CRB2 Loss in Rod Photoreceptors Is Associated with Progressive Loss of Retinal Contrast Sensitivity

Variations in the Crumbs homolog-1 (CRB1) gene are associated with a wide variety of autosomal recessive retinal dystrophies, including early onset retinitis pigmentosa (RP) and Leber congenital amaurosis (LCA). CRB1 belongs to the Crumbs family, which in mammals includes CRB2 and CRB3. Here, we studied the specific roles of CRB2 in rod photoreceptor cells and whether ablation of CRB2 in rods exacerbates the Crb1-disease. Therefore, we assessed the morphological, retinal, and visual functional consequences of specific ablation of CRB2 from rods with or without concomitant loss of CRB1. Our data demonstrated that loss of CRB2 in mature rods resulted in RP. The retina showed gliosis and disruption of the subapical region and adherens junctions at the outer limiting membrane. Rods were lost at the peripheral and central superior retina, while gross retinal lamination was preserved. Rod function as measured by electroretinography was impaired in adult mice. Additional loss of CRB1 exacerbated the retinal phenotype leading to an early reduction of the dark-adapted rod photoreceptor a-wave and reduced contrast sensitivity from 3-months-of-age, as measured by optokinetic tracking reflex (OKT) behavior testing. The data suggest that CRB2 present in rods is required to prevent photoreceptor degeneration and vision loss.


Introduction
The Crumbs protein complex is essential for polarity establishment and adhesion of the retinal neural epithelium. In mammals, the Crumbs family is composed of Crumbs homolog-1 (CRB1), CRB2, and CRB3. CRB1 and CRB2 have a large extracellular domain with epidermal growth-factor-like and laminin-A globular domains, a single transmembrane domain, and an intracellular C-terminal domain of 37 amino acids; CRB3 lacks the extracellular domain [1]. The intracellular domain has a single C-terminal PDZ protein-binding motif and a single FERM-protein-binding motif juxtaposing the transmembrane domain [1]. The Crumbs proteins associate with the adaptor protein PALS1 and one of the multiple PDZ-proteins PATJ or MUPP1 to form the core of the Crumbs complex [2,3]. In the developing mouse retina, the Crumbs proteins localize at the subapical region adjacent to the adherens junctions between retinal progenitor cells [4]. In the mature retina, the Crumbs proteins are present in photoreceptors and Müller glial cells (MGCs) [5,6]. CRB2 protein is present in photoreceptors and MGCs, whereas CRB1 protein is present only in MGCs [7,8].
Loss or reduced levels of the CRB1 or CRB2 proteins in retinal progenitors, immature photoreceptors, or MGCs leads to different retinal phenotypes in mice that mimic the wide spectrum of clinical features described in CRB1-patients, including early and late onset RP and LCA [2]. Loss of either CRB1 or CRB2 in MGCs results in mild retinal dystrophy, without impairing retinal function [6, [15][16][17]. Ablation of Crb2 in retinal progenitor cells, or in immature rod and cone photoreceptor cells, results in progressive thinning and degeneration of the photoreceptor layer, abnormal lamination of immature rods, and loss of retinal function [4,16,18]. Moreover, CRB2 has roles in restricting the proliferation of retinal progenitor cells and the number of rod photoreceptors and MGCs [4]. Mouse CRB2 acts as the modifying factor of CRB1-related retinal dystrophies, since reduction or full ablation of CRB2 in combination with loss of CRB1 results in an exacerbation of the retinal phenotype observed in Crb1 knockout retinas [19][20][21][22]. The specific roles of CRB2 in rod photoreceptor cells still need to be elucidated. We hypothesize that CRB2 in rods is required to maintain proper retinal structure and function.
In fetal human retinas, human iPSC-derived retinal organoids, and adult non-human-primate retinas, CRB1 as well as CRB2 proteins localize at the subapical region in both photoreceptors and MGCs [22,23]. We previously showed proof-of-concept for AAV9-CMV-CRB2, reintroduction of CRB2 into photoreceptors and MGCs rescued the phenotype of Crb2 flox/flox Chx10Cre and Crb1Crb2 F/+ Chx10Cre mouse retinas [24]. Although AAV9 efficiently transduces both mouse photoreceptors and MGCs [25], AAV5 outperforms AAV9 in transducing both human photoreceptors and MGCs in cultured adult human retinal explants and human iPSC-derived retinal organoids [23]; as such, AAV5 the most suitable serotype to be used in the clinics. In mice, AAV5 only infects retinal pigment epithelium and photoreceptors; a new animal model lacking CRB2 specifically in photoreceptor cells that allows us to test the efficacy of the AAV5-CMV-CRB2 vector is required. Therefore, to validate our hypothesis and for the ability of testing the AAV5-CMV-CRB2 vector in the future, we generated mice lacking CRB2 specifically in adult rod photoreceptors (with remaining levels of CRB2 in MGCs and cone photoreceptors), and mice lacking CRB2 specifically in rod photoreceptors and CRB1 specifically in MGCs.
Here, we studied the effects on retinal morphology and function of loss of CRB2 specifically in rod photoreceptors with or without concomitant loss of CRB1. Our data shows that specific ablation of CRB2 in mouse rods leads to RP. The phenotype observed in these retinas was characterized by loss of photoreceptor cells and gliosis in the peripheral and central retina. The retinal degeneration was more severe in the superior than in the inferior retina. Retinal function, measured by ERG, was impaired in 9-month-old animals. Concomitant loss of CRB1 exacerbates the retinal phenotype leading to decreased retinal and visual function from 3-months-of-age. The data suggest that CRB2 in rods is required to maintain cellular adhesion between rods and prevent photoreceptor degeneration and vision loss.

Ablation of CRB2 in Rods with Concomitant Loss of CRB1 Leads to Retinal Dysfunction and Vision Impairment
To study if the specific loss of CRB2 from rod photoreceptors with and without concomitant loss of CRB1 affected retinal function, we performed electroretinography (ERG) in 1-, 3-, 6-, 9-, and 12-month-old Crb2 ∆Rods , Crb1 KO Crb2 ∆Rods , and the respective age-matched controls (Crb2 flox/flox and Crb1 KO Crb2 flox/flox ). One-month-old Crb2 ∆Rods and Crb1 KO Crb2 ∆Rods mice showed similar scotopic and photopic responses to the ones observed in the age-matched controls ( Figure 1C, Supplementary  Figures S2 and S3). In contrast, 3-month-old Crb1 KO Crb2 ∆Rods , but not Crb2 ∆Rods mice, showed slightly reduced amplitudes of a-wave scotopic electroretinography responses, indicating alterations of rod photoreceptor function ( Figure 1A, arrow; Figure 1C). Moreover, 3-month-old Crb1 KO Crb2 ∆Rods also showed reduced b-wave photopic electroretinography (Supplementary Figure S3). The reduction of the scotopic a-wave amplitudes in Crb1 KO Crb2 ∆Rods mice became more evident at 9-and 12-months-of-age ( Figure 1B-D). At these time points, a reduced a-wave was also observed in Crb2 ∆Rods mice (Figure 1B-D; arrows). Twelve-month-old mutants also showed a reduced b-wave ( Figure 1C).
Three-month-old Crb1 KO Crb2 ∆Rods mice showed reduced scotopic a-wave responses, making these mice an interesting and potential suitable CRB1-disease model for future gene therapy rescue experiments. Therefore, we decided to determine whether the visual function was also affected in these mice. To do so, we used an optomotor response test (optokinetic tracking reflex (OKT)) to measure the spatial frequency threshold and contrast sensitivity [28,29]. Spatial frequency was measured by systematically increasing the spatial frequency of the grating at 100% contrast until animals no longer tracked (spatial frequency threshold). A contrast sensitivity threshold was generated by identifying the minimum contrast that generated tracking, over a range of spatial frequencies. Spatial frequency threshold and contrast sensitivity of Crb1 KO Crb2 ∆Rods mice were analyzed at different time points, at 1-, 3-, 7-, and 9-month(s)-of-age. At 1-and 3-months-of-age, no differences were observed in spatial acuity between littermate controls (Crb1 KO Crb2 flox/flox ) and Crb1 KO Crb2 ∆Rods mice ( Figure 2A). However, at 7-and 9-months-of-age the Crb1 KO Crb2 ∆Rods showed a small but statistically significant decrease in spatial frequency thresholds, also described frequently in literature as visual acuity.

Pupil Light Reflex is not Impaired in Crb1 KO Crb2 ∆Rods Mice
To assess if pupil light reflex was affected in Crb1 KO Crb2 ∆Rods mice, we characterized the pupil response to blue and red light stimuli in dark-adapted, light-anesthetized 3-and 9-month-old Crb1 KO Crb2 ∆Rods and in control mice. Pupil response curves, contraction, and dilation after each stimulus were identical in all the experimental groups ( Figure 3A,C). No differences in the maximal pupil contraction were observed between mutant and control(s) mice at any condition analyzed ( Figure 3B,D). These results suggest that pupil light reflex was not affected in the Crb1 KO Crb2 ∆Rods mice.

Loss of CRB2 in Rods Results in Slow Loss of Photoreceptor Cells, Mainly in the Superior Retina
To study if CRB2 specific ablation in rod photoreceptors results in a morphological phenotype, histological analysis of retina sections was performed. All retinal layers were present and displayed normal organization in the Crb2 ∆Rods and Crb1 KO Crb2 ∆Rods mice at 1-month-of-age, suggesting that removal of CRB2 from rod photoreceptors did not affect retinal development and lamination. No morphological abnormalities were observed in the control mice (Crb2 flox/flox ) ( Figure 4A,A') and in the Crb2 ∆Rods ( Figure 4B 15,19] (data not shown). In Crb1 KO Crb2 ∆Rods , disruptions at the outer limiting membrane of the central superior retina were also observed ( Figure 4F,F'; arrow). In some 3-month-old and in all 6-month-old Crb2 ∆Rods ( Figure 4H,H'; arrows) and Crb1 KO Crb2 ∆Rods ( Figure 4F,F'; arrows) retinas, the photoreceptor layer was severely thinned at the peripheral superior area. In 9-and 12-month-old Crb2 ∆Rods and Crb1 KO Crb2 ∆Rods , further thinning of the outer nuclear layer was observed in the peripheral as well as central superior retina ( Figure 5B,B',C,C',E,E',E",F,F',F"). function, we wanted to evaluate if deletion of only CRB2 from rods was sufficient to induce such a deficit. Therefore, the spatial frequency threshold and contrast sensitivity of Cbr2 flox/flox were compared to age-matched Crb2 ΔRods . No differences were observed in spatial frequency threshold. nor in contrast sensitivity ( Figure 2B and Supplementary Figure S4B). The data suggest that the visual function impairment observed in the Crb1 KO Crb2 ΔRods is due to cumulative loss of CRB1 and CRB2 and not to single loss of CRB2 or due to toxicity of iCRE expression in rod photoreceptors.

Pupil Light Reflex is not Impaired in Crb1 KO Crb2 ΔRods Mice
To assess if pupil light reflex was affected in Crb1 KO Crb2 ΔRods mice, we characterized the pupil response to blue and red light stimuli in dark-adapted, light-anesthetized 3-and 9-month-old Crb1 KO Crb2 ΔRods and in control mice. Pupil response curves, contraction, and dilation after each stimulus were identical in all the experimental groups ( Figure 3A,C). No differences in the maximal pupil contraction were observed between mutant and control(s) mice at any condition analyzed ( Figure 3B,D). These results suggest that pupil light reflex was not affected in the Crb1 KO Crb2 ΔRods mice.      To assess and better visualize the photoreceptor cell loss over time, the number of nuclei in rows of photoreceptors was quantified. The number of photoreceptor nuclei in a row was reduced in the entire superior retina and the inferior retina near to the optic nerve head in 6-month-old Crb1 KO Crb2 ∆Rods compared to the Crb2 flox/flox control ( Figure 5G). At 9-months-of-age, the number of photoreceptors in a row were further decreased in almost the entire superior retina and in most of the inferior retina of Crb1 KO Crb2 ∆Rods ( Figure 5H). Also at this time point, the superior retina ( Figure 5H) and some areas of the inferior retina of Crb2 ∆Rods mice became thinner. Twelve-month-old retinas showed almost no photoreceptor nuclei in the far peripheral superior and nearly half in the central superior retina of Crb2 ∆Rods and Crb1 KO Crb2 ∆Rods . At this time also, the inferior retina of Crb2 ∆Rods and Crb1 KO Crb2 ∆Rods mice became thinner compared to the Crb2 flox/flox control ( Figure 5I).

Removal of CRB2 in Rods Results in Loss of Rods Mainly at the Periphery of the Superior Retina
The morphological phenotype observed in the Crb2 ∆Rods and Crb1 KO Crb2 ∆Rods retinas mainly affected the photoreceptor cells. In the far peripheral superior retinas of 9 month-old Crb2 flox/flox control mice, the photoreceptor cells stained positive for recoverin ( Figure 6A). In the far periphery of the superior retinas of Crb2 ∆Rods ( Figure 6B) and Crb1 KO Crb2 ∆Rods ( Figure 6C), no photoreceptors were found. Towards the central superior retina, reduced numbers of recoverin-positive cells were detected, depicting thinning of the outer nuclear layer. Although the number of photoreceptors was reduced, the remaining photoreceptors showed matured inner-and outer-segments. Rhodopsin is normally located in the outer-segments of mature rod photoreceptor cells ( Figure 6D). In areas that showed reduced numbers of photoreceptors in Crb2 ∆Rods and Crb1 KO Crb2 ∆Rods retina, rhodopsin was localized in the outer segments, suggesting that the remaining rods were functional ( Figure 6E,F). Cone photoreceptors can be labelled using an antibody against cone arrestin (CAR) ( Figure 6D). In both knockout lines, cone photoreceptors showed normal morphology ( Figure 6E,F). In the control retinas, peanut-agglutinin (PNA) stained the cone photoreceptor outer-segments and pedicles at the photoreceptor synapses ( Figure 6G). PNA staining was similar in 9-month-old Crb2 ∆Rods ( Figure 6H) and Crb1 KO Crb2 ∆Rods retinas ( Figure 6I). The M-cone photoreceptor outer-segments were stained appropriately with M-opsin antibodies in the control retinas ( Figure 6G) and in both mutant retinas ( Figure 6H,I).
Photoreceptor cell synapses can be stained with MPP4. In the control retinas, MPP4 signal was detected in the photoreceptor synapses at the outer plexiform layer ( Figure 6J,J'). In Crb2 ∆Rods and Crb1 KO Crb2 ∆Rods retinas, MPP4 staining was disrupted and decreased; some ectopic anti-MPP4 labelling was also detected in the outer nuclear layer ( Figure 6K,K',L,L'). Protein kinase (PKC)α is abundant in retinal bipolar cells. In the control retinas, bipolar cells located at the inner nuclear layer and presented normal dendritic arborization in the outer plexiform layer ( Figure 6J,J'). In the knockout retinas, bipolar cells were localized at the correct layer, but their dendritic arborization was affected ( Figure 6K,K',L,L'; arrowheads).

Loss of CRB2 in Rods Leads to Disruption of the Apical Protein Complexes
We previously reported that ablation of CRB2 from both immature cone and rod photoreceptor cells resulted in disruption of the apical protein complexes at the outer limiting membrane [16]. Here, we studied if removal of CRB2 specifically in rod photoreceptors is enough to lead to destabilization of the subapical region and adherens junction protein complexes at the outer limiting membrane using transmission electron microscopy and immunohistochemistry. Photoreceptor cell synapses can be stained with MPP4. In the control retinas, MPP4 signal was detected in the photoreceptor synapses at the outer plexiform layer ( Figure 6J,J'). In Crb2 ΔRods and Crb1 KO Crb2 ΔRods retinas, MPP4 staining was disrupted and decreased; some ectopic anti-MPP4 labelling was also detected in the outer nuclear layer ( Figure 6K,K',L,L'). Protein kinase (PKC)α is abundant in retinal bipolar cells. In the control retinas, bipolar cells located at the inner nuclear layer and presented normal dendritic arborization in the outer plexiform layer ( Figure 6J,J'). In the knockout retinas, bipolar cells were localized at the correct layer, but their dendritic arborization was affected ( Figure 6K,K',L,L'; arrowheads).  At 3-months-of-age, CRB1 protein was present at the subapical region above the adherens junctions in the control ( Figure 8A; arrows) and Crb2 ΔRods ( Figure 8B; arrows) in the superior peripheral retina. In Crb1 KO Crb2 ΔRods , the CRB1 protein was absent, as previously found in Crb1 KO ( Figure 8C). CRB2 localized at the subapical region in the control retinas ( Figure 8D) and in the central At 3-months-of-age, CRB1 protein was present at the subapical region above the adherens junctions in the control (Figure 8A; arrows) and Crb2 ∆Rods (Figure 8B; arrows) in the superior peripheral retina. In Crb1 KO Crb2 ∆Rods , the CRB1 protein was absent, as previously found in Crb1 KO ( Figure 8C). CRB2 localized at the subapical region in the control retinas ( Figure 8D) and in the central superior retina of Crb2 ∆Rods ( Figure 8E) and Crb1 KO Crb2 ∆Rods ( Figure 8F). Other subapical region markers and members of the Crumbs complex, such as PALS1 and MUPP1, were correctly located at the subapical region in the control retina ( Figure 8G). However, mutant retinas showed disruptions of the subapical region, labelled by CRB2, PALS1, and MUPP1 ( Figure 8E,F,H,I; arrows). In the control retinas, β-catenin showed correct localization of the adherens junction ( Figure 8J), while in both mutant retinas, disruptions of the adherens junctions in the central superior retina were observed ( Figure 8K,L). Photoreceptor cell nuclei protruding from the inner-/outer-segment layer were observed at the site of disruption (Figure 8 arrows). In the 9-month-old control retina, CRB2 correctly localized at the subapical region just above the adherens junction marker, β-catenin ( Figure 8M,M'). Also, PALS1 and MUPP1 were correctly located at the subapical region in the control retina ( Figure 8P,P'). However, at the peripheral superior retina of Crb2 ∆Rods and Crb1 KO Crb2 ∆Rods mice, partial disruptions of the subapical region were observed ( Figure 8N,N',O,O',Q,Q',R,R'), whereas adherens junctions were mainly lost in areas with loss of all photoreceptors ( Figure 8N,N',O,O'). PAR3, a member of the PAR complex, and p120-catenin, an adherens junction protein, were lost at sites of disruption of the outer limiting membrane in the central superior retina (Figure 8T,U; arrows).
superior retina of Crb2 ΔRods ( Figure 8E) and Crb1 KO Crb2 ΔRods ( Figure 8F). Other subapical region markers and members of the Crumbs complex, such as PALS1 and MUPP1, were correctly located at the subapical region in the control retina ( Figure 8G). However, mutant retinas showed disruptions of the subapical region, labelled by CRB2, PALS1, and MUPP1 ( Figure 8E,F,H,I; arrows). In the control retinas, β-catenin showed correct localization of the adherens junction ( Figure 8J), while in both mutant retinas, disruptions of the adherens junctions in the central superior retina were observed ( Figure 8K,L). Photoreceptor cell nuclei protruding from the inner-/outer-segment layer were observed at the site of disruption (Figure 8 arrows). In the 9-month-old control retina, CRB2 correctly localized at the subapical region just above the adherens junction marker, β-catenin ( Figure  8M,M'). Also, PALS1 and MUPP1 were correctly located at the subapical region in the control retina ( Figure 8P,P'). However, at the peripheral superior retina of Crb2 ΔRods and Crb1 KO Crb2 ΔRods mice, partial disruptions of the subapical region were observed ( Figure 8N,N',O,O',Q,Q',R,R'), whereas adherens junctions were mainly lost in areas with loss of all photoreceptors ( Figure 8N,N',O,O'). PAR3, a member of the PAR complex, and p120-catenin, an adherens junction protein, were lost at sites of disruption of the outer limiting membrane in the central superior retina (Figure 8T,U; arrows).

Removal of CRB2 from Rods Results in Gliosis in Müller Glial Cells
Müller glial cells extend throughout the entire retina and function to maintain retinal homeostasis and integrity [30]. To study the effect of CRB2 removal from rod photoreceptor cells on the morphology of MGCs, we labelled these cells with glutamine synthetase (GS), SOX9, CD44, and glial fibrillary acidic protein expression (GFAP) antibodies. In 9 month-old control retinas, MGCs stained with glutamine synthetase displayed radial alignment, with well-established apical ends ( Figure 9A,A') and nuclei (SOX9-positive) located in the inner nuclear layer ( Figure 9A,A'). In Crb2 ∆Rods and Crb1 KO Crb2 ∆Rods retinas, the radial alignment of MGCs was disturbed and SOX9-positive nuclei were located most apically ( Figure 9B,B',C,C'; arrows). CD44 is highly expressed in MGC apical villi ( Figure 9D,D'). In the peripheral superior knockout retinas, the radial structure of MGC apical villi was lost ( Figure 9E,E',F,F'). GFAP is a marker for intermediate filaments in MGCs ( Figure 9G,J,M), and an increase in GFAP occurs in gliosis. Knockout retinas showed an upregulation of GFAP in MGCs ( Figure 9H,I,K,L,N,O). Increased GFAP levels were more pronounced in the peripheral ( Figure 9H,I) and central ( Figure 9K,L) superior retina. A moderate increase in GFAP levels was also observed in the inferior retina of Crb2 ∆Rods and Crb1 KO Crb2 ∆Rods (Figure 9N,O).

Discussion
Here, we studied the effect of CRB2 removal specifically from rod photoreceptors in the mouse retina. Furthermore, we evaluated the consequences of concomitant loss of CRB1. Our key findings are: (i) loss of CRB2 from rod photoreceptors results in retinitis pigmentosa, mainly at the peripheral and central superior retina; (ii) CRB2 in rod photoreceptor cells is required to maintain the rod photoreceptor layer and retinal electrical responses; (iii) loss of CRB2 in rods and concomitant loss of CRB1 leads to an exacerbation of the retinitis pigmentosa phenotype; (iv) ablation of CRB2 from rods with concomitant loss of CRB1 results in visual function impairment.
During retinogenesis, mouse CRB2 is located in retinal radial progenitor cells [4]. In the mature retina, the protein is found in rod and cone photoreceptors and in MGCs [7]. We previously demonstrated that retinal development and lamination is affected from embryonic day 15.5, when CRB2 is removed from both immature rod and cone photoreceptors. These Crb2 ∆immPRC mouse retinas mimicked a very early-onset retinitis pigmentosa (Supplementary Figure S5A) [16]. The Crb2 ∆immPRC mouse retinas showed progressive thinning of the photoreceptor layer and mislocalization of retinal cells, which resulted in a severe retinal function impairment, as measured by electroretinography. Moreover, concomitant loss of CRB1 in Crb1 KO Crb2 ∆immPRC retinas exacerbated the retinal phenotype and resulted in an LCA phenotype with thickened superior retina due to abnormal lamination of photoreceptors, intermingled photoreceptor and inner nuclear cell nuclei, and ectopic photoreceptor nuclei in the ganglion cell layer [21].
The Crb1 KO Crb2 ∆Rods retinas described here showed moderate decrease in retinal function and a significant decrease in contrast sensitivity shortly after the onset of morphological retinal degeneration at 3-months-of-age (Supplementary Figure S5A). In these mice, specific ablation of CRB2 from rod photoreceptors was achieved by crossing Crb2 flox/flox mice with Rho-iCre transgenic mice that express iCRE in rod photoreceptors from postnatal day 7, achieving expression in nearly all rods at postnatal day 20 [26]. The onset of CRE expression after retinogenesis might explain the moderate retinal phenotype observed in the Crb2 ∆Rods and Crb1 KO Crb2 ∆rods mice when compared to mice with Cre-mediated ablation in retinal progenitor cells (Crb2 ∆RPC ) or mice with Cre-mediated ablation in immature cone and rod photoreceptor cells (Crb2 ∆immPRC ). We previously demonstrated that short-term depletion of CRB2 from adult retinas using adeno-associated viral delivery of Cre or shRNA against Crb2 leads to sporadic disruptions at foci of the outer limiting membrane, outer plexiform layer, and disruption of adhesion between photoreceptor cells [16]. This, together with the current data reported here, suggests that CRB2 has a function in the regulation of cellular adhesion between rod photoreceptors. Several CRB1-disease mouse models showed a gradient in phenotype severity. The entire retina is affected in Crb2 ∆RPC mice lacking CRB2 from retinal progenitors, MGCs, and photoreceptor cells, or Crb2 ∆immPRC mice lacking CRB2 from the immature rod and cone photoreceptors [4,16]. In Crb1 KO retinas lacking CRB1 in retinal progenitors and MGCs, the phenotype is mainly located in the inferior temporal quadrant [6,15]. Crb1 KO Crb2 ∆immPRC retinas lacking CRB1 and CRB2 from immature photoreceptors showed retinal dystrophy by fusion of the inner and outer nuclear layer throughout the retina. However, larger regions of ectopic photoreceptor cells were observed in the superior retina [21]. The Crb2 ∆Rods and Crb1 KO Crb2 ∆Rods mice reported here presented a retinal phenotype mainly affecting the peripheral and central superior retina. The differences observed in the different mouse CRB1-models might be related to the higher levels of CRB2 at the subapical region in the inferior retina, whereas CRB1 is expressed at higher levels in the superior retina [20]. Several other genes were found to be enriched in the superior retina, amongst these the photoreceptor gene endothelin 2 (Edn2) and the MGC genes ceruloplasmin and glial fibrillary acidic protein (Gfap) [31]. In the Crb2 ∆Rods and Crb1 KO Crb2 ∆Rods mice, GFAP expression was upregulated in the superior retina, with structural changes in the subapical villi of MGC, which suggests the contribution of MGCs to the phenotype observed. Retinal function mediates visual perception, but also other vision-linked reflexes, such as the pupillary light reflex [32][33][34]. Previously, Kostic et al. demonstrated that low-intensity red or blue light stimulus was a good parameter to discriminate rod photoreceptor contribution for the pupil response, while the recovery time after high-intensity blue stimulus was a predictor cone contribution to the pupillary light reflex [33]. Here, we used a vision-linked reflex, the pupillary light reflex, as a retina/vision functional outcome. Crb1 KO Crb2 ∆Rods mice showed a strong reduction in contrast sensitivity from 3-months-of-age onwards, which was not observed in Crb2 ∆Rods mice, and a small decrease in spatial frequency from 7-months-of-age onwards, while the control Crb1 KO Crb2 flox/flox mice did not show a reduction at 9-months-of-age. To our knowledge, this is the first report of loss of visual function as measured by OKT in a CRB1-retinal disease model. The data suggest that visual contrast sensitivity might be a diagnostically accurate and practical test to detect early visual deficit in Crb1 KO Crb2 ∆Rods mice. Moreover, contrast sensitivity tests might potentially be suitable for clinical studies testing candidate medicines for patients with mutations in the CRB1 gene. The correlation between photoreceptor degeneration and OKT contrast sensitivity in different mouse models are poorly described in the literature. However, some studies used different rhodopsin mutant rat strains and demonstrated that OKT spatial frequency and contrast sensitivity did not decline until very late in the photoreceptor degeneration [35,36]. Furthermore, studies performed in mice carrying a missense point mutation in Pde6b (rd10), which are almost deprived of photoreceptor cells at 1-month-of-age, still presented residual spatial frequency responses at 2-months-of-age [37].
Here, we propose that loss of CRB2 leads to loss of rod-cone and rod-MGC adhesion; as a result adherens junctions are disrupted and the cytoarchitecture of the outer nuclear layer is compromised, giving rise to misplaced rod photoreceptor cells. We speculate that misplaced or "loose" rods preferentially degenerate and die or are phagocyted by activated microglial cells. Microglia cell activation will contribute to MGC reactive gliosis, characterized by increased GFAP levels and hypertrophy, by secreting pro-inflammatory mediators. We further speculate that the reactive MGCs from Crb1 KO Crb2 ∆Rods and Crb2 ∆Rods mice present a proliferation of fibrous processes and deposition of proteoglycans at the outer edge of the retina, inhibiting axonal regeneration and exacerbating the loss of rods (Supplementary Figure S5) [38]. Moreover, the clear deficit of contrast sensitivity may allow us to have a rigorous outcome parameter to measure functional vision gain or maintenance after AAV treatment. Therefore, Crb1 KO Crb2 ∆Rods mice might become important resources to test therapies for retinopathies due to mutations in the CRB1 gene.

Electroretinography (ERG)
Dark-and light-adapted ERGs were performed under dim red light using an Espion E2 (Diagnosys, LLC, Lowell MA, USA). ERGs were performed on 1-month-old (1M), 3M, 6M, 9M, and 12M in Crb2 ∆Rods , Crb1 KO Crb2 ∆Rods , Crb2 flox/flox , and Crb1 KO Crb2 flox/flox mice. Mice were anesthetized using 100 mg/kg ketamine and 10 mg/kg xylazine administrated intraperitoneally, and the pupils were dilated using tropicamide drops (5 mg/mL). Mice were placed on a temperature regulated heating pad and reference and ground platinum electrodes were placed subcutaneously in the scalp and the base of the tail, respectively. ERGs were recorded from both eyes using gold wire electrodes. Hypromellose eye drops (3 mg/mL, Teva, The Netherlands) were given between recordings to prevent eyes from drying. Single (Scotopic and Photopic ERG) white (6500 k)-flashes were used. Band-pass filter frequencies were 0.3 and 300 Hz. Scotopic recordings were obtained from dark-adapted animals at the following light intensities: −4, −3, −2, −1, 0, 1, 1.5, 1.9 log cd·s/m 2 . Photopic recordings were performed following 10 minutes light adaptation on a background light intensity of 30 cd·m 2 and the light intensity series used was: −2, −1, 0, 1, 1.5, 1.9 log cd·s/m 2 [39]. . Statistical analysis of the ERG data was performed using a t-test [22]. Responses of the Crb1 KO Crb2 ∆Rods and Crb2 ∆Rods were compared to the Crb2 flox/flox control mice at each time point analyzed.
with an ascending series of ethanol, followed by mixtures of propylene oxide and EPON (LX112). After the infiltration step with pure EPON, the halves of the eyes were positioned on the caps of large BEEM capsules filled with EPON. Ultrathin sections of the eyes, 80 nm thick, were made and stained with uranyl acetate and lead citrate and examined with a FEI Tecnai electron microscope (FEI Tecnai T12 Twin Fei Company, Eindhoven, The Netherlands; camera by OneView, Gatan) operating at 120 Kv. Overlapping images were collected and stitched together into separate images, as previously described [41].

Primary Antibodies
The following primary antibodies were used: β-catenin

Quantification of the Number of Photoreceptor Nuclei in a Row
The number of photoreceptor cell nuclei in a row were measured every 250 µm from the optic nerve head (ONH) in 1-month-old (1M), 3M, 6M, 9M, and 12M Crb2 ∆Rods , Crb1 KO Crb2 ∆Rods , Crb1 KO Crb2 flox/flox and Crb2 flox/flox mice. Three sections in the optic nerve area of 3 retinas from 3-4 independent mice per group were used for the measurements. Retina sections were scanned using a Pannoramic 250 digital slide scanner (3DHISTECH Ltd., Budapest, Hungary) and measurements were performed with CaseViewer 2.1 (3DHISTECH Ltd., Budapest, Hungary). The values of different sections of individual mice were averaged. The number of photoreceptor nuclei from the Crb2 ∆Rods and Crb1 KO Crb2 ∆Rods were compared to the Crb2 flox/flox control mice.

Statistical Analysis
Normality of the distribution was tested by the Kolmogorov-Smirnov test. Statistical significance was calculated by using an unpaired t-test or by using Mann-Whitney U test if the data did not show a normal distribution. All statistical analyses were performed using GraphPad Prism version 7.02 (GraphPad Prism, RRID: SCR_002798). All values are expressed as mean ± SEM. Statistically significant values: * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001.

Conclusions
In conclusion, we found that (i) loss of CRB2 from rod photoreceptors results in retinitis pigmentosa, mainly at the peripheral and central superior retina; (ii) CRB2 in rod photoreceptor cells is required to maintain the rod photoreceptor layer and retinal electrical responses; (iii) loss of CRB2 in rods and concomitant loss of CRB1 leads to an exacerbation of the retinitis pigmentosa phenotype; (iv) ablation of CRB2 from rods with concomitant loss of CRB1 results in visual function impairment. This clear deficit of contrast sensitivity may allow us to have a rigorous outcome parameter to measure functional vision gain or maintenance after AAV treatment. Therefore, Crb1 KO Crb2 ∆Rods mice might become important resources to test therapies for retinopathies due to mutations in the CRB1 gene.

Patents
The Leiden University Medical Center (LUMC) is the holder of patent application PCT/NL2014/050549, which describes the potential clinical use of CRB2. J.W. is listed as inventor on this patent, and J.W. is an employee of the LUMC.