Germline knockout of Nr2e3 protects photoreceptors in three distinct mouse models of retinal degeneration

Significance

Retinitis pigmentosa (RP) is a common form of retinal degenerative disease that can be caused by mutations in any one of dozens of rod photoreceptor genes. The genetic heterogeneity of RP represents a significant challenge for the development of effective therapies. Here, we present evidence for a potential gene-independent therapeutic strategy based on targeting Nr2e3, a gene encoding a transcription factor required for the normal differentiation of rod photoreceptors. Nr2e3-deficient mouse rods express the normal complement of rod photoreceptor genes but also a subset of cone genes. We show that Nr2e3-deficient rods are remarkably resistant to degeneration in three mechanistically diverse RP models, suggesting that the upregulation of cone genes in Nr2e3-deficient rods has a strong neuroprotective effect.

Abstract

Retinitis pigmentosa (RP) is a common form of retinal dystrophy that can be caused by mutations in any one of dozens of rod photoreceptor genes. The genetic heterogeneity of RP represents a significant challenge for the development of effective therapies. Here, we present evidence for a potential gene-independent therapeutic strategy based on targeting Nr2e3, a transcription factor required for the normal differentiation of rod photoreceptors. Nr2e3 knockout results in hybrid rod photoreceptors that express the full complement of rod genes, but also a subset of cone genes. We show that germline deletion of Nr2e3 potently protects rods in three mechanistically diverse mouse models of retinal degeneration caused by bright-light exposure (light damage), structural deficiency (rhodopsin-deficient Rho^−/− mice), or abnormal phototransduction (phosphodiesterase-deficient rd10 mice). Nr2e3 knockout confers strong neuroprotective effects on rods without adverse effects on their gene expression, structure, or function. Furthermore, in all three degeneration models, prolongation of rod survival by Nr2e3 knockout leads to lasting preservation of cone morphology and function. These findings raise the possibility that upregulation of one or more cone genes in Nr2e3-deficient rods may be responsible for the neuroprotective effects we observe.

Retinal degeneration affects millions of people worldwide and can be caused by mutations in any one of more than 250 genes (1–3). Retinitis pigmentosa (RP) is the most common form of retinal degeneration and can result from mutations in dozens of individual genes, many of which are rod-enriched or rod-specific (4, 5). While rods constitute the great majority of photoreceptors in most mammalian retinas including mice and humans, their loss has only relatively mild effects on human visual function which manifest as night blindness. The most severe consequence of rod degeneration is secondary cone loss which occurs via mechanisms that are still actively investigated (6). The death of cones, which mediate daytime vision, is particularly disabling for patients and represents the primary source of morbidity in RP. Therefore, preventing secondary cone loss is a key goal of therapy for patients with RP.

The genetic heterogeneity of RP represents a significant challenge for the development of effective treatments for this disease. For this reason, there is a strong motivation to develop gene-independent strategies that could be used to treat a wide range of genetic forms of RP (7, 8). One promising approach for creating a gene-independent therapy for RP is based on genetic reprogramming of rod photoreceptors. We showed previously that reprogramming adult rods into a cone-like state by acutely knocking out the rod-specific transcription factor Nrl delayed rod death and preserved native cones in a rhodopsin (Rho) knockout mouse model of retinal degeneration (9). Subsequent work extended these findings, showing that acute Nrl knockout makes rods resistant to the effects of mutations in multiple rod-specific genes (9–11). However, as Nrl is a key early-acting transcription factor in the rod transcription network (12, 13), its knockout causes a significant shift in the gene expression profile of rods that ultimately results in changes in their structural and functional properties (9). Moreover, mutations in Nrl itself are a known cause of RP in both mice and humans (14–17). For these reasons, we sought to identify additional genes downstream of Nrl that might be targeted therapeutically but without causing widespread changes in rod gene expression.

For this purpose, we decided to target Nr2e3, a transcription factor downstream of Nrl (12, 18). Nr2e3 mutant mice (Nr2e3^rd7/rd7; subsequently referred to as “rd7”) have hybrid rod photoreceptors that express the full complement of rod genes as well as a subset of cone genes (19, 20). Crucially, for our purpose, the hybrid rod photoreceptors of rd7 mice have nearly normal rod-like morphology and physiology, as was previously shown (19, 21, 22). Accordingly, a future gene therapy approach targeting Nr2e3 would be expected to be much less disruptive than knocking out Nrl and might both preserve rod function and prevent secondary cone loss.

In the present work, we evaluated the feasibility of this idea by characterizing the effects of germline knockout of Nr2e3 in three mouse models of photoreceptor degeneration including light-induced photoreceptor damage (23) and two models of RP which vary in rate of disease progression and mechanism: Pde6b^rd10/rd10 (subsequently referred to as “rd10”) and Rho^−/−(24–27). We show that Nr2e3 knockout slows the degeneration and functional decline of rod cells in all three models, thereby preventing secondary cone death and preserving cone-mediated daylight vision. These experiments establish the feasibility of suppressing the expression Nr2e3 in photoreceptors as a therapeutic strategy for treating a broad range of retinal degenerative disorders.

Results

Photoreceptors of rd7 Mice Are Resistant to the Damaging Effects of Bright Light.

To determine whether mutations in Nr2e3 can protect rods against light-induced degeneration, we performed a series of light-damage experiments on 3-mo-old wild-type (WT) and rd7 littermates. All mice used in these experiments were homozygous for the Leu-450 isoform of RPE65, which makes them more susceptible to light-induced photoreceptor loss (28). We exposed awake animals to bright white light (15 kLux) for 8 h, returned the animals to a normal light/dark cycle for 1 wk, and then analyzed their retinas by histology. This light damage (LD) regimen caused extensive death of rods in WT mice (Fig. 1A). The most severe rod loss (~73%) was observed in the central retina, immediately dorsal to the optic nerve head, whereas a lesser degree of loss was observed in the ventral–central retina (~44%). Overall, the peripheral retina was less affected (Fig. 1C). In stark contrast, rd7 rods showed no appreciable evidence of degeneration after identical light exposure (Fig. 1 B and D). Moreover, cone arrestin immunoreactivity was largely extinguished in WT retinas but well-preserved in rd7 mutants (Fig. 1 A and B).

Fig. 1.

Rods of rd7 mice are resistant to LD. (A and B) Cross-sections of dorsal–central retinas of WT (A) and rd7 (B) mice immunostained with anti-rhodopsin (red) and anti-cone arrestin (green) antibodies. Photoreceptor nuclei were counterstained with DAPI (blue). Four-month-old WT control (A) and rd7 (B) animals were either unexposed (Left) or exposed to bright white light (15 kLux) for 8 h (Right). ONL: outer nuclear layer, INL: inner nuclear layer, GCL: ganglion cell layer. (Scale bar, 50 μm.) (C and D) ONL thickness [quantified as the number of rows of photoreceptor nuclei (ONL rows)] in four regions of the retina from WT (C) and rd7 (D) animals either unexposed (open circles) or exposed to 15 kLux white light for 8 h (closed circles). DP: dorsal–peripheral, DC: dorsal–central, ONH: optic nerve head, VC: ventral–central, VP: ventral–peripheral. Values are means ± SEM (n = 4 in all cases). Statistical significances of the data are presented as **P < 0.01, ***P < 0.001, or NS (not significant, P > 0.05). Numbers show percentages of ONL thickness reduction in light-exposed mice compared to controls.

To extend these findings and correlate them with changes in rod physiological function, we exposed mice to bright white light of three different intensities and evaluated scotopic retinal responses by in vivo electroretinogram (ERG) recordings 1 wk later. We first exposed WT and rd7 mice to a 7.5-kLux light for 5 h (i.e., relatively low LD). One week after light exposure, WT mice experienced a ~40% reduction of the maximal rod-driven ERG a-wave (Fig. 2A) and an accompanying ~25% decline in maximal rod ON bipolar cell-driven ERG b-wave (SI Appendix, Fig. S1), relative to unexposed control animals. Remarkably, the same irradiation did not cause any reduction in the scotopic ERG a-wave responses of rd7 mice (Fig. 2B); there was also no decline in the scotopic ERG b-wave (SI Appendix, Fig. S1). We next exposed mice to a 15-kLux light for 8 h (i.e., intermediate LD). Compared to the 7.5-kLux exposure, this treatment caused more severe reduction of both ERG a-wave (~75%; Fig. 2C) and b-wave (~50%; SI Appendix, Fig. S1) responses in WT animals, consistent with the substantial retinal degeneration observed in this condition (Fig. 1A). However, this brighter light exposure still did not affect either of these ERG components in rd7 mice (a-waves in Fig. 2D). The analysis of cone contribution into dark-adapted ERG response indicated that this light regimen also caused a ~twofold decline of M-cone function in WT mice (from 110 ± 3 µV in unexposed group, n = 12, to 48 ± 18 µV in light-exposed animals, n = 10, *P < 0.05). In contrast, the cone function was fully preserved in rd7 mice under the same conditions (100 ± 16 µV for unexposed, n = 14; 95 ± 2 µV for irradiated animals, n = 10, P > 0.05). Last, we exposed mice to a 40-kLux light for 8 h (i.e., high LD). In WT mice, we observed a dramatic decrease of both scotopic ERG a-waves (Fig. 2E) and b-waves (SI Appendix, Fig. S1), comparable to that observed in the intermediate LD experiment. At this very bright light intensity, we observed reduction of both ERG components in rd7 mice (a-waves in Fig. 2F). Quantitative analysis of the ERG results obtained from the three different light regimens shows that the rods of rd7 mice are substantially more resistant to photodamage than their WT counterparts (Fig. 2 G and H). We conclude that Nr2e3 knockout protects rod integrity and function over a wide range of light exposures, resulting in the structural and functional preservation of cone cells as well.

Fig. 2.

Physiological characterization of rod susceptibility to LD in WT and rd7 mice by in vivo ERG. (A and B) Averaged scotopic intensity–response functions (mean ± SEM) for WT (A) and rd7 (B) mice either unexposed (open diamonds) or exposed to 7.5 kLux white light for 5 h (closed diamonds). n = 10 to 12 (control) or 8 to 14 (rd7) per group. (C and D) Averaged scotopic intensity–response functions (mean ± SEM) for WT (C) and rd7 (D) mice either unexposed (open circles) or exposed to 15 kLux white light for 8 h (closed circles). n = 10 to 12 (control) or 10 to 14 (rd7) per group. (E and F) Averaged scotopic intensity–response functions (mean ± SEM) for WT (E) and rd7 (F) mice either unexposed (open squares) or exposed to 40 kLux white light for 8 h (closed squares). n = 10 to 12 (control) or 10 to 14 (rd7) per group. Data in (A–F) were fitted with hyperbolic Naka-Rushton functions. Error bars for some points in (A–F) are smaller than the symbol size. (G and H) Quantification of maximal scotopic ERG a-wave (G) or b-wave (H) (both recorded with green test flash of 23.5 cd s m⁻²) changes in WT and rd7 mice at different LD conditions. The data were normalized to responses obtained from unexposed mice (i.e., no LD). Values are means ± SEM. Statistical significances of the data in (A–H) are shown as *P < 0.05, ***P < 0.001, or NS (not significant, P > 0.05).

Nr2e3 Knockout Protects Photoreceptors from Degeneration in Rho^−/− Mice.

To determine whether Nr2e3 deficiency can prevent photoreceptor degeneration in a mouse model of RP, we crossed rd7 mice with Rho^−/− mice, which lack rhodopsin and consequently fail to form rod outer segments (29). Rho^−/− mice experience a progressive loss of rods. While cones are not directly affected by the deletion of rhodopsin and initially develop normally, rod death eventually leads to secondary cone loss (26).

Consistent with previous results (26), we observed only a single remaining row of photoreceptor nuclei in the outer nuclear layer (ONL) of 4-mo-old Rho^−/− mice (Fig. 3A). In contrast, the ONL of Rho^−/−;rd7 mice was largely preserved (Fig. 3B) even though rods still lacked outer segments (SI Appendix, Fig. S2). Moreover, while both S- and M-cone opsins were barely detectable in cone outer segments of Rho^−/− mice (Fig. 3 A, Top and Middle rows), both visual pigments were still abundantly expressed in cones of Rho^−/−;rd7 mice (Fig. 3 B, Top and Middle rows). Immunostaining for cone arrestin further confirmed the presence of severe cone degeneration in Rho^−/− mice, while showing relatively preserved cone morphology in Rho^−/−;rd7 mice (Fig. 3 A and B, Bottom rows). To quantify the extent of photoreceptor rescue by rd7, we counted nuclei in the ONL of 4-mo-old Rho^−/− and Rho^−/−;rd7 mice. We found that the ONL of Rho^−/−;rd7 mice contained >fivefold more photoreceptor nuclei (105 ± 14 nuclei/100-µm segment of retina; n = 3) than that of Rho^−/− mice (20 ± 0.5 nuclei/100 µm; n = 3), indicating a powerful neuroprotective effect of the rd7 mutation (***P < 0.001; n = 3; independent-samples t test). Together, these results indicate that Nr2e3 knockout potently prolongs the survival of Rho^−/− rods and prevents secondary cone death.

Fig. 3.

Preservation of cone photoreceptors in 4-mo-old Rho^−/−;rd7 mice. (A and B) Cross-sections of ventral–central (Top) and dorsal–central (Middle and Bottom) retinas of Rho^−/− (A) and Rho^−/−;rd7 (B) mice immunostained with anti-S-opsin (Top), anti-M-opsin (Middle), or anti-cone arrestin (Bottom) antibodies (red). Photoreceptor nuclei in (A and B) were counterstained with DAPI (blue). ONL: outer nuclear layer, INL: inner nuclear layer. (Scale bar, 50 μm.)

Cone Function and Photopic Vision Are Preserved in Nr2e3-Deficient Rho^−/− Mice.

We next used ERG to monitor the progression of photopic functional decline in Rho^−/− and Rho^−/−;rd7 mice. Despite the preservation of rod cell bodies in Rho^−/−;rd7 animals (Fig. 3), the lack of rod visual pigment prevented these cells from responding to light and, as a result, rod-driven ERG responses were absent in this model. Consistent with the morphological changes described above, the M-cone-driven ERG b-wave was normal in 2-mo-old Rho^−/− mice but declined rapidly and was barely detectable by 4 mo of age (Fig. 4 A–D). The cone b-wave responses in Rho^−/−;rd7 animals were substantially larger than those in controls at 2 mo, possibly reflecting synaptic remodeling and plasticity in the retinas of Rho^−/−;rd7 mice in the absence of functional rods, or altered outer retina resistance due to the lack of rod outer segments. Strikingly, M-cone function was largely preserved up to 6 mo of age in Rho^−/−;rd7 mice (Fig. 4 A–D). At 6 mo of age, Rho^−/−;rd7 cone responses were comparable to those in control WT and rd7 mice (Fig. 4D) and significantly higher than those in Rho^−/− mice. Eventually, cone function in Rho^−/−;rd7 mice declined, yet photopic responses remained significantly higher (*P < 0.05) than those in the Rho^−/− group out to 10 mo of age.

Fig. 4.

Preservation of cone function in Rho^−/−;rd7 mice. (A–C) Averaged cone ERG b-wave intensity–response functions (mean ± SEM) for Rho^−/− and Rho^-/-;rd7 mice at 2 mo (A), 4 mo (B), and 6 mo (C) of age. Data in (A–C) were fitted with hyperbolic Naka-Rushton functions. n = 6 to 8 (Rho^−/−) or 6 to 10 (Rho^−/−;rd7) per group. (D) Quantification of cone maximal ERG b-wave (recorded with white test flash of 700 cd s m⁻²) at multiple postnatal time points. n = 6 (WT control), n = 6 to 8 (rd7 control), n = 6 to 8 (Rho^−/−), n = 6 to 10 (Rho^−/−;rd7) per group. Values are means ± SEM. Error bars for some points in (A–D) are smaller than the symbol size. In (A–D), statistical significances of the data (between Rho^−/− and Rho^−/−;rd7 lines) are shown as *P < 0.05, or ***P < 0.001.

We also evaluated visual acuity and contrast sensitivity in the same cohorts of mice by measuring their optomotor head-turning responses to rotating vertical grating stimuli (30, 31). We found that visual acuity was ~3 times greater in 4-mo-old Rho^−/−;rd7 mice than in Rho^−/− animals under bright light (photopic) conditions in which vision is exclusively mediated by cones (Fig. 5A). Photopic visual acuity was well-preserved in Rho^−/−;rd7 mice for up to 10 mo (Fig. 5B), while it deteriorated rapidly in Rho^−/− animals. Notably, photopic visual acuity was preserved to a greater extent than cone ERG responses (Fig. 4D), which might be explained by the compensatory plasticity of overall mouse visual function, as observed previously (32).

Fig. 5.

Long-term preservation of photopic vision in Rho^−/−;rd7 mice. Quantification of changes in photopic visual acuity (A and B) and photopic contrast sensitivity (C and D) at multiple postnatal time points. Raw data, as shown in panels (A) and (C) for 4-mo-old mice, were derived from optomotor responses to rotating gratings under photopic (1.85 log cd m⁻²) background illumination conditions. n = 3 (WT control), n = 4 (rd7 control), n = 4 (Rho^−/−), n = 3 to 5 (Rho^−/−;rd7) mice per group. Values are means ± SEM. Error bars for some points in (B) and (D) are smaller than the symbol size. In (B) and (D), statistical significances of the data (between Rho^−/− and Rho^−/−;rd7 lines) were P > 0.05 (NS) at 2 mo and ***P < 0.001 for all other ages.

The photopic contrast sensitivity of Rho^−/−;rd7 mice was preserved to an even greater extent than visual acuity and was ~35-fold higher than that in Rho^−/− animals at 4 mo of age (Fig. 5C). This difference persisted for up to 10 mo (Fig. 5D). Together, these results demonstrate that Nr2e3 knockout potently preserves photopic visual function in Rho^−/− mice.

Nr2e3 Deficiency Promotes Long-Term Rod and Cone Survival and Function in rd10 Mice.

We next evaluated the neuroprotective effects of Nr2e3 deficiency in another mouse RP model, rd10. These mutant mice have greatly reduced levels of the catalytic subunit of phosphodiesterase 6B and hence impaired rod phototransduction (24, 33). They also experience more rapid rod degeneration than Rho^−/− mice, with secondary cone loss reaching completion by about 2.5 mo of age (34, 35). We crossed rd10 and rd7 mice to generate double mutants and analyze the structure and function of their rods and cones.

Similar to the case with Rho^−/− mice, we found that Nr2e3 knockout greatly suppressed the degeneration of photoreceptors in 4-mo-old rd10 mice, as evidenced by immunohistochemical staining of retinal cross-sections (Fig. 6 A and B). Unlike in rd10 mice, the outer segments of rd10;rd7 rods could be readily observed by anti-rhodopsin immunostaining and appeared healthy. The cones of rd10;rd7 mice were likewise preserved, as demonstrated by cone arrestin staining of retinal flat-mounts (Fig. 6 C and D). To quantify the extent of cellular rescue by rd7, we counted the nuclei in the ONL of 4-mo-old rd10 and rd10;rd7 mice. We found that there were ~13-fold more nuclei in the ONL of rd10;rd7 mice (192 ± 33 nuclei/100 µm; n = 3) compared to that of rd10 (15 ± 2 nuclei/100 µm; n = 3). The difference was highly significant (**P < 0.01, independent-samples t test). The potency of the neuroprotective effect is underscored by the fact that the number of nuclei remaining in the ONL of rd10;rd7 mice was comparable to the number in mice carrying the rd7 mutation alone (189 ± 22 nuclei/100 µm, n = 3).

Fig. 6.

Cone survival in 4-mo-old rd10;rd7 mice. (A and B) Cross-sections of dorsal–central retinas of WT and rd7 controls (A) and rd10 and rd10;rd7 (B) mice immunostained with anti-rhodopsin (red) and anti-cone arrestin (green) antibodies. Photoreceptor nuclei were counterstained with DAPI (blue). ONL: outer nuclear layer, INL: inner nuclear layer, GCL: ganglion cell layer. (Scale bar, 50 μm.) (C and D) Representative retinal flatmounts from rd10 (C) and rd10;rd7 (D) animals immunostained with anti-cone arrestin (red) antibody (Left panels; Scale bar, 1,000 μm). Right panels in (C) and (D) show higher magnification images obtained from dorsal–central regions outlined with white squares. (Scale bar, 50 μm.)

In agreement with these histologic findings, rod-driven ERG a-wave responses were dramatically preserved in rd10;rd7 mice at 4 mo of age relative to rd10 controls (Fig. 7A). Scotopic ERG responses of rd10;rd7 mice were indistinguishable from those of rd7 mice for up to 6 mo. In fact, scotopic function was almost completely preserved up to 10 mo of age (Fig. 7B). As expected, maximal ERG a-waves were reduced by ~30% in rd7 mice as compared to WT animals (Fig. 7B), likely due to the slightly lower number of rods in this strain overall (Fig. 2). The same degree of functional preservation was also observed for scotopic ERG b-waves in rd10;rd7 animals (Fig. 7C). Remarkably, the accompanying decline of photopic ERG b-wave responses in rd10 mice was fully corrected by Nr2e3 knockout as well (Fig. 7D). Importantly, physiological rescue also resulted in preservation of scotopic (Fig. 8 A and B) and photopic (Fig. 8 C and D) visual acuity and contrast sensitivity in rd10;rd7 mice. Collectively, our findings indicate that Nr2e3 deficiency preserves long-term rod- and cone-mediated visual function in rd10 mice.

Fig. 7.

Preservation of photoreceptor function in rd10;rd7 mice. (A) Averaged scotopic ERG a-wave intensity–response functions (mean ± SEM) for rd10 and rd10;rd7 mice at 4 mo of age. Data were fitted with hyperbolic Naka-Rushton functions. n = 10 (rd10) or 12 (rd10;rd7) per group. (B–D) Quantification of scotopic maximal ERG a-wave (B) and b-wave (C) and photopic maximal ERG b-wave (D) (all recorded with white test flash of 700 cd s m⁻²) at multiple postnatal time points. n = 6 (WT control), n = 6 to 8 (rd7 control), n = 8 to 10 (rd10), n = 4 to 12 (rd10;rd7) per group. Values are means ± SEM. Error bars for some points in (B–D) are smaller than the symbol size. In (A–D), statistical significances of the data (between rd10 and rd10;rd7 lines) were ***P < 0.001 for all ages.

Fig. 8.

Long-term preservation of scotopic and photopic vision in rd10;rd7 mice. Quantification of scotopic visual acuity (A), scotopic contrast sensitivity (B), photopic visual acuity (C), and photopic contrast sensitivity (D) at multiple postnatal time points. Data were derived from mouse optomotor responses to rotating gratings under scotopic (−4.45 log cd m⁻²) or photopic (1.85 log cd m⁻²) background illumination conditions. n = 3 (WT control), n = 4 (rd7 control), n = 4 (rd10), n = 3 to 5 (rd10;rd7) mice per group. Values are means ± SEM. Error bars for some points in (A–D) are smaller than the symbol size. In (A–D), statistical significances of the data (between rd10 and rd10;rd7 lines) were ***P < 0.001 for all ages.

Discussion

In the present study, we tested the hypothesis that germline mutation of Nr2e3 (rd7) would render rod photoreceptors resistant to degeneration and thereby prevent secondary cone loss. Indeed, mice with mutations in Nr2e3 show prolonged survival and function of both rods and cones in three mechanistically distinct models of disease: LD, Rho mutation, and Pde6b (rd10) mutation. Together, these findings indicate that the preservation of rods by the deletion of Nr2e3 might offer a viable approach for preventing secondary cone loss and preserving cone function, a key therapeutic goal in patients with RP.

The protective effect of the Nr2e3 mutation is unlikely attributable to the absence of NR2E3 protein per se, but rather to the changes in downstream gene expression that result from its loss. In 2005, three groups reported that mutation of Nr2e3 in mice causes upregulation of a subset of cone genes in rods without significant effects on rod gene expression (19, 20, 36). We subsequently used RNA-seq to compare the transcriptomes of WT and rd7 mouse retinas and found a total of 480 genes to be significantly dysregulated in adult rd7 retinas (37). To more closely evaluate these dysregulated genes, we compared this list with the results of another study in which we used RNA-seq to classify mouse genes as rod-enriched, cone-enriched, or not differentially expressed between the two cell types (38). We were able to map 479 of the 480 rd7-dysregulated genes onto this second dataset (SI Appendix, Table S1). In accordance with the reported role of NR2E3 as a repressor of cone genes in rods, we found that ~37% (154/415) of rd7-upregulated genes are cone-enriched. If we restrict our attention to those genes that are markedly upregulated in rd7 (i.e., ≥threefold), we find that ~61% (48/79) are cone-enriched. The majority of the remaining rd7-upregulated genes (251/415) are not differentially expressed between WT rods and cones, and the extent of their upregulation is generally minimal (i.e., <threefold). Only a small number (47/479) of rd7-dysregulated genes are rod-enriched, and no genes encoding components of the rod phototransduction cascade are affected to a functionally significant extent. Thus, the principal effect of the rd7 mutation is marked (≥threefold) upregulation of a small set of 79 genes, nearly two-thirds (61%) of which are cone-enriched in the adult mouse.

The upregulation of cone genes in rd7 rods raises the possibility that the neuroprotective effects we observed might be attributable, at least in part, to replacement of the defective rod gene by upregulation of its cone equivalent. Indeed, one of the most upregulated cone genes in rd7 rods is Pde6c (SI Appendix, Table S1), which encodes the catalytic subunit of the cone-specific phosphodiesterase. As shown in two separate studies, transgenic expression of the cone PDE6C protein (also referred to as PDE6α′) in mouse rods can restore phosphodiesterase activity and function in rods lacking PDE6β (39, 40). Thus, morphological and functional rescue of rd10 rods (which have markedly reduced levels of PDE6β) by rd7 might be attributable to upregulation of Pde6c expression and consequent restoration of phosphodiesterase activity.

At first glance, it might seem that a similar mechanism could account for rescue of the Rho^−/−phenotype by the rd7 mutation. Transcriptome analysis shows a 2.5-fold increase in blue cone opsin (Opn1sw) expression in rd7 retinas (SI Appendix, Table S1), suggesting that upregulation of cone opsin expression in Rho^−/−rods might restore function and therefore be neuroprotective. However, prior studies showed that there is no upregulation of Opn1sw in rd7 rods, despite the upregulation of various other cone-specific genes in these cells (19). Instead, there is ~twofold increase in the number of blue cones in the rd7 retina, which likely accounts for the modest increase in Opn1sw transcripts detected by whole-retina expression profiling (37). Thus, it appears that rd7-mediated rescue of Rho^−/−rods occurs despite the absence of opsin expression or detectable outer segments in these cells.

Similarly, it is unlikely that a “gene replacement” mechanism mediates the resistance to LD observed in the rd7 retina, as light-induced rod cell death is not caused by deficiency of one specific rod gene. Instead, the potent rescue we observe in three different models of photoreceptor degeneration suggests that either a single general neuroprotective mechanism or multiple model-specific mechanisms are operative in rd7 mice. We propose that these protective effects are likely mediated by the upregulation of one or more cone genes in rd7 rods. Future studies will be directed toward identifying the gene(s) mediating these therapeutic effects.

While promising, these results are only a step toward the development of a successful gene-independent therapy for RP. For obvious reasons, it is not possible to treat human RP patients by inducing germline mutations in NR2E3. It will therefore be necessary to devise a strategy to acutely knock out NR2E3 in mature rods. One strategy would be to use CRISPR-Cas9 technology to achieve knock out. Such a strategy was used previously to knock out mouse Nrl, thereby delaying degeneration in multiple mouse models of RP (9, 11). Alternatively, it might be possible to target NR2E3 using antisense oligonucleotides, an approach which has proven successful in the treatment of neurologic disease (41–43). The transcriptomic and therapeutic effects of acute NR2E3 knockout or knockdown would first need to be evaluated in mice and human retinal organoids.

Another outstanding issue is that loss of Nr2e3 produces distinct phenotypes in mice and humans (19). While rd7 mice show relatively modest changes in rod gene expression and exhibit no evidence of enhanced sensitivity to short-wavelength light, humans with mutations in NR2E3 have enhanced S-cone syndrome characterized by a supranormal ERG response to short-wavelength light (44). These observations suggest that the function of NR2E3 is not entirely conserved between the mouse and human and that NR2E3-mutant human rods may be transfated into supernumerary blue cones. Indeed, a recent analysis of NR2E3-mutant human retinal organoids shows an absence of rhodopsin-expressing rods and a late transfating of “divergent rods” into blue cones (45). Clearly, the therapeutic potential of targeting NR2E3 will require further study in human cells.

A third challenge is that humans with NR2E3 mutations exhibit a late-onset, slowly progressive photoreceptor degeneration. It is therefore possible that therapeutic targeting of NR2E3 in patients will cause a similar degeneration. However, this concern may be unwarranted, because acute knockout of Nr2e3 in mature rods is expected to cause a much milder perturbation of gene expression than germline knockout. Prior studies showed that germline knockout of mouse Nrl alters the expression of thousands of genes and results in a complete transfating of rods into blue cones (46, 47), whereas acute knockout of Nrl in mature rods changes the expression of only 146 genes (11). Thus, acute Nr2e3 knockout in mature rods is likely to yield a much milder phenotype than germline knockout while still retaining therapeutic efficacy.

Two other groups have recently begun to evaluate Nr2e3 as a potential therapeutic target for treatment of RP. Nakamura et al. reported that a small molecule inhibitor of NR2E3 (photoregulin3) prevents photoreceptor degeneration in a mouse model (48, 49). On the other hand, Li et al. suggested that overexpression of Nr2e3 slows disease progression in multiple mouse RP models (50). The results presented here confirm Nr2e3 as a promising therapeutic target, but they contradict some aspects of the prior work. Nakamura et al. found that photoregulin3 causes widespread downregulation of rod genes (48). In contrast, our findings indicate that germline Nr2e3 knockout leads to upregulation of cone genes without major effects on rod gene expression. This finding suggests that photoregulin3 may not act by simply inhibiting NR2E3 function but via a more complicated mechanism. Similarly, Li et al. suggested that adeno-associated virus-mediated overexpression of Nr2e3 preserves ONL thickness in four mouse models of RP (50). However, the histologic images in their paper (Fig. 4A in ref. 50) show no difference in ONL thickness between treated mice and controls. Moreover, our results indicate that Nr2e3 knockout prevents photoreceptor degeneration, a finding that is difficult to reconcile with Li et al.’s reported overexpression results. Resolving these discrepancies is of great significance because efforts are underway to translate this finding into a human therapy (51).

Materials and Methods

Animals.

All experiments were performed in accordance with the NIH Guide for the Care and Use of Laboratory Animals and the Association for Research in Vision and Ophthalmology Statement for the Use of Animals in Ophthalmic and Vision Research and were approved by the Institutional Animal Care and Use Committee of UC Irvine and Washington University in St. Louis. All mice used in this study, with the exception of those used in LD experiments, were homozygous for the Met-450 allele of the Rpe65 gene as determined by a genotyping protocol published elsewhere (52). All mice were confirmed to be free of the Crb1^rd8 mutation (53). The retinal degeneration 7 (rd7) strain (homozygous for the Met-450 allele of Rpe65) was obtained from The Jackson Laboratory (rd7;Rpe65^{Met-450/Met-450}; strain # 004643). For LD experiments, rd7 animals were crossed with 129S2/Sv mice (which carry the Leu-450 variant of Rpe65; Charles River Laboratories) to generate rd7;Rpe65^{Leu-450/Leu-450} mice. These mice were subsequently maintained on a mixed C57Bl6/J x 129S2/Sv genetic background. Mice with a knockout of the rhodopsin gene (Rho^−/−) were described previously (26). Animals with the missense R560C mutation in phosphodiesterase 6 β-subunit (Pde6b^R560C/R560C) causing the retinal degeneration 10 (rd10) phenotype were described previously (33). Both Rho^−/− and rd10 lines were crossed with rd7;Rpe65^{Met-450/Met-450}mice to generate Rho^−/−;rd7 and rd10;rd7 double mutants and corresponding Rho^−/− and rd10 controls on the same genetic backgrounds. All control and experimental mice of either sex were used at 1 to 10 mo of age. Animals were fed with standard chow (LabDiet 5053; Purina Mills) and raised under standard 12-h dark/light cyclic conditions.

Immunohistochemistry.

Preparation of histologic sections was performed as previously described (54). To facilitate identification of the dorsal portion of the retina, the dorsal cornea was marked by cautery (Bovie) immediately prior to enucleation. The eye was then punctured with a 33G needle and fixed in 4% PFA for 5 min. The cornea and lens were then removed, and the eye cup was fixed for an additional 40 min at room temperature. Eyecups were cryoprotected overnight in a solution of 30% sucrose in phosphate-buffered saline and then embedded in Optimal Cutting Temperature compound (Tissue-Tek). Retinal sections were cut at 14-µm thickness using a Leica CM1520 cryostat, mounted on Fisher Superfrost Plus slides, and stored at −20 °C until further use. Immunohistochemical staining was performed with the following primary antibodies: anti-M-opsin (1:200 dilution; Millipore, AB5405), anti-S-opsin (1:600; Millipore, AB5407), anti-rhodopsin (1:200; 4D2 was a kind gift from Robert Molday, University of British Columbia, Vancouver, Canada), and anti-cone arrestin (1:2,000; Millipore, AB15282). For anti-rhodopsin staining, an additional blocking step was performed using AffiniPure donkey anti-mouse Fab fragments (1:10, Jackson Immuno Research) prior to the primary antibody incubation. Secondary antibodies (1:1,000, Invitrogen) included Alexa Fluor 555 donkey anti-mouse and Alexa Fluor 488 or 555 donkey anti-rabbit. Sections were stained for 30 s with DAPI (10 µg/mL, Sigma) and mounted with Vectashield antifade mounting media (vectorlabs). Confocal images at 400× were taken as z-stacks on a Zeiss 880 laser-scanning confocal microscope in the Washington University Center for Cellular Imaging (WUCCI).

Whole-mount retinas were prepared as described (55). Briefly, eyes were marked and removed as with tissue sections. After excising the cornea, a small cut was made to mark the dorsal retina, and the sclera was then removed. Retinas were fixed in 4% PFA for 30 min at room temperature before proceeding with immunohistochemical staining using the primary antibodies listed in above. Before mounting, the lens was removed, and four small incisions were made to permit flattening of the retina. Retinal whole mounts were imaged at 25× magnification on an Olympus BX51 microscope.

Quantification of Photoreceptors.

Photoreceptor nuclei were quantified in a central region of the retina adjacent to the optic nerve head. Images of representative vertical retina cross-sections were obtained from three mice for each genotype at 400× magnification. Nuclei in the ONL were counted across the entire image field (295 µm) using the Cell Counter plugin in ImageJ. The counts were then normalized (nuclei per 100-µm length of ONL) and statistically compared by independent two-tailed Student’s t tests.

LD Experiments.

Pupils of WT (Rpe65^{Leu-450/Leu-450}) and rd7;Rpe65^{Leu-450/Leu-450} mice were dilated with a drop of 1% atropine sulfate, and the animals were placed into cages with a white, light-reflective coating. Each cage was divided into four equal compartments with transparent plexiglass dividers. To prevent grouping, every animal was kept in its own compartment for the duration of the experiment. Mice were exposed to broad-spectrum white light emitted by clusters of LEDs placed on top of each cage, under three different intensity regimens: 7.5 kLux for 5 h, 15 kLux for 8 h, or 40 kLux for 8 h. The animals had full access to food and water during light exposure. During light exposure, atropine sulfate was applied to mouse eyes every 2 h to keep the pupils dilated. After light exposure, animals were returned to their normal cages and kept under standard 12-h light/dark cyclic conditions for 1 wk, prior to ERG recordings and collection of tissue for histologic analysis.

In Vivo Electroretinography (ERG).

Mice were dark-adapted overnight and anesthetized with an intraperitoneal injection of ketamine (100 mg/kg) and xylazine (4 mg/kg). Pupils were dilated with a drop of 1% atropine sulfate. Mouse body temperature was maintained at 37 °C with a heating pad. ERG a-wave and b-wave responses were measured from both eyes by contact corneal electrodes held in place by a drop of Gonak solution (Akorn). Full-field ERGs were recorded with a UTAS BigShot apparatus (LKC Technologies) using Ganzfeld-derived test flashes of calibrated green 530 nm LED light (within a range from 2.2 × 10⁻⁵ cd s m⁻² to 23.5 cd s m⁻²) or white light generated by the Xenon Flash tube (from 80.7 cd s m⁻² to 700 cd s m⁻²), as previously described (56).

Optomotor Responses.

Mouse visual acuity and contrast sensitivity were measured with the OptoMotry system (Cerebral Mechanics) using a two-alternative forced-choice protocol (31), as previously described (57). Optomotor responses were measured under two background illumination conditions: scotopic (−4.45 log cd m⁻²) or photopic (1.85 log cd m⁻²). For scotopic conditions, background monitor luminance was controlled by neutral density filters.

Visual acuity was defined as the threshold for spatial frequency (F_s) of sine-wave gratings stimuli with 100% contrast and measured at a speed (S_p) of 6.3 deg/s for scotopic or 12.0 deg/s for photopic illumination conditions. In this mode, F_s was gradually increased by an automated protocol until its threshold was determined. Temporal frequency (F_t) was automatically adjusted by the computer program, based on the following equation: F_t = S_p·F_s (31). Contrast sensitivity was defined as the inverse of contrast threshold for optomotor responses. In this mode, contrast of the stimuli was gradually decreased by the computer until its threshold was reached. For scotopic illumination conditions, F_s was fixed at 0.128 cyc/deg, F_t was set to 0.8 Hz, and S_p was kept at 6.3 deg/s. For photopic conditions, F_s was fixed at 0.128 cyc/deg, F_t was set to 1.5 Hz, and S_p was kept at 12.0 deg/s.

Statistical Analysis.

For all experiments, data were expressed as mean ± SEM and analyzed with the independent two-tailed Student’s t test (using an accepted significance level of P < 0.05).

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.