AAV-mediated human CNGB3 restores cone function in an all- cone mouse model of CNGB3 achromatopsia

Complete congenital achromatopsia is a devastating hereditary visual disorder. Mutations in the CNGB3 gene account for more than 50% of all known cases of achromatopsia. This work investigated the efficiency of subretinal (SR) delivered AAV8 (Y447, 733F) vector containing a human PR2.1 promoter and a human CNGB3 cDNA in Cngb3−/−/Nrl−/− mice. The Cngb3−/−/Nrl−/− mouse was a cone-dominant model with Cngb3 channel deficiency, which partially mimicked the all-cone foveal structure of human achromatopsia with CNGB3 mutations. Following SR delivery of the vector, AAV-mediated CNGB3 expression restored cone function which was assessed by the restoration of the cone-mediated electroretinogram (ERG) and immunohistochemistry. This therapeutic rescue resulted in long-term improvement of retinal function with the restoration of cone ERG amplitude. This study demonstrated an AAV-mediated gene therapy in a cone-dominant mouse model using a human gene construct and provided the potential to be utilized in clinical trials.


Introduction
In the human retina, photoreceptor cyclic nucleotide-gated (CNG) channels play a pivotal role in phototransduction. Phototransduction relies on the function of CNG channels in both rod and cone photoreceptor outer segment plasma membranes [1][2] . Structurally, CNG channels comprise two structurally related A and B subunits: the rod photoreceptor CNG channel is composed of CNGA1 and CNGB1 subunits, whereas the cone photoreceptor is composed of CNGA3 and CNGB3 subunits [2][3][4][5][6] . Recent studies have shown that mutations in CNGA3 or CNGB3 cause clinically indistinguishable forms of congenital achromatopsia (ACHM) [7] . ACHM or rod monochromatism is an autosomal recessive hereditary visual disorder characterized by the absence of functional cone photoreceptors in infancy, affecting one in approximately 30 000 individuals worldwide [7] . Affected individuals exhibit pendular nystagmus, poor visual acuity, and photophobia [8] . The first signs of ACHM in infancy are nystagmus and photophobia as evidenced by squinting in bright light [8] . To date, mutations in six different genes have been identified as responsible for ACHM: CNGA3, CNGB3, GNAT2, PDE6C, PDE6H, and ATF6. Clinically, approximately 50% of ACHM patients are associated with CNGB3 mutations, about 25% with CNGA3 mutations and smaller fractions with mutations in the cone transducin or phosphodiesterase genes in European populations [7][8][9][10][11][12] . Previous studies demonstrated that the CNGA3 is the ion-conducting subunit, while the CNGB3 fails to form functional homomeric channels in heterologous expression systems when expresses alone [13][14] . Because the B subunits are only thought to confer specific biophysical properties to the CNG channel complex whereas the A subunits contribute to the principal channel properties [9,15] . Therefore, the A subunits are thought to be the primary subunits while the B subunits play a modulatory role for the function of both rod and cone CNG channels [9,15] . Hence, knock-out of the gene encoding Cngb3 in mouse retina would cause a moderate reduction of the cone function [16] .
Recent clinic trails on patients with Leber congenital amaurosis (LCA) demonstrated the successful expression of the RPE65 gene with adenoassociated viral (AAV) vectors [17][18][19] , which provided a possible treatment for retinal degeneration including ACHM. Restoration of cone function and improvement in photopic vision were achieved in mouse and dog models of ACHM by subretinal (SR) delivery of AAV vectors [20][21] . Progresses in AAVbased gene replacement therapy to restore conemediated function in Cngb3 deficient mouse models provided a foundation for the development of clinical trials for human ACHM patients [21][22] . In these studies, AAV vectors containing human CNGB3 cDNA were delivered subretinally in Cngb3 −/− mice, which carry a naturally occurring mutation in the Cngb3 gene.
We chose the Cngb3 −/− /Nrl −/− mouse model because it possessed a genetically and phenotypically wellcharacterized cone degeneration. This double knockout mouse had a retinal phenotype similar to its single knock-out Cngb3 −/− mouse with impaired cone function and cone degeneration. Cngb3 −/− /Nrl −/− mice lacked scotopic light response due to Nrl deficiency. The expression levels of cone arrestin and S-opsin were reduced in Cngb3 −/− /Nrl −/− mice [16] . In addition, we speculate that this mouse model will respond well to SR gene-based therapy since rescue has been achieved in Cngb3 −/− mice. Here, human CNGB3 expression was achieved using AAV8 (Y447, 733F) vector driven by a cone-specific promoter to test whether AAV8-(Y447, 733F)-PR2.1-hCNGB3 can rescue Cngb3 −/− /Nrl −/− cones when delivered to the SR space.

Materials and methods
Animals C57BL/6J mice were obtained from the Jackson Laboratory (Bar Harbor, USA). Cngb3 −/− [23] and Nrl −/− [24] mice were kindly provided by Dr. Anand Swaroop at NEI/NIH. Cngb3 −/− /Nrl −/− mice were generated as described previously [16] . All animals were maintained under standard laboratory conditions (18 °C -23 °C, 40%-65% humidity) with food and water available ad libitum in the University of Florida Health Science Center Animal Care Service Facilities a 12-hour/12-hour light/dark cycle with <15 ft-c environmental illumination. All experiments were approved by the Institutional Animal Care and Use Committee of the University of Florida and conducted in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and National Institutes of Health regulations.

Construction of AAV vectors
Human CNGB3 cDNA (purchased from American Type Culture Collection, USA) was cloned under the human PR2.1 promoter which has been shown to target transgene expression to mammalian L/M cones [25] to make AAV-PR2.1-hCNGB3 construct. The construct was packaged in AAV8 (Y447, 733F). The vectors were purified and tittered according to previously published methods [26] .

SR injections
At postnatal day 14 (P14), one microliter of AAV8 (Y447, 733F)-PR2.1-hCNGB3 vector (10 13 vector genome/mL) was injected subretinally into one eye of each Cngb3 −/− /Nrl −/− mouse. Transcorneal SR injections [27] were performed with a 33 gauge blunt needle mounted on a 5 mL Hamilton syringe. First, an entering hole was introduced at the edge of the corneal by a 30 gauge disposable needle, then the Hamilton syringe loaded with 1 μL of viral vector mixed with fluorescein dye was glided through the cornea opening into the subretinal space with previously described methods [28][29] . The other eye remained untreated as a control. All eyes used for further evaluation had no apparent surgical complications and more than 80% of retinal detachment.

Electroretinography
Electroretinography was performed at six months following SR injections. A UTAS Visual Diagnostic System with a Big Shot Ganzfeld (LKC Technologies, USA) was used for ERG recording. All mice were anesthetized by ketamine (72 mg/kg)/xylazine (4 mg/kg) by intraperitoneal injection. The pupils were dilated with 1% atropine and 2.5% phenylephrine hydrochloride. Fifty recordings were averaged for light-adapted ERG measured light intensity of 1.4 log cd·s/m 2 in the presence of 30 cd·s/m 2 background light with inter-stimulus intervals of 0.4 seconds. Scotopic and photopic b-wave amplitudes were averaged and used to generate SEM. ERG data were presented as mean±SEM. Statistical analysis was performed with unpaired t-test and significance was defined as a P value of less than 0.05.

Tissue preparation and immunohistochemistry
Eyes were enucleated at six months following SR injection. Retinal sections and whole mounts were prepared according to previously described methods [28,30] . Briefly, the limbus of eyes was marked at "12 o'clock" with a hot needle immediately after sacrifice. Then eyes were enucleated and fixed in 4% paraformaldehyde overnight at 4 °C. The cornea and lens were removed from the eyes. For retinal frozen sections, the remaining eyecups were rinsed with PBS and cryoprotected 30% sucrose in PBS for 4 hours at room temperature, then embedded in cryostat compound (Tissue TEK OCT, Sakura Finetek USA, Inc., USA) and frozen at −80 °C. For immunohistochemistry, retina sections or whole mounts were rinsed in PBS and blocked in 0.3% Triton X-100, 5% BSA in PBS for 60 minutes at room temperature. Sections or whole mounts were incubated with lectin peanut agglutinin (PNA) conjugated to an Alexa Fluor 488 (1:200, Invitrogen, USA), M-or S-cone opsin (1:400, Millipore, USA), human CNGB3 (1:100, ThermoFisher, USA) primary antibodies diluted in 5% BSA in PBS overnight at 4 °C, then washed by PBS for three times, incubated with IgG secondary antibody tagged with Alexa-594 diluted 1:500 in PBS at room temperature for two hours and washed with PBS. Sections were mounted with Vectashield Mounting Medium for Fluorescence (H-1400, Vector Laboratories, USA) and coverslipped. Whole mounts and sections were analyzed with a Zeiss CD25 microscope fitted with Axiovision Rel. 4.6 software.
Expression of the S-opsin in the treated Cngb3 −/− /Nrl −/− eyes was detected across the entire retinas by whole mount immunostaining and maintained at least 6 months after P14 SR injection of AAV8 (Y447, 733F) vector with therapeutic gene (Fig. 3F, J and N). Immunostaining of ventral retinal sections was also performed to check the cellular localization of S-opsin expression. In addition, S-opsin expression in the cone outer segments was evidenced by co-localization with positive PNA staining (Fig. 4). The positive S-opsin staining confirmed that SR-AAV8 (Y447, 733) treatment prevented cone degeneration (Fig. 3).

Discussion
The goal of this study was to test gene therapy using a new all-cone mouse model for treating human CNGB3-achromatopsia. In the human retina, cones account for only ~5% of the photoreceptors [33] . They are primarily concentrated in macula and responsible for day vision, central visual acuity and color vision [7] . Both cone and rod CNG channels exhibit a 3:1 stoichiometry between A and B subunits [3][4][5]13] . The cone photoreceptor CNG channel is composed of A3 and B3 subunits, whereas the rod photoreceptor CNG channel is formed by A1 and B1 subunits [4][5][6]15] . CNGB3 is known as the modulator. A CNGA3 homomeric channel is fully functional in heterologous expression systems [14,34] . Although CNGB3 shares a common topology with CNGA3 and possesses a poreforming region, the B3 subunit does not form a functional channel without A3 subunits [14,34] . Therefore, Cnga3 −/− mice show a complete loss of cone response [23] , while deficiency of Cngb3 leads to residual cone response in mouse retina [35] . Fortunately, mutations in CNGB3 are clinically responsible for approximately half of all cases of ACHM patients which present moderate cone degeneration [36] , to permit a relatively long period for therapy intervention. In this study, AAV delivered human CNGB3 expression in this all-cone mouse model is promising for the treatment of CNGB3-achromatopsia and other cone-specific diseases.
Nrl −/− retinas played shorter cone OS than wildtype [24,31] . Knock-out of the Cngb3 and Nrl gene partially mimicked the fovea of ACHM patients. A subsequent study showed that Cngb3 −/− /Nrl −/− mice had a retinal phenotype similar to the sum of their respective single knock-out counterparts, i.e. impaired cone function and cone degeneration [16] . Effective gene therapy for ACHM requires efficient transgene expression in cones. Here, the human CNGB3 introduced by AAVs and expressed in the cone OS and S-opsin were restored following AAV-mediated CNGB3 expression, and AAV-mediated expression of the human CNGB3 could restore light responses in the The AAV serotype is one of factors that contributes to an efficient transgene delivery. Although the AAV serotype 5 transduced a significantly greater number of photoreceptor cells compared with serotype 2 [37] , the AAV8 had higher inherent photoreceptor transduction efficiency than AAV2 or 5 [37][38][39][40] , making it the most suitable vector for ACHM gene therapy. Moreover, comparing with the traditional AAV vector, site-directed tyrosine to phenylanine (Y-F) mutagenesis of selected tyrosine residues on AAV capsid also increased transduction efficiency by protecting vector particles from proteasomal degradation [30,[40][41] . The specific promoter is another important factor for efficient gene therapy. The PR2.1 promoter contains 2.1 kb of nucleotides, including the locus control region (LCR) and other sequences upstream of the L-opsin coding region. This promoter was reported to direct high level of human Cngb3 expression in both L/M and S cones in mouse retinas [22] . Accordingly, we designed AAV8 (Y447, 733F)-PR2.1-hCNGB3 vector and obtained a robust and longer-term rescue in the Cngb3 −/− /Nrl −/− mouse.
Our study showed very slow cone degeneration in the Cngb3 −/− /Nrl −/− mice, and the photopic b-wave amplitudes were still 38% of Nrl −/− mice at the age of 6.5 months (Fig. 1B). The factor to consider was the detachment of the fragile structure of retinas after SR injection. In LCA2 human clinical trial, SR injections that detached the fovea area resulted in loss of fovea cone OS and cone cell death, and it was concluded that SR injection under the fovea may cause damage rather than benefit and should be approached with prudence [18,[42][43] . Since IV delivered AAV8 (Y447, 733F) could efficiently target cone photoreceptors and rescued the cone functions in Cngb3 −/− /Nrl −/− [44] , we expect similar cone rescue in Cngb3 −/− /Nrl −/− mice by AAV8 (Y447, 733F)-PR2.1-hCNGB3 vector. We even expect it to be utilized in clinical trials due to the reduced physical barrier to vectors intended for cones from the vitreous.
In summary, AAV-mediated human CNGB3 expression restored cone function in Cngb3 −/− /Nrl −/− mouse which can be served as a good model for further study of gene therapy on CNGB3achromatopsia.

Acknowledgments
This work was supported by NIH (Grant No.