AAV-Mediated Clarin-1 Expression in the Mouse Retina: Implications for USH3A Gene Therapy

Usher syndrome type III (USH3A) is an autosomal recessive disorder caused by mutations in clarin-1 (CLRN1) gene, leading to progressive retinal degeneration and sensorineural deafness. Efforts to develop therapies for preventing photoreceptor cell loss are hampered by the lack of a retinal phenotype in the existing USH3 mouse models and by conflicting reports regarding the endogenous retinal localization of clarin-1, a transmembrane protein of unknown function. In this study, we used an AAV-based approach to express CLRN1 in the mouse retina in order to determine the pattern of its subcellular localization in different cell types. We found that all major classes of retinal cells express AAV-delivered CLRN1 driven by the ubiquitous, constitutive small chicken β-actin promoter, which has important implications for the design of future USH3 gene therapy studies. Within photoreceptor cells, AAV-expressed CLRN1 is mainly localized at the inner segment region and outer plexiform layer, similar to the endogenous expression of other usher proteins. Subretinal delivery using a full strength viral titer led to significant loss of retinal function as evidenced by ERG analysis, suggesting that there is a critical limit for CLRN1 expression in photoreceptor cells. Taken together, these results suggest that CLRN1 expression is potentially supported by a variety of retinal cells, and the right combination of AAV vector dose, promoter, and delivery method needs to be selected to develop safe therapies for USH3 disorder.


Introduction
Usher syndrome (USH) is an autosomal recessively inherited disorder responsible for more than half of combined deafness and blindness in humans [1,2]. It represents a group of clinically and genetically heterogeneous disorders, classified into three major types, USH1, USH2, and USH3, depending on the onset and severity of the symptoms, and the presence or absence of vestibular function [3][4][5][6]. It affects both photoreceptors and cochlear hair cells, the sensory blocked by other binding partners, preventing a reliable identification of its cellular localization in the retina [33]. One immunohistochemistry study showed that the protein is present in mouse photoreceptor cells [35], while other researchers concluded that its mRNA originates exclusively from the inner retina [33]. In this work, we used an adeno-associated vector (AAV) to examine in vivo the preferred subcellular localization of CLRN1 in a variety of cell types following intravitreal or subretinal delivery in adult mice. Although this approach does not identify the specific retinal cell type(s) where CLRN1 is endogenously produced, the subcellular localization of AAV-expressed CLRN1 should provide useful information about its function. In addition, since light-driven translocation of phototransduction proteins was previously used to detect retinal defects in several mouse models of retinal disease [36][37][38][39], we also investigated whether Clrn1 KO mice displayed any abnormalities in the compartmentalization of arrestin-1 and α-transducin following light exposure.

Ethics Statement and Animals
The generation and characterization of Clrn1 KO mice have been previously described [33]. Adult Clrn1 KO and corresponding age-matched wild-type isogenic controls (at least 3-months of age), as well as adult C57BL/6 mice were used in all experiments. Animals were maintained on a cycle of 12 hour of light (200 lux) and 12 hour of darkness in the University of Florida Health Science Center Animal Care Services Facility (Gainesville, FL). All experiments were approved by University of Florida Institutional Animal Care and Use Committees and conducted in accordance with the Association for Research in Vision and Ophthalmology (ARVO) Statement for the Use of Animals in Ophthalmic and Vision Research.

Preparation and Delivery of the CLRN1-AAV vector
A self-complementary AAV2 quadruple capsid tyrosine mutant (Y272,444,500,730F) vector (scAAV2quadYF) was used for packaging either a C-terminal Venus-tagged human clarin-1 (CLRN1-Venus, from Dr. Imanishi at Case Western University) or an HA-tagged CLRN1 cDNA under the control of the ubiquitous smCBA (small chicken β-actin) promoter. An AAV vector was also generated containing the reporter mCherry driven by a proximal 2000 nucleotide mouse Clrn1 promoter region [40]. The viral vectors were produced and purified according to previously reported methods [41]. Viral titer was determined by real-time PCR. Subretinal and intravitreal injections were performed in adult wild-type mice under anesthesia as described [42,43]. In brief, the nasal cornea was penetrated with a 30.5-gauge disposable needle and a 33-gauge unbeveled, blunt-tip needle on a Hamilton syringe was introduced into the subretinal space. For intravitreal injections, a 33-gauge beveled needle was passed through the sclera, at the equator, next to the limbus, and into the vitreous. Each right eye received 1 μl of AAV vector (approximately 1x10 13 genome copies/ml), while the left eye remained uninjected. At approximately 6 to 8 weeks post-injection, the CLRN1-Venus was detected directly by visualization of Venus (a variant of YFP) fluorescence in retinal sections, while the CLRN1-HA tagged protein was visualized by immunostaining with a high-affinity anti-HA-Fluorescein conjugated antibody (3F10, Roche Diagnostics, Indianapolis, IN). A mouse monoclonal antibody (B6-30) against rod opsin [44] was used to label the outer segments.

Electroretinographic analysis (ERG)
Treated C57BL/6 mice were dark-adapted (>12 hr) and anesthetized with a mixture of ketamine (72 mg/kg)/xylazine (4 mg/kg) by intraperitoneal injection, and ERG experiments were conducted at 2 months following subretinal injections with the AAV2 quad YF vector carrying the CLRN1-HA-tagged transgene as described above. For toxicity assays, one eye in each mouse received one microliter of either a full strength titer of approximately 10 13 vector genome/ml (vg/mL) or serially diluted vectors of 10 12 , 10 11 , or 10 10 vg/mL. Pupils were dilated with topical agents (1% atropine sulfate, 2.5% phenylephrine hydrochloride) under dim red light. Full-field ERGs were recorded as previously described [42], using a UTAS Visual Diagnostic System (LKC Technologies, Gaithersburg, MD). Dark-adapted ERGs were elicited with 0.02, 0.2 and 2 scot-cdÁsecÁm -2 stimuli. Differences in maximum b-wave amplitudes between AAV CLRN1-HA -injected and uninjected contralateral left eyes were analyzed by Student t test (GraphPad Prism 6.0, GraphPad Software, San Diego, CA), and considered statistically significant if p<0.05. All ERG data are presented as mean ± SEM.

Light-driven translocation experiments and immunohistochemical analysis
All mice were dark adapted for at least 12 hours prior to the light adaptation. Dim red light was used for all procedures in the dark room. For light exposure, unanesthetized Clrn1 KO and wildtype mice were placed in a clear plastic box and exposed to light (1000 lux intensity at the cage level) for 1 hour. Following euthanasia, the eyes were enucleated, placed into vials containing 4% paraformaldehyde solution, and fixed for at least 4 hours prior to moving them into 1× phosphate-buffered saline (PBS). The eyes were processed for paraffin embedding and sectioned at a thickness of 4 μm, at the same retinal locations for both wild-type and knock-out eyes. For immunostaining, deparaffinized tissue sections were dewaxed in xylene and rehydrated in a graded series of ethanol, and then incubated with 0.5% Triton X-100 for 15 min, followed by blocking with a solution of 1% albumin for 1 hour. Wild-type and Clrn1 KO retinal sections were processed simultaneously under identical conditions. Arrestin-1 was detected either with mouse monoclonal antibody SCT-128 [45], or C10C10 [46] at 1:200 dilution. The α-transducin subunit was detected using a rabbit polyclonal antibody (sc-389, Santa Cruz Biotechnology, Dallas, TX) at 1:1000 dilution. Secondary antibodies conjugated to Alexa-594 or Alexa-488 fluorophore (Molecular Probes/ Invitrogen, Eugene, OR) were used at 1:500 dilution in PBS. All sections were examined by fluorescence microscopy, using either a Leica TCS SP2 laser scanning confocal microscope (Leica, Heidelberg, Germany) or an Axiophot microscope (Zeiss, Thornwood, NY). Vertical densitometric profile plots representing the average fluorescent signal intensity across the photoreceptor layer were generated with ImageJ's Plot Profile function for both arrestin-1 and α-transducin subunit, using the DAPI nuclear stain to determine the outer plexiform layer (OPL) edge as described [47].

Statistical Analysis
The fluorescence intensity of the immunolabeled proteins was analyzed as previously described [48,49]. ImageJ software (National Institutes of Health) was used for measuring the relative intensity of the fluorescence signal in at least three different randomly located regions from the mid-central retinas of wild-type and Clrn1 KO mice. The relative signal intensity in the OPL was expressed as a percentage of the total intensity in the photoreceptor layer. Statistical differences between the two groups were assessed by unpaired t test (GraphPad Prism 6.0, GraphPad Software, San Diego, CA). Error bars represent the mean ± standard error of mean (SEM). Results are considered statistically significant if Ã p<0.05.

Electron microscopy
For ultrastructural analysis of murine photoreceptor cells, enucleated eyes from wild-type and Clrn1 KO mice were fixed in 2.5% glutaraldehyde in 0.1 M cacodylate buffer containing 0.1 M sucrose for 1.5 h at 4°C in total. After 30 min fixation the cornea and lens were removed and the eye cups were fixed for additional 1 h. Subsequently, eye cups were washed with 0.1 M cacodylate buffer containing 0.1 M sucrose for 30 min. Next, eye cups were fixed with 2% osmium tetroxide in 0.1 M cacodylate buffer containing 0.1 M sucrose for 1 h at room temperature, followed by dehydration in ethanol (30-100%) and embedding in Renlam M-1 resin (Serva Electrophoresis, Heidelberg, Germany). Ultrathin sections, counterstained with ethanolic uranyl acetate and lead citrate, were analyzed in transmission electron microscope (Tecnai 12 BioTwin, FEI, Netherlands). Images were obtained with a CCD camera (charge-coupleddevice camera; SIS MegaView3; Surface Imaging Systems, Herzogenrath, Germany) and processed with Adobe Photoshop CS (Adobe Systems).

Results
An AAV2 quad YF vector carrying a CLRN1-Venus transgene cDNA driven by a smCBA promoter (scAAV2quadYF-smCBA-CLRN1-Venus) was delivered to adult wild-type mouse eyes via either the subretinal or intravitreal route in order to examine the pattern of CLRN1 distribution by direct fluorescence imaging. We determined that CLRN1 expression is supported by a variety of cells, including RPE, rod and cone photoreceptors, amacrine, bipolar, horizontal, ganglion and Müller cells (Figs 1 and 2). Of these, photoreceptors have a unique compartmentalized morphology that makes them amenable for sub-cellular localization studies. Subretinal delivery, which primarily targets photoreceptors and the RPE, led to robust CLRN1-Venus protein expression at the inner segments, outer nuclear cell body membrane and the outer plexiform layer (Fig 1).
Strong CLRN1-Venus fluorescence following intravitreal delivery was found in the outer and inner plexiform layers where the synaptic contacts are formed in the retina (Fig 2). CLRN1-Venus was detected in various retinal neurons, including horizontal, bipolar and amacrine cells, identified based on their location and morphology (Fig 2A). In addition, CLRN1 expression in Müller glia was apparent throughout the retina, concentrating particularly at their apical microvilli that extend upward from the outer limiting membrane, as well as in Müller cell processes surrounding the outer plexiform layer (Fig 2B). CLRN1 was also detected prominently at the nerve fiber layer formed by the ganglion cell axons and extended towards and into the optic nerve ( Fig 2C). Although previous studies in cochlear hair cells showed that CLRN1 cellular localization is not affected by the presence of relatively large reporter proteins such as YFP (27kDa) on the C-terminal end of CLRN1 [31,50], we also performed experiments in which CLRN1 was tagged with the much smaller hemagglutinin (HA) sequence and its expression was detected by immunostaining. Subretinal and intravitrealy delivered CLRN1-HA vector led to a similar pattern of expression as CLRN1-Venus. Intravitrealy delivered CLRN1-HA led to strong labeling of numerous Müller cells spanning the entire thickness of the retina, and inner retinal neurons with their extensive dendrites in the inner plexiform layer (Fig 2D). Subretinal CLRN1-HA vector led to expression at the photoreceptors inner segments, outer nuclear cell body membrane and synaptic regions, as well as the RPE apical membrane (Fig 3B).
Full field electroretinographic (ERG) analysis was performed to determine if the AAVmediated CLRN1-HA overexpression as driven by the strong CBA promoter had potential detrimental effects on retinal function ( Fig 3A). There was a significant decrease in the ERG amplitudes in treated versus untreated mouse eyes following subretinal injections with the full titer AAV2 quad YF vector carrying the CLRN1-HA-tagged transgene. For example, AAVinjected eyes which received a full vector dose of approximately 10 10 vg, had a b-wave amplitude average (at 2 scot-cdÁsecÁm -2 ) of 315.5±39.16 μV, compared to untreated controls, with a maximum b-wave amplitude average of 539.5± 33.28μV (average ± SEM, n = 7, p = 0.0005). At a much lower vector dose of approximately 10 7 vg, there was no statistically significant difference between control uninjected and AAV-CLRN1 injected eyes. Examination of subretinally injected CLRN1-HA-treated retinas following treatment with one of the lower vector doses (10 8 vg) showed strong CLRN1 expression at the inner segment region, with weak labeling of the OPL, and normal outer nuclear layer thickness (Fig 3B).
We also tested a construct containing the 2kb region from the endogenous mouse Clrn1 promoter driving the mCherry reporter protein following subretinal delivery. This 2000 nucleotide long CLRN1 promoter sequence is located upstream of exon 0, the translation start site of the primary CLRN1 splice variant [40]. We determined that this promoter does not restrict transgene expression specifically to either photoreceptors or RPE cells, as mCherry fluorescence was detected in both cell types (S1 Fig). However, transgene expression was low and localized mainly near the injection site, where vector particles would have had the highest concentration (S1 Fig, right panel), suggesting that this promoter is relatively weak.
Since mutations in other Usher proteins have been shown to cause defects in protein translocation in response to light [37,38], we assessed whether the lack of CLRN1 protein in KO mice leads to any abnormalities in the translocation pattern of arrestin-1 and α-transducin, which move in opposite directions between the inner and outer segments within photoreceptor cells in response to light. This movement of key phototransduction proteins through the connecting cilium plays an essential role in photoreceptor maintenance and survival, and can be used as a sensitive tool to assess subtle pathological changes in diseased photoreceptors that appear normal before degeneration starts [51,52]. Arrestin-1 is an abundant soluble cytoplasmic protein that undergoes significant movement from the inner compartments of the photoreceptor cells towards outer segments in response to light in order to associate with lightactivated phosphorylated rhodopsin with high affinity and specificity, preventing any further interaction with the G-protein transducin [53,54]. In the dark, arrestin-1 was localized primarily to the inner segments, outer nuclear layer and synapses of photoreceptor cells, while αtransducin was found in the outer segments in both Clrn1 KO and wild-type mouse photoreceptors (S2 Fig). In contrast, light-dependent translocation of Arrestin-1 but not α-transducin in Clrn1 KO and isogenic wild-type controls, assessed by using previously described methods [47,48], showed distinct differences. Following light exposure, arrestin-1 displayed the expected directional movement from the inner photoreceptor compartments into the outer segments in both wild-type and Clrn1 KO retinas (Fig 4). Although the direction of arrestin-1 redistribution is similar in both types of mouse retinas, the immunostaining pattern in the Clrn1 KO mice was different than wild-type, with more arrestin-1 remaining in the outer plexiform layer. We used a semiquantitative evaluation of arrestin-1 immunoreactivity to determine if there were significant changes in arrestin-1 OPL localization in KO relative to wild-type retinas ( Fig 4B). The relative signal intensity of arrestin1 in the OPL was significantly greater in the Clrn1 KO group compared to wild-type control mice (17.17±0.93 in KO vs. 10.45±0.52 in WT, p<0.0001). Light-exposed wild-type mice displayed a weak residual fluorescence staining signal in the OPL, suggesting that arrestin-1 is never entirely removed from its synaptic location following light exposure. However, the majority of Clrn1 KO mice examined under identical conditions had a much higher arrestin-1 immunoreactive signal at regions comprising the photoreceptor portion of the OPL. Rod transducin-α however translocated similarly towards the inner compartments and OPL of photoreceptors in both Clrn1 KO and wild-type of mice (S3 Fig). In contrast to arrestin-1, the mean relative intensity of α-transducin immunoreactivity in the OPL of Clrn1 KO mice was not significantly different from the wild-type (16.95±1.57 in KO vs 15.97±1.63 in WT, p = 0.659, Fig 4B). In conclusion, α-transducin translocation in rods is not affected by the lack of CLRN1 under our experimental light conditions.
The finding that more arrestin-1 remained confined within the OPL of Clrn1 KO mice relative to wild-type controls following light exposure prompted us to examine in more detail the ultrastructure of the ribbon synapses here using electron microscopy. The presynaptic ribbon  of photoreceptor cells constitutes a large, dynamic, electron dense structure that is covered by hundreds of synaptic vesicles and displays a high degree of structural plasticity [55]. No obvious structural differences in the ribbon synapses of wild-type and Clrn1 KO mice were observed (Fig 5), suggesting that the ultrastructural morphology of photoreceptor synapses is not affected by CLRN1 deficiency.

Discussion
CLRN1 was originally suggested to have a synaptic role based on its sequence homology with stargazin, also a tetraspanin, which regulates the synaptic targeting of glutamate receptors in the brain [9]. In the cochlea, CLRN1 was shown to be important for maintaining the stability of the F-actin rich stereocilia [31][32][33], as well as playing a role in the synaptic maturation of hair cells [56]. However, the role of CLRN1 in the retina remains poorly understood mainly because USH3 mouse models have normal retinal function and morphology, and reports on its endogenous cellular localization are conflicting. Previous studies indicate that Clrn1 is expressed either in the inner or outer retina, depending on the method used for analysis. One study found that Clrn1 mRNA is expressed in a developmentally regulated manner in Müller cells, and its expression declines to undetectable levels from the inner retina as determined by in situ hybridization [33]. Interestingly, we determined that a 2kb region from the endogenous mouse Clrn1 promoter resulted in non-uniform transgene distribution, concentrated mainly at the injection site in photoreceptors and RPE cells, with undetectable expression levels in the inner retina. Although this promoter region does not contain all the elements required for the endogenous transcriptional control of CLRN1 expression in various cell types, its in vivo characterization suggests that its activity is considerably weaker than CBA, in agreement with previous cell culture experiments [40], and that CLRN1 may indeed be absent or only expressed at low levels in the inner retina. One study using a custom-made antibody indicated that the protein was expressed at the base of the connecting cilium, inner segment and ribbon synapses of photoreceptors [35]. In contrast, RT-PCR experiments from retinal extracts with or without photoreceptor cells concluded that Clrn1 mRNA originates exclusively from the inner retina [33]. Siegert et al. created an online gene expression profile database by using several transgenic mice and showed that Clrn1 mRNA was found predominantly in starburst amacrine cells [57]. In the zebrafish, the protein was detected both in the inner and outer retina, with strong expression seen at the lateral contacts between photoreceptor inner segments, outer limiting membrane (OLM), and subapical processes from Müller cells, a region containing proteins involved in cell junctions important for photoreceptor stability [58]. Several USH proteins have been localized at the actin-rich OLM [24], including protocadherin 15 [59], with which CLRN1 was previously shown to interact [60].
In this study, we opted for an AAV-based approach to express a CLRN1-Venus fusion protein and examine its distribution by direct imaging of Venus fluorescence in mouse retinal sections. We show that following AAV-mediated delivery, a variety of cells can support CLRN1 expression, including RPE, photoreceptors, ganglion, amacrine, and Müller glia. This is in contrast to the behavior of other AAV-expressed transgenes described previously, in which a similar CBA promoter was used and the vector targeted multiple cell types [61]. Those studies showed that following AAV transfer, transgene expression usually occurred only within those cells from which it was naturally derived, a phenomenon thought to result from the instability of a specific mRNA or protein within the environment of non-endogenous cells [42,61,62]. The presence of AAV-expressed CLRN1 in a variety of cell types therefore suggests that off-target cells are capable of producing and maintaining this protein when using a ubiquitous promoter, which has important implications for the design of USH3 gene therapy studies, and for developing a safe CLRN1 vector delivery system to prevent the photoreceptor cell degeneration in USH3 patients. Taken together, previous studies based on RT-PCR indicate that CLRN1 is endogenously expressed at low levels [33,57]. In order to avoid potential toxic effects associated with overproduction of CLRN1 in photoreceptor cells, the right combination of AAV vector serotype, titer, and promoter needs to be identified.
Previous expression studies in HEK293 cells demonstrated that CLRN1 is a plasma membrane protein, showing little colocalization with ER markers, and is enriched in specific cell membrane regions including microvilli, lamellipodia, and cellular protrusions rich in F-actin [29,31]. In our in vivo study, we found that the AAV-expressed CLRN1 is present at the inner segments and synaptic regions of photoreceptor cells. Subcellular localization of proteins is essential for their biological function, and, as shown by previous studies, photoreceptor membrane proteins lacking specific targeting information within their amino acid sequence tend to accumulate predominantly in the outer segment compartment [51,63]. Consequently, CLRN1 may contain distinct amino acid motifs for its specific targeting to inner photoreceptor compartments, avoiding trafficking towards the outer segment.
In the retina, the USH protein network is thought to play a major role in the intracellular trafficking processes within photoreceptor cells through the connecting cilium [25,26,64]. In addition, recent studies have shown that the USH interactome is molecularly linked to other proteins whose mutations lead to ciliopathies, including RPGR and CEP290 [65][66][67]. Many USH proteins have also been localized either pre-or post-synaptically at the specialized ribbon synapses of photoreceptors and second order neurons, suggesting a potential role for these proteins in maintaining the synaptic plasticity and transport of cargo vesicles to the synapse [17,18,20,23,68]. In spite of their presence at the OPL, no functional evidence of photoreceptor synaptic defects resulting from deficiency in any USH protein was found in the currently available USH animal models [69], including our Clrn1 KO mice shown here [33]. The scaffold proteins SANS and whirlin have been shown to assemble a protein network composed of USH1 and USH2 members at the periciliary region, situated at adjacent membranes of the inner segment collar and the connecting cilium, with SANS linking the macromolecular complex to the microtubule cytoskeleton [26]. One study showed that CLRN1 can interact directly with the USH1F protein, protocadherin-15 (USH1F protein) [60]. CLRN1 may thus play a structural role in the ciliary-periciliary region of photoreceptors as a member of the USH interactome, potentially impacting protein transport processes as well. The increase of arrestin-1 remaining in the OPL upon light exposure in Clrn1 KO compared to wild-type controls is consistent with a delay in the arrestin-1 translocation towards the outer segments. In the dark, arrestin-1 is anchored within the inner compartments of photoreceptors through interactions with other proteins [70]. These interactions include microtubules, which are abundant not only within the synaptic region, but also within the inner segments of photoreceptor cells [71,72]. Cytoskeletal elements, such as microfilaments and microtubules, play both structural and transport roles in photoreceptor cells, in addition to maintaining the morphology and functional plasticity of ribbon synapses [73]. CLRN1 was shown to exhibit a potential association with either tubulin or F-actin in previous studies [29,35] and thus, CLRN1 deficiency may destabilize the network of proteins that structurally support a normal cytoskeleton in photoreceptor cells.
In summary, we show that a variety of cells can support CLRN1 expression in the retina following AAV delivery, and this has important implications for designing gene therapy studies. We found that the compartmentalization of AAV-expressed CLRN1 in photoreceptors is similar to that of known Usher proteins, localizing primarily to the inner segment region and outer plexiform layer. In view of its apparently low levels of endogenous expression as determined by previous studies, vector optimization remains a critical aspect for developing safe therapies for USH3 disorder.