Ablation of retinal ciliopathy protein RPGR results in altered photoreceptor ciliary composition

Cilia regulate several developmental and homeostatic pathways that are critical to survival. Sensory cilia of photoreceptors regulate phototransduction cascade for visual processing. Mutations in the ciliary protein RPGR (retinitis pigmentosa GTPase regulator) are a prominent cause of severe blindness disorders due to degeneration of mature photoreceptors. However, precise function of RPGR is still unclear. Here we studied the involvement of RPGR in ciliary trafficking by analyzing the composition of photoreceptor sensory cilia (PSC) in Rpgrko retina. Using tandem mass spectrometry analysis followed by immunoblotting, we detected few alterations in levels of proteins involved in proteasomal function and vesicular trafficking in Rpgrko PSC, prior to onset of degeneration. We also found alterations in the levels of high molecular weight soluble proteins in Rpgrko PSC. Our data indicate RPGR regulates entry or retention of soluble proteins in photoreceptor cilia but spares the trafficking of key structural and phototransduction-associated proteins. Given a frequent occurrence of RPGR mutations in severe photoreceptor degeneration due to ciliary disorders, our results provide insights into pathways resulting in altered mature cilia function in ciliopathies.


Proteomic analysis of PSC.
To test the role of RPGR in regulating protein trafficking into the cilia, we assessed the composition of the PSC of the Rpgr ko mice as compared to wild type mice. We performed MS/MS analysis of the PSC at 2 months and 4 months of age (Supplementary Tables S1 and S2 online, respectively). Each experiment consisted of 6 retinas from wild type and 6 retinas from Rpgr ko mice. Data are average of three biological replicates at each age. At 2 months of age, we detected 1366 proteins in wild type and Rpgr ko PSC. At 4 months, we detected a total of 1614 proteins in wild type and Rpgr ko PSC. As predicted, RPGR was not detected in the Rpgr ko PSC.
We then analyzed our MS/MS data to identify proteins that were increased or decreased in the in the Rpgr ko PSC as compared to wild type and categorize them based on their relative abundance. To this end, we applied the following criteria: (i) the average emPAI values had to be <0.5 fold if the abundance of the protein decreased and >2 fold if abundance increased and (ii) the number of unique peptides had to be greater than 3, if a protein is not detected in the other genotype. Based on these criteria, the proteins that increased or decreased are listed in Tables 1-4. At 2 months of age, we detected 28 proteins that decreased in amount by less than 0.5 fold in Rpgr ko PSC as compared to wild type. On the other hand, we detected 30 proteins at 4 months of age that showed less than 0.5-fold reduction in Rpgr ko PSC. Among the category of proteins whose relative abundance increased by >2-fold in Rpgr ko PSC at 2 months of age, we detected 17 proteins, whereas at 4 months of age, we found 24 proteins in Rpgr ko PSC.
It is becoming clear that the TZ also acts as a size-exclusion barrier for soluble proteins [36][37][38] . Interestingly, soluble proteins that reduced in abundance by 0.5-fold in Rpgr ko PSC decreased from 13 out of 28 at 2 months of age (46%; average molecular weight: 51.7 kDa) to 6 out of 30 at 4 months of age (20%; average molecular weight of 80 kDa) ( Table 5). Whereas 2 out of 17 proteins that increased in amount at 2 months of age (11.7%; average molecular weight of 99.5 kDa) are soluble proteins, 7 out of 24 proteins increased in amount at 4 months of age (~30%; average molecular weight of 99 kDa) are soluble proteins (Table 5). We therefore, suggest that the barrier for entry or retention of soluble proteins into the PSC is affected in the absence of RPGR.

Phototransduction proteins and known ciliopathy-associated proteins in Rpgr ko PSC.
Previous studies showed a reduction in photoreceptor function in the Rpgr ko mice 32 . We therefore, tested if this phenotype was due to an earlier defect in reduced translocation of selected phototransduction proteins to the OS. Our MS/MS analysis as well as validation by immunoblotting revealed no significant difference in the amount of arrestin, ROM1, and cyclic nucleotide gated channel CNGB1 in the Rpgr ko PSC as compared to wild type mice. These results suggest that photoreceptor dysfunction in the absence of RPGR in mice is not likely due to a defect in the trafficking of the tested phototransduction-associated proteins.
Alterations in BBS-associated proteins are associated with photoreceptor dysfunction among other ciliopathy disorders. Our analysis revealed BBS1, BBS2, BBS5, and BBS7 in the PSC; however, their levels did not alter between age-matched wild type and Rpgr ko PSC. Similarly, we did not detect differences in IFT polypeptides or associated Kinesin motor subunits in the Rpgr ko PSC.  Table 2. Proteins with reduced abundance in PSC of Rpgr ko at 4 months. S1, S2, and S3 are the three biological repeat samples used in the present study.

Proteins with altered abundance in Rpgr ko PSC. A majority of proteins that decreased in amount
in Rpgr ko PSC at both stages belonged to the category of ubiquitin-proteasome system (UPS) and cilia function (Tables 1 and 2 and Supplementary Figure S1). As the UPS is enriched at the cilium and has been detected in the PSC in earlier proteomics analyses (discussed later) 10,39 , we selected PSMD3 (26 S proteasome non-ATPase regulatory subunit 3) and Insulin degrading enzyme (IDE) for further analysis. Immunoblot analysis of PSC from wild type and Rpgr ko showed a reduction in the amount of both PSMD3 and IDE1 in the Rpgr ko PSC (Fig. 3). Total amount of these proteins did not alter, indicating that there was no change in the expression levels of PSMD3 and IDE in Rpgr ko retina. It was recently shown that proteasomal function is modulated by ciliary-centrosomal protein OFD1 39 . As we detected reduced abundance of OFD1 in the Rpgr ko PSC at 4 months of age, we performed immunoblot analysis using anti-OFD1 antibody. Our results validated the MS/MS data and showed reduced amount of OFD1 amount in Rpgr ko PSC (Fig. 3). No change in the total protein levels of OFD1 was detected in these analyses.
We found high representation of the category of membrane trafficking and intracellular transport among the proteins that were increased in Rpgr ko PSC (Tables 3 and 4 Figure S2). Specifically, we detected more than 5-fold increase in IQ-domain GTPase Activating Proteins IQGAP1, which is predicted to be involved in intracellular trafficking and neuronal regulation 40 . Given a crucial role of small GTPases and their regulation in the maintenance of PSC 28,41,42 , we validated the levels of IQGAP1 by immunoblot analysis. We found that indeed, amount of IQGAP1 is increased in the Rpgr ko PSC relative to the wild type (Fig. 3). No change in the total levels of IQGAP1 was detected.

and Supplementary
Localization of altered proteins in the retina. Although all the proteins identified in our dataset have been previously reported as part of the mouse PSC proteome 10 , the localization of IDE, PSMD3, OFD1 and IQGAP1 has not been examined in mammalian retina. To corroborate our findings, we performed immunofluorescence analysis of these proteins using adult mouse retina. Our analysis revealed that in addition to the inner segment (IS) and outer plexiform layer (OPL), IDE, PSMD3, OFD1, and IQGAP1 localize to the TZ, as determined by co-staining with ciliary markers acetylated α -tubulin (AcT) or detyrosinated tubulin (DeTyr) ( Fig. 4 and Supplementary Figure S3). Interestingly, PSMD3 was also detected in the OS. Consistent with a prominent role of OFD1 at the basal body, OFD1 co-localizes with γ -tubulin in photoreceptor IS (Supplementary Figure S3).
We next examined the effect of loss of RPGR on the localization of these proteins. Immunofluorescence analysis of age-matched Rpgr ko mouse retinas revealed interesting observations (Fig. 4). We found reduced IDE and OFD1 associated signal reduced in the TZ with a concomitant mislocalization in the IS and outer nuclear layer (ONL) of Rpgr ko retina (Fig. 4A,B). PSMD3 staining revealed strikingly reduced staining in the OS and increased in TZ/IS of Rpgr ko retina (Fig. 4C).

Discussion
Although RPGR is widely expressed and regulates ciliary trafficking, it is puzzling that RPGR mutations in humans do not result in typical cilia-associated systemic and developmental or early-onset defects. Rather, RPGR mutations are one of the most common causes of severe photoreceptor degenerative diseases of adulthood 19,43 . Photoreceptor cilia are unique with respect to their demands of maintenance of ciliary function. They undergo periodic shedding of distal tips of cilia with concomitant renewal of membrane and proteins at the base. It is estimated that about 10% of the OS tips are shed each day 44,45 . This is accompanied by massive transport of opsin and other OS proteins to the cilium 46,47 . Hence, it is conceivable that maintenance of ciliary function plays a crucial role in photoreceptor health. Our findings suggest that RPGR is one such regulator of functional maintenance of mature photoreceptors. RPGR accomplishes this task by likely modulating the docking and trafficking of key proteins involved in proteasomal function and intracellular protein trafficking.
Selecting 2 and 4 months of age of Rpgr ko mice provided a unique opportunity to ascertain defects in photoreceptor protein trafficking that are likely to cause photoreceptor degeneration rather than a secondary effect of degeneration. Earlier studies using immunofluorescence analyses had revealed mistrafficking   Table 4. Proteins with increased abundance in PSC of Rpgr ko at 4 months. S1, S2, and S3 are the three biological repeat samples used in the present study.
of opsins in Rpgr ko retina at 6 months of age 32 . Our results indicate that loss of RPGR may not directly affect trafficking of opsin or other phototransduction proteins and the mislocalization of opsin observed in the Rpgr ko retina could have been an effect of degenerating photoreceptors. Support of this hypothesis comes from several previous studies showing opsin mislocalization as a common phenotype across multiple gene mutations associated with retinal degeneration due to photoreceptor dysfunction 48 .
Membrane proteins targeted to cilia may dock at the periciliary membrane and then are loaded onto axonemal cargo carriers for delivery to the OS via the TZ 38 . However, fate of soluble proteins is not completely understood. It was reported that small soluble molecules such as GFP could equilibrate between inner and outer segments in frog photoreceptors 49 . Moreover, it was shown that arrestin and transducin diffuse through the TZ 50-52 . These findings suggest an absence of a diffusion barrier in photoreceptors, at least for smaller proteins. We detected a specific alteration in the levels of soluble proteins in the Rpgr ko PSC. We found an overall increase in higher molecular weight soluble proteins at 4 months of age as compared to 2 months. As ciliary base is considered analogous to the nuclear pore complex and the role of another retinal ciliopathy protein RP2 in maintaining the function of nuclear pore proteins at the base of cilia 53,54 , it is conceivable that there is a size-exclusion barrier of soluble proteins at the TZ of photoreceptors. This barrier is likely maintained by RPGR and its interacting proteins. Additional analyses of the effect of RPGR and its multiprotein complexes on the integrity of such a barrier should provide crucial insights into the precise role of TZ of cilia.
Our data point to an important role of UPS in the pathogenesis of RPGR-associated photoreceptor degeneration. In addition to identifying several UPS components that show varying levels in Rpgr ko PSC as compared to wild type, validation of three proteins that are directly or indirectly involved in UPS provide further support to the association of the UPS in cilia-dependent XLRP pathogenesis. We provide four scenarios to support this hypothesis: (i) among the UPS components identified in our PSC proteome, PSMD3 showed reduced levels at both 2 and 4 months of age in the Rpgr ko PSC. A previous report on the involvement of another PSMD subunit PSMD13 in modulating RPE65 levels to regulate retinal health further supports the role of PSMD proteins in retinal health 55 . Moreover, UPS dysfunction has been reported in several retinal degenerative diseases 56 . The UPS is concentrated at the PSC and is implicated in regulating levels of phototransduction and other PSC proteins to cope with immense oxidative stress in photoreceptors 10 . Decrease in PSMD3 levels indicates that Rpgr ko photoreceptors have reduced ability to cope with oxidative stress that accumulates over time; (ii) the UPS regulates insulin signal transduction by modulating degradation of key components, such as insulin receptor substrates 57 . Insulin, in turn, can inhibit proteasome and this action is dependent upon IDE 58 , which is also reduced in the Rpgr ko PSC; (iii) PSMD3 modulates insulin resistance in association with polyunsaturated fatty acids (PUFAs) 59 , which are a key component of photoreceptor OS membranes [60][61][62] ; (iv) OFD1, which is also reduced in Rpgr ko PSC, is shown to modulate proteasome function 39 . Given that an intronic mutation in OFD1 is associated with XLRP 63 , we reckon that RPGR and OFD1 play overlapping roles in the manifestation of XLRP pathogenesis. Validation of these hypotheses warrants further investigations.
Vesicular trafficking by small GTPases in photoreceptors is critical for their function and survival 28,41 . Our analyses revealed abundant quantities of small GTPase regulators, such as IQGAP1 in the PSC of Rpgr ko mice. IQGAPs are not only involved in intracellular trafficking, they are also implicated in actin-microtubule interaction and regulation of cytoskeletal dynamics. Commensurate with this, our analysis revealed alterations in levels of proteins belonging to the families of cytoskeleton-based vesicular trafficking to cilia. These results corroborate previous findings that RPGR regulates protein trafficking likely by modulating the activity of GTPases or regulators of GTPase activity 28 .
It should be noted that there is not much overlap between proteins whose abundance changes between 2 and 4 months. We posit that the proteins that show increased abundance at 2 months in the PSC may not be as abundant at 4 months due to progression of disease condition and associated subtle variations that may affect their abundance in the PSC in Rpgr ko mice. Hence, such proteins would not make the cutoff of the threshold in our analysis. On the other hand, proteins that show reduced abundance at 2 months but are no longer detected at 4 months may reflect reduction in overall expression levels of such proteins. This would reduce to such an extent even in a wild type mouse retina that the variations are no longer significant in Rpgr ko are and hence, do not make the cut off in our analyses.  Table 5. Soluble proteins with altered in abundance in Rpgr ko PSC. No predicted post-translational modifications to allow membrane association were detected in these proteins. MW: predicted molecular weight; kDa: kilo Daltons. Regulators of protein trafficking, such as RPGR are involved in maintaining the structure and function of mature cilia. Such functions are specifically crucial in mature neurons, which are under immense oxidative stress. Defects in RPGR result in subtle but incremental insult, which result in neurodegeneration. Our studies thus, provide clues to understanding pathways involved in the maintenance of mature cilia, which will also assist in delineating the pathogenesis of retinal degeneration in systemic ciliopathies.

Methods
Mice. Rpgr ko mice (in C57BL6/J background) were procured from Dr. Tiansen Li (National Eye Institute) and were characterized earlier 32 . Control C57BL6/J mice were obtained from The Jackson Laboratories (Bar Harbor, ME). Both strains of mice were reared in the same animal facility at UMASS Medical School. All methods were carried out in accordance with the approved guidelines. All experimental protocols were approved by Institutional Animal Care and Use Committee and Institutional Biosafety Committee of UMASS Medical School.

PSC preparation.
Mouse PSC was prepared essentially as described 10,64 . Briefly, fresh mouse eyes were dissected and retinas were placed in a tube with 150 μ l of 8% OptiPrep (Nycomed, Oslo, Norway) prepared in Ringer's buffer (130 mM NaCl, 3.6 mM KCl, 2.4 mM MgCl 2 , 1.2 mM CaCl 2 , 10 mM HEPES, pH 7.4, containing 0.02 mM EDTA) and vortexed for 1 min. The samples were centrifuged at 200 × g for 1 min, and the supernatant containing the PSC was transferred to fresh eppendorf tube. The resultant pellet was dissolved in 150 μ l of 8% OptiPrep, vortexed, and centrifuged again at 200 × g for 1 min. These steps were repeated five times. The PSC was pooled (∼ 2 ml), overlaid on a 10-30% continuous gradient of OptiPrep in Ringer's buffer, and centrifuged for 50 min at 26,500 × g. PSC were collected from second band (about two-thirds of the way from the top), diluted three times with Ringer's buffer, and centrifuged for 3 min at 500 × g to remove the cell nuclei. The supernatant containing PSC was transferred to a new tube and centrifuged for 30 min at 26,500 × g. The pelleted material contained pure intact PSC.
Immunofluorescence and Immunoblotting. For staining of the isolated PSC complex preparations, 10 μ l of the suspension was spotted on glass slides and fixed in 4% paraformaldehyde in PBS (pH 7.4) for 15 minutes. Mouse retina was processed for immunofluorescence as described 26,65,66 . For staining with anti-PSMD3 and anti-IDE antibodies, consistent results were obtained when retinas were processed using paraffin-embedding 65,66 . Slides containing PSC and mouse retinal sections were washed with PBS, permeabilized, and blocked for 1 hour with blocking solution containing 5% normal goat serum with 0.5% Triton X-100 in PBS in a humidifying chamber at RT. Primary antibodies were added and slides were incubated overnight at 4 °C. After washing three times with PBS, samples were incubated for 1 hour with goat anti-rabbit (or mouse) Alexa Fluor 488 nm or 546 nm secondary antibody at room temperature. Hoechst 33342 (Life Technologies Corp.) was diluted with PBS to final 1 μ g/mL and used to label the nuclei. Samples were washed with deionized water and then mounted (Fluoromount; Electron Microscopy Sciences, Hatfield, PA) under glass coverslips and visualized using a microscope (Leica TCS SP5 II laser; Leica Microsystems).
For Immunoblotting, equal amount of protein samples (50 μ g) were denatured at 95ºC for 5 min in 10 mm Tris-HCl, pH 6.8, 4% SDS, 20% sucrose, 4% β -mercaptoethanol and separated on SDS-polyacrylamide gels. Proteins were transferred onto Immobilon-FL membranes (Millipore, Bedford, MA). Blots were blocked in 5% nonfat milk in Tris-buffered saline containing 0.1% Tween 20 (TBST) for 30 min and labeled with the primary antibody in 5% nonfat milk prepared in TBST for overnight. The blots were washed with TBST and labeled with a secondary anti-mouse or anti-rabbit antibody conjugated to HRP and processed for chemiluminescence reaction.
Proteomic Analysis. Analyses of three biological replicates were performed using 6 retinas each from wild type and Rpgr ko mice (total 18 retinas from each genotype). Samples were run ~1.5 cm into the resolving gel of 10% minigels (BioRad, Hercules, CA) and protein-containing regions were excised, destained, and cut into 1 × 1 mm pieces. Gel pieces were then placed in 1.5-ml eppendorf tubes with 1 ml of water for 30 min. The water was removed and 100μ l of 250 mM ammonium bicarbonate was added. For reduction 20 μ l of a 45 mM solution of 1, 4 dithiothreitol (DTT) was added and the samples were incubated at 50 C for 30 min. The samples were cooled to room temperature and alkylated by adding 20 μ l of a 100 mM iodoacetamide solution for 30 min. Gel slices were washed twice with 1 ml water aliquots. The water was removed and 1 ml of 50:50 (50 mM Ammonium Bicarbonate: Acetonitrile) was placed in each tube and samples were incubated at room temperature for 1 hr. The solution was then removed and 200 μ l of acetonitrile was added to each tube at which point the gels slices turned opaque white. The acetonitrile was removed and gel slices were further dried in a Speed Vac. Gel slices were rehydrated in 50 μ l of 2 ng/μ l trypsin (Sigma) in 0.01% ProteaseMAX Surfactant (Promega): 50 mM Ammonium Bicarbonate. Additional bicarbonate buffer was added to ensure complete submersion of the gel slices. Samples were incubated at 37 o C for 21hrs. The supernatant of each sample was then removed and placed in a separate 1.5 ml eppendorf tube. Gel slices were further dehydrated with 100 μ l of 80:20 (Acetonitrile: 1% formic acid). The extract was combined with the supernatants of each sample. The samples were then dried down in a Speed Vac. Samples were dissolved in 25 μ l of 5% Acetonitrile in 0.1% trifluroacetic acid prior to injection on LC/MS/MS. LC/MS/MS on Q Exactive. A 3 μ l aliquot was directly injected onto a custom packed 2 cm x 100 μ m C 18 Magic 5 μ particle trap column. Peptides were then eluted and sprayed from a custom packed emitter (75 μ m x 25 cm C 18 Magic 3 μ m particle) with a linear gradient from 95% solvent A (0.1% formic acid in water) to 35% solvent B (0.1% formic acid in Acetonitrile) in 90 minutes at a flow rate of 300 nl per minute on a Waters Nano Acquity UPLC system. Data dependent acquisitions were performed on a Q Exactive mass spectrometer (Thermo Scientific) according to an experiment where full MS scans from 300-1750 m/z were acquired at a resolution of 70,000 followed by 12 MS/MS scans acquired under HCD fragmentation at a resolution of 35,000 with an isolation width of 1.2 Da. Data Analysis. Raw data files were processed with Proteome Discoverer (version 1.3) prior to searching with Mascot Server (version 2.4) against the Uniprot_Mouse database. Search parameters utilized were fully tryptic with 2 missed cleavages, parent mass tolerances of 10 ppm and fragment mass tolerances of 0.05 Da. A fixed modification of carbamidomethyl cysteine and variable modifications of acetyl (protein N-term), pyro glutamic for N-term glutamine, and oxidation of methionine were considered. Search results were loaded into the Scaffold Viewer (Proteome Software, Inc.) for comparisons of sample results between wild type and Rpgr ko PSC. Prior to additional analyses, major contaminants such as keratins, ribosomal and mitochondrial proteins were removed from our PSC proteome to increase the specificity of identified proteins.
We used previously reported stringency criteria to enrich specifically altered proteins. All MS/MS samples were analyzed using Mascot search engine (Matrix Science, London, UK; version 2.4.0). Mascot was set up to search the SwissProt_062613 database (selected for mice) with a fragment ion mass tolerance of 0.050 Da and a parent ion tolerance of 10.0 PPM. Scaffold (version Scaffold_4.3.4, Proteome Software Inc., Portland, OR) was used to validate MS/MS based peptide and protein identifications. Peptide identifications were accepted if they could be established at greater than 95.0% probability by the Peptide Prophet algorithm 67,68 with Scaffold delta-mass correction. Protein identifications were accepted if they could be established at greater than 95.0% probability and contained at least 3 identified peptides. We put false discovery rate (FDR) cutoff ~<1, which means that more than 99% of the proteins in our list were true proteins.
Quantitative proteomic analysis. Spectral counting has been used for differentiating the abundance of protein amount in complex mixture of different biological samples. However protein quantification by spectral counting would be biased by length of proteins because larger proteins generate more spectral counts as compared to smaller proteins. For our quantitative proteomic analysis, we adopted a previously described method called exponentially modified protein abundance index (emPAI) 69 . emPAI is widely used method in quantitative comparative proteomics and offers approximate, label-free, relative quantitation of the proteins in a mixture based on protein coverage by the peptide matches in a database search result. PAI (protein abundance index) of each protein was calculated by dividing number of observed peptides to the number of observable peptides per protein. which is proportional to protein content in a protein mixture. Solubility prediction of identified proteins was calculated using Expropriator (http://mips.helmholtz-muenchen.de/proso/proso.seam) 70 .
Functional Analysis. After finding proteins, which are differently expressed, we performed pathway analysis by using DAVID (Database for Annotation, Visualization and Integrated Discovery) classification system. We analyzed differential expression of proteins based on >2 fold change, for both down regulated and up regulated, from overall wild type and knockout datasets.