Gene therapy prevents photoreceptor death and preserves retinal function in a Bardet-Biedl syndrome mouse model

Abstract

Patients with Bardet-Biedl syndrome (BBS) experience severe retinal degeneration as a result of impaired photoreceptor transport processes that are not yet fully understood. To date, there is no effective treatment for BBS-associated retinal degeneration, and blindness is imminent by the second decade of life. Here we report the development of an adeno-associated viral (AAV) vector that rescues rhodopsin mislocalization, maintains nearly normal-appearing rod outer segments, and prevents photoreceptor death in the Bbs4-null mouse model. Analysis of the electroretinogram a-wave indicates that rescued rod cells are functionally indistinguishable from wild-type rods. These results demonstrate that gene therapy can prevent retinal degeneration in a mammalian BBS model.

Bardet-Biedl syndrome (BBS) is clinically diagnosed by the presence of at least four of the following signs: retinal dystrophy, polydactyly, obesity, learning disabilities, male hypogonadism, and renal anomalies (1). Of these, the visual phenotype is particularly devastating: the average BBS patient will progress to legal blindness before his or her 16th birthday. There are currently 15 genetic loci (Online Mendelian Inheritance in Man #209900) that are known to be associated with BBS, and although inheritance of this disease was once thought to follow a classic autosomal recessive pattern, recent evidence suggests that a more complex pattern, termed “triallelic inheritance,” is involved at certain BBS loci (2). Only within the last decade has evidence emerged that BBS proteins play a role in ciliary function (3). More specifically, Bbs4 and the other BBSome components seem to be involved in both recruitment of cargo toward the ciliary basal body (4, 5) and in intraflagellar transport along the cilium (5–7). Several reports have elucidated the mechanisms by which disruption of BBS proteins give rise to the individual phenotypes seen in this highly pleiotropic syndrome (8–11).

Retinal degeneration is a central feature of all BBS mouse models generated to date (8, 12–16). In rod and cone photoreceptors, the connecting cilium is a highly specialized ciliary structure that serves as the sole conduit from the inner segment to the outer segment. Because protein synthesis occurs proximal to the outer segment, rhodopsin and other visual proteins must be trafficked through the connecting cilium to reach their site of action in the outer segment. Abd-El-Barr et al. (11) showed that when Bbs4 was deleted in mice, rhodopsin and cone opsins became grossly mislocalized in rod and cone photoreceptors, respectively. This was followed by apoptotic photoreceptor death and deterioration of the electroretinogram (ERG) a- and b-waves (14). Ultrastructural analysis of rods from young animals revealed normal-appearing connecting cilia and basal body structures. This latter finding has important implications for gene therapy: because the structural transport apparatus seems to develop normally in Bbs4-null photoreceptors, we hypothesized that postnatal supplementation of the Bbs4 gene in rods would rescue rhodopsin mislocalization and prevent rod photoreceptor death.

To test our hypothesis, we chose a pseudotyped adeno-associated virus (AAV) as a delivery vector. AAV has many desirable features as a retinal gene therapy vector, including low immunogenicity (17), high transduction efficiency (18), long-term transgene expression (19), and specific serotype-dependent cellular tropisms (20). Furthermore, data from three landmark phase I clinical trials suggest that AAV has an excellent safety profile in humans (21–23). Here, we describe the results of AAV-mediated Bbs4 delivery into the rods of Bbs4-null mice. Our findings demonstrate that gene therapy can rescue rhodopsin mislocalization in this BBS mouse model and maintain nearly normal rod outer segment morphology. Moreover, our treatment prevents photoreceptor death, improves electrophysiological function of the retina, and preserves visually evoked behavioral responses. These data from a mammalian model suggest that BBS-associated retinal degeneration can be treated.

Results

Retinal Degeneration in the Bbs4-Null Mouse.

We performed anatomical and functional studies in Bbs4-null mice and wild-type littermates to obtain baseline values before our rescue experiments. In the wild-type retina, immunohistochemical rhodopsin staining (red) localized to the outer segment region of rod photoreceptors and was completely absent from the outer nuclear layer (ONL) (Fig. 1A, 8 wk of age). Rhodopsin was also found in the outer segment region of age-matched Bbs4-null (^−/−) retinas, but this was accompanied by mislocalized rhodopsin molecules that tended to form annular patterns (Fig. 1B, arrowheads) around individual photoreceptor nuclei (blue). Fig. 1C illustrates the progressive loss of photoreceptor nuclei in the BBS retina, which others have shown to occur through apoptosis (11, 12). At 2 wk of age the ^−/− retina already showed evidence of photoreceptor loss (seven to eight nuclei per ONL column) compared with an age-matched ^+/+ control whose ONL was 10 to 11 nuclei thick. Photoreceptor death in the ^−/− retina progressed steadily over time, and only one to two nuclei per ONL column remained at 16 wk. Outer segments in the ^−/− retina were shorter and more heterogeneous than ^+/+ outer segments at the earliest time points examined and were entirely unrecognizable by 16 wk.

Fig. 1.

Rhodopsin mislocalization and retinal degeneration in Bbs4-null mice. Immunolocalization of rhodopsin (red) in 8-wk-old wild-type (A) and Bbs4^−/− (B) retinas. Topro-3 (blue) marks nuclei, and PNA (green) identifies cone sheaths. Arrowheads in B indicate rod photoreceptors that exhibit mislocalized rhodopsin. (C) Photomicrographs illustrate the progressive loss of photoreceptors in the ONL of Bbs4^−/− retinas, with wild-type sections at both ends for comparison. (D–F) Attenuation of the ERG accompanies photoreceptor loss. At 8 wk, Bbs4^−/− (red) responses are much smaller than wild-type littermates (black) at flash intensities eliciting either rod-only (D) or mixed rod–cone (E) responses. F is a high-magnification view of a-wave leading edges at two different intensities, with a photoresponse model fitted to the data (dotted traces). Flash intensities in log scot cd*s/m² are −3.65, −2.93, −2.08, and −0.89 for D; 3.45 for E; 2.10 and 3.45 for F. (Scale bars, 20 μm in A and B, 10 μm in C.)

We performed ERGs to measure retinal function in Bbs4-null mice over time. At all ages and flash intensities examined, a- and b-wave amplitudes were severely attenuated in ^−/− mice (red traces) compared with ^+/+ controls (black traces). Representative 8-wk-old data are shown for rod-only (Fig. 1D) and mixed rod–cone (Fig. 1E) responses. To parameterize the rod photoresponses, we performed an ensemble fit of the Cideciyan-Jacobson photocurrent model (24) against ERG a-wave data from two different flash intensities. Fig. 1F shows the a-wave leading edges from a representative ^−/− eye (Right) whose rod response is lower in both amplitude and sensitivity compared with an age-matched wild-type control eye (Left).

Expression of AAV Transgenes.

Previous characterization of this mouse model has shown that cone opsin mislocalization is severe by 2 wk of age (11), and cone ERGs are attenuated by 88% at 4 wk (14). For these reasons, we chose to direct our gene therapy efforts toward rescue of rod photoreceptors, which have a slower course of degeneration. We used a pseudotyped self-complementary adeno-associated virus (scAAV5) containing the mouse opsin promoter (mOP) to drive strong expression in rods (25). Two viral constructs were used in this study: a control virus containing GFP (henceforth referred to as scAAV5-mOP-GFP) and a therapeutic virus containing the mouse Bbs4 gene with an N-terminal HA tag (scAAV5-mOP-BBS4). In our hands, subretinal injection of 1.1 μL of scAAV5-mOP-GFP into a wild-type mouse eye resulted in a localized region of GFP expression (Fig. 2A) that covered 4.8% ± 1.1% of the retina (mean ± SEM, n = 6). Transduction of rod cells within those regions was extremely efficient, with nearly 100% of rod cells expressing GFP close to the injection site (Fig. 2Bi). In contrast, cone photoreceptors (arrowheads in Fig. 2Bi, red signal in Fig. 2Bii) expressing GFP were present but rare. Cones were identified by examining serial optical sections from flat-mounted retinas stained with rhodamine-conjugated peanut agglutinin (PNA) (Movie S1).

Fig. 2.

Transgene expression after subretinal injection of AAV. (A) Retinal flatmount 4 mo after scAAV5-mOP-GFP injection revealed localized gene expression only near the site of injection. (Scale bar, 1 mm.) (Bi) Confocal image of the top of the ONL showing rods (green from GFP) and cones (dark cells marked by arrowheads). Cones were identified by rhodamine-PNA staining in serial optical sections. (Scale bar, 10 μm.) (Bii) Inner segments belonging to three cones (PNA) are the only ones not expressing GFP. (Scale bar, 5 μm.) (C) Western blot analysis of a retinal sample injected with AAV carrying the HA-Bbs4 construct shows an HA-immunoreactive band near 59 kDa that is absent in the uninjected control sample. (D) AAV-mediated Bbs4 expression localizes primarily to the inner segments and proximal outer segments (asterisk), whereas no such staining is seen in uninjected controls. (Scale bar, 20 μm.) Right: A rhodopsin signal is shown to delineate the OS/IS border. All experiments were performed in phenotypically normal controls (Bbs4^+/+or Bbs4^+/−). Topro-3, nuclear stain; PNA, cone sheath marker; OS, outer segment; IS, inner segment; OPL, outer plexiform layer.

Injection of our scAAV5-mOP-BBS4 construct into the subretinal space resulted in a single HA-immunoreactive band by Western blot (Fig. 2C). This band, which was absent in the uninjected control sample, was located at ≈59 kDa, the expected molecular mass for Bbs4 (UniProt# Q8C1Z7) with an HA tag. To determine subcellular localization of the viral Bbs4 protein, we performed immunohistochemistry on 40-μm vibratome sections. Retinas injected with scAAV5-mOP-BBS4 demonstrated a clear band of punctate HA-positive staining (asterisk) that was not seen in uninjected control retinas (Fig. 2D). This band of staining was most dense in the inner segment region located between the photoreceptor nuclei (ONL) and the rhodopsin-rich outer segments (Fig. 2D, Right); however a few punctae were observed at the proximal margin of the outer segment layer. These data are consistent with tissue localization of the endogenous protein (4, 11) as well as observations from cell culture experiments that BBS4 can be found both inside the cilium (5) and at the centriolar satellites (4, 5).

Rescue of Rhodopsin Mislocalization.

After establishing the subcellular localization of viral Bbs4, we next determined whether our gene therapy treatment could prevent rhodopsin mislocalization. Two-week-old Bbs4-null mice were treated by subretinal injection of scAAV5-mOP-BBS4 into the temporal hemiretina of one eye. At 12 wk of age, horizontal cryostat sections were cut that contained a virus-treated region in the temporal half of the section and an untreated control region in the nasal half. Immunohistochemistry indicated that the percentage of photoreceptors with annular mislocalized rhodopsin staining (arrowheads) was lower in treated regions (Fig. 3A) compared with untreated regions (Fig. 3B). The percentages in treated vs. untreated regions for the three retinas examined (Fig. 3C) were 0.30% vs. 4.19% (P < 0.0001), 0.23% vs. 1.58% (P < 0.01), and 0.09% vs. 2.30% (P < 0.0001). At least 600 ONL cells were counted in each region, and significance was calculated using a χ² test for a 2 × 2 contingency table.

Fig. 3.

scAAV5-mOP-BBS4 rescues rhodopsin mislocalization and restores outer segment morphology. Treated (A) and untreated (B) regions of the same retina demonstrate clear differences in the percentage of photoreceptors with mislocalized rhodopsin (arrowheads). The red signal immediately below the ONL represents nonspecific staining characteristic of this antibody (Fig. 1A). Note also that the ONL is considerably thicker in the treated retina (A). Red, rhodopsin; blue, nuclear stain Topro-3. (Scale bars, 20 μm.) (C) In all three retinas examined at this age, the percentage of photoreceptors exhibiting rhodopsin mislocalization was significantly lower in treated vs. untreated regions (*P < 0.01; **P < 0.0001; significance calculated by χ² test for a 2 × 2 contingency table). (D) Regions of Bbs4^−/− retina treated with scAAV5-mOP-BBS4 contained outer segments that were strikingly similar to WT outer segments. In untreated control Bbs4^−/− retina, outer segments are not recognizable at 16 wk of age. (Scale bar, 10 μm.) OS, outer segment; IS, inner segment; INL, inner nuclear layer.

Outer segments in Bbs4-null retinas also showed improvement after treatment with scAAV5-mOP-BBS4. Treated ^−/− mice killed at 16 wk demonstrated patches of dense, homogeneous outer segments in the temporal hemiretina measuring 20.7 ± 0.52 μm in length (mean ± SEM, n = 12) (Fig. 3D, Right). These were similar to wild-type outer segments (Fig. 3D, Left) in both appearance and length (22.6 ± 0.21 μm, n = 12). Outer segments from untreated ^−/− retinas, in contrast, were sparse, heterogeneous, and short (13.1 ± 0.44 μm, n = 12) at 4 wk and unrecognizable at 16 wk (Fig. 3D, Center Left and Center Right).

Photoreceptor Rescue.

To determine whether gene therapy could prevent photoreceptor cell death in the Bbs4-null retina, we counted the number of nuclei remaining in cellular columns of the ONL. Fig. 4 shows both eyes from a ^−/− mouse treated at 2 wk and killed at 16 wk. The left eye received a temporal scAAV5-mOP-BBS4 injection, whereas the right eye received a temporal scAAV5-mOP-GFP injection, leaving the nasal portions of both retinas unexposed to either virus. Heat maps of ONL thickness (Fig. 4A) clearly demonstrate a localized region in the temporal half of the left eye, where the ONL was four to six photoreceptor nuclei in thickness. This region corresponds precisely to the portion of the retina exposed to the therapeutic Bbs4 virus. Regions not receiving the therapeutic virus (i.e., the nasal half of the left eye and both halves of the right eye) had undergone unmitigated degeneration that left the photoreceptor layer only one to two nuclei thick. Photomicrographs (Fig. 4B) from the five locations indicated in the heat maps illustrate the stark contrast between ^−/− retina with and without gene therapy treatment. Treated regions (i and ii) have a much thicker ONL along with nearly normal-appearing outer segments, whereas untreated regions (iii–v) have a homogeneously thin ONL and no outer segments.

Fig. 4.

Treatment with scAAV5-mOP-BBS4 prevents photoreceptor death. (A) At 16 wk, photoreceptor survival was assessed by outer nuclear layer (ONL) thickness in horizontal sections containing optic nerve. Heat maps indicate thick outer nuclear layer (red) at the site of scAAV5-mOP-BBS4 injection, and thin outer nuclear layer (blue) in all other regions. (B) Photomicrographs from treated (i and ii) and untreated (iii–v) locations indicated in A. (Scale bar, 10 μm.) Outer segments in Bii became bent during fixation and appear in cross-section. IS, inner segment; OS, outer segment; INL, inner nuclear layer.

The pattern of localized ONL thickness in the region treated with scAAV5-mOP-BBS4 was consistent across all retinas examined. This included two animals killed at 12 wk (Figs. S1 and S2), two animals killed at 16 wk (Fig. 4 and Fig. S3), and one animal killed at 45 wk (Fig. S4). This oldest animal was of particular interest because the untreated eye contained absolutely no photoreceptors, whereas the temporal region of the treated eye had six to seven rows of photoreceptors at its thickest point.

Functional Improvement of Vision.

To assess rod function in the Bbs4-null retina after gene therapy, we performed ERGs and parameterized a-wave responses using least-squares fit of the Cideciyan-Jacobson photocurrent model (Eq. 1). This analysis produced an amplitude (R_max) and sensitivity (σ) value to describe the rod response of each eye. Fig. 5A shows actual a-wave data (solid traces) at two different intensities and fitted curves (dashed traces) for a 12-wk-old ^−/− mouse with one eye treated (Left) and the other untreated (Right).

Fig. 5.

Analysis of ERG a-wave after scAAV5-mOP-BBS4 treatment. (A) Representative 12-wk a-wave leading edges and fitted model traces (dotted lines) for treated (Left) and control (Right) eyes of a Bbs4^−/− mouse. (B) The model-fitting procedure returned amplitude (R_max) and sensitivity (σ) parameters for eyes tested at 4, 8, 12, and 16 wk. Individual eyes are shown as open symbols (blue, scAAV5-mOP-BBS4; red, scAAV5-mOP-GFP or no injection), and filled symbols represent averages (error bars = SEM). Group averages are statistically different at 8, 12, and 16 wk (Table 1). (C) Average R_max and σ values from B were used as inputs to the photocurrent model, and output traces are shown for each of the four time points examined.

After treatment at 2 wk of age, ERG a-waves were analyzed at 4, 8, 12, and 16 wk. Data are displayed in Fig. 5B, with both individual eyes (open symbols) and average values (filled symbols) shown. At all four time points, treated eyes (scAAV5-mOP-BBS4) on average had both higher amplitude and higher sensitivity than control eyes (scAAV5-mOP-GFP or uninjected). The difference between treated and control eyes was small at 4 wk but grew wider and reached statistical significance at all three subsequent time points (Table 1). Of particular interest is the fact that the average sensitivity of treated eyes actually increased from 12 to 16 wk, reaching a value of 3.80 ± 0.077 log scot cd⁻¹m²s⁻³ (mean ± SEM, n = 15). This value was virtually indistinguishable from the age-matched wild-type value of 3.83 ± 0.064 (n = 6). Average values of R_max and σ from Fig. 5B were substituted into Eq. 1 to produce average a-wave leading edges for both groups (Fig. 5C) at all four time points (4 wk Left, 16 wk Right). Note the faster rise time and greater amplitude of the treated (blue) traces at all times. Interestingly, the amplitude difference between treated and control remained relatively constant over time, even though both traces decreased in absolute amplitude (Discussion).

Table 1.

Electroretinogram a-wave analysis

Age (wk)	Cohort	Rmax (μV)	σ (log scot cd−1m2s−3)	n	P
4	Treated	119.4 ± 11.4	3.66 ± 0.075	12	0.583
	Control	105.8 ± 7.63	3.61 ± 0.046	16
8	Treated	72.9 ± 4.97	3.62 ± 0.040	18	0.003
	Control	59.4 ± 3.07	3.42 ± 0.041	22
12	Treated	42.6 ± 3.45	3.60 ± 0.061	20	<0.0001
	Control	31.1 ± 1.62	3.25 ± 0.044	24
16	Treated	28.5 ± 2.91	3.80 ± 0.077	15	<0.0001
	Control	17.5 ± 1.60	3.08 ± 0.100	21

Rod response amplitudes (R_max) and sensitivities (σ) from Bbs4^−/− eyes injected at 2 wk of age. Treated eyes received scAAV5-mOP-BBS4, and control eyes received either scAAV5-mOP-GFP or no injection. P values were calculated using Hotelling's t-squared statistic.

We also performed optokinetic response (OKR) testing to determine whether the limited region of preserved retina was sufficient to elicit a visually evoked behavioral response. A total of nine mice were tested (ages 9–11 mo), each animal having received a monocular scAAV5-mOP-BBS4 injection at 2 wk of age. Preliminary testing under scotopic conditions revealed that four of the nine treated eyes could elicit a positive OKR, whereas no such response was seen for any of the contralateral untreated eyes. Further behavioral testing is currently under way.

Discussion

Here we present evidence that gene therapy can prevent the major pathologic hallmarks that are found in the Bbs4-null retina (11, 13, 14). Specifically, viral Bbs4 expression in rod photoreceptors rescues rhodopsin mislocalization and maintains healthy-appearing rod outer segments. Photoreceptor death is prevented in regions receiving gene therapy treatment, whereas untreated regions degenerate rapidly. We confirmed these anatomical findings by electroretinography and behavior. ERG rod responses demonstrated greater amplitude and higher sensitivity in treated eyes vs. untreated controls, and positive OKRs were elicited from some treated eyes but no untreated controls. Collectively, these results suggest that gene therapy is a feasible approach for treating blindness in human BBS patients. On the basis of our success in supplementing the Bbs4 gene, it seems very likely that other forms of BBS, especially those that are molecularly linked to the BBSome (5), will also be suitable targets for gene therapy. The same may hold true for other retinal ciliopathies, including Usher syndrome, Joubert syndrome, Senior-Loken syndrome, and certain forms of Leber congenital amaurosis (26). Furthermore, because obesity in BBS is also thought to occur by an analogous protein mislocalization mechanism involving the leptin receptor in hypothalamic neurons (9), it is possible that some forms of syndromic obesity may also be candidates for gene therapy.

In our hands, subretinal injection of scAAV5-mOP-GFP resulted in GFP expression with an average retinal coverage of only 4.8%. Although this is less area than others have estimated, we achieved extremely high rod transduction within the coverage area (approaching 100% near the site of injection) and minimized adverse events resulting from mechanical trauma to the tissue. The viral coverage pattern was beneficial for our study: it provided convenient internal controls when examining treated (temporal) and untreated (nasal) regions of transverse retinal sections. Moreover, because the transduction of rod cells was nearly 100% locally and absent in distant regions, the histological effects of the virus were readily apparent. For reasons previously mentioned, we chose the mouse opsin promoter to focus our efforts on rescuing rod photoreceptors. As a result, our therapeutic viral construct led to clear improvement in the rod system but little cone improvement. However, given a wider therapeutic time window and an appropriate promoter (27–29), there is no obvious reason why cones could not also be rescued. Unlike the mouse used in this study, the Bbs1 M390R knockin has relatively normal cone ERG function at 11 wk of age (16) and may therefore be an appropriate model for future evaluation of cone rescue.

Although the precise molecular role of Bbs4 and other BBS proteins in photoreceptors is not completely understood [see Zaghloul and Katsanis (30) for a review], much of the literature points to a role in intracellular protein trafficking and a localization in the vicinity of the ciliary basal body. Our observation of viral Bbs4 expression in the inner segment and proximal outer segment is entirely consistent with this idea. ScAAV5-mOP-BBS4 was able to reverse the prototypical rhodopsin mislocalization phenotype seen in several BBS mouse models (11, 12, 15), as well as other models in which ciliary proteins were deleted (31, 32). It is not entirely clear whether rhodopsin mislocalization in this model is a direct result of Bbs4 deletion or rather a secondary insult caused by photoreceptor degeneration. This distinction, however, is of little consequence for the present study. Rhodopsin mislocalization was found to precede apoptotic photoreceptor death in the Bbs2- and Bbs4-null mouse models (11, 12), so on the basis of our finding that gene therapy rescues rhodopsin mislocalization, we fully expected that gene therapy would also prevent photoreceptor death. This was indeed the case: localized injection of scAAV5-mOP-BBS4 at 2 wk of age resulted in localized photoreceptor rescue that persisted even until the last time point examined (45 wk). Conversely, untreated regions steadily degenerated until photoreceptor loss was complete. We have therefore established that a therapeutic window exists in the Bbs4-null mouse, during which the retinal disease process is reversible.

Analysis of the full-field ERG a-wave using the Cideciyan-Jacobson model (24) proved to be remarkably sensitive in this study. Without having to use a multifocal ERG technique we were able to observe a highly significant (P < 0.0001) improvement in rod function even though our subretinal injections covered only 5% of the retina. The corneal ERG potential is a summation of all radial extracellular currents in the eye, so in our cohort of eyes treated with scAAV5-mOP-BBS4 we hypothesized that the ERG represents a linear combination of 5% healthy (treated) retina and 95% degenerating (untreated) retina. The maximal a-wave amplitude, R_max, is related to the total circulating dark current in the outer segments at the time of light onset (33) and is therefore expected to decline in this mouse model as outer segments are lost owing to photoreceptor death. Our a-wave data (Fig. 5C) show that R_max in treated eyes declines from 4 to 16 wk but is larger than R_max from untreated control eyes at every time point. More importantly, the amplitude difference between the two experimental groups is nearly constant. The constant difference suggests that treated eyes have a small population of rescued outer segments that remain healthy over time (Fig. 4 Bi and Bii) and steadily contribute to the a-wave. This, however, is superimposed on a large background signal originating from unhealthy outer segments that degenerate over time.

Our linear summation hypothesis is also supported by changes in the a-wave sensitivity (σ) over time. Shady et al. (34) concluded from human retinitis pigmentosa patients that rod degeneration resulted in progressive reduction of transduction amplification, a quantity closely related to the sensitivity parameter reported in the present study. We show that rod sensitivity in untreated Bbs4-null eyes decreases monotonically from 4 to 16 wk. In treated eyes, however, sensitivity decreases from 4 to 12 wk, followed by a notable increase from 12 to 16 wk. It is of particular interest that the sensitivity of treated eyes at 16 wk is very similar to the sensitivity of age-matched wild-type eyes. In the context of our linear summation model, these data suggest that the jump in sensitivity between 12 and 16 wk represents the gradual disappearance of the 95% unhealthy rod signal and an emerging dominance of the 5% healthy (treated) rods with relatively normal rod phototransduction parameters. The preservation of visually evoked behavioral responses from four of nine treated eyes further confirms that a population of functional rods survived as a result of our treatment. The fact that any treated mice exhibited these behavioral responses was somewhat surprising, considering the limited area of rescued photoreceptors. Further studies are currently under way to correlate anatomical, electrophysiological, and behavioral rescue at the level of individual eyes.

From an experimental perspective our mean coverage factor of 5% per viral injection was more than sufficient to observe anatomical and functional benefit, thus validating the efficacy of our viral construct. In large animal models, retinal coverage of 20–30% is readily achievable by subretinal injection (35, 36). If translated to a human subject, 30% coverage containing the macular region would have the potential to preserve nearly normal visual acuity. An important prerequisite for gene therapy, however, is a timely genetic diagnosis before extensive photoreceptor death has occurred. Clinical studies suggest that among BBS1 patients, only half will have visual acuity of 20/60 or better at the time of diagnosis; the situation seems to be worse for those with BBS10 (37, 38). This underscores the importance of BBS awareness in the ophthalmology and broader medical communities. Our results are an encouraging first step toward preserving vision in BBS patients, but ultimate success of such a therapy will require a collaborative effort between translational researchers and vigilant clinicians.

Materials and Methods

Viral Constructs.

Full-length cDNA for the mouse Bbs4 gene (accession no. BC145771) was obtained from Open Biosystems. An HA epitope tag (YPYDVPDYA) was added to the N terminus of Bbs4, followed by ligation of HA-Bbs4 into the sc-MOPS500-hGFP plasmid in place of hGFP. The plasmid transfection method using HEK293 cells as previously described (39) was used to produce and purify scAAV2/5 vectors carrying either Bbs4 or GFP. Viral titers were determined by real-time PCR as previously described (40) and were 6.82 × 10¹² VG/mL for scAAV5-mOP-BBS4 and 3.64 × 10¹² VG/mL for scAAV5-mOP-GFP.

Animals and Subretinal Injections.

Bbs4-null mice were generated as previously described (8). All animal procedures were approved by the Institutional Animal Care and Use Committee at Baylor College of Medicine. Details of the transscleral subretinal injection technique are given in SI Materials and Methods.

Histology and Immunoblotting.

Methods were performed as previously described (11) with minor modifications. Ten-micrometer frozen sections were stained with 1:1,000 anti-rhodopsin mAb (RET-P1 clone; Neomarkers) and DAPI. For HA-BBS4 identification, a 1:1,000 dilution of anti-HA mAb was used (16B12 clone; Covance). For light microscopy, 1-μm Epon sections were stained with methylene blue and basic fuchsin. To create heat map representations, digitized images were evenly divided into 32 regions per hemiretina. The thickness value for each region was determined by averaging three counts of nuclei found in an ONL column.

Functional Vision Testing.

Mice were anesthetized by i.p. injection of 51 mg/kg ketamine, 10 mg/kg xylazine, and 0.86 mg/kg acepromazine, and ERGs were recorded as previously described (11). A-wave leading edges were modeled using the method described by Cideciyan and Jacobson (24). Briefly, dark-adapted ERGs were recorded in response to two bright flashes (2.10 and 3.45 log scot*cd*s/m²), and the following equation was fit to both curves simultaneously using least-squares minimization:

where I is the flash intensity; R_max, the maximum response amplitude; σ, sensitivity; and t, time after stimulus onset. More information about Eq. 1, curve fitting software, and OKR testing can be found in SI Materials and Methods.