New mouse models for recessive retinitis pigmentosa caused by mutations in the Pde6a gene

The heterotetrameric phosphodiesterase (PDE) 6 complex, made up of a , b and two g subunits, regulates intracellular cGMP levels by hydrolyzing cGMP in response to light activation of G protein coupled receptors in cones and rods, making it an essential component of the visual phototransduction cascade [Zhang, X. and Cote, R.H. (2005) cGMP signaling in vertebrate retinal photoreceptor cells. Front. Biosci. , 10, 1191–1204.]. Using a genetic positional candidate cloning strategy, we have identiﬁed missense mutations within the catalytic domain of the Pde6a gene in two mouse models from an ethyl nitrosourea chemical mutagenesis screen. In these ﬁrst small rodent models of PDE6A, signiﬁcantly different biochemical outcomes and rates of degeneration of murine photoreceptor cells were observed, indicating allelic variation and previously unrecognized structure–function relationships. In addition, these new models reveal that the mutations not only affect the function of the PDE6A protein itself, but also the level of PDE6B within the retina. Finally, we show that the variation of the disease phenotype by background modiﬁer genes may be dependent upon the particular disease allele present.


INTRODUCTION
Retinitis pigmentosa (RP), characterized by progressive photoreceptor degeneration, vision impairment and often blindness, affects approximately 1 in 4000 individuals worldwide (1). According to RetNet (http://www.sph.uth.tmc.edu/retnet/ home.htm), a database that summarizes heritable retinal disease loci reported to date, 21 autosomal recessive RP genes have been identified, many of which encode proteins important in the rod photoreceptor visual transduction cascade. Central among these is the heterotetrameric phosphodiesterase (PDE) 6 complex, consisting of a, b and two g subunits. Each subunit is essential for photoreceptor function and maintenance, as mutations within PDE6A and PDE6B have been reported to cause recessive RP in humans (2 -4), and mutations within Pde6g have been reported to cause retinal degeneration in mice (5,6). Several mouse models with mutations in Pde6b, such as rd1 (7), rd10 (8), rd1 2J (nmf137) (8) and those from the Medical Research Council, Human Genetics Unit, Western General Hospital (Edinburgh, Scotland, UK) (9) are available. The Cardigan Welsh corgi dog, however, is the only animal model reported to date with a PDE6A mutation (10,11). With the exception of a brief clinical report (2,4) and a report describing the mutation in PDE6A (10,11), studies to examine the biochemical consequences of a mutation in PDE6A have yet to be carried out.
Here, we report two new mouse models with Pde6a mutations identified in a G3 ethyl nitrosourea (ENU) mutagenesis screen. The original residues mutated in these Pde6a models encode for amino acids that are highly conserved across species both at the nucleotide and amino acid level. Despite the fact that both mutations are located within the catalytic domain of the PDE6A protein, the biochemical consequences of the mutations differ. cGMP levels are elevated in Pde6a nmf282/nmf282 mutants but reduced in Pde6a nmf363/nmf363 mice when compared with their respective controls.
Photoreceptor degeneration also differs between the mutants. Photoreceptor degeneration occurs at a faster rate in Pde6a nmf282/nmf282 mice. Our data suggest that the valine 685 residue is more critical to PDE6A function than aspartic acid 670. We also show that genetic background modifier(s) are able to delay photoreceptor cell death in dark-reared Pde6a nmf363/nmf363 but not in Pde6a nmf282/nmf282 mutants. The availability of this allelic series of Pde6a mutations will allow for detailed in vivo studies of the structure and function of PDE6A and provides a means to examine genetic modifying influences on disease progression.

RESULTS
nmf282 and nmf363 are caused by missense mutations in Pde6a Mice with an abnormal fundus pigmentation were observed by indirect ophthalmoscopy in a colony of ENU mutagenized G3 A.B6-Tyr þ /J mice in the Neuroscience Mutagenesis Facility (NMF) at The Jackson Laboratory. At 3 months of age, three out of 11 G3 offspring from a sibship were affected, suggesting a recessive inheritance of the disorder. Outcrosses of an affected mutant, named nmf282, to wild-type strain A.B6-Tyr þ /J and an intercross of the resulting F1 mice confirmed the recessive inheritance pattern as no F1 mice were affected but 25% of F2 mice had a grainy fundus appearance upon examination. To map the disease-causing mutation, affected mice were outcrossed to strain DBA/2J and the resultant F1 mice were intercrossed. DNA samples from the segregating F2 intercross were used in a genome wide scan in which a DNA pooling strategy was employed, followed by genotyping of individual DNA samples with simple sequence length polymorphic (SSLP) markers to confirm the linkage. The nmf282 disease locus was mapped to Chromosome 18 between markers D18Mit51 and D18Mit81. Subsequently, all F2 mice were phenotyped for the grainy retinal appearance by indirect ophthalmoscopy and individually genotyped for the flanking markers, D18Mit51 and D18Mit81, to construct a fine structure map for the region containing nmf282. All unaffected mice carrying a recombination between the flanking markers were crossed to homozygous nmf282/nmf282 animals to determine whether the recombinants carried the nmf282 allele. The region encompassing nmf282 was narrowed to 1.26 Mb in size between markers D18Mit51 and SNP rs8237402 (Fig. 1A, Supplementary Material, Fig. S1 and Table S1).
The critical nmf282 region identified by the fine structure map contained 12 genes, according to the Ensembl database (http:// www.ensembl.org), including Pde6a, an excellent candidate gene. Sequencing of both Pde6a cDNA and genomic DNA from A.B6-Tyr þ /J-Pde6a nmf282/nmf282 mice revealed a single nucleotide G to A missense mutation in exon 16 (Fig. 1B), predicted to cause an amino acid change from valine to methionine (V685M). The valine residue is highly conserved among zebrafish, mouse, dog, human and bovine, suggesting its importance in the catalytic function of PDE6A (Fig. 1C).
During the screening of additional G3 offspring of C57BL/ 6J ENU mutagenized mice, another mutant with a grainy-appearing retina was observed in the NMF colony. Clinical examination of the entire sibship revealed four out of eight G3 mice from the family were affected. Subsequent outcrosses to wild-type C57BL/6J mice and an F2 intercross of normal F1 progeny revealed that five out of 25 F2 mice were affected, indicating a recessive mode of inheritance. In a genome-wide scan, the disease locus named nmf363 was also mapped to Chromosome 18 between markers D18Mit51 and D18Mit81. Subsequently, a complementation test with A.B6-Tyr þ /J-Pde6a nmf282/nmf282 was carried out to determine if nmf363 and nmf282 were alleles of Pde6a. All compound heterozygous mice (C57BL/6J;A.B6-Tyr þ / J-Pde6a nmf282/nmf363 ) had grainy retinas when examined by indirect ophthalmoscopy at P38, indicating that nmf363 and nmf282 were allelic.
To confirm the results of the complementation test above, cDNA and genomic DNA from C57BL/6J-Pde6a nmf363/ nmf363 mice were directly sequenced. An A to G missense mutation in exon 15 of Pde6a was identified and is predicted to cause an amino acid change from aspartic acid to glycine (D670G) in PDE6A (Fig. 1D). This Pde6a nmf363 mutation, located in the catalytic domain of PDE6A, is 15 amino acids from the Pde6a nmf282 mutation. The aspartic acid residue at position 670 is also highly conserved among zebrafish, mouse, dog, human and bovine (Fig. 1C).
Mutations within the Pde6a subunit lead to rapid photoreceptor degeneration By histological analysis of retinal sections from A.B6-Tyr þ / J-Pde6a nmf282/nmf282 mice and age-matched controls stained with hematoxylin-eosin, photoreceptor degeneration was clearly observable at P12 with an 20% reduction in the outer nuclear layer (ONL) thickness and an 50% reduction in the outer segment (OS) thickness ( Fig. 2A and B). At P14, the thickness of the ONL and IS decreased to 50% of that observed in heterozygote mice and all OS were degenerated. Nearly all photoreceptors were gone by P21. In contrast, at P12, retinal sections of C57BL/6J-Pde6a nmf363/nmf363 mutants and controls are indistinguishable ( Fig. 2C and D). By P14, an 30% reduction in ONL thickness is observed in mutant mice. The thickness of the IS and OS is almost comparable to that of C57BL/6J-Pde6a nmf363/þ controls until P16. Degeneration of OS is clearly seen in C57BL/6J-Pde6a nmf363/ nmf363 mutants at P19. However, even at P36, a single layer of cell bodies still remained in the ONL of the mutant mice. Therefore, the photoreceptor degeneration while rapid has a later onset and is prolonged in C57BL/6J-Pde6a nmf363/nmf363 mutants compared to A.B6-Tyr þ /J-Pde6a nmf282/nmf282 mutant mice. Functional analyses by ERG are consistent with these histological observations (Fig. 2E). While the ERG response is extinguished in A.B6-Tyr þ /J-Pde6a nmf282/nmf282 mutants at P14, some residual activity is still observed in C57BL/ 6J-Pde6a nmf363/nmf363 mutants at P21.  (Fig. 3A). As anticipated, the activity of cGMP-specific PDE in the Pde6a nmf282/nmf282 mutant retinas was significantly reduced in comparison to heterozygous controls at P7, P10 and P12. Trypsinization is reported to maximally activate cGMP-PDE activity (12) by cleavage of the PDE complex component, PDE6G, the inhibitory subunit of PDE6 and PDE6B (13). Therefore, the cGMP-specific PDE activities of trypsinized retinal cytosolic and membrane fractions of Pde6a nmf282/nmf282 at P12 were measured (Fig. 3B). An activation of cGMP-PDE activity was not observed following trypsinization of the retinal fractions of the mutant mice, whereas trypsinization dramatically activated both cytosolic and membrane cGMP-PDE activity of the heterozygous controls. This suggests that the residue mutated in A.B6-Tyr þ / J-Pde6a nmf282/nmf282 mice is absolutely essential for the catalytic function of PDE6A.

Retinas of
The cGMP-specific PDE activities in C57BL/ 6J-Pde6a nmf363/nmf363 were also significantly lower than those of heterozygous controls at P12, P16 and P19 (Fig. 3C). However, the reduction in activity of retinal cGMP-specific PDE of C57BL/6J-Pde6a nmf363/nmf363 was less than that observed for A.B6-Tyr þ /J-Pde6a nmf282/nmf282 . Also, in contrast to results obtained for A.B6-Tyr þ / J-Pde6a nmf282/nmf282 mice, trypsinization activated both cyto-solic and membrane cGMP-PDE activity in the retina of the C57BL/6J-Pde6a nmf363/nmf363 mutants (Fig. 3D). However, the activity after trypsinization in retina of the mutant mice remained significantly lower than that observed in the retinas of heterozygous controls.
To determine if the reduction in PDE activity was attributable to a reduction in PDE6A levels, western analysis was carried out. Immunoblots showed that PDE6A and PDE6B were abundant in the membrane fraction with significantly lower levels in the cytosolic fraction of homogenized retinas from heterozygous control mice. In contrast, the levels of both proteins were clearly reduced, and when corrected for loading, the cytosolic fraction appeared to contain more PDE6A and PDE6B than the membranous fraction in the retinas of A.B6-Tyr þ /J-Pde6a nmf282/nmf282 and C57BL/ 6J-Pde6a nmf363/nmf363 mutant mice (Fig. 4). PDE6G was detected exclusively in the membrane fraction of the retinal extracts at equivalent levels in both heterozygous controls and mutant mice carrying either mutation.
It has been previously shown that cGMP content increases in mice carrying mutations affecting PDE6B activity (14,15). To determine if the observed reduction in cGMP-PDE activity also lead to the accumulation of cGMP in A.B6-Tyr þ /J-Pde6a nmf282/nmf282 and C57BL/ 6J-Pde6a nmf363/nmf363 mutant mice, the retinal cGMP content of mutants and their respective controls were measured between postnatal day (P) 7 through P21 ( Fig. 5A and B). To map the mutations, genetically, affected mutants were mated to DBA/2J mice, and the resulting F1 progeny were either intercrossed to produce F2 offspring or backcrossed to an affected mutant. In the subsequent high-resolution intercrosses, uninformative recombinant F2 mice were progeny-tested by crossing them to mice homozygous for the mutation to confirm that the recombinant animal carried the mutant allele. The nmf282 locus was mapped between markers D18Mit51 and SNP rs3705875. The position of Pde6a is shown by the arrow between the markers D18Mit51 and SNP rs8237402.    In A.B6-Tyr þ /J-Pde6a nmf282/þ (Fig. 5A), the retinal cGMP level was 1.5 + 0.2 pmol/mg protein at P7, but gradually increased as photoreceptor OSs matured. In the retina of mutant mice, cGMP levels were slightly but significantly higher than that observed in controls at P7, similar to controls at P8, and lower than controls at P10. After P11, mutant cGMP levels rose significantly, reaching 28.8 + 3.2 pmol/mg protein at P12, which is approximately 2-fold higher than that observed in control retinas. The cGMP content in retinas of C57BL/6J-Pde6a nmf363/þ control and C57BL/ 6J-Pde6a nmf363/nmf363 mutant mice were measured between P12 through P21 (Fig. 5B). The cGMP level in the heterozygous control retina was 18.9 + 1.7 pmol/mg protein at P12, and gradually increased as photoreceptor OSs matured. However, unlike A.B6-Tyr þ /J-Pde6a nmf282/nmf282 mutants, cGMP levels of C57BL/6J-Pde6a nmf363/nmf363 mutant mice were similar to the control at P12, but declined to significantly lower levels than controls between P14 through P21. To determine if the lower cGMP content was due to a reduction in the generation of cGMP, the activity of the key enzyme necessary for its generation, guanylate cyclase (GC), was assessed. The basal activity of GC in the membranous fraction of retinal extracts in the presence of 30 mM EDTA and absence of exogenous guanylate cyclase activating proteins (GCAPs) was not different in heterozygote controls and mutants at P10, but GC activity was significantly reduced in Pde6a nmf282/nmf282 mutants compared to heterozygous controls at P12 and P14 (Fig. 6A). In C57BL/6J-Pde6a nmf363/nmf363 mutants, GC activity did not differ from heterozygous control until P13, but at P16, GC activity was significantly lower than that of heterozygous controls (Fig. 6D). The reduction in activity was reflected in the levels of GC protein which was reduced in both Pde6a nmf282/nmf282 retinas at P14 ( Fig. 6B and C) and C57BL/6J-Pde6a nmf363/nmf363 retinas at P16 ( Fig. 6E and F).
In an effort to normalize the data above with respect to the actual amount of OSs present, immunoblots of retinal lysates from the different mutants and controls were probed with antirhodopsin antibody in a longitudinal analysis. Interestingly despite significant shortening of OSs ( Fig. 2B and D), rhodopsin levels did not differ across the time points described above for cGMP-PDE activity and cGMP level assays (data not shown). This led us to examine rhodopsin localization in the Pde6a mutants. Rhodopsin is localized not only in OS but also the perinuclear region of ONL in A.B6-Tyr þ /J-Pde6a nmf282/nmf282 at P11 and C57BL/6J-Pde6a nmf363/nmf363 mutant retinas at P14 (data not shown). The same phenomenon was also observed in C3H-Pde6b rd1/rd1 mice at P11 (data not shown).
Natural disease progression and biochemical differences between A.B6-Tyr 1 /J-Pde6a nmf282/nmf282 and C57BL/ 6J-Pde6a nmf363/nmf363 are due to allelic variance and not to the influence of genetic background To determine whether the biochemical differences and rate of photoreceptor degeneration between A.B6-Tyr þ / J-Pde6a nmf282/nmf282 and C57BL/6J-Pde6a nmf363/nmf363 raised in standard vivarium conditions were attributable to allelic

182
Human Molecular Genetics, 2009, Vol. 18,No. 1 or genetic background influences, we intercrossed male and female F1 C57BL/6J;A.B6-Tyr þ /J-Pde6a nmf282/nmf363 compound heterozygous mice to generate F2 mice homozygous for either allele and compound heterozygotes on a mixed C57BL/6J and A.B6-Tyr þ /J segregating background. The photoreceptor degeneration observed at P14 in mutants in a mixed genetic background was similar to that observed for the same alleles in their original fixed inbred backgrounds ( Fig. 7A and B). The reduction of the thickness of ONL and OS in compound heterozygous mutants lay between the two homozygous parental alleles in the same segregating background. In addition, the pattern of retinal cGMP levels at P14 in mixed background animals of either variant was also similar to that observed in those obtained for the original inbred background (Fig. 7C). These data strongly indicate that the differences between the two Pde6a mutants are due to the specific mutations and not to genetic background influences.

No activation of calpain or caspase-3 is observed in Pde6a mutants
High intracellular cGMP concentrations have been shown to lead to an increased intracellular Ca 2þ influx (16), presumably due to the opening of cGMP-gated channels. High intracellular Ca 2þ concentrations are in turn thought to activate calpain, one of the calcium-activated neutral proteases, and/or caspase-3 (17,18). To test whether calpain and/or caspase-3 are activated in A.B6-Tyr þ /J-Pde6a nmf282/nmf282 mutants, in which positive ONL TUNEL staining was observed at P11 (Fig. 8A) or C57BL/6J-Pde6a nmf363/nmf363 at P14, we examined the cleavage of a-fodrin, a substrate for both calpain and caspase-3. a-fodrin, with a molecular weight of 240-280 kDa, is cleaved to 145 and 150 kDa forms by calpain and to 145 and 120 kDa by caspase-3 (16,18). As anticipated for the C57BL/6J-Pde6a nmf363/nmf363 mutant mice in which cGMP levels were not elevated, cleavage of fodrin and caspase-3 was not observed. Similarly, in A.B6-Tyr þ / J-Pde6a nmf282/nmf282 mutant retinas cleavage of fodrin and caspase-3 was also not observed, despite the increased cGMP levels in this mutant. To confirm the results, immunoblots probed with anti-cleaved caspase-3 antibody showed an absence of cleaved caspase-3 in both mutants ( Fig. 8B and C).
Dark-rearing slows the progression of photoreceptor degeneration in C57BL/6J-Pde6a nmf363/nmf363 but not in A.B6-Tyr 1 /J-Pde6a nmf282/nmf282 mutant mice and is due to genetic background influences Progression of photoreceptor degeneration in some RP mouse models can be retarded by dark-rearing, suggesting that the exposure to light can precipitate or hasten the degenerative process (7). To determine if this was also the case with Pde6a mutants, both strains were bred and dark-reared until sacrifice at P14 (Pde6a nmf282/nmf282 ) or P19 (Pde6a nmf363/ nmf363 ). At P14, the ONL thickness of dark-reared A.B6-Tyr þ /J-Pde6a nmf282/nmf282 mice did not differ from that of mutants reared under standard lighting conditions ( Fig. 9A and B). The ONL thickness at P19 of dark-reared C57BL/6J-Pde6a nmf363/nmf363 mutant mice, however, was significantly greater than that observed for mutants reared under standard lighting conditions. The functional analyses by ERG were consistent with the histological observations. At P21 when the ERG a-wave is extinguished in C57BL/ 6-Pde6a nmf363/nmf363 mutants reared under standard lighting condition, a clear a-wave, whose amplitude is smaller than normal C57BL/6J (Fig. 2E), is still observed in dark-reared mutants (Fig. 9C). The b-wave amplitude of dark-reared animals is also larger than that of the mutants reared under standard lighting conditions. To test whether this protection is caused by genetic background influences, dark-reared F2 mice homozygous for either allele on a mixed segregating background were examined for protection from photoreceptor loss at P19. As anticipated, no protection from photoreceptor loss was observed in 9 F2 Pde6a nmf282/nmf282 mutants. Interestingly, photoreceptor loss similar to the animal reared under standard lighting conditions was observed in all eight dark-reared F2 Pde6a nmf363/ nmf363 mutants examined (Fig. 10). These results suggest that the partial protection from photoreceptor loss seen in dark-reared C57BL/6J-Pde6a nmf363/nmf363 is probably a result of genetic background influences. Although a relatively small cohort, because none of the mice that were genotypically nmf363/nmf363 on the segregating background were protected, it suggests that the genetic modifiers act in a recessive manner or that multiple modifiers act in concert to contribute to the observed protection.

DISCUSSION
The PDE6 complex is an essential component of the visual phototransduction cascade. It regulates intracellular cGMP levels by hydrolysis of cGMP as a consequence of light activation of cone or rod G protein coupled receptors (20). Mutations within PDE6A and PDE6B have been identified in humans, accounting for 3 -4% of recessive RP in North America (4). A small rodent model for PDE6A mutations had, however, not yet been reported. The two new mouse models, generated by ENU chemical mutagenesis, C57BL/ 6J-Pde6a nmf363/nmf363 and A.B6-Tyr þ /J-Pde6a nmf282/nmf282 , described here provide a unique resource to study both the biochemical and pathological consequences of different mutations within the PDE6A gene.
In the present study, we identify two missense mutations within Pde6a that are predicted to alter highly conserved amino acids in the catalytic domain of the protein. Previous studies indicate that the aspartate residue at position 670, which is mutated in C57BL/6J-Pde6a nmf363/nmf363 , is conserved across the PDE superfamily. Although a crystal structure of PDE6A is currently not available, an X-ray crystallography determined structure of PDE5A. Because of the high similarity of the two proteins, by inference, the mutant residue in Pde6a nmf363/nmf363 is predicted to be positioned very near the bound GMP in the crystal structure of the PDE5a catalytic domain (21) when GMP is bound (Fig. 12). This suggests that this residue may have functional importance across the PDE superfamily. In contrast, the valine

184
Human Molecular Genetics, 2009, Vol. 18, No. 1 at position 685, which is mutated in A.B6-Tyr þ /J-Pde6a nmf282/ nmf282 , while conserved within the PDE6 family, including PDE6B and PDE6C, varies considerably across the PDE superfamily. In fact, residue 685 is a methionine in the PDE2 family. Therefore, residue 685 may have an important functional role that is specific for the PDE6 family and its unique function in phototransduction. The importance of this residue for PDE6A is highlighted by the more severe retinal phenotype observed when it is mutated, when compared with the mutation at aspartate 670. As anticipated, both mutations lead to a decrease in cGMP-PDE activities. However, unexpectedly, the cGMP-PDE activity differed between the two mutants with C57BL/6J-Pde6a nmf363/nmf363 mutants having a milder reduction in activity than that observed in A.B6-Tyr þ / J-Pde6a nmf282/nmf282 mutant mice. The total cGMP levels also differed. In A.B6-Tyr þ /J-Pde6a nmf282/nmf282 , the retinal cGMP content from P11 through P14 was almost twice as high as that observed in the heterozygous controls. In contrast, the cGMP content in the C57BL/6J-Pde6a nmf363/nmf363 retinas was significantly lower than that observed in the heterozygous controls between P14 and P21. The levels and activity of GC, the key enzyme necessary for the generation of cGMP, were reduced in both mutants. However, while the cGMP-PDE activity was almost gone in the A.B6-Tyr þ /J-Pde6a nmf282/ nmf282 retinas at P10, it was still partially active in the C57BL/6J-Pde6a nmf363/nmf363 retinas at P16. Therefore, it is possible that the residual activity of cGMP-PDE is involved in the lower cGMP content observed in C57BL/ 6J-Pde6a nmf363/nmf363 retinas. Alternatively, since 90% of the total cGMP in OSs in WT animals is bound to the GAF domains of PDE6 (22), the reduction of cGMP levels in Pde6a nmf363 homozygotes may be due primarily to the decrease in PDE6 levels. The cytoplasmic cGMP levels may not be dramatically altered. In contrast, in Pde6a nmf282 homozygotes, a high level of cGMP in addition to greatly reduced PDE6 levels suggests that cytoplasmic levels of cGMP must be much higher than normal. This difference may partially account for the difference in severity and in effects of darkrearing between the two alleles.
A reduction in retinal cGMP content in mice heterozygous for the Pde6b rd1 mutation (23,24) has been observed in association with mildly lower retinal cGMP-PDE activities compared to wild-type controls (19). Likewise, a similar difference in cGMP levels was observed between null and hypomorphic alleles of aryl hydrocarbon receptor-interacting protein-like 1 (AIPL1), a protein essential for the biosynthesis of PDE6. While Aipl1 null mutants were reported to have an elevated retinal cGMP content (25), Aipl1-hypomorphic mice were shown to have a lower retinal cGMP content than controls (26). It is interesting to note that AIPL1 plays an essential role in processing of farnesylated proteins in retina (27). The carboxyterminus of PDE6A is farnesylated (28), and this modification is necessary for proper anchoring of the PDE6 complex to the OSs (29). Presumably, the observed changes in cGMP levels in Aipl1 mutants are in part due to its role in generating a fully functional PDE6 protein complex localized to the OSs. In both Pde6a mutants, the PDE6A content of the cytosolic fraction was greater than that for the membranous fraction suggesting that PDE6A is not bound to the OS membranes as observed for the wild-type controls. Perhaps, the mutations play roles in the pathological consequences of the disease phenotype by a combination of indirect effects on defective binding to the membranous fraction and the reduction in PDE6B levels, and of the direct effects on catalytic activity of the PDE6A subunits.
Through genetic studies, we have shown that the differential reduction in retinal cGMP-PDE activity between the two mutants is due to allelic and not due to genetic background influences. Trypsinization which is reported to activate cGMP-PDE activity maximally (12) was unable to increase cGMP-PDE activity in the retinal lysates of A.B6-Tyr þ / J-Pde6a nmf282/nmf282 mice, whereas trypsinization activated both cytosolic and membrane cGMP-PDE activity from the retinas of the C57BL/6J-Pde6a nmf363/nmf363 mice. This suggests that the nmf282 mutation renders the PDE6A protein non-functional, and that the valine at position 685 is important for the enzymatic activity of PDE6A. Alternatively, the mutation may affect both the PDE6A catalytic activity as well as the binding of PDE6G to PDE6A, rendering the trypsinization superfluous. However, the V685M amino acid change does not fall within the residues 751 -763 of PDE6A which have been previously reported to be important for the binding between PDE6A and PDE6G (13).
The mutation that A.B6-Tyr þ /J-Pde6a nmf282/nmf282 carries causes an accumulation of cGMP that could potentially lead to excessive calcium influx through cGMP-gated cation channels. Although intracellular calcium levels were not determined in the present study, intracellular cGMP (14,15) and calcium levels (16) have been reported to be elevated at P10 prior to the onset of apoptosis in Pde6b rd1/rd1 mice. Calcium levels have also been shown to be elevated in light-induced photoreceptor damage models (30), suggesting that calcium may play a central role in retinal degenerations. It is still largely unknown how the sustained increase of intracellular calcium leads to photoreceptor degeneration. However, it has been suggested that calpain, one of the calcium-activated neutral proteases, and caspase-3 are activated and lead to the cleavage of a-fodrin, an essential cytoskeletal protein. Cleavage of a-fodrin is thought to contribute to the cellular collapse in apoptotic cells (18). This phenomenon has been reported in Pde6b rd1/rd1 mutants (16) in which cGMP levels are elevated 4 -5-fold and calpain activity increased significantly during retinal degeneration. Both changes are closely associated with the death of photoreceptor cells (31,32). In contrast, in neither of the Pde6a mutants examined was cleavage of a-fodrin or caspase-3 observed. Therefore, a mechanism other than increased calcium influx may underlie the retinal degeneration observed in these particular mutants in which cGMP levels are either 2-fold elevated or reduced by 50% in comparison to wild-type controls.
Since dark-rearing of C57BL/6J-Pde6a nmf363/nmf363 mutants was able to slow the photoreceptor degeneration, oxidative stress may potentially be involved in the mechanism of cell death in this model. Alternatively, if the lower cGMP levels observed in C57BL/6J-Pde6a nmf363/nmf363 mice play an important role in the degeneration, dark-rearing may provide some protection by increasing the cGMP levels. Further investigation to clarify the mode through which apoptosis occurs is necessary.
Mutations in the PDE6A gene have been described in families with autosomal recessive RP (2), and the ENU-induced mouse models described here are the first small rodent models of RP

186
Human Molecular Genetics, 2009, Vol. 18, No. 1 induced by mutations within Pde6a. To date, the Cardigan Welsh corgi dog has been the only animal model available to study the disease pathology and progression as a result of a mutation within PDE6A (9,10). The Pde6a mouse models described here reveal complex changes, not only within the protein itself but in other molecules of the phototransduction cascade, and the allelic variation demonstrated by these models will yield further insight into the structure and function of the mutant residues. Additionally, the modification of the disease phenotype by dark-rearing will be valuable for identify-ing molecules that may be able to delay RP caused by PDE6A mutations. Finally, these models will be useful for testing various treatment modalities such as the modification of environmental effects and the use of pharmacological agents or gene therapy that are expected to delay or ameliorate photoreceptor degeneration.

Animals, ENU mutagenesis and crosses
Breeding and housing of the mice were conducted in the NMF or in the Research Animal Facility at The Jackson Laboratory under specific pathogen-free conditions. The mice were provided an autoclaved 6% fat chow diet (5K52, Lab Diet, PMI International, Brentwood, MO, USA) and acidified (pH 2.5; to retard bacterial growth) water in a vivarium with a 12:12 h dark -light cycle. To generate chemically mutanized mice, male A.B6-Tyr þ /J or C57BL/6J mice were treated with three doses of ENU, and immediately mated with untreated females of the same background strain to produce G1 offspring. If, only after 3 months from ENU administration, mutagenized G0 mice were able to sire offspring, the mutagenesis was considered successful. For recessive mutations, a three-generation screen was used in which male G1 mice from successfully mutagenized G0 males were mated to wild-type female mice of the same strain to produce G2 offspring. Female G2 mice were backcrossed to their G1 sires to produce G3 offspring. G3 mice were screened by indirect ophthalmoscopy. Once a potential mutant was identified, affected G3 mice were crossed with wild-type mice of the same genetic background to generate F1 offspring, and the F1 mice were intercrossed. If the F1 offspring were unaffected and one-fourth of the individuals in the F2 generation were affected, the mutation was considered to be heritable in a recessive fashion. The mutant mouse colonies were maintained by heterozygote Â heterozygote, mutant Â heterozygote and mutant Â mutant matings. To genetically map the mutations, affected mutants were mated with DBA/2J mice, and the resulting F1 progeny were intercrossed to produce F2 offspring or backcrossed to the affected parental strain to produce a BC population. In the subsequent high-resolution intercrosses, uninformative recombinant mice were progeny-tested by crossing them to mice homozygous for the mutation to confirm that the recombinant animal carried the mutant allele.
Chromosomal localization, fine structure mapping and genotyping Tail DNA was isolated from mice generated in the intercrosses described above, according to Buffone and Darlington (33). DNA pools of affected and unaffected F2 offspring were subjected to a genome-wide scan with 80 robust SSLP markers known to differ between the parental strains, A.B6-Tyr þ /J or C57BL/6J, and DBA/2J (34). Once a significant skewing in allele abundance between the two DNA pools was observed, DNA samples were tested individually using SSLP MIT and SNP markers (Supplementary Material, Table S1) to confirm the linkage. To screen for a mutation within the coding region of the candidate gene, Pde6a, which mapped to the region, we designed PCR primers based on complementary and genomic DNA sequences of the gene. Both cDNA and genomic DNA were used for templates in PCR reactions. To extract total RNA, wild-type, heterozygous and mutant Figure 9. (A) Histological analysis of retinas from A.B6-Tyr þ /J-Pde6a nmf282/þ , A.B6-Tyr þ /J-Pde6a nmf282/nmf282 , C57BL/6J-Pde6a nmf363/þ and C57BL/6J-Pde6a nmf363/nmf363 reared under standard lighting condition and dark-reared at P14 (A.B6-Tyr þ /J-Pde6a nmf282/þ and A.B6-Tyr þ /J-Pde6a nmf282/nmf282 ) or P19 (C57BL/6J-Pde6a nmf363/þ and C57BL/6J-Pde6a nmf363/nmf363 ). All sections were stained with hematoxylin-eosin and photographed at the same magnification (200Â). Scale bar represents 50 mm. (B) ONL (i), IS (ii) and OS (iii) thickness was measured at P14 (A.B6-Tyr þ /J-Pde6a nmf282/þ and A.B6-Tyr þ / J-Pde6a nmf282/nmf282 ) or P19 (C57BL/6J-Pde6a nmf363/þ and C57BL/6J-Pde6a nmf363/nmf363 ). The ONL of dark-reared C57BL/6J-Pde6a nmf363/nmf363 mice at P19 was thicker than those reared under standard lighting conditions. The thickness of the ONL at P14 in dark-reared A.B6-Tyr þ /J-Pde6a nmf282/nmf282 mice was almost the same as that of mutants reared under standard lighting conditions. Ã P , 0.05, between heterozygote and homozygote by Student's t-test. (C) The a-wave (i) and b-wave (ii) amplitudes of dark-adapted electroretinograms of C57BL/6J-Pde6a nmf363/nmf363 reared under standard lighting conditions (n ¼ 4) and dark-reared C57BL/6J-Pde6a nmf363/nmf363 at P21 (n ¼ 4). While the ERG a-wave is extinguished in C57BL/6J-Pde6a nmf363/nmf363 mutants reared under standard lighting conditions, a-wave amplitude is clearly observed in dark-reared C57BL/6J-Pde6a nmf363/nmf363 mutants. The b-wave amplitude of the dark-reared animals is also larger than that of the animals reared under standard lighting conditions. retinas at postnatal day (P) 10 were excised and treated with Trizol reagent (Invitrogen, Carlsbad, CA, USA). Respective cDNA pools were generated by reverse transcription from total RNA using the Retroscript w kit (Ambion, Austin, TX, USA) as instructed by the manufacturer. For PCR amplifications, 25 ng of DNA was used in a 25 ml volume using Ampli-Taq DNA polymerase (Applied Biosystems, Foster City, CA, USA), Advantage 2 polymerase mix (Clontech, Mountain View, CA, USA) or HotMaster taq (Eppendorf AG, Hamburg, Germany) according to the manufacturer's instruction. PCR products were either separated by electrophoresis on a 1% agarose gel or 4% MetaPhor (Cambrex, East Rutherford, NJ, USA) agarose gel, and visualized under UV light after staining with ethidium bromide, or used for direct sequencing. DNA sequencing was performed using the Applied Biosystems 3730xl DNA Analyzer (Applied Biosystems, Foster City, CA, USA) by the DNA sequencing service in The Jackson Laboratory. After the mutation was identified, mutagenically separated PCR (35) assays were carried out to determine the genotype of the animals. Table 1 shows the sequences of the primer sets and annealing temperatures used.

Histological methods
Eyes from heterozygous and mutant mice sacrificed at P11 through P36 by carbon dioxide asphyxiation were enucleated, and fixed overnight at 48C in a solution of methanol:acetic acid (3:1) diluted to 50% with 1X phosphate buffered saline (PBS). Fixed eyes were embedded in paraffin, and sectioned at 4 mm intervals. After staining with hematoxylin and eosin, the sections were examined for photoreceptor morphology and scored for degree of photoreceptor degeneration by assessing the number of photoreceptor cell bodies present. A Polaroid DMC1 digital camera (Polaroid, Waltham, MA, USA) connected to a Leica DM LB microscope (Leica Microsystems,

Measurement of cGMP content in retina
After mice were dark-adapted for at least 2 h, they were sacrificed by carbon dioxide inhalation, and their eyes were enucleated. Retinas were dissected under dim red illumination and homogenized in a 6% trichloroacetic acid solution to extract cGMP. The samples were centrifuged at 16 000g for 10 min at 48C. Both the supernatant and the pellet were saved for the cGMP and protein assays, respectively. Trichloroacetic acid in the supernatant was then removed with watersaturated diethyl ether. The levels of cGMP in the samples were determined by radioimmunoassay using a cGMP assay kit (RPA525, GE Healthcare, Buckinghamshire, UK) according to the manufacturer's instructions. The total protein in the retinas precipitated by trichloroacetic acid was determined using a micro BCA protein assay kit (Pierce Biotechnology, Rockford, IL, USA).

Measurement of cGMP-PDE activity
Retinas were dissected and immediately homogenized in 500 ml of PDE assay buffer comprised of 40 mM Tris -HCl (pH 7.4), 5 mM MgCl 2 , 1 mM DTT, 0.1 mM sodium orthovanadate, 50 mM sodium fluoride, 0.1 mM PMSF and 1 mM leupeptin. The homogenate was centrifuged at 16 000g for 30 min at 48C. The pellet resuspended in 200 ml of PDE assay buffer served as retinal crude membrane fraction. For trypsinization experiments, the samples were incubated with 10 mg/ml L-1-Tosylamide-2-phenylethyl chloromethyl ketone-treated trypsin (Sigma) at 48C for 30 min, and the reactions were stopped with the addition of 60 mg/ml soybean trypsin inhibitor (Sigma). PDE activity was measured using an assay mixture containing 0.4 mM [ 3 H]-cGMP and 10-30 mg of cytosolic and retinal crude membrane fraction in a total volume of 100 ml. Samples were incubated for 15 min at 378C and the reaction was terminated by incubation at 1008C for 3 min. Calf intestine alkaline phosphatase, 0.4 U/sample (New England Biolabs, Ipswich, MA, USA) was added to each sample to cleave the phosphate group from [ 3 H]-5 0 -GMP generated by PDE. After incubation for 10 min at 378C, 1.0 ml of a suspension of AG1-X2 resin (50 -100 mesh, Cl 2 form, mixed 1:3 with distilled water; Bio-Rad, Hercules, CA, USA) was added to each sample to remove unreacted [ 3 H]-cGMP and tubes were first rotated for 1 h at room temperature, and then centrifuged for 5 min at 3500g. The radioactivity contained in 500 ml of supernatant was assessed in a liquid scintillation counter. A micro BCA protein assay kit (Pierce Biotechnology) was used to determine the total protein contained in each sample. PDE activity was calculated using the retention rate of guanosine determined by spectrophotometry.

Guanylate cyclase assay
Retinas were dissected and immediately homogenized in 500 ml of GC homogenizing buffer comprised of 50 mM Tris -HCl (pH 7.4) and 3.3 mM MnSO 4 . The homogenate was centrifuged at 16 000g for 30 min at 48C. The pellet resuspended in 200 ml of GC homogenizing buffer served as the crude retinal membrane fraction. To measure GC basal activity, 100-300 mg of crude retinal membrane was incubated in 50 mM Tris -HCl (pH 7.4), 3.3 mM MnSO 4 , 100 mM 3-Isobutyl-1-methylxanthine (Sigma), 40 mM phosphocreatine (Sigma), 75 mg creatine phosphokinase (Sigma) and 1 mM GTP. Total assay volume was 100 ml. To measure background levels of cGMP, samples in which 20 ml of 30 mM EDTA was added were also prepared. Samples were incubated for 10 min at 378C and the reaction was terminated by adding 400 ml of 6% trichloroacetic acid. The samples were centrifuged at 16 000g for 10 min at 48C. Both the supernatant and the pellet were saved for cGMP and protein assays, respectively. Trichloroacetic acid in the supernatant was then removed with water-saturated diethyl ether. The levels of cGMP in the samples were determined by radioimmunoassay using a cGMP assay kit (RPA525, GE Healthcare) according to the manufacturer's instructions. The total amount of the pellet precipitated by trichloroacetic acid was determined using a micro BCA protein assay kit (Pierce Biotechnology). The GC activity was calculated as follows; the activity (pmol/mg protein/min)¼(cGMP content in the intact sample/protein content in the intact sample2background GMP content/ protein content in the background sample)/10.

Electroretinography
After dark-adaptation for at least 2 h, mice were anesthetized with ketamine (80 mg/kg) and xylazine (16 mg/kg). The pupils were dilated by the administration of 1% cyclopentolate. Body temperature was maintained at 378C using a heating pad. Retinal responses were recorded as described previously (36).

Statistical analysis
Values from the quantitative assays are represented as mean + SEM. Student's t-test was used to determine if the differences between experimental groups were significant. Tukey -Kramer test was used for multiple comparisons. A statistical difference was considered significant at P , 0.05.

SUPPLEMENTARY MATERIAL
Supplementary Material is available at HMG Online.