Leber congenital amaurosis linked to AIPL1: A mouse model reveals destabilization of cGMP phosphodiesterase

Methods

Generation of Aipl1-Targeted Mice. Genomic fragments containing the flanking regions of exon 1 (5.7 kb) and exon 6 (4.1 kb) of Aipl1 from genomic DNA of mouse strain 129Sv were amplified by PCR. To construct the targeting vector, we cloned the 5′ genomic flanking sequence (5.3 kb) as an HpaI/KpnI fragment and cloned the 3′ genomic flanking sequence (2.9 kb) as a PmeI/NotI fragment in pCG4619 targeting vector (gift of R. Palmiter, University of Washington). The targeting vector was linearized and electroporated into R1 embryonic stem (ES) cells, and neomycin-resistant colonies were selected. A correctly targeted clone, VR4, was identified by PCR analysis and confirmed by Southern blotting. ES cells heterozygous for the targeted mutation (VR4) were microinjected into C57BL/6J blastocysts to generate chimeras. The chimeras were mated with C57BL/6J females to obtain progeny that were heterozygous for the mutant allele. Heterozygous animals were inbred to generate homozygous animals that lack a functional Aipl1 gene. The targeted allele was maintained in 129/SvJ and C57BL/6 mixed backgrounds. Mice carrying the mutant allele were identified by PCR. Primers used to amplify wild-type allele from the N terminus of Aipl1 were F1 (5′-CTGGGGAGGTCAAAAGGTCATCAAAGT-3′) and R1 (5′-AAGAACCACGGGGTCCTATCT-3′) and from the C terminus were R2 (5′-CTGCTGAAGAAGGAGGAGTACTATG-3′) and F2 (5′-GGGCTAGAAGGCCTGTAGTCAT-3′). Primers used to amplify the knockout allele from the N terminus were F1 and V1 (5′-CTTTCCACACCCTAACTGACACAC-3′) and from the C terminus were V2 (5′-ATCGCCTTCTATCGCCTTCTTGAC-3′) and F2.

RNA Isolation and RT-PCR. Total RNA from a whole eye was isolated by TRIzol reagent (Invitrogen), following the manufacturer's guidelines. Oligo(dT)-primed reverse transcription reactions were performed with 1 μg of total RNA by using SuperScript II (Invitrogen) to obtain cDNA, which was used as a template in PCRs. Aipl1 was amplified with primers 5′-CTAGCCCAGCATGCTCCGGCA-3′ and 5′-CAGAAAGAGCCACAGCCTCTTGTC-3′, which yielded a product of 200 bp. PCR conditions for all of the reactions were 95°C for 1 min followed by 95°C for 45 sec, 55°C for 45 sec, and 72°C for 45 sec, for 30 cycles. Hprt, which was used as a control, was amplified by using primers 5′-CAAACTTTGCTTTCCCTGGT-3′ and 5′-CAAGGGCATATCCAACAACA-3′, and yielded a 250-bp product. For analyzing Pdeα transcripts, we used primers 5′-CTGGTTCTTTAACTGTCCAGTGCCA-3′ and 5′-ATGAGGAGATTACACCCATGCT-3′, which amplified a 300-bp product. Primers 5′-ACCAAATTGCTATAGGCAGAGTCC-3′ and 5′-AGCTGACGAGTATGAGGCCAAAGTCAAG-3′ amplified a 300-bp fragment of Pdeβ. Full-length Pdeγ was amplified by using 5′-CTGACAGAGTCCAGAAGCTAAGG-3′ and 5′-CTAGGGACTCAGGCTCAGGTTT-3′. All amplified products were confirmed by sequencing. For all experiments described in this study, we used littermates from heterozygous matings for heterozygous or wild-type controls.

Histology. Mouse eyes were enucleated under room illumination and fixed in Carnoy's fixative overnight at 4°C. Fixed whole eyes were embedded in paraffin and sectioned by using an ultramicrotome.

Immunohistochemistry. Mouse eyes were enucleated under room illumination and incubated in PBS with 4% paraformaldehyde at room temperature. Retinas were dissected after 10 min, and the incubation in 4% paraformaldehyde was continued at room temperature for 1 h. Fixed retinas were cryoprotected in 30% sucrose/PBS overnight at 4°C and frozen in optimal cutting temperature compound (OCT) on dry ice. Sections (10 μm) were cut and mounted on Superfrost Plus slides (Fisher Scientific) and stored at –20°C. For immunohistochemistry, slides were washed in PBS and incubated for 2 h in primary antibody. After extensive washing in PBS, slides were incubated in secondary antibody (Alexa anti-mouse 488 and Alexa anti-rabbit 568, 1:1,000; Molecular Probes) for 1.5 h. All incubations were done at room temperature in 5% goat serum/0.2% Triton X-100/PBS. 4′,6-Diamidino-2-phenylindole (DAPI) was used to visualize nuclei, and slides were cover-slipped with Fluoromount-G (Southern Biotechnology Associates).

Electron Microscopy. Enucleated mouse eyes were fixed with a mixture of 2.5% glutaraldehyde/2% paraformaldehyde in phosphate buffer (0.1 M, pH 7.4) for 10 min at room temperature. Eyecups were made by removing the cornea and the lens, and were fixed overnight at 4°C. The eyecups were postfixed in osmium tetroxide and embedded in Epon. Sections (0.1 μm) were cut from these blocks and used for electron microscopy.

Electroretinography. A gold wire/contact lens electrode held on the cornea with 2–3% methylcellulose was used with a reference electrode in the mouth to record electroretinograms (ERGs). Mice were anesthetized with xylazine and ketamine and maintained on a heating block at 37°C. Light flashes from a photographic flash unit were delivered through a fiber optic bundle with a lens focused on the cornea. Stimuli were attenuated with neutral density filters. The output from the two intensities of white light flashes used in Fig. 3 were 300 nJ/cm² and 300 μJ/cm² measured at the cornea. ERG potentials were amplified ×1,000, filtered between 0.1 and 3,000 Hz, and digitized at 5,000 Hz. Data were collected in igor pro (Wavemetrics, Lake Oswego, OR) with an Instrutech ITC-16 analog to digital interface using a library of custom acquisition routines written by Fred Rieke (University of Washington).

Western Blotting. Whole eyes or retinas were homogenized in ROS buffer (20 mM Hepes, pH 7.4/60 mM KCl/2 mM MgCl₂/30 mM NaCl/100 μM PMSF/1 mM DTT/5 mM 2-mercaptoethanol) and a mixture of protease inhibitors (Roche Diagnostics). Homogenized tissues were solubilized in 1× SDS lysis buffer (62.5 mM Tris·HCl, pH 6.8/2% SDS/10% glycerol/0.005% bromophenol blue/5% 2-mercaptoethanol) by heating the samples at 100°C for 5 min. Equal amounts (50 μg) of sample were separated by SDS/PAGE gel electrophoresis, transferred to poly(vinylidene fluoride) (PVDF) membranes (Immobilion-P, Millipore), and probed with antibodies (4). Total protein concentration was estimated by Bradford assay (Bio-Rad). The antibodies and the dilution used in this study were: AIPL1 polyclonal antibody at 1:5,000 dilution (4), Rho monoclonal antibody (4D2; gift of R. Molday, University of British Columbia, Vancouver) at 1:4,000 dilution, guanylyl cyclase (GC-E) and GC-F antibody (gift of D. L. Garbers, University of Texas Southwestern Medical Center, Dallas) at 1:5,000 dilution, transducin γ (Tγ) antibody (gift of N. Gautam, Washington University School of Medicine, St. Louis) at 1:1,000 dilution, RK monoclonal antibody (G8; Affinity BioReagents) at 1:5,000 dilution, recoverin antibody at 1:5,000 dilution, PDE antibody (MOE; gift of B. Fung, School of Medicine, Los Angeles) at 1:4,000 dilution, and Prkca antibody (Oxford Biomedical Research, Oxford, MI) at 1:5,000 dilution.

Measurement of PDE Activity and cGMP Levels. Retinas were dissected and homogenized in ROS buffer (three retinas in 100 μl of buffer). Homogenates (2 μg of total protein) were treated with 0.05 μg of trypsin for 4 min at room temperature, and then quenched with 2 μg of soybean trypsin inhibitor. PDE activity was measured in 80 mM Tris, pH 8/4 mM MgCl₂/7.2 mM cGMP/10 nM CaCl₂/0.2 μl of [³H]cGMP (specific activity, 15 Ci/mmol; 1 Ci = 37 GBq). The reaction mixture was incubated for 4 min at 30°C, and then stopped by heating for 2 min at 100°C. Two microliters of 5 mM GMP was added to each sample, and the samples were spun for 5 min at 16,000 × g in a microfuge. Five microliters of the reaction supernatant spotted on a TLC plate (PEI cellulose; EM Science), and GMP and cGMP were separated by using 0.2 M LiCl. Excised GMP spots were eluted in 1 ml of 2 M LiCl and counted in a Beckman scintillation counter. All reactions were performed in triplicate, and the experiment was repeated at least twice. We extracted cGMP from mouse retinas by boiling in 0.1 M HCl as described by Farber and Lolley (13). An aliquot of each extract before boiling was used to measure the protein concentration by Bradford assay (Bio-Rad) using BSA as the standard. The amount of cGMP in the extract was determined by a RIA kit (Amersham Pharmacia). Each reaction was performed in duplicate, and the experiment was repeated at least four times. To minimize the individual variations between mice, retinas were pooled from different mice for each time point assayed.

Results

Generation of AIPL1-Deficient Mice. To create a mouse model of LCA caused by mutations in AIPL1, we deleted most of the coding region (exons 2–5) of Aipl1 in the mouse genome by homologous recombination (Fig. 1a). A PCR-based method (Fig. 1b) and Southern blotting (data not shown) confirmed the deletion. RT-PCR analysis showed that Aipl1 mRNA is undetectable in retinas from Aipl1^–/– mice (Fig. 1c). Immunoblots of retinal homogenates confirmed the absence of AIPL1 in the homozygous knockout and reduced AIPL1 protein levels in heterozygotes (Fig. 1d). The AIPL1 mutation is inherited with normal Mendelian distribution, indicating that lack of AIPL1 does not affect viability. Aipl1^–/– and Aipl1^+/– mice are healthy, fertile, and have no gross morphological abnormalities. Heterozygotes up to 8 months old do not exhibit any photoreceptor abnormalities detectable by ERG or histology (data not shown).

Retinal Structure and Rapid Degeneration of Photoreceptors. Deletion of Aipl1 does not cause any obvious morphological developmental changes in the retina. All retinal lamina develop appropriately, and there is no difference in the thickness of the outer nuclear layer before postnatal day 11 (P11, 10 rows of nuclei) (Fig. 2a and see Fig. 6a, which is published as supporting information on the PNAS web site). Immunofluorescence showed that the developing photoreceptor layer contains both rods and cones (Fig. 2 d and e). Phosphohistone immunostaining of retinas from P2 and -4 showed no difference in the number of mitotic cells between wild-type and Aipl1 knockout mice (data not shown). This finding suggests that AIPL1 does not play a role in cell proliferation.

Degeneration in Aipl1^–/– retinas can first be detected by light microscopy at day 12, and it proceeds rapidly thereafter. By P14, the thickness of the photoreceptor layer is reduced by half (see Fig. 6b). At P18, only a single layer of photoreceptor nuclei remains (Fig. 2 b and f) and photoreceptor degeneration is complete by 4 weeks (Figs. 2c and 6c). Both rods and cones degenerate at a similar rate (Figs. 2 f and g and 6c). The degeneration is independent of light (data not shown) and specific to the photoreceptor layer (Fig. 2 b and c). As in other cases of photoreceptor degeneration (14, 15) glial fibrillary acidic protein is up-regulated in Aipl1^–/– retinas (data not shown). Photoreceptor degeneration is more severe in central than peripheral retina. This finding indicates that degeneration proceeds from center to periphery (data not shown), consistent with previous models of retinal degeneration (16, 17). Staining of retinal sections with propidium iodide, an indicator of early stages of cell death (18) (see Fig. 7, which is published as supporting information on the PNAS web site), showed brightly stained apoptotic nuclei in the outer nuclear layer beginning at P9. This finding coincides with the initial elongation of photoreceptor outer segments. There are no signs of apoptosis before P9 (see Fig. 7).

Although the essential biochemical role of AIPL1 has not been established definitively, one function that has been suggested is protein transport (19). We evaluated this suggestion by examining the subcellular localization of several retinal proteins, including Rho, GC-E, RK, Tγ, PDE, and recoverin (Fig. 2d, supporting information, and data not shown). Each of these proteins localizes normally in Aipl1^–/– photoreceptors before they degenerate. Therefore, it is unlikely that AIPL1 plays a general role in protein transport.

Electrophysiological Responses. We performed electroretinography to evaluate the ability of Aipl1^–/– and Aipl1^+/+ retinas to respond to light. At an early stage (P12) in retinal development, flash stimuli evoke small but reliable a-wave responses in Aipl1^+/+ and Aipl1^+/– retinas, indicating the onset of photoreceptor activity at the time of formation of outer segments (Fig. 3a). By P18, a- and b-waves in Aipl1^+/+ retinas are strong (Fig. 3c). Responses from Aipl1^+/– and Aipl1^+/+ mice are indistinguishable at this age and at ages up to 90 days (data not shown). A paired flash analysis confirmed that cone responses are also normal in Aipl1^+/– retinas (data not shown). In contrast, we were unable to evoke ERG responses from Aipl1^–/– mice at any age with any illumination condition (Fig. 3 b and d). This finding indicates the absence of rod and cone activity in Aipl1^–/– retinas and is consistent with the finding that children affected by the Aipl1 form of LCA have no recordable ERG within 1 year of birth.

Disorganization of Photoreceptor Outer Segments. The absence of ERG responses prompted us to examine the ultrastructure of Aipl1^–/– photoreceptor outer segments by transmission electron microscopy. At P8, when outer segments first appear, Aipl1^+/+ and Aipl1^–/– retinas are indistinguishable (data not shown). At P11, photoreceptor outer segments of Aipl1^–/– mice are shorter than normal, highly disorganized, and fragmented (Fig. 4 a–d). Inner segments are enlarged and disorganized (Fig. 4 b and d). Vacuolar inclusions (indicated by asterisks), similar to those found in rd/rd mice, are present in the inner segments of photoreceptors in Aipl1^–/– retinas (17) (Fig. 4 d and e).

PDE Is Not Active in Mice Lacking AIPL1. We quantified the relative amounts of retinal proteins by immunoblot analysis. At P18, no photoreceptor-specific proteins are detectable because most of the photoreceptor layer has degenerated (Fig. 5a). At P8 and -10, well before significant degeneration has occurred, the levels of Rho, GC-E, RK, T_βγ, and recoverin are similar in Aipl1^+/+ and Aipl1^–/– retinas (Fig. 5a and data not shown). A striking exception is that the levels of all three subunits of PDE (α, β, and γ) were reduced by 90% in Aipl1^–/– mice (n = 3). Reduced but detectable levels of all three PDE subunits are apparent at P8, the first day that they are normally expressed; however, the mRNA levels of all three PDE subunits are normal (Fig. 5b). This finding contrasts with the phenotype of the rd/rd mouse in which PDE-β protein is completely absent and mRNA is significantly reduced (20–23).

To confirm the reduced levels of PDE, we assayed PDE activity after limited trypsin proteolysis, a treatment that specifically destroys the inhibitor but not the catalytic subunits of PDE (24). We found no significant PDE activity in Aipl1^–/– retinas compared to Aipl1^+/+ retinas at P8-P12 (Fig. 5c and Inset). This finding suggests that the low amount of PDE (10% of Aipl1^+/+) detected immunologically in Aipl1^–/– retinas at this age is not enzymatically active. We also measured the cGMP content of retinas. Consistent with the absence of cGMP hydrolytic activity in Aipl1 ^–/– retinas, cGMP accumulates to higher levels (7- to 9-fold higher at P12) in Aipl1^–/– retinas than in Aipl1^+/+ retinas (Fig. 5d). The elevation of cGMP levels precedes the retinal degeneration.

Rapid retinal degeneration and elevated levels of retinal cGMP are characteristic phenotypes of an extensively studied mouse mutant, rd/rd (17). However, there are distinct differences between the phenotypes of Aipl1^–/– and rd/rd. Inactivation of the PDE-β subunit gene in rd/rd reduces the level of PDE-β mRNA and causes complete loss of the PDE-β subunit (20–23). In contrast, PDE mRNA levels are normal, and all three PDE subunits are destabilized in Aipl1^–/– retinas before the retina degenerates. Full-length PDE-β subunit can be detected at P8–P10 of Aipl1^–/– retinas (Fig. 5a). Full-length PDE-β is undetectable at this age in rd/rd retinas because PDE-β is truncated by a stop codon in the rd gene. Aipl1^–/– mice also differ from rd/rd mice in that no detectable ERG response can be recorded from Aipl1^–/– mice at early stages of retinal development (P12), whereas rd/rd mice produce reduced but significant responses at that age (17). Retinal degeneration in Aipl1^–/– mice has remained completely linked to the inactivated Aipl1 locus on chromosome 11 through three generations of outcrosses (data not shown). The PDE catalytic subunits are on different linkage groups. Only the PDE-γ gene is in linkage group 11. To eliminate the possibility that a spurious mutation in PDE-γ subunit could contribute to retinal degeneration, we sequenced the entire gene and found no mutations. The PDE-β gene linked to rd is on chromosome 5. To make certain that the rd mutations were absent from our strains, we sequenced the entire PDE-β gene from our Aipl1^–/– mice. Neither of the two PDE-β mutations linked to rd, Y347Stop or the proviral insertion in intron 1 (25–27), was present in the Aipl1^–/– mice that we characterized (see Fig. 8, which is published as supporting information on the PNAS web site).

Discussion

Our analysis of Aipl1^–/– mice shows that normal expression of AIPL1 is required for photoreceptor survival. Although rods and cones form in the absence of AIPL1, they do not respond to light. Degeneration of both rods and cones is complete within 3 weeks of birth. The PDE holoenzyme is destabilized by the AIPL1 deficiency. Degeneration appears to be caused by this destabilization. Support for this idea comes from the observation that direct inactivation of the PDE-β subunit gene in rd/rd mice also causes photoreceptor degeneration. Like the PDE mutation in rd/rd mice, the Aipl1^–/– mutation also reduces cGMP hydrolytic activity and elevates cGMP, causing rapid degeneration of photoreceptor cells (17).

What is the link between AIPL1 and PDE stability? Our previous study revealed that AIPL1 can enhance protein farnesylation (4). At least three proteins in photoreceptors are known to be farnesylated, PDE-α, Tγ, and RK (6–9). Of these, only PDE-α requires farnesylation for its stability (9, 11). A previous study showed that mutations that block farnesylation cause degradation of PDE-α protein in cultured cells without loss of PDE-α mRNA (11). Loss of protein without loss of mRNA is a phenotype specific to defects in farnesylation of PDE-α (11). In our study of Aipl1^–/– retinas, we found that mRNA levels for all three PDE subunits are normal. In contrast, mutations that block geranylgeranylation of the PDE-β subunit reduce the stability of both protein and mRNA (11). Altogether, these observations indicate that AIPL1 directly influences the farnesylation of PDE-α. In the absence of AIPL1, PDE-α is not efficiently farnesylated and is unstable and the entire trimeric PDE complex degrades. It is also possible that AIPL1, through its interaction with the farnesylated PDE-α subunit, may act as a chaperone to promote the folding and assembly of PDE subunit complex. In this case, absence of AIPL1 results in defective subunit assembly, and the whole complex is destabilized. It is possible that AIPL1 plays a role in promoting the proper intracellular localization of PDE in outer segments. However, it is unlikely that AIPL1 is involved in the transport of PDE because the residual PDE found in AIPL1 knockout retinas localizes to photoreceptor outer segments (Fig. 9, which is published as supporting information on the PNAS web site). Further studies are needed to distinguish the mechanism by which AIPL1 affects the stability of the PDE heterotrimer. Irrespective of the mechanism by which AIPL1 affects the stability of active PDE, reduction in PDE results in the consequent elevation of cGMP, causing rapid apoptosis of photoreceptor cells.

Our analysis of the phenotype of the Aipl1^–/– mouse has revealed a molecular pathway that can contribute to photoreceptor degeneration in LCA linked to mutations in AIPL1. We have shown that the Aipl1^–/– mouse is a suitable animal model with which to further enhance the understanding of LCA. In humans, mutations in AIPL1 cause LCA, a disease that affects both rods and cones. LCA has a more severe effect on the retina and an earlier onset than retinitis pigmentosa linked to mutations in PDE subunits (28, 29). This finding suggests that there may be significant functions for AIPL1 in addition to stabilizing PDE. Further studies will be needed to identify those additional activities.

Supplementary Material