Retinitis pigmentosa: impact of different Pde6a point mutations on the disease phenotype

Abstract

Mutations in the PDE6A gene can cause rod photoreceptors degeneration and the blinding disease retinitis pigmentosa (RP). While a number of pathogenic PDE6A mutations have been described, little is known about their impact on compound heterozygous situations and potential interactions of different disease-causing alleles. Here, we used a novel mouse model for the Pde6a R562W mutation in combination with an existing line carrying the V685M mutation to generate compound heterozygous Pde6a V685M/R562W animals, exactly homologous to a case of human RP. We compared the progression of photoreceptor degeneration in these compound heterozygous mice with the homozygous V685M and R562W mutants, and additionally with the D670G line that is known for a relatively mild phenotype. We investigated PDE6A expression, cyclic guanosine mono-phosphate accumulation, calpain and caspase activity, in vivo retinal function and morphology, as well as photoreceptor cell death and survival. This analysis confirms the severity of different Pde6a mutations and indicates that compound heterozygous mutants behave like intermediates of the respective homozygous situations. Specifically, the severity of the four different Pde6a situations may be categorized by the pace of photoreceptor degeneration: V685M (fastest) > V685M/R562W > R562W > D670G (slowest). While calpain activity was strongly increased in all four mutants, caspase activity was not. This points to the execution of non-apoptotic cell death and may lead to the identification of new targets for therapeutic interventions. For individual RP patients, our study may help to predict time-courses for Pde6a-related retinal degeneration and thereby facilitate the definition of a window-of-opportunity for clinical interventions.

Introduction

Retinitis pigmentosa (RP) is a hereditary neurodegenerative disease of the retina which affects photoreceptors and is a major cause of early-onset blindness in the industrialized world (1).

The genetic mutations triggering RP usually lead to a disturbance of the phototransduction cascade, often associated with an elevation of cyclic guanosine mono-phosphate (cGMP) (2). In the dark, high guanylate cyclase activity increases cGMP levels (3). Phototransduction starts with a light-induced, conformational change of opsin molecules, causing sequential activation of transducin and phospho-diesterase-6 (PDE6), the latter of which subsequently reduces the intracellular cGMP concentration. High cGMP levels maintain cyclic-nucleotide-gated (CNG) cation channels in the open state, allowing Ca²⁺ influx (4). PDE6 activation reduces cGMP levels, thus closing CNG channels and causing hyperpolarization and signal transmission to second order neurons.

Genetic mutations affecting PDE6 function lead to an excessive accumulation of cGMP and subsequent rod photoreceptor death (5,6), followed by a mutation independent, secondary death of cone photoreceptors (7). In rod photoreceptors, PDE6 is composed of two catalytic subunits—alpha (A) and beta (B)—and in the inactive state associates with two inhibitory gamma subunits. When all mutations affecting any of the three PDE6 alpha, beta or gamma subunits are considered together, these are responsible for up to 4–8% of human RP patients (8–10). Previously, PDE6 dysfunction was investigated primarily in animal models affected by mutations in the Pde6b gene, such as the rd1 (11) or the rd10 mouse (12). A variety of Pde6a mutations are also known to cause RP (13) but, so far these have been relatively little studied.

RP animal models are normally studied in the homozygous state (e.g. as Pde6b^rd1/rd1) even though, in human RP patients, in outbred populations, homozygosity is relatively rare and, in fact, compound heterozygosity, where two different disease-causing alleles come together in one individual, is more frequent (e.g. PDE6A^V685M/R562W). To account for this, we have used various Pde6a mutant mice and studied the progression of retinal degeneration both in the homozygous and compound heterozygous situations. To facilitate a comparison with the human situation, we focussed on Pde6a mouse mutants for which homologous RP patients have previously been identified. This relates to point mutations resulting in an amino acid exchange in various positions of the PDE6A protein, i.e. R562W, D670G and V685M. For the sake of clarity, in the following these animal models will be referred to by the position of their respective mutations.

The previously reported Pde6a D670G and V685M mouse mutants have been generated by N-ethyl-N-nitrosourea (ENU) mutagenesis and were identified through a screen for altered fundus pigmentation (13). In this study, we generated and studied an additional Pde6a R562W-knock-in mutant. The generation of this mutant followed the identification of a human RP patient compound heterozygous for the c.1684C>T/p.Arg562Trp and c.2053G>A/p.Val685Met mutations in PDE6A. Therefore, the generation of the R562W-knock-in mutant gave the opportunity to study the exact homologous genotype of the patient through crossbreeding of the R562W-knock-in with the V685M mutant. We then used both homozygous and compound heterozygous Pde6a mutant animals to assess the relative impact of different genetic insults on the progression and the severity of retinal degeneration. The data generated here may serve as a reference for further pre-clinical studies in RP animal models and—extrapolated to the human situation—may guide future clinical trials for RP therapy development.

Material and Methods

Animals

All Pde6a mutants used were generated and maintained on the C57BL6/J strain background (wild type; wt), were housed under standard white cyclic lighting, had free access to food and water and were used irrespective of gender. The V685M (A.B6-Tyr+/J-Pde6a^{nmf282/nmf282}) and D670G (C57BL/6J- Pde6a^nmf363/363) animals were obtained from the Jackson Labs (Bar Harbor, MA, USA).

The R562W-knock-in mutant was generated by GenOway (Lyon, France) using standard procedures of homologous recombination in murine embryonic stem (ES) cells. Briefly, left arm and right arm homology fragments of 4066 bp (covering exon 10 to exon 12 and parts of the flanking introns of Pde6a) and of 3830 bp (covering exon 13 and parts of the flanking introns of Pde6a), respectively, were polymerase chain reaction (PCR) amplified from a C57BL/6-derived BAC clone carrying Pde6a sequences. To generate the knock-in mutation the ‘CGG’ arginine codon was replaced by a ‘TGG’ tryptophan codon at the respective position in exon 13 in the right arm homology fragment by in vitro mutagenesis. The cloned fragments were verified by Sanger sequencing and then assembled for the final targeting construct that comprised a neomycin positive selection cassette flanked by loxP sites and inserted in intron 12 between the left and right arm homology fragments, and a diphtheria toxin expression cassette for negative selection. The targeting construct was electroporated into C57BL/6-derived ES cells according to GenOway's electroporation procedures (i.e. 10⁸ ES cells in the presence of 100 µg of linearized plasmid, 260 V, 500 µF). Positive selection was started 48 h after electroporation, by addition of 200 µg/ml of G418 (150 µg/ml of active component, Life Technologies, Inc.). Resistant clones were isolated and amplified in 96-well plates. Duplicates of 96-well plates were made. The set of plates containing ES cell clones amplified on gelatine were genotyped by both PCR and Southern blot analysis and the presence of the mutation was verified by sequencing. One fully characterized ES clone was used for injection into blastocysts of C57BL/6J-Tyr^c-2J/J mice and two highly chimeric male mice were obtained. These mice were mated with C57BL/6 cre deleter female mice to excise the neomycin selection cassette. The F1 progeny was tested for proper excision of the neomycin cassette by means of PCR and Southern blotting. Since the used C57BL/6-derived ES cells carry the rd8 allele (14), we mated the F1 with C57BL/6J animals and then crossbred F2 animals to obtain homozygous R562W mice devoid of rd8.

For all animals used in this study, day of birth was considered as post-natal day (P) 0. All procedures carried out on animals were reviewed and approved by the competent authority (Regierungspräsidium Tuebingen). All efforts were made to minimize the number of animals used and their suffering.

Functional analysis of the p.R562W mutant gene

We used an established system for the functional expression of PDE6 protein based on a PDE6C/PDE5 fusion construct and the use of the baculovirus/Sf9 insect cell system for recombinant protein expression (15,16). The mutated construct bearing a tryptophan codon instead of an arginine codon at the site homologous to p.R562 of Pde6a was generated by in vitro mutagenesis (Quik Change in vitro Mutagenesis Kit from Agilent, Waldbronn, Germany). Generation of recombinant bacmids, transfection of Sf9 insect cells and viral amplifications were carried out according to the manufacturers’ recommendations (Life Technologies/Invitrogen, Darmstadt, Germany). Lysis of cells and purification of recombinant protein was carried out as previously described (15) and PDE activity was measured using [³H]-cGMP (GE Healthcare, Munich, Germany) as substrate (17). Briefly, 5 µg purified wt and mutant protein, respectively, were incubated in a total volume of 40 µl of 20 mm Tris–HCl pH8.0, 50 mm NaCl, 15 mm MgSO₄, 2 mm β-mercaptoethanol, 0.1 µM shrimp alkaline phosphatase and 5 µM [³H]cGMP (100 000 cpm) for 10 min at room temperature. The reaction was stopped by the addition of 0.5 ml AG1-X2 anion exchange resin (Bio-Rad) in a 20% bed volume suspension. Samples were incubated with the resin for 10 min with occasional vortexing and then centrifuged at 9000g for 2 min. Two hundred and fifty microliter aliquots of the clear supernatant were removed and measured in a scintillation counter (Beckman LS 6000, Beckman Coulter GmbH, Krefeld, Germany).

RT-PCR and allelic quantification

Homozygous and heterozygous Pde6a R562W-knock-in mice were sacrificed at the age of P14 and P80, respectively. In addition, we used P13 and adult C57BL6 wt mice as control. Retinas were dissected from enucleated eyes and used to prepare total RNA. The tissue was lysed mechanically in a Precellys homogenizer (Peqlab Biotechnologie GmbH, Erlangen, Germany) and total RNA isolated through affinity chromatography on silica membrane (peqGold Total RNA Kit; Peqlab Biotechnologie GmbH).

Single-stranded cDNA was synthesized from 1–2 µg of total RNA by reverse transcription applying random hexamers for priming (Transcriptor First Strand cDNA Synthesis Kit, Roche, Mannheim, Germany). For qualitative reverse transcriptase-polymerase chain reaction (RT-PCR), we amplified 1/10 volume of the first-strand cDNA with primers MmPde6a_cDNA_ex12Fnw (5-ACGCGGAGTCATACGAAATC-3) and MmPde6a_cDNA_ex15Rnw (5-ATGATGCCTTTCCAAGATGG-3) or Pde6a_cDNA_Ex11 (5′-AGAGGTGTACGGCAAAGAGC-3′) and Pde6a_cDNA_Ex14 (5′-GTTGTTCGTGCCTCTGTGGT-3′) applying standard PCR conditions.

RT-PCR products were cloned into pCR2.1 using the TA Cloning Kit (Invitrogen/Life Technologies) and plasmid DNA purified from single bacterial clones was sequenced applying cycle sequencing and BigDye Terminator V1.1 chemistry (Applied Biosystems/Life Technologies). Sequencing products were separated on an ABI 3130XL capillary sequencer (Applied Biosystems/Life Technologies). Raw sequences were processed using Sequencing Analysis Software V5.2 (Applied Biosystems/Life technologies) and assembled into contigs using SeqMan (Lasergene, Madison, WI, USA).

Relative quantification between the R562W-knock-in allele and the wt allele was done by pyrosequencing. For relative quantification of allelic Pde6a transcripts, we amplified 1/10 volume of first-strand retinal cDNA of heterozygous mutant mice with primers Pde6a_Mm_Ex12–13_F (5′-TTCCACATCCCGCAAGAG-3′) and a 5′-biotinylated reverse primer Pde6a_Mm_Ex12–13_R (5-Bio-CCAGCAAGGAGAACATGGTC-3′) applying standard PCR conditions. For comparison, we amplified an exon 13 internal fragment of the Pde6a gene from genomic DNA of heterozygous R562W-knock-in mice (extracted from ear punches) with primers Pde6a_Mm_Ex13-DNA_F (5′-GTTCACAGGCCCTGGTGC-3′) and 5′ biotinylated Pde6a_Mm_Ex12–13_R (see above). PCR products were purified by immobilization on streptavidin coated sepharose beads (GE Healthcare) and vacuum filtration (Vacuum Manifold, Qiagen, Hilden, Germany). Single-stranded PCR products were sequenced with primer Pde6a_Mm_Ex12–13_S (5′-GAATCACTTACCACAACTG-3′) on a Pyromark Q96 instrument according to the manufacturer's recommendations (Qiagen).

Minigene splicing assays

Minigene constructs were generated by cloning both allelic products of a PCR with primers EcoRI-PDE6A-MMf (5′-aaGAATTCTGTATCAGTACGACCCAAGAC-3′; EcoRI recognition site underlined) and BamHI-PDE6A-MMr (5′-aaGGATCCGAGTTGTATACTTCTCTATTCTTGGAA-3′; BamHI recognition site underlined) with genomic DNA of a heterozygous R562W-knock-in mutant into the EcoRI and BamHI sites of the pSPL3 vector. The amplified fragment from the wt allele encompasses exon 13 of Pde6a, 424 bp of upstream intron 12 sequence, and 308 bp of downstream intron 13 sequence. The fragment from the knock-in allele covers an additional 96 bp insert (comprising a single loxP site and flanking multiple cloning site sequences derived from the targeting vector) inserted 247 bp upstream of exon 13. Further constructs were obtained by introducing the c.1684C>T/p.R562W mutation into the wt allele construct and by reverting the mutation in the knock-in allele construct through in vitro mutagenesis according to the manufacturer's protocol (Quik Change Mutagenesis Kit; Agilent). The inserts of the minigene constructs were verified by Sanger sequencing as outlined above. Purified plasmid DNA of the constructs were used to transfect HEK293 and 661W as follows: Cells were seeded in 6-well plates in Dulbecco's modified Eagle's medium (DMEM) (Gibco/Life Technologies) with 10% fetal bovine serum (Biochrom GmbH, Berlin, Germany) and the following day, at 80–90% confluency, cells were transfected with 4–10 µg plasmid DNA of the minigene construct using 20 µl Lipofectamine 2000 per well and OptiMEM^® supplemented with GlutaMAX^TM (Gibo/Life Technologies) as diluents and medium. After 6 h incubation, cells were harvested by trypsinization with 0.05% trypsin-ethylenediaminetetraacetic acid (Gibco/Life Technologies), centrifuged at 1500 rpm for 5 min and transferred to a 6 cm dish with DMEM supplemented with 10% fetal calf serum and antibiotics (10 ml/l of penicillin–streptomycin solution [P4333; Sigma-Aldrich Chemie GmbH, Munich, Germany] and 10 ml/l Amphotericin B [250 µg/ml in water; Biochrom AG]). Twenty-four hours post-transfection, cells were lysed and total RNA was extracted applying the peqGOLD Total RNA Kit (Peqlab Biotechnologie GmbH). Single-stranded cDNA was synthesized as outlined above and cDNA was amplified with primers SA2 (5′-ATCTCAGTGGTATTTGTGAGC-3′) and SD6 (5′-TCTGAGTCACCTGGACAACC-3′) applying standard PCR conditions. RT-PCR products obtained from transfected HEK293T cells were cloned into pCR2.1 using the TA Cloning Kit (Invitrogen/Life Technologies) and plasmid DNA isolated from single bacterial clones sequenced as outlined above.

Histology, immunohistochemistry and immunofluorescence

Animals were sacrificed in the morning (10–11 am), their eyes enucleated and fixed in 4% paraformaldehyde (PFA) in 0.1 M phosphate buffer (pH 7.4) for 45 min at 4°C. PFA fixation was followed by cryoprotection in graded sucrose solutions (10, 20, 30%). Unfixed eyecups were directly embedded in cryomatrix (Tissue-Tek, Leica, Bensheim, Germany). Sagittal 12 µm sections were obtained and stored at −20°C.

For immunofluorescence, sections were incubated overnight at 4°C with primary rabbit antibody against PDE6A (Novus Biologicals, Cambridge, UK; # NBP1–87312), rabbit antibody against caspase-3 (Cell Signaling, Danvers, MA; #9664), or sheep antibody against cGMP (1:400; kindly provided by Harry Steinbusch, Maastricht, The Netherlands), then washed in PBS and incubated for 1 h with Alexa Fluor 488-conjugated, matching secondary antibodies (Molecular Probes, Inc. Eugene, USA). Negative controls were carried out by omitting the primary antibody. Sections were mounted with Vectashield (Vectorlabs, Burlingame, CA, USA) for microscopy.

Calpain activity and TUNEL assay

Calpain activity was investigated using an enzymatic in situ assay (18). Unfixed cryosections were incubated for 15 min in calpain reaction buffer (CRB; 25 mm HEPES, 65 mm KCl, 2 mm MgCl2, 1,5 mm CaCl2, 2 mm DTT) and then incubated at 35°C for 1 h in the dark in CRB with 2 mm fluorescent calpain substrate 7-amino-4-chloromethylcoumarin, t-BOC-Leucyl-l-methionine amide (Life technologies, Darmstadt, Germany; #A6520). Fluorescence was uncaged by calpain-dependent cleavage of t-Boc-Leu-Met-CMAC.

Terminal deoxynucleotidyl transferase dUTP nick end labelling (TUNEL) assay was performed using an in situ cell death detection kit (Fluorescein or TMR; Roche Diagnostics GmbH, Mannheim, Germany). For controls terminal deoxynucleotidyl transferase, enzyme was either omitted from the labelling solution (negative control), or sections were pre-treated for 30 min with DNAse I (Roche, 3 U/ml) in 50 mm Tris–HCl, pH 7.5, 1 mg/ml BSA to induce DNA strand breaks (positive control). While negative control gave no staining, positive control stained all nuclei in all layers of the retina (19).

Microscopy, cell counting and statistical analysis

Light and fluorescence microscopy were usually performed at room temperature on an Axio Imager Z.1 ApoTome Microscope, equipped with a Zeiss Axiocam MRm digital camera. Images were captured using Zeiss Axiovision 4.8 software; representative pictures were taken from central areas of the retina using a 20×/0.8 Zeiss Plan-APOCHROMAT objective. Adobe Photoshop CS3 (Adobe Systems Incorporated, San Jose, CA) was used for primary image processing.

For the quantifications of positively labelled cells (TUNEL, cGMP), pictures were captured on three entire sagittal sections for at least three different animals for each genotype and age using Mosaic mode of Axiovision 4.8 at 20× magnification. The average area occupied by a photoreceptor cell (i.e. cell size) for each genotype and age was determined by counting DAPI-stained nuclei in nine different areas (50 × 50 µm) of the retina. The total number of photoreceptor cells was estimated by dividing the outer nuclear layer (ONL) area by this average cell size. The number of positively labelled cells in the ONL was counted manually. We considered cells as positively labelled only if they showed a strong staining of either the photoreceptor nuclei (e.g. for TUNEL) or perinuclear areas (e.g. for cGMP, caspase-3). Values obtained are given as fraction of total cell number in ONL (i.e. as percentage) and expressed as mean ± standard error of the mean (SEM).

For the quantification of PDE6A protein expression, fluorescent pictures were loaded into ImageJ (Vers. 1.44; Wayne Rasband, National Institute of Mental Health, Bethesda, MA) and the line plot option was used to obtain maximum intensity values for the ONL and outer segment (OS) areas. The (unstained) ONL values were taken for background subtraction, the average pixel intensity in wt OS was arbitrarily set to 1; the mutant values were expressed as a function of that. For statistical comparisons, the unpaired Student t-test as implemented in Prism 5 for Windows (GraphPad Software, La Jolla, CA) was employed.

Non-invasive in vivo imaging and functional testing

A baseline characterization with electroretinography (ERG), spectral domain optical coherence tomography (SD-OCT) and scanning-laser ophthalmoscopy (SLO) was performed at 4 weeks of age. The groups were composed of four animals of each genotype.

Electroretinography

ERGs were recorded binocularly from different Pde6a mutants as described previously (20). Mice were dark-adapted overnight and anaesthetized using a combination of ketamine (66.7 mg/kg body weight) and xylazine (11.7 mg/kg body weight). Their pupils were dilated and single-flash ERG responses were obtained under scotopic (dark-adapted) and photopic (light-adapted with a background illumination of 30 cd/m², starting 10 min before recording) conditions. Single white-flash stimuli ranged from −4 to 1.5 log cd·s/m² under scotopic and from −1.0 to 1.5 log cd·s/m² under photopic conditions. Ten responses were averaged with inter-stimulus intervals of 5 s (for −4 to −0.5 log cd·s/m²) or 17 s (for 0 to 1.5 log cd·s/m²).

Spectral-Domain optical coherence tomography

Retinal structures of the still anesthetized animals were visualized via OCT imaging with a Spectralis^TM HRA + OCT (Heidelberg Engineering GmbH, Heidelberg, Germany). This device features a superluminescent diode at 870 nm as low coherence light source. Scans are acquired at a speed of 40 000 scans per second and each two-dimensional B-scan contains up to 1536 A-scans (21–23). The images were taken with the equipment set of 30° field of view and with the software Heidelberg Eye Explorer (HEYEX version 5.3.3.0, Heidelberg, Germany). Resulting images were exported as 8-bit colour bitmap files and processed with CorelDraw X3 (Corel corporation, Ottawa, ON, Canada).

Scanning-laser ophthalmoscopy

Eyes were kept moisturized with Methocel (Omnivision, Puchheim, Germany) so that SLO imaging was performed in the same session as OCT. It was carried out with a HRA 1 system (Heidelberg Engineering) according to previously described procedures (24). Briefly, the HRA 1 system features lasers in the short (visible) wavelength range (488 nm in both and 514 nm), and also in the long (infrared) wavelength range (795/830 nm and 785/815 nm). The 488 and 795 nm lasers are used for fluorescein (FLA) and indocyanine green (ICG) angiography, respectively. GFP excitation was detected in the autofluorescence mode at 488 nm with a 500 nm barrier filter.

Results

Functional analysis of the p.R562W mutation in PDE6A

To verify the pathogenicity and to assess the functional consequence of a recently identified new missense variant (p.R562W) in the PDE6A gene, we expressed the mutant PDE6 protein in Sf9 insect cells by using an established chimeric PDE6C/PDE5 construct (15). Purified recombinant protein was analyzed for cGMP hydrolysis activity. In comparison with wt protein, the catalytic activity of R562W mutant protein was reduced to ∼10% (Fig. 1).

Figure 1.

The R562W mutant PDE6 protein shows reduced catalytic activity. cGMP hydrolysis activity of wild-type and R562W mutant chimeric PDE6C/PDE5 protein was determined in vitro on purified recombinant protein. The bars were normalized to wild-type enzymatic activity (100%).

Generation of a new mouse model for the R562W mutation

In order to study the retinal phenotype caused by the R562W mutation, we generated a Pde6a:R562W-knock-in mouse mutant applying state-of-the art homologous recombination in mouse ES cells. The knock-in allele bore the actual c.1684C>T/p.R562W mutation in exon 13 and, in addition, upstream, in intron 12, remnant sequences of the targeting process including a single loxP site flanked by 62 bp of sequence of the multiple cloning site derived from the targeting vector. We used heterozygous R562W-knock-in animals to test for proper expression of the mutant allele in the retina. Upon allelic quantification of Pde6a RT-PCR products, we found an excess of wt allele derived transcripts in heterozygous animals (Fig. 2A).

Figure 2.

Pde6a cDNA analysis in the R562W-knock-in mouse mutant. (A) Pyrosequencing-based relative quantification of wild-type and R562W knock-in Pde6a alleles at the genomic DNA level (top) and in reverse transcribed retinal cDNA of a heterozygous Pde6a:R562W-knock-in mutant. Note that the assayed variant nucleotide position is the actual c.1684C>T mutation and that allelic quantification of cDNA was done on RT-PCR products with primers in exons 12 and 13. (B) Qualitative analysis of the integrity of retinal Pde6a transcripts in the Pde6a:R562W-knock-in mouse mutant. RT-PCR products with primers in exons 12 and 15 of Pde6a were amplified from retinal RNA of a homozygous Pde6a:R562W-knock-in mutant and a C57BL/6 wild-type control. The larger sized product of 403 bp is derived from correctly spliced transcripts and the smaller product represent mis-spliced transcripts lacking exon 13. Controls ‘no RNA’: w/o RNA in the DNaseI digestion and subsequent RT and PCR reactions, ‘RTC’: w/o RNA in the cDNA synthesis and subsequent PCR reaction, ‘NTC’: no template control for the PCR reaction. One hundred base pair ladder size standard is shown in the rightmost lane.

Further RT-PCR experiments with primers located in exon 12 and 14 revealed two major products in homozygous R562W mutants. The larger RT-PCR product was identical in size (403 bp) to the RT-PCR obtained from control animals, and the other ∼100 bp smaller (Fig. 2B). Sequencing of cloned products from homozygous R562W mutant animals revealed that the smaller cDNA fragment lacks exon 13. Additionally, we performed RT-PCR experiments in heterozygous R562W animals. Sequencing of cloned RT-PCR products showed the absence of exon 13 in the smaller cDNA fragment while the larger cDNA fragment represented a mixture of correctly spliced wt and knock-in allele derived transcripts, respectively (Supplementary Material, Fig. S1). These findings indicated an incompletely penetrant splicing defect caused by the knock-in allele. Bioinformatic analysis suggested that the C>T transition created an exonic splicing silencer (ESS) site which most likely explained the impaired exon 13 splicing (data not shown). Skipping of exon 13 resulted in an in-frame deletion of the open reading frame and translated into a PDE6A protein shortened by 36 amino acid residues (p.541_576del), which lacks the amino-terminal portion of the catalytic domain including the first metal binding motif.

Since we could not rule out an effect of the resident modifications (loxP site and multiple cloning site sequences; loxP/MCS insert) in intron 12 in the knock-in allele on transcript splicing, we generated minigene constructs that include the murine exon 13 and 424 bp of upstream intron 12 and 309 bp of downstream intron 13 sequences inserted in to the pSPL3 exon trapping vector. Four different variants of the minigene constructs were cloned: (i) wt, (ii) wt exon 13 and the 96 bp loxP/MCS insert in intron 12, (iii) c.1684C>T/p.R562W mutation in exon 13 with wt intron 12 portion, and (iv) the knock-in allele (c.1684C>T/p.R562W and loxP/MCS insert in intron 12) (Supplementary Material, Fig. S2). Constructs were transfected into HEK293T cells and total RNA was prepared 24 h post-transfection. RT-PCR for the expression construct with primers located in flanking exons of the pSPL3 vector indicated that the splicing defect observed in the knock-in mouse mutant was caused by the c.1684C>T/p.R562W mutation and not primed or influenced by the modifications in intron 12 (Supplementary Material, Fig. S3). Given that a considerable proportion of exon 13 skipped transcripts were also expressed from the wt minigene construct in HEK293T cells, we re-performed minigene assays for the c.1684C>T/p.R562W mutant construct in the mouse cone photoreceptor-like cell line 661W. RT-PCR with RNA from transfected 661W cells showed a strongly reduced proportion of transcripts with exon 13 skipping from the wt minigene construct and roughly equal proportions of correctly spliced and exon 13 skipped transcripts from the mutant construct (Supplementary Material, Fig. S4). This corresponded to the findings in mutant mouse retina (Fig. 2B).

Decreased PDE6A expression causes cGMP accumulation and photoreceptor death

We first assessed the expression of the PDE6A protein in the retina of wt and Pde6a mutants, at post-natal day (PN) 11, i.e. a time-point just before the onset of photoreceptor degeneration and widespread destruction of the ONL. In wt retina PDE6A is present almost exclusively in photoreceptor OS. In contrast, in V685M, V685M/R562W and R562W PDE6A is essentially undetectable, with a faint, dot-like staining pattern in R562W OS indicative of a very low PDE6A protein expression. In D670G retina, PDE6A protein is detectable in OS; however, the amount of protein detected is clearly lower than in wt (Fig. 3A–E).

Figure 3.

Loss of PDE6A expression causes cGMP accumulation. In the PN11 wild-type (wt) retina (A), immunostaining for PDE6A shows strong protein expression in photoreceptor OS. In contrast, at the same age, in V685M retina (B), the PDE6A protein is undetectable. Similarly, in the compound heterozygous V685M/R562W retina (C), the protein is essentially absent. In R562W retina, (D) small immunoreactive dots at the level of the OS may indicate a minimal protein expression. In the D670G mutant (E), however, there is a clear PDE6A protein expression; albeit at reduced levels when compared with wt. At PN11, wt retina is essentially negative for cGMP immunoreactivity (F). All Pde6a mutants, however, display individual rod photoreceptor cells that have accumulated large amounts of cGMP (G–J). The quantification of cGMP-positive cells in the ONL and the PDE6A pixel intensity in the OS (arbitrary units; AU) shows an inverse correlation (K). Images are representative for immunostaining performed on retinal sections from at least three independent animals for each genotype.

The decreased detection of PDE6A in mutant retinas correlated with an increased accumulation of cGMP (Fig. 3F–J). Wt retina did not show signs of important cGMP accumulation (wt: 0.003% cGMP-positive cells ± 0.003 SEM, n = 6), while in all four Pde6a mutants, the numbers of rod photoreceptors showing high levels of cGMP was elevated already at PN11 (V685M: 0.61% ± 0.04; V685M/R562W: 0.19% ± 0.06; R562W: 0.17% ± 0.08; D670G: 0.007% ± 0.002; n = 3 for all).

To further assess the progression of retinal degeneration in the various Pde6a mutants, we used the TUNEL assay to label dying photoreceptors and, conversely, quantified the number of rows of surviving photoreceptors (Fig. 4). In the V685M ONL, the percentage of TUNEL-positive cells peaks already at PN12, something that is also evidenced by the early and rapid loss of photoreceptor rows (Fig. 4B, G and L). In the compound heterozygous V685M/R562W and in the R562W, cell death peaked at PN15 although the peak amplitude was slightly lower in the R562W mutant (Fig. 4C, D, H, I, M and N). When compared with the V685M situation, this corresponded to a delay in the onset and progression of degeneration of ∼2 days in V685M/R562W retina and 4 days in R562W retina. The D670G mutant showed a relatively small peak of cell death at PN21 and also the slowest overall progression of degeneration among the four Pde6a mutant genotypes (Fig. 4E, J and O).

Figure 4.

Photoreceptor cell death and survival. The TUNEL assay in the wt retina (A, F and K) occasionally labelled cells dying due to developmental processes. In contrast, in all Pde6a mutants (B–E, G–J and L–O), photoreceptor cell death was dramatically increased. The images show the situation at P12, P15 and P21, time-points corresponding to the peak of cell death for the different models. The line graph at the bottom left (P) illustrates the progression of photoreceptor cell death as evidenced by the TUNEL assay in the different Pde6a mutants. The peak times as well as the peak amplitudes correspond to the speed of retinal degeneration, which is illustrated by the loss of photoreceptors (Q). Images are representative for TUNEL assays performed on retinal sections from at least three independent animals, quantifications in P, Q include data from 3–7 animals per genotype and timepoint.

Retinal degeneration has frequently been associated with the execution of classical apoptosis and the activity of caspase-type proteases (25,26). However, more recent studies point to the activity of non-apoptotic and caspase-independent mechanisms of cell death in hereditary retinal degeneration (27) which may instead rely on the activity of Ca²⁺ activated calpain-type proteases (2,28). To address this question for the Pde6a mutants used here, we performed an immunostaining for activated caspase-3, a key effector in classical apoptosis. Since the chances for a positive detection of caspase-3 activity are highest when cell death is high, we focussed on the peaks of degeneration as assessed by the TUNEL assay. In the early post-natal wt, a low level of caspase-3 activity was present, likely relating to developmental cell death in rodent retina (29). Remarkably, none of the Pde6a mutants showed any significant increase of caspase-3 activity when compared with wt (Fig. 5A–E; quantification in K). An in situ assay for Ca²⁺ activated calpain-type proteases (28) revealed very low levels of activation in wt retina. In Pde6a mutants however, the numbers of photoreceptors cells showing calpain activation was dramatically increased (Fig. 5F–J). When we then assessed the progression of calpain activity over time, we found a strong temporal correlation between calpain activity and cell death (Fig. 5L).

Figure 5.

Photoreceptor degeneration in Pde6a mutants correlates with calpain not caspase activity. Immunofluorescence for activated caspase-3—a key marker for classical apoptosis—reveals no major differences between wild-type and mutant retinae (A–E). This is also illustrated by the quantification of caspase-3-positive cells, which shows only extremely low numbers of cells at the respective peaks of degeneration, with no significant (n.s.) differences to wt (K). In contrast, an in situ calpain activity assay reveals strong differences between wild-type (F) and all four Pde6a mutant genotypes (G–J). The progression of calpain activity was analyzed over time (K) and showed a strong correlation to the extent of cell death and the progression of retinal degeneration (cf. Fig. 4). Images shown are representative for at least three different stainings obtained at the respective mutant's peak of degeneration. Note that the differences in the progression of Pde6a mutant degenerations are also illustrated here by the differences in retinal ONL sizes. Data shown in K and L was obtained from observations made on 4–6 independent specimens for each genotype and timepoint.

Characterization of Pde6a mutants in vivo

The four Pde6a mutant lines were morphologically and functionally characterized in vivo by means of SLO and OCT imaging and ERG recording at the age of 4 weeks. In SLO imaging, overall fundus appearance was visualized with the green laser at 514 nm (Fig. 6A, F, K and P); the retinal vasculature was studied with angiography applying ICG (Fig. 6B, G, L and Q) and fluorescein (Fig. 6C, H, M and R) and the appropriate laser wavelengths (795 and 488 nm, respectively). Retinal layering was studied by means of OCT imaging (Fig. 6D, E, I, J, N, O, S and T). D670G mice revealed a heavily spotted fundus appearance in the native fundus imaging (Fig. 6A) as well as in the fluorescein angiography mode (Fig. 6C) whereas in the OCT analysis, a very thin ONL was visualized depicting the presence of only few photoreceptor rows (Fig. 6D and E). A spotty fundus was also found in the R562W mice (Fig. 6F) together with a further decrease in the retinal thickness where no outer retina was detected (Fig. 6I and J). Compound heterozygous V685M/R562W mice showed a heavily affected retinal fundus with large areas of degeneration (Fig. 6K and M). In these mutants, the retinal thickness was strongly decreased, which resulted in enhanced visibility of choroidal structures in each SLO imaging mode (Fig. 6K–M). Accordingly, the OCT analysis revealed a highly degenerated retina (Fig. 6N and O). V685M was the mouse line with the strongest degeneration, retinal and choroidal structures were difficult to distinguish due to the severe decrease in the retinal thickness (Fig. 6S and T). Altogether, a different degree of retinal degeneration was detected and a gradient based on the severity (from less to more affection) of the disease was established: D670G>R562W>V685M/R562W>V685M.

Figure 6.

Retinal morphology in Pde6a mutants in vivo. D670G (A–E), R562W (F–J), V685M/R562W (K–O) and V685M (P–T) mutant mice were examined with SLO and SD-OCT at PN30. Examinations included fundus native imaging (A, F, K and P), ICG angiography (B, G, L and Q) and fluorescein angiography (C, H, M and R). For the OCT analyses, horizontal scans through the optic disc head are shown (E, J, O and T; magnifications; D, I, N and S). All of the mutants analysed revealed retinal degeneration. D670G mice showed several spots spread over the fundus (A), which were also visible in fluorescein angiography examination (C). Retinal thickness was also affected and a very thin ONL was detected (D and E). A similar fundus appearance (F–H) but a more severe decrease in the retinal thickness was found in the R562W mutants (I and J). V685M/R562W compound heterozygous mice showed large areas of retinal degeneration in native fundus imaging (K). The loss of retinal layers demonstrated in OCT imaging (O and N), allowed the enhanced visualization of choroidal structures (L and M). V685M exhibited a very severe retinal degeneration in both SLO and OCT imaging (P–T). GCL, ganglion cell layer, IPL, inner plexiform layer; INL, inner nuclear layer; ONL, outer nuclear layer; RPE, retinal pigment epithelium; d, dorsal; n, nasal; t, temporal; v, ventral.

The in vivo morphological findings correlated well with the functional data obtained with ERG. Full-field ERG measurements under both scotopic and photopic conditions allow the assessment of retinal function (Fig. 7A and B). Depending on the extent of morphological alterations, different ranges of ERG recordings could be observed. Retinal function was mostly preserved in the D670G animals [blue box and whisker plot (B&W)], greatly reduced in the R562W mice (green B&W) and completely missing in the V685M mutant (red B&W). The V685M/R562W, as an intermediate mutant line, is positioned between the R562W and V685M variants. Scotopic and photopic ERG traces at the highest stimulus intensity (Fig. 7C) were similar to those of a Rho^−/− animal used as a functional control for cone-only responses (20,30). A comparison of the size of ERG amplitudes of different Pde6a mutants with age-matched C57/BL6 wt mice revealed that amplitudes of Pde6a mutants were considerably smaller and represent one-third to one-fourth of wt animals (20). Taken together, these in vivo morphological and functional observations were closely matching the results obtained in ex vivo quantification of photoreceptor cell death and survival (Fig. 4).

Figure 7.

Functional assessment of different Pde6a mutants based on electroretinography. Quantitative evaluation of the scotopic [dark-adapted (A)] and photopic [light-adapted (B)] b-wave amplitude data, acquired at PN30 and shown as box-and whisker-plot (B&W), for the different Pde6a mutant lines. In B&W, boxes indicate the 25–75% quantile range, whiskers the 5 and 95% quantiles and asterisks the medians of the data. Scotopic (top) and photopic (bottom) single-flash ERG at 1.5 log (cd·s/m²) maximal light intensity (C). ERGs from D670G mutant (blue) are similar to ERGs from a Rho^−/− mouse, a valid model for pure cone function. D670G (blue), R562W (green), V685M/R562W (orange) and V685M mice (red).

Discussion

In this report, we investigated the pathologic consequences of three different Pde6a point mutations for retinal photoreceptor degeneration. We analyzed the effects of these mutations on PDE6 activity, photoreceptor cGMP accumulation and progression and mechanisms of retinal degeneration. Previously, a variety of RP animal models have been generated and studied that carried human gene mutations homozygously (31–33). Here, we created an in vivo model carrying two different point mutations (Pde6a^V685M/R562W), exactly homologous to a human RP patient genotype, representing one of the first attempts to create a compound heterozygous, patient-matched basis for the development and assessment of individualized RP therapies.

Loss of PDE6 function, cGMP signalling and cell death

PDE6 is the main photoreceptor cGMP hydrolysing enzyme and displays the highest activity levels among all PDE's (34). Although excessive accumulation of cGMP has been found in various animal models for RP, it is particularly prominent in animals that carry mutations in PDE6 (2). However, it is still unclear to what extent different PDE6 mutations cause cGMP accumulation and how this correlates with progression speed and severity of retinal degeneration. In all Pde6a mutants studied here the genetic defects led to a reduced PDE6 function as evidenced by an accumulation of cGMP. Compared with mutations where PDE6 function is abolished entirely, such as in the rd1 mouse (35), the accumulation of cGMP in the Pde6a point-mutants examined here was less prominent.

High levels of cGMP are thought to cause an over-activation of CNG channels leading to excessive amounts of Ca²⁺-influx (28,36). This is compatible with the observed increase in the numbers of photoreceptor cells showing a high degree of Ca²⁺-activated calpain-type protease activity. Calpain activity is often associated with non-apoptotic forms of cell death (37) and in all Pde6a mutants it is strongly correlated in time with the progression of photoreceptor cell death. The relatively low amount of calpain activity-positive photoreceptors, when compared with the rd1 mouse (38), likely reflects the lower extent of cGMP accumulation and thus lower Ca²⁺-influx. This may be particularly evident in the V685M mutant where the numbers of dying, TUNEL-positive cells exceeds the numbers of calpain activity positive cells by ∼2:1. In the more slowly degenerating mutants the relative extent of calpain activity is more important, and reaches a ratio of ∼1:1 in the D670G mutant. In either case, these results are suggestive of a causal involvement of calpain activity in the degenerative process.

Remarkably, in all four mutant genotypes the numbers of photoreceptors displaying high caspase-3 activity was very low, at wt levels, and not correlated to the progression of the mutation-induced degeneration. Since caspase-3 is key to the execution of numerous terminal proteolytic events in classical apoptotic cell death (27), this result indicates that apoptosis does not play any major role in Pde6a mutation-induced retinal degeneration. This is in agreement with a recent study on the prevalence of non-apoptotic cell death in 10 different rodent retinal degeneration models (2). Similar observations were also made in monkey retinal cell cultures subjected to hypoxia/reoxygenation where calpain proteolytic activity was in fact found to inactivate caspase-type proteases (39)

Overall, our results may provide important insights for potential therapy developments, suggesting the targeting of non-apoptotic processes, such as excessive cGMP-signalling or calpain activity as a feasible treatment approach.

Variable speed of progression, functional and degeneration phenotypes

The four different Pde6a mutants used here were found to have a variable pace of degeneration, corresponding to the effect of the genetic mutations on cGMP accumulation and—as deduced from this—PDE6 function.

Remarkably, there is a clear discrepancy between the extent of cGMP accumulation observed in Pde6b mutant rd1 photoreceptors and Pde6a V685M photoreceptors. Even though the speed of photoreceptor loss is comparable in both RP models, and both PDE6 protein isoforms have been suggested to be enzymatically equivalent (40), in rd1 retina 6–15% of photoreceptors show dramatic cGMP accumulation (6) compared with only 0.6% in V685M retina. The extent of rd1 cGMP accumulation might be explained by a complete loss of PDE6 expression caused by the rd1 mutation. In the V685M mutant a minor amount of residual PDE6 activity could prevent such a dramatic accumulation of cGMP; however, even a more moderate elevation of cGMP (undetectable by immunohistochemistry) would still result in rapid photoreceptor degeneration.

The phenotype of the novel R562W mutant is intermediate between that of the V685M and the D670G mutant. Unexpectedly, we found that the pathogenic effect of this underlying c.1684C>T substitution is twofold and at distinct levels: pre-mRNA splicing and protein enzymatic activity. The C>T transition (+64 of exon 13) is predicted to create an ESS site (–TGGTGG–) which results in an in-frame skipping of exon 13. The splicing defect induced by the mutation is only partially penetrant and therefore correctly spliced transcripts derived from the mutant allele, which accounts for ∼50–60% of transcripts (Fig. 2B), will be translated into a full-length protein carrying the R562W substitution. This R562W mutant protein exhibits a reduction of its catalytic activity to ∼10% compared with the WT protein. The shortened p.541_576del protein derived from Pde6a transcripts devoid of exon 13, lacks a strongly conserved segment of the protein's catalytic domain, including one of the two metal ion binding motifs which are important for functional activity (Supplementary Material, Fig. S5) (41). Therefore, we reason that the shortened protein may essentially be non-functional or with very strongly reduced activity. Moreover, protein stability of such shortened proteins is often reduced as also documented for other mouse models for retinal dystrophies (42). The findings with the R562W mutant complements prior reports of splicing defects in retinal PDE6 genes caused by single nucleotide exonic mutations such as the c.1814G>A/p.N605S in Pde6b in the atrd3 mouse model (43) or the c.2368G>A/p.E790 K in the human PDE6C in a family with achromatopsia (44). It also emphasizes the need for a thorough transcript analysis of the mutant gene for a proper assessment of its consequences.

Typical for RP, the loss of the primarily affected rod photoreceptors was followed by a secondary loss of cone photoreceptors via a well-known but still aetiologically unresolved mechanism (45). Differences in rod loss among the mutants thus corresponded well to the functional data in Ganzfeld ERG at PN30, although these essentially represent cone function. In particular, ERG amplitudes in the D670G mutants were similar to Rho^−/− animals (30), while no more responses were left in V685M mutants. The results in the other lines were in between these extreme values. This correlation allows to provide an estimate of the minimal number of rod photoreceptors that still need to be physically present (i.e. surviving) in order to maintain residual cone photoreceptor function at a recordable level. In our setting, at least one intact row of rod photoreceptors was required in order to still preserve cone ERG at P30.

V685m/R562W: a new patient-matched, compound heterozygous animal model for RP

Current research on the mechanisms underlying recessive RP usually employs homozygous animal models, such as the Pde6b^rd1/rd1 mouse (11,46). In RP patients, in outbred populations, homozygosity is, however, less common than compound heterozygous mutations with two different disease-causing alleles (47,48). This creates incongruence between the currently used animal models for RP and the actual patient situation, in particular because the pathological consequences of compound heterozygous versus homozygous mutations are to date only poorly understood.

According to Muller (49), the action of recessive mutations may be explained be either constituting an amorph (classical null mutation) or a hypomorphic mutation, the latter being associated with lower amounts of the gene product or reduced function or activity. If the amounts of the gene product or its activity is strictly additive and proportional to the phenotypic outcome, then one would expect for compound heterozygous mutations an intermediate disease phenotype, which lies between the two respective homozygous disease phenotypes. However, this must not be necessarily true, for instance in traits that develop in a threshold-dependent manner. Moreover, one needs to consider situations of intra-allelic complementation in which different mutations in the same gene may ‘neutralize’ each other, a phenomenon most likely to occur in proteins that act as oligomers (50). Thus, comparative studies between homozygotes and compound heterozygotes bear important insights into the mode of action of individual mutations, potential interactions between different mutant proteins, and the correlation of the disease phenotype with the amount and functional activity of the mutant gene product.

Our results with the V685M/R562W retinal degeneration suggest that for the Pde6a gene a classic additive model for the interaction of the two mutations is valid, resulting in a proportional relationship between the amount or the functional activity of the mutant gene product and the phenotypic outcome. If generalized, this model enables to predict the outcome of an unknown genotype in terms of severity and disease progression if the phenotype of two adjoined genotypes in Pde6a is known. Our findings also demonstrate that a genotype–phenotype correlation with high predictability exists for Pde6a-associated retinal degeneration at least in murine mutants raised on the same genetic background. Therefore, the lack of consistent genotype–phenotype correlation sometimes observed in human RP patients is most likely due to secondary genetic factors and/or differences in lifestyle (51,52).

Nevertheless, our study provides a rational basis for predictions on human RP phenotypes and disease progression in compound heterozygous situations, provided the pathological consequences of at least one of the disease-causing alleles are known. This knowledge could be very valuable for patient counselling and also—if suitable therapeutic approaches become available in the future—for the determination of the best possible time-frame for therapeutic interventions.

The large genetic diversity in RP causing mutations—even if affecting the same gene—may cause very different degeneration phenotypes. Our detailed characterization of different homozygous and compound heterozygous mouse models for RP will provide a basis for further investigations, in particular on the development of future individualized therapies. The wide availability of a large variety of animal models will likely facilitate the generation of ‘personalized’ compound heterozygous animal models that match the human disease condition very closely. Such models could greatly speed up the development of a more personalized medicine in the field of retinal degenerations. For the individual RP patient, the prediction of time-courses for Pde6a-related retinal degeneration is currently of particular value. Upon availability of causative or symptomatic treatment options, the definition of a window-of-opportunity for clinical interventions will gain further importance.

Funding

This work was supported by the Kerstan Foundation (RD-CURE), the 2nd People's Hospital Of Yunnan Province and 4th Affiliated Hospital Of Kunming Medical University/China, Deutsche Forschungsgemeinschaft (DFG PA1751/4-1;7-1), Alcon Research Institute, and the European Commission (DRUGSFORD: HEALTH-F2-2012-304963; EyeTN: FP7-People-2012-ITN-317472).

Acknowledgements

We thank K. Masarini, N. Rieger and Britta Baumann for skilful technical assistance. We thank Dr Muayyad Al-Ubaidi (University of Oklahoma, Oklahoma City, USA) for kindly providing the 661W cell line.

Conflict of Interest statement. None declared.

Author notes

†

The authors wish it to be known that, in their opinion, the first three authors should be regarded as joint First Authors.

‡

Present address: Department of Obstetrics and Gynaecology, Medical University of Vienna, Waehringer Guertel 18-20, 1090 Vienna, Austria.

¶

Present address: CeGaT GmbH, Tübingen, Germany.

§

These authors should be considered as equivalent corresponding authors.