Gene therapy rescues cone function in congenital achromatopsia

Abstract

The successful restoration of visual function with recombinant adeno-associated virus (rAAV)-mediated gene replacement therapy in animals and humans with an inherited disease of the retinal pigment epithelium has ushered in a new era of retinal therapeutics. For many retinal disorders, however, targeting of therapeutic vectors to mutant rods and/or cones will be required. In this study, the primary cone photoreceptor disorder achromatopsia served as the ideal translational model to develop gene therapy directed to cone photoreceptors. We demonstrate that rAAV-mediated gene replacement therapy with different forms of the human red cone opsin promoter led to the restoration of cone function and day vision in two canine models of CNGB3 achromatopsia, a neuronal channelopathy that is the most common form of achromatopsia in man. The robustness and stability of the observed treatment effect was mutation independent, but promoter and age dependent. Subretinal administration of rAAV5-hCNGB3 with a long version of the red cone opsin promoter in younger animals led to a stable therapeutic effect for at least 33 months. Our results hold promise for future clinical trials of cone-directed gene therapy in achromatopsia and other cone-specific disorders.

Figure 1.

Normal rod function and loss of cone function in CNGB3^m/m- and CNGB3^−/−-mutant dogs and short-term restoration of cone ERG signals following a single subretinal treatment with rAAV5-PR2.1-hCNGB3. Representative, ERG traces evoked by full-field white flashes under dark-adapted (rod and mixed cone-rod responses) and light-adapted (cone 1 and 29 Hz) conditions are shown. Compared with an age-matched normal wild-type dog, the treated eye of the CNGB3^−/− dog (M606; see Table 1) showed restoration of cone function as elicited by single and 29 Hz flicker light flashes 7 weeks after subretinal injection. The smaller amplitude of the restored cone function compared with the normal dog can be explained by the fact that the subretinal bleb covered ∼30% of the entire retina.

Figure 2.

Presence of L/M- and S-cones in CNGB3-mutant canine retinas, expressing markers that characterize the differentiated state. Comparison of canine retinal structure and cone-specific protein expression in normal (A1—A4; 12 weeks), CNGB3^−/− (B1—B4; 17 weeks), and CNGB3^m/m (C1—C4; 12 weeks) dogs. The H&E-stained sections show normal outer retinal structure independent of disease status (A1, B1, C1). The expression and distribution of L/M- and S-opsin is not affected by the CNGB3-mutation. L/M-opsin labeling (green) co-localizes with cone arrestin expression (hCAR labeling in red) in the outer segments of the L/M-cones (A2, B2, C2). In contrast, cone arrestin labeling is weak in S-cone outer segments; hence, the co-localized signal is dominated by the green S-opsin labeling (arrows in A3, B3, and C3). Closer analysis of cone arrestin distribution shows that the hCAR labeling is much weaker in the S-cone outer segments compared with the L/M-cones of the canine WT retina (arrows in A4). hCAR labeling is essentially not visible in the S-cone outer segments of the CNGB3-mutant dogs (arrows in B4 and C4). Calibration bars = 20 µm. RPE, retinal pigment epithelium; OS, outer segment; IS, inner segment.

Figure 3.

Promoter-dependent robustness of GFP-transgene expression in CNGB3^−/−-mutant cone photoreceptors. (A) 25 weeks after subretinal injection, GFP expression is cone-specific, but weak when using the 3LCR-PR0.5 promoter; only one GFP-positive cone is seen without immunohistochemical labeling. (B) Thirty-eight weeks following injection with 3LCR-PR0.5 promoter, immunohistochemistry using a GFP antibody shows that most cones are GFP positive. (C) Four weeks after subretinal injection, GFP-transgene expression was most robust with the PR2.1 promoter, and GFP protein specifically located in L/M-cones (red) and visible as native fluorescence. (D) Wide-field fundus photograph shows the posterior segment of a canine eye with the triangular shaped yellow-green tapetal fundus. It contains the round subretinal bleb superior to the optic nerve head visible immediately following successful injection. (E) Schematic representation of treated area in right eye 4 weeks after subretinal injection with PR2.1-GFP vector. The vector bleb (blue) is located within the canine tapetal fundus (green triangle) and covers part of the horizontally elongated area centralis (black), the site of maximal cone density. Robust GFP expression is visible without immunohistochemical enhancement in the center of the original subretinal vector bleb (E1). Transgene expression tapers off beyond the periphery of the bleb area (E2, E3). Calibration bars = 40 µm.

Figure 4.

Long-term restoration of cone retinal function after a single subretinal treatment with rAAV5-PR2.1-hCNGB3. (A) Representative 29 Hz cone flicker responses recorded from a CNGB3^m/m- and CNGB3^−/−-mutant dog over 33 months after subretinal injection. The successful restoration of cone function was sustained in both animals without any long-term deterioration of the rescue effect. (B) Long-term restoration of the 29 Hz cone flicker responses are shown for four treated dogs (see also Table 1). There was no deterioration of the ERG amplitudes over time.

Figure 5.

Restoration of day vision evaluated by objective behavioral testing with an obstacle avoidance course. The graph shows the transit time in seconds for dogs navigating a 3.6 m obstacle course as a function of ambient light intensity. CNGB3^m/m- and CNGB3^−/−-mutant animals were combined in this figure. Compared with untreated CNGB3-mutant dogs, transit times are significantly shorter in the unilaterally treated dogs at ≥25 lux. At higher light intensities of 65 and 646 lux, there were significant differences between the untreated and treated dogs, with the transit times in treated animals being close to the normal control values. Even though the transit time significantly improved with gene replacement therapy, it did not completely normalize, probably because only ∼30% of the retina was treated in only one eye. See Supplementary Material, Movie S2, for examples of dogs navigating the obstacle avoidance course. Data of normal controls and untreated CNGB3-mutants taken from Ref. (42). P-values: *P < 0.01; **P < 0.001; ***P < 0.0001.

Figure 6.

Restoration of normal protein localization following subretinal treatment with rAAV5-hCNGB3 and rescue of cone function. Compared with the normal wild-type retina (A1, A3), both alpha transducin (GNAT2) and CNGA3 are not visible in the cone outer segments of the untreated canine CNGB3^−/− (B1, B3) or CNGB3^m/m (C1, C3) retinas. However, the localization of L/M-opsin is unaffected by disease (A2, A4, B2, B4, C2, C4). Normal cone outer segment localization of both GNAT2 (D1, E1) and CNGA3 (D3, E3) was restored in vector-treated regions of both the CNGB3^−/− and CNGB3^m/m retinas. Localization of L/M-opsin was unaffected by the treatment (D2, D4, E2, E4). Calibration bar = 20 µm. (F) Immunoblot of heterologously expressed CNGA3 subunits and canine retina homogenates using polyclonal anti-canine CNGA3-antibody. Despite the absence of CNGA3-labeling by immunohistochemistry in untreated CNGB3-mutant cones (B3, C3), a western blot shows that CNGA3 is present in both the canine CNGB3^m/m and CNGB3^−/− retinas. For comparison, cultured tSA-203 cells were transfected with canine CNGA3 (right panel). The molecular weight of the major CNGA3 band is 103 kDa; the faint band is 98 kDa.

Figure 7.

Representative histogram showing relative mRNA expression levels of the hCNGB3 transgene as determined by qRT-PCR following subretinal injection with rAAV5-hCNGB3. (A) While no hCNGB3 transgene expression was found in the wild-type (wt) and untreated achromatopsia (ACHM) retinas, all retinas injected with the therapeutic vector had detectable transgene expressions. The expression was highest with the PR2.1 promoter. The expression levels were significantly higher in the successfully PR2.1-injected retinas (with rescued cone function) compared with non-successfully 3LCR-PR0.5-injected retinas (with no/transient rescue of cone function). For comparison, expression levels of L/M-, S- and rod opsin are shown. The expression levels of both rod and cone opsins are similar between wt and ACHM-affected treated and untreated retinas. Only the S-opsin mRNA expression was significantly higher in the wt compared with the unsuccessfully treated ACHM retinas. P-values: *P < 0.05. (B) For comparison, a scatter plot shows the expression levels of the cone-specific genes cCNGA3, cCNGB3, L/M- and S-opsin in untreated normal wild-type (wt, n = 7), carrier (CNGB3^+/m, n = 3; CNGB3^+/−, n = 4) and ACHM-affected dogs (CNGB3^m/m, n = 3; CNGB3^−/−, n = 3). Except for the non-detectable cCNGB3 in the CNGB3^−/− dogs, the expression levels did not differ between animal groups. (C) Cone ERG flicker amplitude increased with higher hCNGB3 transgene expression: comparison of detectable retinal hCNGB3 transgene expression levels with the cone flicker ERG amplitudes measured before tissue collection in 12 eyes revealed a highly significant nonparametric Spearman rank correlation of 0.85 (P < 0.0001). Relative gene expression for each gene compared to 18S was calculated as .