Complex Expression Pattern of RPGR Reveals a Role for Purine-Rich Exonic Splicing Enhancers

Abstract

purpose. To examine the mechanism underlying transcript heterogeneity in the gene for the retinitis pigmentosa GTPase regulator (RPGR).

methods. Transcript heterogeneity was analyzed by reverse transcription-polymerase chain reactions (RT-PCR), rapid amplification of cDNA ends (RACE), and transient expression of minigene constructs. Protein variants were identified by immunoblot analysis and by immunocytochemistry.

results. RPGR transcripts terminated either uniformly at the end of exon 19, producing the constitutive transcript with few variants, or at variable sites downstream from exon 15. The latter transcripts resembled the previously described open reading frame (ORF)14/15 variant, but the ORF14/15 exon was not found in full length. Instead, various portions of a purine-rich region were removed as introns. Numerous splice site combinations were used, giving rise to innumerable variants. Analysis of the purine-rich region found multiple exonic splicing enhancers (ESEs) known to promote splicing through interaction with serine-arginine repeat (SR) proteins. Antibodies targeting different regions of RPGR detected a multitude of RPGR proteins in photoreceptors, concentrated in the connecting cilium. Predominant ORF14/15-encoded RPGR polypeptides migrated at approximately 200 kDa and were photoreceptor specific.

conclusions. The exceptional heterogeneity in RPGR transcript processing results primarily from a novel form of alternative RNA splicing mediated by multiple exonic splicing enhancers. RPGR is composed of a population of proteins with a constant N-terminal core encompassing the RCC1 homology domain followed by a C-terminal portion of variable lengths and sequences.

Mutations in the gene coding for the retinitis pigmentosa GTPase regulator (RPGR) cause X-linked RP, ^{1 2} a form of hereditary photoreceptor degeneration that leads to blindness. The in vivo function of RPGR is not fully understood. The N-terminal sequence of RPGR is similar to the regulator of chromatin condensation (RCC1), a nuclear protein that catalyzes guanine nucleotide exchange for the small GTPase Ran and regulates nuclear import and export. ³ In photoreceptors, RPGR is concentrated in the connecting cilium ⁴ through binding to an RPGR interacting protein (RPGRIP). ⁵ The connecting cilium is a junctional structure that separates two cellular compartments with vastly different protein compositions. In mice without RPGR, cone photoreceptors exhibit ectopic localization of opsin in the cell bodies and synapses early on. Subsequently, both cone and rod photoreceptors degenerate. The presence of an RCC1 homology domain in RPGR, its localization in the connecting cilium, and the retinal phenotype in the mutant mice are consistent with the proposal that RPGR plays a role in maintaining polarized protein distribution across the connecting cilium by regulating directional transport and/or restricting redistribution. ⁴  

Considerable complexity in the expression of RPGR has been reported. The initial study on RPGR identified a transcript consisting of 19 exons. ¹ This transcript is expressed in a wide variety of tissues ^{1 2 6 7} and is referred to in this report as the constitutive transcript. Coding sequence for the RCC1 homology domain spans exons 1 to 11. ¹ The remaining sequence does not contain any recognizable motifs. Extensive alternative splicing in the constitutive transcript has been reported. ^{6 8} In addition to the constitutive transcript, a novel human RPGR variant discovered recently terminates in intron 15 of the RPGR gene. ⁹ This alternative terminal exon, referred to as ORF15, consists of the constitutive exon 15 and part of intron 15. ⁹ A region of purine-rich repetitive sequence is found in this exon and encodes alternating glycine and glutamic acid residues. In the mouse retina, the analogous exon is referred to as ORF14/15 because intron 14 is not spliced out, forming a contiguous terminal exon composed of exon 14, intron 14, and the ORF15 exon. ⁹ A large number of disease-causing mutations were identified in the ORF15 exon in patients with X-linked RP. ^{9 10 11} This contrasts with the absence of disease-causing mutations in exons 16 to 19 unique to the constitutive transcript, suggesting that the ORF15 transcript (or ORF14/15 transcript in the mouse) may be the functionally significant RPGR isoform. Most of these mutations reside in the purine-rich repetitive region and result in a shift of the reading frame. 

Little is known about how RPGR transcript heterogeneity relates to its expression at the protein level, which has been elusive to detection with antibodies. This is illustrated by the fact that so far only three studies have identified any RPGR protein from native tissues. ^{4 12 13} In this study, we set out to determine the major RPGR transcript and protein isoforms in photoreceptors. We found an unexpected exceptional level of heterogeneity in RPGR pre-mRNA splicing, producing innumerable RPGR variants. 

Materials and Methods

RNA Preparation

Total RNA was isolated with extraction reagent (TRIzol, Invitrogen, San Diego, CA). Poly-A(+) RNA was isolated by oligo-dT cellulose chromatography. To enrich for cytoplasmic RNA, retinas or cells were lysed at 4°C in a buffer containing 50 mM Tris-Cl (pH 8.0), 140 mM NaCl, 1.5 mM MgCl_2, 0.5% Nonidet P-40, 1 mM dithiothreitol (DTT), and 1000 U/mL RNase inhibitor. Cell nuclei were removed by centrifugation at 1000g. The cytoplasmic RNA was isolated from the supernatant using a kit (RNeasy; Qiagen, Valencia, CA). Unless otherwise noted, tissues from C57BL/6 mice were used for RNA as well as protein analyses. 

Rapid Amplification of cDNA Ends

Rapid amplification of cDNA ends (RACE) was performed with a kit (SMART; Clontech Laboratory, Palo Alto, CA), with either total or poly-A(+) RNA used as templates. DNA products were gel purified, cloned into a vector (pGEM-T Easy; Promega, Madison, WI), and analyzed by DNA sequencing. 

Reverse Transcription-Polymerase Chain Reaction (RT-PCR)

Total RNA was treated with RNase-free DNase I (Ambion, Austin, TX) to remove all contaminating genomic DNA and used as templates for cDNA synthesis. Two reverse transcriptases were used (Superscript II, used at 42°C; GibcoBRL, Gaithersburg, MD, and ThermoScript, used at 70°C, a heat-stable variant of avian myeloblastosis virus [AMV] reverse transcriptase; Invitrogen). PCR was performed with commercial polymerase (Long Expand polymerase; Roche Biochemicals, Indianapolis, IN). Human gene primers ABCRP1 and ABCRP2R, and mouse gene primers mABCRP1 and mABCRP2R were used to amplify the human and mouse ABCR cDNAs as controls for template quality (Table 1) . For cloning, PCR products were gel purified and ligated into the pGEM-T Easy plasmid vector. 

Antibodies, Immunoblot analysis, and Immunofluorescence

Peptides corresponding to residues 321 to 338 (ORF1; CEKVSLETEHLQRAQGK) and 707 to 725 (ORF2; GKVADNESDRKQSPKVS) of the mouse RPGR ORF14/15 sequence ⁹ were synthesized, conjugated to keyhole limpet hemocyanin, and used to immunize rabbits. The specific antibodies were purified from antisera by affinity chromatography, with immobilized peptides used as ligands. Other antibodies were raised in rabbit, with recombinant proteins expressed in Escherichia coli used as immunogens. Their lengths and locations along the RPGR polypeptide sequence are shown in the antibody map in Figure 5A . All antibodies were used after affinity purification. For immunoblot analysis, tissue samples were homogenized in SDS-protein sample buffer. An axoneme-enriched preparation from retinas was prepared as described. ⁵ Immunofluorescence staining was performed as described. ^{4 5} As we have previously reported, retinas must be cryosectioned without prior fixation to ensure successful staining. ⁵  

Construction of RPGR Minigenes

Two RPGR minigene constructs were generated (see Fig. 4A ). To assemble these constructs, exons 1 to 13 were amplified from mouse retinal cDNA. Intronic regions, exon 16, and the full length ORF14/15 exon were amplified from mouse genomic DNA. PCR primers were designed based on published RPGR sequences and/or sequences obtained from a proprietary database (Discovery System; Celera Genomics, Rockville, MD). The minigene constructs were cloned into the mammalian expression vector pcDNA3 (Invitrogen). 

Transient Transfection

COS7 cells were grown in six-well culture plates and transfected with cDNA expression constructs or RPGR minigene constructs, with the use of a transfection reagent (Lipofectamine 2000; Invitrogen). At 48 hours after transfection, cells were washed with phosphate-buffered saline and harvested. For immunoblot analysis, the cells were solubilized in 1× SDS protein sample buffer, and the lysates were analyzed by probing with RPGR antibodies. For RT-PCR, cytoplasmic RNA was isolated as described earlier and used as the template. 

Results

Screening for RPGR Transcript Variants

5′ and 3′ RACE analyses were performed with anchoring primers in the constant regions of RPGR, the rationale being that these were the only assays that could provide a relatively unbiased screening for unknown transcript variants (Fig. 1A) . 5′ RACE analysis with an anchoring primer at the end of exon 13 demonstrated that the RPGR transcript was relatively uniform from exons 1 through 13 (Fig. 1B) . 5′ RACE with an anchoring primer at the end of the ORF14/15 exon, however, yielded heterogeneous products, providing the initial indication that the ORF14/15 exon was heterogeneously processed (Fig. 1B) . Consistent with the 5′ RACE results, 3′ RACE using anchoring primers upstream from the point of divergence yielded a diverse population of products (Fig. 1C) . Cloning and sequencing of major 3′ RACE products provided some unexpected findings (Fig. 1D) . First, most of these transcripts were ORF14/15-like by virtue of having some ORF14/15 exon sequence. However, various portions of the purine-rich region were removed as if they were introns. Second, a substantial fraction of these ORF14/15-like transcripts retained intron 13, resulting in early truncation of the ORF. Third, these ORF14/15-like transcripts had several different 3′ ends. In each case, a polyadenylation signal sequence was found upstream, suggesting that these were bona fide 3′ ends. A major transcription termination site was found 933 bases downstream from the ORF14/15 exon. The latter termination site would be consistent with a 6-kb ORF14/15-like transcript found in a recent study. ⁷ A much longer transcript, also described in the report of that study, would not be readily detected by 3′ RACE because of the size limitations of this technique. 

Exceptional Heterogeneity of RNA Splicing within the ORF14/15 Exon

RT-PCR analyses were performed to examine RNA splicing more closely in the ORF14/15 region (Fig. 2) . Using primer pairs spanning the ORF14/15 exon, we obtained products over a wide range of sizes (Fig. 2A) . These products were all smaller than would be expected for the full-length product. PCR with a primer pair spanning the most highly purine-rich region confirmed that this was where most of the variability originated. In addition to performing reverse transcription at the conventional temperature (42°C), a heat-stable reverse transcriptase was used to perform the reactions at a higher temperature (70°C) to reduce the likelihood that the reactions may fail because of unusual secondary structures assumed by the purine-rich sequence. Similar results were obtained with both enzymes (Fig. 2A) . Sequence analysis of the PCR products showed that portions of the purine-rich sequence were removed in an extremely variable manner. From 40 independent clones analyzed by sequencing, no two transcripts were spliced in exactly the same way (Fig. 2B) . An extensive search for the full-length ORF14/15 exon, either by RT-PCR or by screening a retinal cDNA library, found no evidence that a full-length form existed. The corresponding genomic region were readily amplified by PCR (Fig. 2A) and these PCR products were cloned and propagated in E. coli. Thus, neither PCR difficulty nor instability in E. coli can account for our inability to isolate the full-length ORF14/15 variant. These data therefore support the notion that an exceptional degree of alternative RNA splicing occurs within the ORF14/15 exon. In fact, the ORF14/15 exon is not a single exon but two exons joined together by poorly defined splice junctions. 

By sequence analysis, all three reading frames downstream from the splice junctions were found. Among those transcripts retaining a longer purine-rich region, there appeared to be a bias toward maintaining the original reading frame of the full-length ORF14/15 exon (Fig. 2C) . Among smaller spliced products, the reading frames downstream from the splice junctions appeared random. 

Heterogeneous splicing within the ORF14/15 exon was found in several strains of mice (C57Bl, BALB/C, and 129 strains; data not shown) and in the ORF15 exon of the human RPGR transcripts (Fig. 2D) . No full-length ORF15 exon was found in the human RPGR transcripts. Thus, heterogeneous processing of the RPGR transcripts appears to be a general phenomenon in mammalian retinas. 

A substantial fraction of the mouse ORF14/15-like transcripts retained intron 13. These may represent splicing intermediates that had not been transported out of the nuclei. By performing RT-PCR on templates enriched for cytoplasmic mRNA, intron 13-containing transcripts were found abundantly represented in the cytoplasmic RNA pool (Fig. 2E) . Therefore, they were mature RPGR transcripts that would be available for translation. Retention of intron 13 thus contributes additionally to the complexity of the murine RPGR transcripts. It may, however, be species specific. Unlike the human RPGR transcript, the murine constitutive RPGR transcript is spliced from exon 13 directly to exon 16, indicating that the 3′ splice site of intron 13 (and of intron 14) is suboptimal. Inefficient removal of intron 13 could result from the fact that the ORF14/15-like transcripts terminate before the emergence of exon 16 and fall short of the preferred splice junction. 

Multiple ESEs in the Purine-Rich Region

Exonic splicing enhancers (ESEs) are cis-acting elements that interact with auxiliary splicing factors such as the serine-arginine repeat (SR) proteins. They facilitate recognition of upstream 3′ splice acceptor sites and inclusion of the exons in which they reside. ^{14 15 16 17} The role of ESEs is most apparent in alternative exons defined by weak splice sites, and disruption of ESEs results in the skipping of these exons. A major type of ESEs are purine-rich sequences consisting of GAR repeats, where R is a G or an A. We searched the mouse and human RPGR purine-rich region for well-characterized ESEs ¹⁸ and found multiple copies of ESEs in this region. For example, GAGGAAGAA is an ESE originally identified in the troponin T alternative exon. ¹⁸ This sequence element is present at 21 copies in the mouse ORF14/15 purine-rich region. Given the diversities of ESEs found in different genes and the degeneracy in the ESE consensus sequences, the entire purine-rich region might be considered a tandem array of ESEs. Binding to ESEs at different locations within the ORF14/15 region by SR-related splicing factors provides the most plausible explanation for the observed heterogeneous splicing. 

Given the well established relationship between splicing of a terminal exon and promotion of transcriptional termination, ^{19 20} efficient termination of ORF14/15-like transcripts within or downstream from the purine-rich region is most likely related to the splicing events within the ORF14/15 region. However, multiple copies of the consensus sequence for binding to the zinc finger protein MAZ were also identified in the purine-rich region. ^{21 22} MAZ binding sites are found between a number of closely spaced genes. These sites bind MAZ proteins and cause RNA polymerase complex to pause and to terminate transcription. ²² In these situations, MAZ sites are thought to prevent transcription read-through and interference with the downstream gene. It is possible that MAZ binding sites located in the purine-rich region of the RPGR gene also play a role in the alternative transcriptional termination. 

Limited Variability of the Constitutive Transcript

We examined alternative splicing of the constitutive RPGR transcript by RT-PCR (Fig. 3A) . A major product migrating at 2.3 kb was obtained. Cloning and sequencing of the PCR product found three variants, referred to as constitutive variants 1 to 3, among which constitutive variant 1 is identical with the previously reported murine RPGR transcript. ⁶ Variants 2 and 3 have different reading frames from variant 1 downstream from the site of alternative splicing. An alternative exon 15a was previously reported to be retina specific. ⁸ In our assays, transcripts incorporating exon 15a appears to be rare (Fig. 3B) . Three protein variants, readily detected by antibodies (described later), appear to be encoded by these transcript variants. 

Repression of the ORF14/15 Exon by Splicing of Exon 16

RPGR transcripts either terminate within or just downstream from ORF14/15 or terminate after exon 19. In the latter (constitutive) transcripts, ORF14/15 is always skipped. To help understand how splicing choices are made with regard to synthesizing one class of variants versus the other, we transiently transfected minigene constructs in COS7 cells and analyzed splicing of the transcripts (Fig. 4) . We hypothesized that the presence of the splice acceptor site of exon 16 in the primary transcript represses inclusion of the ORF14/15 exon, because the splice junctions bordering exons 13 and 16 are stronger and would override other splice sites between. Retention of the ORF14/15 exon and heterogeneous RNA splicing within the ORF14/15 exon were found in transcripts from the minigene without exon 16 (minigene −E16). In a second construct which extended the sequence further downstream to include exon 16 (minigene+E16), exon 13 was spliced to exon 16 directly, and the ORF14/15 exon was skipped. Thus, the alternative splice sites in the ORF14/15 region were weak and repressed if the splice acceptor site of exon 16 was present in the pre-mRNA, leading to the direct joining of exons 13 and 16 and synthesis of the constitutive transcript. 

Diversity of RPGR Protein Products

We analyzed the RPGR protein with six antibodies targeting different positions along the RPGR polypeptide sequence (Fig. 5A) . On immunoblots, RPGR polypeptides migrating over a range of sizes were identified. With an antibody (S3) targeting epitopes unique to the constitutive variant, bands were detected migrating at approximately 90 to 100 kDa. Three closely spaced individual bands were resolved on longer separation on gels (Fig. 5B) . This triplet was absent in the RPGR^−/− (knockout [KO]) mouse retinas, confirming that they were bona fide protein products of the constitutive transcript. With an antibody (S1) targeting epitopes common to all known RPGR variants, an additional 200-kDa band was detected (Fig. 5B) . This band exhibits a broad pattern, indicating a collection of polypeptides of similar sizes. In addition, minor RPGR polypeptides of lower molecular weights were detected after longer exposure of the immunoblots. The signal intensity of all RPGR variants was greatly diminished in the rd mouse retinas devoid of photoreceptors, suggesting that in the retina the RPGR proteins are derived primarily from photoreceptors. Probing with antibodies (ORF1 and ORF2) specific to the ORF14/15-like transcripts showed directly that proteins in the 200-kDa size range were encoded by these transcripts (Fig. 5C) . Because the ORF2 antibody gave a very weak signal, a crude ciliary axoneme fraction enriched for RPGR ⁵ was required for immunoblot analysis. 

The apparent molecular weight of the large ORF14/15-related polypeptides (200 kDa) was higher than that calculated from our average cDNA clones. To determine whether the ORF14/15-related cDNAs are capable of encoding proteins in this molecular weight range, we expressed one representative clone (T12) by coupled in vitro transcription-translation and by transient transfection in COS7 cells (Fig. 5D) . This clone had 654 base pairs (approximately two thirds) of its purine-rich sequence spliced out, but maintained the original reading frame of the full-length ORF14/15 exon. Both assays produced a protein approximating 200 kDa in molecular mass. An abundance of glutamic acid residues in the polypeptide is the likely explanation for the higher apparent molecular mass. ²³ Thus, the group of polypeptides migrating at 200 kDa appear to be encoded by transcripts that are shortened by several hundred base pairs in the purine-rich region because of splicing. 

Expression in COS7 cells produced, in addition, proteins of smaller sizes. RT-PCR analysis of the RPGR transcripts from T12-transfected COS7 cells showed that the remaining purine-rich region in clone T12 underwent further splicing, producing smaller transcripts. Expression of a cDNA clone retaining intron 13 produced a protein migrating at 70 kDa (data not shown), suggesting that RPGR protein variants from the retina in this size range may be derived from transcripts that retain intron 13. Transient transfection in COS7 cells followed by immunoblot analysis with region-specific antibodies S1 and S2 helped to establish a correlation between the constitutive protein variants and their cognate transcripts (Fig. 5E) . 

To begin addressing the functional significance of having a myriad of RPGR variants in photoreceptors, it would be informative to determine whether there is a difference in subcellular localization among variants that may indicate different physiological roles. Using antibodies specific either for the constitutive or the ORF14/15 variants, we found both concentrated in the connecting cilium (Fig. 6) . Because the RCC1 homology domain of RPGR mediates localization in the cilium and because most variants retain the RCC1 homology domain, it is likely that all RPGR protein variants are concentrated in the connecting cilium. 

Discussion

Our work confirmed the previous report ⁹ that the ORF14/15 region is used as a major alternative terminal exon in retinal photoreceptors. However, the results of RT-PCR assays showed that RPGR transcript retaining the full-length ORF14/15 exon was not the prevalent form. Splice site combinations were so variable that the total number of splice variants could not be enumerated. In contrast to the ORF14/15 variant, variability in the constitutive RPGR transcript was limited and unremarkable compared with neuronal gene expression in general. ²⁴ Thus, the complexity of RPGR expression is due primarily to the diverse population of ORF14/15-like transcripts. The finding of heterogeneous splicing in the ORF14/15 exon suggests that certain sequence changes in the genomic DNA in this exon may affect splicing in unpredictable ways and lead to production of a defective transcript. This possibility should be kept in mind when evaluating disease-causing mutations in this region. 

There appears to be less heterogeneity of RPGR proteins than might be expected from the heterogeneity of the ORF14/15-like transcripts. In this regard, it should be noted that a large portion of the ORF14/15-like transcripts retain intron 13, leading to early truncation of the coding sequence. The polypeptide encoded by these transcripts, with an estimated molecular mass of 70 to 80 kDa, does not accumulate to a high level (Fig. 5) . Transfection experiments in COS cells also showed that cDNA clones that truncated early consistently produced smaller amount of proteins (D.-H. Hong, T Li; unpublished observation, 2002). Thus, differential stability of RPGR proteins favoring those that are in frame and retain more of the Glu-rich region may contribute to the preferential accumulation of larger (200 kDa and larger) RPGR proteins. We also cannot rule out the possibility that the RT-PCR reaction may favor amplification of the smaller transcripts, leading to their overrepresentation. 

Heterogeneous splicing in the purine-rich region is most likely caused by the multiple ESEs. The SR family of splicing factors promote the spliceosome assembly by providing bridging interactions among the small nuclear ribonuceloprotein particles (snRNPs), splicing factors, and RNA motifs. One class of sequences bound by the SR family of proteins ¹⁸ or SR-related proteins ¹⁷ are purine-rich ESEs, which promote inclusion of weak and alternative exons in which they reside. The multiple repeat units of ESEs present in the purine-rich region could readily explain the exceptional variability of splice site selection, because random binding of SR proteins to different ESEs would generate a vast repertoire of splice site combinations. The factor(s) that interact with the ESEs in the RPGR primary transcript in vivo remains to be identified. 

We propose a model for RPGR synthesis in which recognition of the ESEs and splicing within the ORF14/15 region occur cotranscriptionally (Fig. 7) . The kinetic lag in the appearance of exon 16 in the primary transcript, separated from the ORF14/15 exon by a large intronic region (>10 kb), provides a time window in which this event may occur. Recognition of alternative terminal exons strongly promotes transcriptional termination. ^{19 20 25} If splicing takes place within the ORF14/15 region before emergence of exon 16, it would be coupled efficiently to transcriptional termination to produce an ORF14/15-like transcript. If not, the splicing machinery would preferentially join exons 13 and 16 directly and remove all intervening sequence, leading to synthesis of the constitutive transcript. The observation that the ORF14/15-like transcripts are enriched in photoreceptors would therefore indicate that these cells are more efficient in using the ESEs than other cell types. Existing evidence suggests that the splicing machinery in photoreceptors may differ from that of other tissues. For example, mutations in three pre-mRNA splicing factors cause dominantly inherited RP. ^{26 27 28 29} Because these factors are used in all cells, the selective involvement of photoreceptors in disease may be explained by the intrinsic differences between photoreceptors and other cells in terms of abundance and activities among different splicing factors. It would be interesting to determine whether RPGR is one of the downstream target genes affected by mutations involving those splicing factors and whether this is part of the pathogenic process. 

The splicing phenomenon described herein appears to be novel and to our knowledge has not been described for any other gene. One example that comes close to the splicing complexity of RPGR is the calcium-activated potassium channel expressed in the chick cochlea. It expresses a large number of splice variants, but these variants are derived from combinations of well defined exons rather than the continuously variable splicing seen in RPGR. In terms of functional significance, the repertoire of potassium channel variants expressed in hair cells changes with cell position along the basilar papilla. This parallels changes in the response frequencies of the cells and is thought to be at least partly responsible for the electrical tuning of the hair cells. ³⁰ In the case of RPGR, the necessity for a myriad of variants is not clear. Because all RPGR protein variants are concentrated in the photoreceptor connecting cilium, they are unlikely to perform distinct functions in different cellular compartments. It is an intriguing possibility that a collection of RPGR variants is required to fulfill the diverse trafficking and regulatory role of the connecting cilium. Alternatively, some of these variants may be functionally redundant. In this regard, introduction of a single RPGR splice variant into mutant animals carrying null alleles of RPGR would provide useful clues. It should be noted, however, that any attempt at transgenic expression of an ORF14/15-like variant should take into account splicing in the purine-rich region. How photoreceptors make splicing choices in the ORF14/15 exon when presented in the context of a cDNA construct is unknown at present. Aberrant splicing may lead to the production of deleterious protein products. The latter possibility should sound a cautionary note in the design of somatic gene therapy designed to correct the photoreceptor disease caused by mutations in RPGR. 

 Supported by National Eye Institute Grant EY10309, the Foundation Fighting Blindness, and the Chatlos Foundation, Longwood, Florida.

 Submitted for publication February 14, 2002; revised June 27, 2002; accepted July 16, 2002.

 Commercial relationships policy: N.

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Table 1.

 

PCR Primers

Gene	Name	Sequence (5′-3′)
mRPGR	m2	ATGCCAAGAGGGTCGCGATGG
	RAP1	GCAGTTCAATTGATTGCCTGTGGTGG
	RAP2	CTGCGGCGAAGAGAGCGGGAGAGACCCC
	RAP3	GAGAAAGTGATGGAAAGTACACCGTGCAC
	RAP4	ATGATACGACCAGCCGAGATTCTGGAAGC
	RAP6	ATACCAGAGGAGCAGGAAGGACCAGAGG
	RAP14	CCTCTGAATCAATGGAGCCACTGGACTC
	RAP20	GATGACTACGAGTTCCAATGAGAAG
	ORFP4	CTGAGGGTGACGGGGATCAAATCTG
	R2	TGTGCTCGGTGACCATTGAAAC
	R16P2R	TTAAGGACTTCATCTAGCTTCTCATCAGTAATATC
	RAP20R	CTAATAACTTCTCATTGGAACTCGTAGTCATC
	RAP3R	GTGCACGGTGTACTTTCCATCAC
	ORFP4R	CAGATTTGATCCCCGTCACCCTCAG
	ORFPbR	CTGTAATATCTCCTTTCTTCTTTTCTAGCTCCTCTCCTT
	ORFP14R	GCAACTTTTCCAATTGCCTTCTTGTATTC
	ORF11R	GGACTCCATTGGCATTTTAGACGGC
	P7	GAGATGACTTCCCTGTTACTTCAATTCCAG
	P9	GGACAGAACACAAGATATGAGCTGAGGAGG
hRPGR	hRPGR2	CGTACTGCCCGTGGCATGAGG
	hEX11	AGGACCTCATGCAGCCAGAGGAACCAG
	h3R	TCTGACTGGCCATAATCGGGTCAC
mABCR	mABCRP1	CAAGGCGATGCCTTCAGCAGGACTG
	mABCRP2R	GATGACAGTGATTCCATACTCCTCAGGG
hABCR	ABCRP1	AAGGCGATGCCCTCAGCAGGAATGC
	ABCRP2R	CTGGAGGTGCTTGGATTTGTTCACC

Figure 1.

 

RACE analysis of RPGR transcripts in the mouse retina. (A) Diagram of the mouse RPGR gene structure based on published sequences. ^{7 9} Positions of the 5′ and 3′ RACE anchoring primers are shown below the diagram. TAA, translational stop codon. Introns are not drawn to scale. (B) 5′ RACE shows uniformity of splicing between exons 1 and 13 and heterogeneity within the ORF14/15 exon. (C) 3′ RACE confirms heterogeneous processing within the ORF14/15 exon. (D) Cloning and sequencing of major 3′ RACE products found three types of clones: those that terminated early in the ORF14/15 exon, those that have undergone heterogeneous splicing within the ORF14/15 exon, and those representing the constitutive transcript. A major polyadenylation site is 933 bases downstream from the ORF14/15 exon. Transcripts retaining intron 13 were found frequently.

Figure 2.

 

RT-PCR analysis of the ORF14/15-like transcripts in the retina. (A) Amplification of the ORF14/15 region gave products over a range of sizes but none was full length. The strong band appearing below the 1-kb marker was resolved into a collection of bands on longer run (not shown). They were derived from RNA templates, as shown by the absence of amplification if reverse transcription was omitted (−RT). Amplification of genomic DNA template is shown as a positive control for PCR reactions. RT reactions were performed with reverse transcriptase at 42°C or with a heat-stable AMV reverse transcriptase at 70°C. The sizes and positions of the expected full-length products are shown in the schematic diagram. (B) Sequence analysis shows that numerous splice site combinations were used. A small number of representative samples are shown. (C) All reading frames downstream from the splice junction were found, but larger transcripts tended to maintain the original reading frame of the full-length ORF14/15 exon. (D) Heterogeneous splicing in the purine-rich region in human RPGR ORF15 transcripts. Amplification of the ABCR (ATP-binding cassette transporter) transcript (>5 kb) indicated that the cDNA template was of sufficient quality to amplify long targets. (E) Mouse RPGR ORF14/15-like transcripts retaining intron 13 were well represented in the cytoplasmic RNA pool. The exon-intron structure of the amplified fragments are schematically represented on the right. Arrows: Primer locations.

Figure 3.

 

RT-PCR analysis of the constitutive RPGR variants. (A) Amplification of full-length RPGR constitutive variants. Diagram of the murine RPGR gene structure is shown at the top. Thin arched lines: segments removed by alternative splicing. Arrows: primer locations. (B) RPGR exon 15a variant accounted for a very minor fraction of the total. Const, constitutive.

Figure 4.

 

Analysis of transcript splicing using minigene constructs expressed in COS7 cells. (A) Diagrams of the minigene constructs. (B) RT-PCR analysis of RNA isolated from COS7 cells after transient transfection with the minigenes. In minigene −E16 which carries an obligate terminal ORF14/15 exon, splicing occurred within the purine-rich region. In minigene+E16, inclusion of ORF14/15 was repressed and exon 13 was joined directly to exon 16, generating the equivalent of a constitutive transcript.

Figure 5.

 

Analysis of RPGR protein products with a battery of antibodies. (A) Six RPGR antibodies targeting different regions are shown schematically. (B) RPGR proteins were detected over a wide range of sizes. The closely spaced bands around 95 kDa were coded by the constitutive transcripts. Products derived from the ORF14/15-like transcripts ranged in size from 70 to 200 kDa, with the highest abundance at 200 kDa. The membranes were stripped and reprobed with anti-synaptotagmin antibody as the loading control. (C) Detection with antibodies ORF1 and ORF2 provided direct confirmation that some of the proteins migrating around 200 kDa were encoded by the ORF14/15-like transcripts. (D) Expression of one ORF14/15-like variant, T12, produced a protein with an apparent molecular weight of approximately 200 kDa. (E) Correlation of the constitutive RPGR protein variants with their cognate coding transcripts was established by analysis with region-specific antibodies. Splicing of exon 17 to exon 19 led to a frame-shift and truncation of the C-terminus. Therefore, const. 3 was not recognized by the S2 antibody. a.a., amino acids; Const, constitutive variant; KO, knockout; WT, wild type.

Figure 6.

 

Immunofluorescence staining for RPGR protein variants (red). Frozen retinal sections from WT and RPGR KO (not shown) mice were stained with primary antibodies as indicated. The S1 antibody was common to all RPGR variants, whereas the S3 and ORF1 antibodies were specific for the constitutive and ORF14/15-like variants, respectively. Both were shown to be concentrated in the connecting cilia. Sections were counterstained with Hoechst dye 33342 to highlight the nuclear layers (blue). GC, ganglion cell layer; INL, inner nuclear layer; IPL, inner plexiform layer; IS, inner segment; ONL, outer nuclear layer; OS, outer segment; RPE, retinal pigment epithelium.

Figure 7.

 

A model for RPGR expression. Splicing decisions depend on whether the ESE elements are recognized within a limited time window before the emergence of the exon 16 in the primary transcript. Splicing within the ORF14/15 exon promotes transcriptional termination. Numerous ORF14/15-like variants are generated. If the ORF14/15 exon is not recognized in time, emergence of exon 16 then commits the splicing machinery to join exons 13 and 16 and produce the constitutive transcript.