 kb band was seen in both tissues, but not the 3 kb product. Additionally, a weak smaller band of 1.9 kb was present in human testis. Sequencing of individual clones derived from testis RNA revealed two alternative transcripts (Fig. 3, isoforms 1 and 2). The 1.9 kb band represents an isoform where exon 13 is spliced to exon 16. The 2.1 kb fragment corresponds to an isoform co-linear with the known cDNA consisting of 19 exons, as described previously (6,8). Exon 14 was spliced to exon 15, but exon 14a was not present. However, exon14, intron 14 and exon 15 of human RPGR also contain a continuous ORF.
Detection of a retina-specific exon
Clones derived from human retina RNA revealed a transcript co-linear with the known cDNA consisting of 19 exons, except for a 39 bp insertion that was present between exons 15 and 16 (Fig. 3, isoform 3, EMBL accession no. AJ238395). All of the six clones analysed contained this exon, designated 15a. It encodes 11 amino acids, followed by a premature stop codon. The deduced protein sequence is truncated by 169 amino acids, compared with the variant lacking this exon.
To search for the homologous mouse exon 15a, we performed RT-PCR on mouse eye cDNA with primers in exons 15 and 16. Again, a 39 bp insertion was found. At the nucleotide level, the sequences of human and mouse exon 15a are nearly 77% identical (Fig. 4D). The homologous mouse exon 15a encodes six amino acids preceding a premature stop codon. To obtain the full-length transcripts of mRpgr including exon 15a, we performed long-range PCR on mouse eye cDNA with forward and reverse primers in exon 15a and flanking primers in exons 1 and 19. Cloning and sequencing of the PCR products identified two isoforms in the 5′ direction of the cDNA (exons 1–15a) resulting from alternative splicing of exons 14, 14a and 15 (Fig. 2B, isoforms 11 and 13). In the 3′ direction (exons 15a to 19), two variants were present, which showed skipping of exons 18a and 19a, or of exon 19a only (Fig. 2B, isoforms 12 and 14).
To look for the expression pattern of exon 15a, we performed RT-PCR with primers in exons 15 and 16 on several mouse and human tissues (Fig. 4A-C). Expression of exon 15a was detectable only in human retina and in mouse eye.
The retina-specific exon is deleted in a patient
In order to search for mutations in the two newly identified exons, we performed single-strand conformation polymorphism (SSCP) analyses in 23 unrelated X-linked RP patients. In two of them, bandshifts were detected in the human homologue of mouse exon 14a. The same shift was present in five of 47 controls and was therefore considered to be a polymorphism.
SSCP analyses of exon 15a detected no bandshifts. However, the exon 15a fragment was not amplifiable from DNA of patient 2557, who had been shown in a previous study to carry a 6.4 kb deletion in the 3′ portion of the RPGR gene (7). To characterize the position of the deletion further, we hybridized EcoRI-digested DNA of the patient and a healthy control with probes from exons 15, 15a and 16 (Fig. 5A). Indeed, exon 15a gave no signal in the patient's DNA, while the two flanking exons detected the deletion breakpoint fragment.
To exclude an effect of the deletion on mRNA stability as the primary cause of the disease, we performed northern blot analysis with poly(A)+ RNA from lymphoblastoid cell lines of the patient and a healthy control (Fig. 5B). In both RNA samples, hybridization of a full-length RPGR cDNA probe showed similar signal intensities. RT-PCR with primers in exons 15 and 16 and subsequent sequencing of the PCR products again revealed no differences in the patient and the control, indicating correct splicing of the mRNA in extraocular tissue. To determine the exact position of the breakpoint in patient 2557, we amplified a breakpoint-spanning fragment from genomic DNA by long-range PCR with primers in exons 15 and 16. Cloning and sequencing of the 4 kb PCR product identified the proximal deletion breakpoint in intron 15, ∼1.5 kb downstream of exon 15. The distal deletion breakpoint is located 70 bp downstream of the splice donor site of exon 15a, nucleotide position 826 in EMBL sequence no. X94767 (Fig. 5C). These results demonstrated that only exon 15a and flanking intronic sequences were deleted in the patient.
Ophthalmological examination of the deletion patient
Patient 2557 was born in 1957 as the youngest brother in a sib-ship of four. Three older brothers, his mother and his maternal grandfather had RP. The patient complained about difficulties seeing in dim light from the age of 10 years. At his first examination (11 years), it was not possible to demonstrate any ocular pathology besides slightly irregular macular reflexes. Two years later, a slight reduction in visual acuity (VA) was noted and the macular reflexes were characterized as ‘metallic’. In the following years, he was unavailable for follow-up examinations. At 33 years ofage, 1 year after his driving licence was suspended, he was re-examined. At this time, his visual problems had worsened considerably. VA on the right and left eye was 0.3 and 0.4, respectively. He had an upper ring scotoma in both visual fields (Goldmann perimetry IV/4e and I/4e) and the slope of the dark adaptation record was abnormal with a nearly linear rod phase and a slightly elevated rod threshold (1.8 × 10−5 cd/m2). His dark-adapted Ganzfeld ERG showed a severe reduction of the flicker responses with prolonged cone implicit times and a less severe affection of the rods. During the following 4 years, very rapid visual deterioration was observed. At the last examination in 1994, VA had dropped to 0.01 and 0.04 for the right and left eyes, respectively, and 2° tunnel vision was measured with a large object by Goldmann perimetry. The ocular fundus (Fig. 5D) revealed a moderate constriction of the arterioles, a homogeneous paleness of the optic discs, diffuse as well as a widespread patchy atrophy of the retinal pigment epithelium and spicule hyperpigmentations, mainly located in the lower retinal quadrants. His dark adaptation was nearly extinct, with a 4.5 log unit threshold elevation to 1.8 × 10−1 cd/m2. The ERG showed cone responses near the lower detection limit (<5 µV) and a residual rod response below the 10th percentile. In conclusion, this patient demonstrated a retinal dystrophy with a relatively late debut of night blindness around the age of 10 years. Significant objective signs were not present before the age of 20 years. From 30 years of age, very rapid deterioration took place leading to blindness by the age of 37. Electrophysiologically, this retinopathy appeared as a cone-rod dystrophy.
In order to explore the transcription pattern of RPGR and its orthologous mouse gene (mRpgr), we performed northern blot hybridization and RT-PCR experiments with RNA from various mouse and human tissues, including retina, brain and testis. At least 12 splice variants were identified in the mouse, one of which had been described previously (13; GenBank accession no. AF044677). Alternative splicing is confined to the 3′ half of the gene, while exon usage at the 5′ end, which encodes the RCC1-homologous domain, is rather uniform. The size of isoforms 1–6 correlates with signals observed upon northern blot analysis (2.7–3.6 kb; Fig. 1A). Isoforms 7–10 (1.6–2.1 kb), which lack large parts of the mRpgr cDNA, cannot clearly be assigned to bands on the blot, indicating a low level of expression. Within the 3′ end of the RPGR gene, two novel exons were identified, one in mouse testis and another in mouse eye and human retina. Exon 14a of mRpgr is expressed predominantly in testis and represents intron 14 of the human gene. Although this sequence is conserved in three species (human, mouse and dog) and contains an ORF, expression was not detectable in human testis. In the mouse, exon 14a was not used as a separate exon, but only together with exons 14 and 15. This may indicate a difference in exon usage between mouse and man. The second novel exon 15a is located in intron 15. It is only expressed in the mouse eye and human retina and contains a premature stop codon. Consequently, the protein-coding capacity of exons 16–19 is not utilized in the retina, and the RPGR gene product contains 646 instead of 815 amino acids (Fig. 6). The importance of this exon for RPGR function in the retina is illustrated by three lines of evidence: (i) it is expressed exclusively in the human retina and mouse eye (Fig. 4); (ii) a deletion of exon 15a gives rise to RP in a patient who retained all the flanking exons (Fig. 5) and (iii) all previously identified mutations are located upstream of this exon (Fig. 6).
The RP phenotype in the deletion patient, which is characterized by a relatively late onset of night blindness and an ERG reminiscent of a cone-rod dystrophy, could be explained by an unstable transcript, resulting from the deletion. However, at least in lymphoblastoid RNA of the patient, a normal expression level and a correctly spliced lymphoblastoid isoform were detected upon northern blot hybridization and RT-PCR. Exon 15a in man and mouse encodes 11 and six amino acids, respectively, and introduces an alternative stop codon. So far, no mutations have been described in exons 16–19 of RPGR, which give rise to the RP phenotype (Fig. 6). These findings suggest that the C-terminus of the ubiquitous variant, including the isoprenylation site at the very end, is not essential for retinal function. Deletion of the isoprenylation site was previously shown to abolish the Golgi localization of the mRpgr protein in transiently transfected COS cells (13). Therefore, the introduction of a premature stop could lead to a different subcellular localization of the protein in the retina. Similar observations have been reported for an alternatively spliced isoform of BRCA1. Exon 11 of this gene contains at least one functional nuclear localization signal and, when absent, the protein is located in the cytoplasm whereas it is transported into the nucleus when exon 11 is present (24).
Alternative splicing allows expression of multiple protein isoforms from a single gene in a tissue- or time-dependent manner, which is important for cell type-specific function or developmental processes. The identification of tissue-specific alternative splice variants of otherwise ubiquitously expressed genes may explain why mutations in those genes lead to phenotypes restricted to a single organ or tissue. Recently it was shown for the hairless gene, which is involved in congenital alopecia, that exon 17 is lacking in a skin-specific transcript, whereas other cell types predominantly express mRNA molecules containing this exon (25). Similarly, the retina-specific exon 15a of RPGR may be a clue to why mutations in this ubiquitously expressed gene lead to a degeneration of the retina. Mutations could interfere with the function of the retina-specific RPGR isoform whereas other cell types expressing the mutant protein are either insensible or able to complement their deleterious effect. So far, little is known about tissue-specific factors required for alternative splicing, but they may contribute significantly to variation of the disease phenotype, as shown recently for CFTR (cystic fibrosis transmembrane conductance regulator) (26).
Mutations in RPGR were identified in 20% of all patients with X-linked RP. According to linkage data, <60% of the families map to the RP3 locus. The existence of additional coding sequences may provide an explanation for the apparent paucity of mutations. Splice variants containing further not yet identified exons might be the targets for mutations in the remaining families. The detection of exon 15a, that is deleted in an RP patient, supports this hypothesis. Moreover, northern blot analysis of the homologous mouse gene indicated additional splice variants in the eye, ranging in size from 5 to 20 kb (Fig. 1A). Also, mutations could affect non-coding sequences required for and regulating alternative splicing. It has been shown for the cardiac troponin T gene (cTNT) that mutation of intronic splicing enhancers can significantly reduce the usage of a muscle-specific exon (27). The identification of a retina-specific RPGR exon raises the possibility that mutations in the remaining families might also be located in intronic elements required for the regulation of the spatial and temporal expression of alternative RPGR transcripts.
Materials and Methods
RNA isolation and northern blot analysis
Total RNA was isolated from cells and tissues with the Clontech Atlas Pure RNA Isolation Kit (Clontech, Heidelberg, Germany). Poly(A)+ RNA was obtained using Dynabeads oligo(dT)25 (Dynal, Hamburg, Germany) according to the manufacturer's instructions. After formaldehyde gel electrophoresis of the RNA (28), gels were washed three times for 7 min in diethylpyrocarbonate-treated H2O and blotted onto Hybond-N+ membrane (Amersham, Braunschweig, Germany) using 10× SSC as transfer buffer. RNA was cross-linked using a Stratalinker (Stratagene, Amsterdam, The Netherlands) and by baking the membrane for 2 h at 80°C. A 25 ng aliquot of the DNA probes was labelled by random primer extension using [32P]dCTP, and hybridization was carried out in QuickHyb solution (Stratagene) at 60°C according to the manufacturer's instructions.
Total RNA was digested with DNase I (Gibco BRL, Eggenstein, Germany) for 15 min at room temperature. DNase I was inactivated by addition of EDTA to a final concentration of 2.5 mM and incubation at 65°C for 10 min. A 5 µg aliquot of DNase I-digested total RNA was used directly for reverse transcription which was carried out using random hexanucleotide priming and SUPERSCRIPT II (Gibco BRL) in a 20µl reaction according to the protocol provided. PCR with mRpgr/RPGR-specific primers was performed taking 1 µl of the reverse transcription reaction as template in either a standard PCR reaction setup with AmpliTaq (Perkin Elmer, Weiterstadt, Germany) or following the instructions of the Expand Long Template PCR System (Roche, Mannheim, Germany) manual. In each experiment, a sample without reverse transcriptase was amplified under the same conditions as the reverse-transcribed RNA in order to confirm the absence of contamination. In the case of mouse eye and testis, and human retina, testis and placenta RNA, the SMART PCR cDNA Synthesis Kit from Clontech was used according to the manufacturer's instructions to generate cDNA from 1 µg of total RNA. Subsequently, 1 µl out of 100 µl of cDNA was used as template in a PCR with mRpgr/RPGR-specific primers. For sequencing, PCR products either were cloned directly into pGEM-T Easy (Promega, Mannheim, Germany) or, in the case of weak bands, were re-amplified after purification from agarose gels. Sequencing was carried out on a Li-cor Long Readir 4200 sequencer as described previously (29).
Molecular characterization of the deletion in patient 2557
DNA from the patient and a control was isolated from lymphoblastoid cell lines according to standard procedures. For Southern blot analysis, 10µg of DNA were digested with EcoRI according to the manufacturer's instructions and the fragments were electrophoresed and blotted as described previously (7). Probes were labelled with [32P]dCTP by random primer extension and hybridized to the membrane following standard protocols. For long-range PCR, 100 ng of genomic DNA of the patient was used to amplify a 4 kb breakpoint-spanning fragment with primers in exons 15 and 16. PCR was carried out with the Expand Long Template PCR System (Roche) according to the manufacturer's instructions. The fragment was cloned in pGEM-T Easy (Promega) and sequenced on a Li-cor Long Readir 4200 sequencer as described (29).
Exon 14a was amplified in three fragments from 100 ng of genomic DNA using the following primer pairs: 5′-TTC ATG ACGCAG CCAGCT ACGA-3′ and 5′-AGT TGCCAA AGA AAG GAC TCA TC-3′; 5′-GCA GAA ATA GCA GGT ATG AAG GA-3′ and 5′-CCA TAC CGT ATG TTT TGG TCA GT-3′; 5′-TCC AGC AGC CTG AGG CAA TAG-3′ and 5′-CCA TGC ACC TTC ACA TTT TCC TC-3′. PCR was carried out in a 25 µl reaction with 1.5 mM MgCl2 and a 61°C annealing temperature. Exon 15a was amplified in one fragment from 100 ng of genomic DNA using the following primer pair: 5′-CTG TGA TCC TAA CGC AAG AC-3′ and 5′-TAG AAA TCA CAT CAT AGC AC-3′. PCR was carried out in a 25µl reaction with 2.5 mM MgCl2 and a 58°C annealing temperature. SSCP analysis of the fragments was performed as described previously (9). Fragments revealing a bandshift were sequenced on ABI automated sequencers using dye terminator chemistry.
The authors are very grateful to the patients and their families for contributing to this study. We would like to thank Hannelore Madle and Susanne Freier for cell culturing, and Christina Zeitz and Silke Schneider for their technical assistence in PCR amplification and sequencing of cDNAs. Anneke den Hollander is thanked for making available total RNA from human retina. This work was supported by the Foundation Fighting Blindness, USA (R.K.) and by the Deutsche Forsch-ungsgemeinschaft (grant Be 1559/2–1 to S.L.).