Cone Phosphodiesterase-6α′ Restores Rod Function and Confers Distinct Physiological Properties in the Rod Phosphodiesterase-6β-Deficient rd10 Mouse

Phosphodiesterase-6 (PDE6) is the key effector enzyme of the vertebrate phototransduction pathway in rods and cones. Rod PDE6 catalytic core is composed of two distinct subunits, PDE6α and PDE6β, whereas two identical PDE6α′ subunits form the cone PDE6 catalytic core. It is not known whether this difference in PDE6 catalytic subunit identity contributes to the functional differences between rods and cones. To address this question, we expressed cone PDE6α′ in the photoreceptor cells of the retinal degeneration 10 (rd10) mouse that carries a mutation in rod PDEβ subunit. We show that adeno-associated virus-mediated subretinal delivery of PDE6α′ rescues rod electroretinogram responses and preserves retinal structure, indicating that cone PDE6α′ can couple effectively to the rod phototransduction pathway. We also show that restoration of light sensitivity in rd10 rods is attributable to assembly of PDE6α′ with rod PDE6γ. Single-cell recordings revealed that, surprisingly, rods expressing cone PDE6α′ are twofold more sensitive to light than wild-type rods, most likely because of the slower shutoff of their light responses. Unlike in wild-type rods, the response kinetics in PDE6α′-treated rd10 rods accelerated with increasing flash intensity, indicating a possible direct feedback modulation of cone PDE6α′ activity. Together, these results demonstrate that cone PDE6α′ can functionally substitute for rod PDEαβ in vivo, conferring treated rods with distinct physiological properties.


Introduction
Rod and cone photoreceptor cells share a similar phototransduction pathway but exhibit strikingly different physiological properties. Rods, responsible for scotopic vision, are highly light sensitive. Cones, responsible for photopic vision, are intrinsically less sensitive, have faster response kinetics, and adapt to a wider range of light intensities (Pugh and Cobbs, 1986;Fu and Yau, 2007). One of the key unresolved questions is how the physiolog-ical differences between rods and cones can be correlated with the distinctive properties of their phototransduction proteins. Previous studies have shown that the lower thermal stability of cone pigments is likely to contribute to the lower sensitivity of cones but, once activated, rod and cone pigments can couple equally efficiently to rod or cone transducin (Kefalov et al., 2003(Kefalov et al., , 2005Shi et al., 2005Shi et al., , 2007Fu et al., 2008). Thus, consistent with our previous research (Deng et al., 2009) and other studies (Ma et al., 2001), the signaling properties of rod and cone transducin ␣-subunit do not contribute to the difference in light sensitivity between rods and cones (but see Chen et al., 2010). As a result, the expression levels and molecular properties of phototransduction components downstream of transducin are likely to play an important role in defining the distinctive physiological properties of rods and cones.
The photoreceptor cyclic nucleotide phosphodiesterase-6 (PDE6) plays an essential role in phototransduction by regulating the cGMP levels in rods and cones (Fu and Yau, 2007). The most obvious distinction between rod and cone PDE6 is that rod PDE6 is composed of two distinct catalytic subunits ␣, ␤ (PDE6〈, PDE6〉) and two inhibitory subunits ␥ (PDE6G), whereas cone PDE6 is composed of two identical catalytic subunits ␣Ј (PDE6C) plus two cone-specific inhibitory subunits ␥Ј (PDE6H) (Gillespie and Beavo, 1988;Hamilton and Hurley, 1990;Li et al., 1990). Each of the catalytic subunits of PDE6 consists of two N-terminal regulatory cGMP binding GAF (for cGMP-specific phosphodiesterases, adenylyl cyclases, and FhlA) domains (GAFa and GAFb) and a catalytic domain located in the C-terminal region. The catalytic domains are highly conserved among rod and cone PDE6 subunits and exhibit equivalent enzymatic activities (Mou and Cote, 2001;Muradov et al., 2010). Among the GAF domains, rod PDE6 GAF displays a higher affinity toward cGMP than cone PDE6 (Gillespie and Beavo, 1989). It has been suggested that the differences in GAF binding affinities toward cGMP and PDE6␥ might contribute to the higher efficiency of cone PDE6 activation by transducin ␣-subunit (Muradov et al., 2010).
In this study, we tested whether PDE6 catalytic subunit identity contributes to the functional differences between rods and cones by expressing cone PDE6␣Ј in the retinal degeneration 10 (rd10) photoreceptor cells, which carry a mutation in the ␤-subunit of rod PDE6 (Chang et al., 2007). We show that cone PDE6␣Ј can restore rd10 rod function by assembling with rod PDE6␥. Furthermore, it confers rods with distinct physiological properties.

Materials and Methods
Animals. rd10 mice and wild-type (WT) C57BL/6J controls were obtained from The Jackson Laboratory. The mice of either sex were bred and maintained in the University of Florida Health Science Center Animal Care Services Facilities in a continuously dark room, except for husbandry at ϳ400 lux illuminance. All experiments were approved by the local Institutional Animal Care and Use Committees at the University of Florida and Washington University and conducted in accordance with the Association for Research in Vision and Ophthalmology Statement for the Use of Animals in Ophthalmic and Vision Research and National Institutes of Health regulations.
Subretinal injections. Postnatal day 14 (P14) rd10 pups raised in the dark were brought to a normal illuminated room for injection and then returned back to dark. A total volume of 1 l of AAV8 Y733F-smCBA-PDE6␣Ј vector (4.25 ϫ 10 12 vector genomes/ml) was injected subretinally into the left eyes, and the right contralateral eyes served as untreated controls. Subretinal injections were performed as described previously (Pang et al., 2006(Pang et al., , 2008. Briefly, a 33 gauge blunt needle mounted on a 5 l Hamilton syringe was introduced through the corneal opening made by 30 gauge needle, and injections were visu-alized by fluorescein-positive subretinal bleb. One percent atropine eye drops and neomycin/polymyxin B/dexamethasone ophthalmic ointment were given after injection. Electroretinogram analyses. At 5 weeks after injection, rod-and conemediated electroretinograms (ERGs) were recorded separately using a UTAS Visual Diagnostic System equipped with Big Shot Ganzfeld (LKC Technologies) according to protocols described previously with minor modifications (Pang et al., 2010). Scotopic rod recordings were performed with three increasing light intensities at Ϫ1.6, Ϫ0.6, and 0.4 log cds/m 2 . Ten responses were recorded and averaged at each light intensity. Photopic cone recording were taken after mice were adapted to a white background light of 30 cds/m 2 for 5 min. Recordings were performed with four flash intensities at 0.1, 0.7, 1.0, and 1.4 log cds/m 2 in the presence of 30 cds/m 2 background light. Fifty responses were recorded and averaged at each intensity. Scotopic and photopic b-wave amplitudes from untreated, treated rd10, and WT controls at each intensity were averaged and used to generate an SD. The differences between recordings from untreated and treated eyes were analyzed by the paired t test.
Morphology and immunohistochemistry. Treated rd10 mice were killed and enucleated 2 d after ERG recordings for morphological and immunohistochemical analysis. The eyecups were fixed in a mixture of 4.0% paraformaldehyde and 0.5% glutaraldehyde for 3 h at room temperature and then paraffin embedded and sectioned at 4 m through the optic nerve for hematoxylin and eosin (H&E) staining. Retinal sections for immunohistochemistry were prepared according to previously described methods (Deng et al., 2009(Deng et al., , 2012. Briefly, eyes were fixed in 4% paraformaldehyde. Cornea, lens, and vitreous were removed from each eye without disturbing the retina. The remaining eyecup was rinsed with PBS and then cryoprotected by placing it in 30% sucrose in PBS for 4 h at 4°. Eyecups were then embedded in cryostat compound (Tissue TEK OCT; Sakura Finetek) and frozen at Ϫ80°C. Retinal tissue cryosections were sectioned at 12 m thickness, rinsed in PBS, and blocked in 2% normal goat serum and 0.3% Triton X-100 in 1% BSA in PBS for 1 h at room temperature. Anti-PDE6␣Ј (3184P) (Kirschman et al., 2010), rhodopsin, or red/green-cone opsin (Millipore Bioscience Research Reagents) antibodies (all 1:1000 dilutions) were diluted in 0.1% Triton X-100 and 1% BSA in PBS and incubated with sections overnight at 4°. The sections were then washed three times with PBS, incubated with IgG secondary antibody tagged with Alexa Fluor-594 (Invitrogen) at 1:500 dilution and lectin peanut agglutinin (PNA) conjugated to Alexa Fluor-488 (Invitrogen) at 1:200 dilution in PBS at room temperature for 1 h, and washed with PBS. Sections were mounted with Vectashield Mounting Medium for Fluorescence (H-1000; Vector Laboratories) and coverslipped. Sections were analyzed with a Carl Zeiss CD25 microscope fitted with Axiovision release 4.6 software.
Western blot analyses. Untreated, AAV8 Y733F-smCBA-PDE6␣Јtreated rd10 and WT eyes (five eyes each) were carefully dissected, and the eyecups were pooled and homogenized by sonication in a buffer containing 0.23 M sucrose, 5 mmol/L Tris-HCl, pH 7.5, and protease inhibitors (Roche Complete). After centrifugation, aliquots of the ex- Figure 1. Detection of PDE6␣Ј expression after delivery of AAV8 Y733F-smCBA-PDE6␣Ј in rd10 retinas. A, immunofluorescence of PDE6␣Ј (labeled as red) expression in both rods (arrows) and cones (arrowheads) in injected retinas. PDE6␣Ј expression can only be detected in cones in WT control. Only spotty staining can be detected in untreated eyes because of significant retina degeneration. Cones were labeled by PNA (green). Scale bar, 20 m. B, Western blot analysis of untreated, treated rd10, and WT control retinas. Injected rd10 retinas showed robust protein expression resulting from the AAV-mediated PDE6␣Ј expression in both rods and cones as driven by the smCBA promoter. In contrast, the WT control displays a weaker immunoreactive band as a result of the presence of PDE6␣Ј in cones only. INL, Inner nuclear layer; ONL, outer nuclear layer.
tracts containing equal amounts of protein (50 g) were analyzed by electrophoresis on 10% polyacrylamide-SDS gels, transferred, and probed with a PDE6␣Ј antibody (Kolandaivelu et al., 2011). An antibody against ␣-tubulin (rabbit polyclonal ab4074; Abcam) was used as an internal control. Visualization of specific bands was performed using the Odyssey Infrared Fluorescence Imaging System (Odyssey; Li-Cor).
Immunoprecipitation. Frozen retinal eyecups from untreated, AAV8 Y733F-smCBA-PDE6␣Ј-treated rd10 and WT (five each) were homogenized in 400 l of immunoprecipitation (IP) buffer (in mM: 10 Tris-HCl, pH 7.5, 100 KCl, 20 NaCl, and 1 MgCl 2 ) containing protease and phosphatase inhibitors and 10 mM iodoacetamide using a pellet pestle (VWR) in a 1.5 ml Eppendorf tube on ice (15 s for three times). After homogenization, Triton X-100 was added to a final concentration of 1% (500 l total volume). The homogenized retinal extracts were precleared by addition of 10 l of immunopure immobilized protein A plus beads (Thermo Fisher Scientific) by incubating at 4°C for 1 h. Supernatants were collected by centrifuging at 10,000 ϫ g (Eppendorf 5424) for 5 min at 4°C. IP was performed with supernatants (400 l) using mouse monoclonal ROS-1 antibodies. We used 1.5 g of ROS-1 antibody for each pull-down experiment. Bound proteins were eluted by boiling with 50 l of 1ϫ Laemmli's sample buffer and separated by 4 -20% SDS-polyacrylamide gel (Bio-Rad) and transferred to Immuno-Blot LF PVDF membrane (Bio-Rad). Immunoblot analyses were performed with individual rod PDE6␣, PDE6␤, and PDE6␥ subunits and cone PDE6␣Ј (3184p)-specific primary antibodies according to our previously published method (Kolandaivelu et al., 2011).
Single-cell recordings. Mice kept in darkness for at least 12 h were killed by CO 2 , and the eyes were removed under dim red light. Under infrared light, the retina was cut into small pieces and then finely chopped. Isolated pieces of retina were stored in Locke's solution at 4°C until use. The perfusion Locke's solution (in mM: 112 NaCl, 3.6 KCl, 2.4 MgCl 2 , 1.2 CaCl 2 , 10 HEPES, 20 NaHCO 3 , 3 Na 2 -succinate, 0.5 Naglutamate, and 10 glucose, pH 7.4) was equilibrated with 95% O 2 /5% CO 2 bubbling and heated to 34 -37°C. Glass capillaries were pulled and heat polished to fit the rod outer segment (ROS) diameter and then filled with electrode solution containing the following (in mM): 140 NaCl, 3.6 KCl, 2.4 MgCl 2 , 1.2 CaCl 2 , 3 HEPES, and 10 glucose, pH 7.4. A rod photoreceptor was drawn into the electrode to record the inward current of the outer segment (OS). The dark current was amplified by a current-to-voltage converter (Axopatch 200B; Molecular Devices), low-pass filtered by an eight-pole Bessel filter with a cutoff frequency of 30 Hz (Krohn-Hite), digitized at 1 kHz, and recorded with pClamp 8.2 software (Molecular Devices). Ten-millisecond flashes were delivered from a calibrated light source via computer-controlled shutters. Light intensity and wavelength were changed with neutral density and interference ( max ϭ 500 nm) filters (Edmund Optics). Intensity-response data were fit by the Hill equation: R R max ϭ I n I n ϩ I o n , where R is the transient-peak amplitude of response, R max is maximal response amplitude, I is flash intensity, and I o is flash intensity to generate half-maximal response.
Visual acuity test. Scotopic visual acuity of 2-month-old mice was determined using a two-alternative forced-choice protocol (Umino et al., Figure 2. ERG responses, retinal morphology, immunohistochemistry, and scotopic visual acuity of rd10 mice after AAV8 Y733F-smCBA-PDE6␣Ј delivery at 5 weeks after injection (5-wk-pj). A, B, Representative examples of dark-adapted ERG traces (A) and light-adapted ERG traces (B) from an rd10 mouse at 5 weeks after injection. C, Dark-adapted ERG was partially restored in injected rd10 eyes. Statistical analysis demonstrated a significant difference between uninjected and fellow vector-treated eyes for dark-adapted b-waves at Ϫ1.6, Ϫ0.6, and 0.4 log cds/m 2 (*p Ͻ 0.01). D, Light-adapted ERG responses were improved in treated rd10 eyes compared with their contralateral controls as a result of rod function rescue and rod cell survival (*p Ͻ 0.02). Error bars are mean Ϯ SD. E, Comparison of ERG responses between PDE6␤-treated and PDE6␣Ј-treated rd10 eyes 5 weeks after injection. There were no significant differences in dark-adapted b-wave amplitudes at three light intensities tested between PDE6␤-treated and PDE6␣Ј-treated rd10 eyes (all p Ͼ 0.1). Bar graph representing the mean Ϯ SEM. b-Wave amplitudes at indicated flash intensities were compared by repeated-measures ANOVA, with the Bonferroni's post hoc test for ANOVA ( p Ͻ 0.1) used to compare means at individual flash intensities. F, Restoration of scotopic visual acuity in PDE6␣Ј-treated rd10 mice 5 weeks after injection. Data were derived from mouse optomotor responses to rotating gratings under background monitor luminance of Ϫ4.45 log cd/m 2 . Bar graphs are mean Ϯ SEM. 2008). The Optomotry system (Cerebral Mechanics) consisted of a square array of four computer monitors with a pedestal in the center where the mouse was placed. An infraredsensitive television camera and a round array of six infrared light-emitting diodes mounted above the animal were used to observe the mouse but not the monitors. Using a staircase paradigm, rotating sine-wave vertical gratings were applied on the monitors where they formed a virtual cylinder around the animal (Prusky et al., 2004). The mice responded to the stimuli by reflexively rotating their head in either clockwise or counterclockwise direction. Optomotor responses were measured under monitor background illumination of Ϫ4.45 log cd/m 2 , which was set by neutral density film filters.
Visual acuity was defined as the threshold for spatial frequency (F s ) of gratings with 100% contrast and measured at the speed (S p ) of 6.0°/s. F s was gradually altered by the computer protocol until its combined threshold for both stimuli directions was determined. Temporal frequency (F t ) was automatically adjusted by the computer software, based on the following equation: F t ϭ S p ϫ F s . Data were analyzed using independent twotailed Student's t test, with an accepted significance level of p Ͻ 0.05.

Results
Expression of cone PDE6␣ in rd10 mouse retinas An AAV8 Y733F capsid-tyrosine mutant vector containing the mouse PDE6␣Ј cDNA driven by a ubiquitous smCBA promoter was delivered subretinally to one eye of rd10 mice at P14, whereas the contralateral eyes remained uninjected and served as controls. PDE6␣Ј expression in treated retinas was analyzed by immunostaining (Fig. 1A) and Western blot analysis (Fig. 1B) at 5 weeks after injection. PDE6␣Ј expression was found in both rods and cones of treated rd10 mice after immunostaining with a cone-specific PDE6␣Ј antibody, whereas it was found exclusively in the cones of WT control mouse retinas based on colocalization with a cone OS sheath-specific PNA marker. Photoreceptor cells in untreated retinas were significantly degenerated at this age, and only a weak spotty staining was detected for residual cones (Fig. 1A). Low levels of expression were also observed in the inner retina most likely as a result of nonspecificity of the PDE6␣Ј antibody because similar labeling was observed in untreated and treated rd10, as well as in the WT sections (Fig.  1A). Western blot analysis using the same antibody (Fig. 1B) detected abundant PDE6␣Ј expression in injected rd10 retinas compared with WT controls in which PDE6␣Ј was expressed predominantly in cones. This result provides evidence that PDE6␣Ј is robustly expressed in rd10 rods after AAV8 treatment because rods comprise the majority (97%) of photoreceptor cells in the mouse retina. PDE6␣Ј expression was reduced to almost undetectable levels in retina from uninjected rd10 animals (Fig.  1B), presumably because of the degeneration of cones caused by the loss of PDE6␤-deficient rods.

Functional and structural retinal preservation in PDE6␣-treated rd10 eyes
To determine whether exogenously expressed cone PDE6␣Ј can rescue rod function in rd10 mice, full-field scotopic and photopic ERG responses were recorded from uninjected rd10 mice, injected rd10 mice 5 weeks after injection, and age-matched WT controls. Rod-mediated ERG responses were undetectable in rd10 mice at this age (7 weeks old), whereas vector delivery of PDE6␣Ј to rd10 rods led to significant restoration of rod-driven ERG responses (Fig. 2 A, C). The average rod-mediated b-wave amplitude at a flash intensity of Ϫ1.6 log cds/m 2 in treated eyes was 109 Ϯ 39 V (mean Ϯ SD), whereas it was undetectable in contralateral untreated eyes (n ϭ 3, p Ͻ 0.01). The treated eye rod ERG b-wave amplitude was ϳ35% of the WT level. Conemediated ERG amplitudes in injected eyes also showed some improvement compared with uninjected controls (Fig. 2 B, D), presumably as a result of better preservation of the cones after restoration of rod function and rod survival (Fig. 3C). The average cone b-wave amplitude was 44 Ϯ 8 V (mean Ϯ SD) in injected eyes versus 27 Ϯ 10 V in contralateral untreated eyes at 1.4 log cds/m 2 (n ϭ 3, p Ͻ 0.02). We also recorded ERG responses from some rd10 mice injected with vector expressing PDE6␤ and observed no significant differences between PDE6␤ and PDE6␣Ј treatments (Fig. 1E), suggesting that therapy in the rd10 mouse was equivalent whether we used the heterologous rod subunit or the homologous cone subunit. Finally, the scotopic visual acuity of PDE6␣Ј-treated rd10 mice improved significantly (0.349 Ϯ 0.088, n ϭ 8) over that of untreated controls (0.069 Ϯ 0.024, n ϭ 5) (Fig. 2F ), although rod visual performance still remained subpar compared with WT mice (0.776 Ϯ 0.072).
Three rd10 mice exhibiting significant ERG rescue were killed 2 d after the recordings, and retinal morphology was analyzed by H&E staining (Fig. 3A). Only one layer of photoreceptor nuclei remained in the outer nuclear layer of untreated rd10 retinas with no evident outer or inner segments. In contrast, retinal structure was partially preserved in injected eyes, with five to seven layers of nuclei remaining compared with 12 layers in WT controls. Additionally, treated retinas retained ϳ20 -50% of the normal OS length. Uninjected, injected, and WT retinal sections were also stained with rhodopsin antibody (Fig. 3B) and cone opsin-specific (Fig. 3C) antibody to further confirm the morphological rescue. Expression of both rhodopsin and cone opsin was evident and much more abundant in treated rd10 eyes compared with their spotty staining observed in uninjected controls.
We also recorded ERG responses from rd10 mice at 5 months after treatment (Fig. 4 A, B). The average rod-driven ERG b-wave amplitude (Fig. 4A) at a flash intensity of Ϫ1.6 log cds/m 2 in treated eyes was 81 Ϯ 15 V (mean Ϯ SD), and it was significantly higher than the undetectable ERGs in uninjected eyes (n ϭ 3, p Ͻ 0.005). The cone-mediated ERG responses (Fig.  4B) were also undetectable in untreated eyes at this age, whereas the average b-wave amplitude in treated eyes at 1.0 log cds/m 2 was 51 Ϯ 13 V (mean Ϯ SD) (n ϭ 3, p Ͻ 0.03). PDE6␣Ј expression was still evident in both rods and cones in treated eyes at 5 months after injection (Fig. 4C). Transducin was strongly expressed in the treated eyes as determined by immunostaining, whereas it was undetectable in untreated eyes at 5 months after injection (Fig. 4D). Thus, PDE6␣Ј-mediated rescue of rd10 rod structure and function persisted even months after the AAV injection.

Cone PDE6␣ binds to rod PDE6␥ to restore rod function
Restoration of the light-dependent rod response in rd10 animals suggested that PDE6␣Ј expressed by AAV is capable of forming a functional complex with rod PDE6␥. Before testing this idea, we investigated the levels of various subunits of rod PDE6 holoenzyme. Uninjected rd10 animals with advanced stage of rod degeneration lacked all three subunits of rod PDE6 (Fig.  5A). Despite preservation of five to eight layers of photoreceptor cells in injected animals, we observed destabilization of both rod . ERG rescue and immunohistochemistry of rd10 mice 5 months after injection (5 mon pj). A, Dark-adapted ERG from uninjected, contralateral injected rd10, and WT control eyes at light intensity of 0.4 log cds/m 2 (*p Ͻ 0.005). B, Light-adapted ERG from the same mice at light intensity of 1.0 log cds/m 2 (*p Ͻ 0.03). Error bars are mean Ϯ SD. C, PDE6␣Ј expression 5 months after delivering AAV8 Y733F-smCBA-PDE6␣Ј in rd10 retinas. Immunofluorescence of PDE6␣Ј (labeled as red) can be detected in both rods (arrows) and cones (arrow heads) in injected rd10 retinas. Onlyspottystainingwasfoundinuntreatedeyeasaresultofsignificant retina degeneration. Cones were labeled by PNA (green). D, Immunostaining of uninjected, injected rd10, and WT retinas with transducin antibody (red). Green, Cone OS sheath-specific PNA. Scale bar, 20 m. INL, Inner nuclear layer; ONL, outer nuclear layer. PDE6 catalytic subunits (Fig. 5A). Compared with age-matched WT controls, minor amounts of PDE6␣ or PDE6␤ were expressed in total retinal extracts from injected animals. In contrast, there was a dramatic upregulation in PDE6␣Ј expression in these retinas. Although lower than in WT controls, we observed robust expression of rod PDE6␥ in treated animals likely as a result of complex formation with the virally introduced PDE6␣Ј. To directly test whether the formation of a complex between cone PDE6␣Ј and rod PDE6␥ existed, we performed IP with a monoclonal antibody, ROS-1, that exclusively recognizes assembled and functional PDE6 complex from both rods and cones (Kolandaivelu et al., 2009(Kolandaivelu et al., , 2011. As expected, we observed assembled rod and cone PDE6 subunits in ROS-1 pull-downs from WT controls (Fig. 5B). Assembled PDE6␣Ј was also observed from surviving cones in uninjected animals. In treated rd10 animals, we detected a complex of PDE6␣Ј and PDE6␥ indicating that the restoration of light sensitivity in rd10 rods is attributable to the function of cone PDE6␣Ј assembled with rod PDE6␥ (Fig. 5B).

Single-cell recordings from injected rd10 rods
To gain additional insight into the light responses generated by rods expressing cone PDE6␣Ј, we performed single-cell recordings from injected rd10 rods and WT controls. For comparison, we also obtained responses from rd10 rods treated with vector expressing rod PDE6␤. Although all retinas of PDE6␣Ј-and PDE6␤-treated rd10 mice were still subject to some level of degeneration, we were able to find areas with healthy ROS in portions of the retina in which AAV vectors seemed to have been successfully delivered. We obtained photoresponses from 14 PDE6␣Ј-treated rods (from two animals) and 22 PDE6␤-treated rods (from three animals). No significant differences were found between the photoresponses of WT and PDE6␤-treated rd10 rods (compare with Fig. 6 A, C, Table 1), indicating that the exogenous expression of PDE6␤ by AAV infection into rods of rd10 mice successfully rescued rod physiological functions. The dark currents, measured from saturated photoresponses, were comparable among WT, PDE6␤treated, and PDE6␣Ј-treated rods (Table  1). Thus, PDE6␣Ј ectopically expressed in rod photoreceptors could form a functional complex with rod PDE6␥ and maintain normal spontaneous activity and dark cGMP levels. However, we also observed several unusual features in the responses of PDE6␣Ј-treated rods. First, PDE6␣Ј-treated rods had higher sensitivity and produced larger single-photon responses than WT rods (Fig. 7A, Table 1). Consistent with this result, intensity-response relationships of dark-adapted rods showed that the flash intensity required for half-saturating response of the PDE6␣Ј-treated rods was approximately twofold lower than that of WT rods (Fig. 6D, inset, Table 1). Second, the time-topeak and integration time of dim-flash responses were substantially prolonged in PDE6␣Ј-treated rd10 rods (Fig. 7A, Table 1). The rising phase of dim-flash response was similar among WT, PDE6␤-treated, and PDE6␣Ј-treated rd10 rods (Fig. 7A), which indicates that light-induced cGMP hydrolysis activated by PDE occurred at comparable rates. However, the response recovery phase was substantially delayed in PDE6␣Ј-treated rods, indicating that the deactivation of cone PDE6␣Ј was less effective. This slower than normal response shutoff could potentially explain the increased sensitivity and single-photon response amplitude in PDE6␣Ј-treated rd10 rods. Third, unlike in WT and PDE6␤treated rd10 rods, the response kinetics in PDE6␣Ј-treated rods accelerated substantially with increasing flash strength (Fig. 6B), and the resulting intensity-response curves appeared shallower than these of WT and control PDE6␤-treated rd10 rods. Both of Figure 5. Cone PDE6␣Ј subunit expressed in rods associates with rod PDE6␥ inhibitory subunit. A, Immunoblot analysis with indicated antibodies show the total levels of PDE6 subunits in retinal extracts from WT control, rd10 uninjected (Rd10 Un Inj), and rd10-injected (Rd10 Inj) animals. B, IP of assembled PDE6 subunits with ROS-1 antibody from retinal extracts of WT control, uninjected rd10, and injected rd10 animals. After ROS-1 IP, immunoblots were probed with indicated antibodies. Compared with total extracts (A), IP samples (B) were 10 times more concentrated.

Discussion
In this study, we expressed cone PDE6␣Ј subunit exogenously in the retinas of rd10 mice to investigate its biochemical and light signaling properties in a rod cell environment. Our results demonstrate that cone PDE6␣Ј can functionally substitute for rod PDE6␣␤ to mediate light signaling in rods, as shown by full-field ERG analysis, behavioral experiments, and single-cell recordings. Rod PDE6 catalytic subunits are destabilized in PDE6␣Ј-injected retinas despite the functional and morphological rescue of rods, and restoration of rod light sensitivity is mediated by assembly of cone PDE6␣Ј with rod PDE6␥. Rods with cone PDE6␣Ј are approximately two times more sensitive to light than WT cells, and this difference is likely the result of the slower shutoff of their light responses. The slower rate of deactivation indicates that inhibition by rod PDE6␥ or the hydrolysis of T␣*-GTP on PDE6␣Јtransducin ␣ complex by regulator of G-protein signaling-9 (RGS9) is less efficient than normal.
We demonstrated previously that AAV-mediated subretinal delivery of rod PDE6␤ transgene confers long-term rescue of visual function and morphological preservation of the rd10 retinas (Pang et al., 2011). In the present study, AAV8 Y733F cone PDE6␣Ј-treated rd10 retinas showed comparable levels of rescue in gross morphology, amplitudes of rod-driven full-field ERG signals, and the maximal amplitude of single-photon responses, clearly demonstrating that cone PDE6␣Ј can couple effectively to the rod visual signaling pathway in response to light. Our work complements the previous finding of the ability of rod PDE6 to substitute for cone PDE6 to mediate visual signaling in Nrl Ϫ/Ϫ cpfl1 mouse model (Kolandaivelu et al., 2011). Although the PDE6␣Ј transgene was driven by an smCBA ubiquitous promoter, we detected most PDE6␣Ј in photoreceptor cells in which it is normally expressed. We observed similar phenomenon of endogenous cell-specific expression in the cases of RPE65 (Pang et al., 2006), transducin (Deng et al., 2009), and PDE6␤ (Pang et al., 2011 proteins when using the ubiquitous CBA promoter. The significant scatter in the sensitivity of the AAV-treated rods (Fig. 6D, inset) most likely reflects the variability of AAVmediated PDE6 expression.
We also showed that vector-expressed cone PDE6␣Ј localized properly in ROSs. It has been suggested that binding of PDE6 to OS membranes is essential for rapid activation by transducin (Liebman et al., 1987). Cone PDE6␣Ј ectopically expressed in rods of Xenopus laevis was shown to colocalize with endogenous PDE6 on disc rim regions in RODs (Muradov et al., 2009). The similar rising phases of dimflash responses between PDE6␣Ј-treated rd10 and WT rods as shown with single-cell recordings suggest that the activation rate of the catalytic cone PDE6 subunit in the rod environment is comparable with that of rod PDE6 and that exogenously expressed cone PDE6␣Ј is appropriately localized to ROS disk membranes. We further show that the restoration of light sensitivity in rd10 rods is attributable to the assembly of cone PDE6␣Ј with rod PDE6␥. Robust expression of PDE6␥ was observed in injected retinas, most likely as a result of complex formation with the virus-introduced PDE6␣Ј. The presence of cone PDE6␣Ј or restoration of rod cells did not help in preserving rod PDE␣, which was degraded without its PDE␤ partner. It appears that, regardless of cell type, cone PDE6␣Ј forms homodimers to be functional in vivo. The same holds true for rod PDE6 in the sense that PDE6␣ and PDE6␤ are obligated to function as heterodimers (Kolandaivelu et al., 2011). Apparently, the state of association is determined by the properties of the subunits rather than the photoreceptor cell type. All families of vertebrate cyclic nucleotide phosphodiesterases function as homodimers, and, although the reason behind the heterodimerization of rod PDE6 is not known, it presumably exists as a mechanism to control the amount of functional PDE6 enzyme present in rods (Kolandaivelu et al., 2011).
The equivalent rate of activation between WT rods and rd10 rods expressing cone PDE6␣Ј or PDE6␤ suggests that activated T␣*-GTP can effectively release the inhibitory constrain of rod PDE6␥ from cone PDE6␣Ј catalytic domain. However, the slower shutoff of PDE6␣Ј-treated rods indicates that deactivation of cone PDE6␣Ј by inhibitory rod PDE6␥ or the hydrolysis of ␣-subunit-bound GTP on PDE6␣Ј-transducin complex is less efficient. The GAFa domains also bind to the inhibitory ␥ subunits and play a role in the dimerization of the PDE6 catalytic subunits (Muradov et al., 2004). The strength of interaction between PDE6␥ and GAF domains is modulated by cGMP binding to GAF domain. cGMP binding induces an allosterical GAF conformational change and enhances PDE6␥ binding affinity, and, in a reciprocal manner, binding of ␥-subunit to PDE6 catalytic dimer increases the binding affinity of cGMP to the GAF domains (Yamazaki et al., 1982;Cote et al., 1994). Accordingly, dissociation of either one weakens the binding of the other. Based on a structural study of PDE6␥ (Barren et al., 2009), it has been suggested that the interaction between T␣*-GTP and PDE6␥ induces a hinge-like movement of the last 10 residues away from  Mean Ϯ SEM integration time was calculated by dividing the area of dim-flash response by its amplitude. Amplitude of single-photon response was estimated from variance/mean ratio of dim-flash responses evoked by consecutive identical stimuli. One-way ANOVA with the post hoc Tukey's HSD test determined significant differences (*p Ͻ 0.05 vs WT and PDE6␤). No significant difference was found between WT and PDE6␤ parameters.
the enzyme active site without the T␣*-GTP/PDE6␥ complex completely disassociating from the PDE6 holoenzyme. The inactivation of T␣*-GTP by its intrinsic GTPase activity is the ratelimiting step to restore the photoresponse to a dark-adapted state and its regulator RGS9-1 associates with PDE6␥ to accelerate the GTPase activity of T␣*-GTP (Arshavsky and Burns, 2012). The multiple interactions of PDE6␥ with PDE6␣␤, T␣*-GTP, and RGS9-1 complex are likely to occur in a precisely controlled temporal sequence that coordinates the activation and deactivation of PDE6 (Zhang et al., 2012). The major sequence difference between cone and rod PDE6 resides in the GAF domains, with cone PDE6 displaying a lower affinity toward cGMP. The relative affinity of rod PDE6␥ binding to T␣*-GTP versus the PDE6 catalytic subunits may be defined by the state of cGMP occupancy on the GAF domains of PDE6␣Ј. Likewise, the affinity of rod PDE6␥ for vector-expressed cone PDE6␣Ј may be lower than that for the rod PDE6␣␤. These differences may contribute to the slower inactivation of the cone PDE6␣Ј expressed in rods. It would be interesting to study the effects of replacing rod PDE6␥ with cone PDE6␥Ј or the entire rod PDE holoenzyme with cone PDE6, because PDE6␥ critically regulates phototransduction through on and off interactions with PDE6␣␤, T␣*-GTP, and RGS9-1. Overall, it is difficult from our results to gain a clear view of the role of PDE in the differences in sensitivity or kinetics between rods and cones. Interestingly, although rods and cones share the same GAP complex, cones express RGS9 at higher levels (Zhang et al., 2003). This observation, together with the slow inactivation of cone PDE6␣Ј in rods observed by us indicate that, perhaps, the timely T␣*-GTP/PDE complex inactivation in cones requires higher GAP activity than in rods. Finally, our single-cell recordings from cone PDE6␣Ј-treated rd10 rods demonstrated an unusual response acceleration with increasing flash strength. This, together with a shallower intensity-response curve for these rods indicates a potential acceleration of cone PDE inactivation with increased phototransduction activation. A direct modulation of PDE activity was recently suggested as an additional adaptation mechanism in mouse rods , although it has not been directly demonstrated. Notably, however, we did not observe substantial response acceleration in WT or PDE6␤-treated rd10 rods, suggesting that this is a cone PDE-specific phenomenon. Such a negative feedback modulation of cone PDE6 is an exciting novel concept and represents a potential mechanism for extending the functional range of cones. Future studies should help elucidate the mechanism(s) that regulates cone PDE6 activity and how this phenomenon affects cone light adaptation.