A variety of rod opsin mutations result in autosomal dominant retinitis pigmentosa and congenital night blindness in humans. One subset of these mutations encodes constitutively active forms of the rod opsin protein. Some of these dominant rod opsin mutant proteins, which desensitize transgenic Xenopus rods, provide an animal model for congenital night blindness. In a genetic screen to identify retinal degeneration mutants in Drosophila, we identified a dominant mutation in the ninaE gene (NinaEpp100) that encodes the rhodopsin that is expressed in photoreceptors R1-R6. Deep pseudopupil analysis and histology showed that the degeneration was attributable to a light-independent apoptosis. Whole-cell recordings revealed that the NinaEpp100 mutant photoreceptor cells were strongly desensitized, which partially masked their constitutive activity. This desensitization primarily resulted from both the persistent binding of arrestin (ARR2) to the NINAEpp100 mutant opsin and the constitutive activity of the phototransduction cascade. Whereas mutations in several Drosophila genes other than ninaE were shown to induce photoreceptor cell apoptosis by stabilizing a rhodopsin-arrestin complex, NinaEpp100 represented the first rhodopsin mutation that stabilized this protein complex. Additionally, the NinaEpp100 mutation led to elevated levels of Gqα in the cytosol, which mediated a novel retinal degeneration pathway. Eliminating both Gqα and arrestin completely rescued the NinaEpp100-dependent photoreceptor cell death, which indicated that the degeneration is entirely dependent on both Gqα and arrestin. Such a combination of multiple pathological pathways resulting from a single mutation may underlie several dominant retinal diseases in humans.
Retinitis pigmentosa (RP) is a group of genetic disorders that exhibits abnormal electroretinograms and gradual vision loss attributable to the death of the rod and then cone photoreceptors in humans (Rattner et al., 1999; Phelan and Bok, 2000). Several X-linked and autosomal dominant and recessive forms of RP have been identified (Wang et al., 2001), with rhodopsin mutations being a prevalent cause of autosomal dominant RP (Kaushal and Khorana, 1994; Gal et al., 1997). Rhodopsin is a seven transmembrane, G-protein-coupled receptor that is activated by a photon of light to initiate the phototransduction cascade. Various models describe dominant rhodopsin-mediated degeneration mechanisms, including the following: (1) defective maturation and/or trafficking of rhodopsin (Roof et al., 1994; Sung et al., 1994), (2) constitutively active rhodopsin and persistent stimulation of the phototransduction cascade (Robinson et al., 1992; Dryja et al., 1993; Rao et al., 1994), and (3) abnormal interactions between rhodopsin and the phototransduction components (Min et al., 1993; Li et al., 1995; Rim and Oprian, 1995).
The Drosophila ninaE gene encodes the rhodopsin protein (NINAE) that is expressed in six (R1-R6) of the eight adult photoreceptor cells in each ommatidium (O'Tousa et al., 1985; Zuker et al., 1985). The NINAE protein is activated by blue light, which stimulates the dgq-encoded Gqα subunit of the heterotrimeric G-protein (Lee et al., 1990, 1994; Scott et al., 1995). The photoactivated metarhodopsin is inactivated by phosphorylation and arrestin (ARR2) binding (Byk et al., 1993; Dolph et al., 1993; Kiselev and Subramaniam, 1994; Alloway and Dolph, 1999). Orange light photoconverts metarhodopsin to rhodopsin, which dissociates from arrestin and is dephosphorylated by the rdgC-encoded Ca2+-dependent serine/threonine phosphatase (Byk et al., 1993). In the absence of arrestin dissociation, the persistent NINAE-arrestin complexes are endocytosed and induce apoptosis of photoreceptor cells (Alloway et al., 2000; Kiselev et al., 2000). Recessive ninaE mutants exhibit a defective light response and photoreceptor cell death (Leonard et al., 1992). Dominant ninaE mutations were identified as suppressors of either rdgC-mediated (Kurada and O'Tousa, 1995) or rdgB-dependent (Colley et al., 1995) retinal degeneration. The rdgB gene encodes a novel phosphatidylinositol transfer protein that is required for the light response and photoreceptor viability (Vihtelic et al., 1991, 1993). All of the previously isolated dominant ninaE mutants are defective in the maturation and/or trafficking of rhodopsin in the endoplasmic reticulum (ER) (Colley et al., 1995; Kurada et al., 1998). Thus, dominant ninaE mutations likely mimic the degeneration mechanisms underlying rhodopsin-mediated autosomal dominant RP in humans.
We report a new dominant ninaE mutant (NinaEpp100) that exhibits retinal degeneration and constitutively active and strongly desensitized photoreceptor cells. The photoreceptor cell desensitization and death resulted from both the NINAEpp100 mutant rhodopsin forming a stable complex with arrestin and the persistent localization of the Gqα in the cytosol. Genetic experiments reveal that the arrestin and Gqα mechanisms are mutually exclusive. Furthermore, the predominant localization of an active Gqα protein in the cytosol stimulates photoreceptor cell death through a novel mechanism.
Materials and Methods
Isolation of the dominant NinaEpp100 mutant. Wild-type males were starved for 6 hr and then fed 25 mm ethyl methanesulfonate (Sigma, St. Louis, MO) in a 1% sucrose solution overnight (Lewis and Bacher, 1968). The males were mated en masse to SM1/Sco; TM2/Sb virgin females, and the F1 progeny were raised in constant light for 10 d. The F1 flies were screened for the absence of a wild-type deep pseudopupil, which is a virtual image of the rhabdomeres from several adjacent ommatidia (Franceschini, 1972).
Characterizing the retinal degeneration phenotype. Flies were raised under constant light and scored daily for the presence of the deep pseudopupil as described previously (Paetkau et al., 1999). The average percentage of flies possessing a wild-type deep pseudopupil and SDs were determined from at least four replicates (100-150 flies per trial) of each genotype and age. Flies raised in constant darkness were collected, aged until a designated day, scored for the presence or absence of a wild-type deep pseudopupil, and then discarded.
For light and electron microscopy of retinal sections, flies were raised in a 12 hr light/dark cycle to the desired age and then decapitated. The heads were bisected, fixed, and embedded in Polybed 812 as described previously (Lee et al., 1994). Sections for light and electron microscopy were generated and processed as described previously (Paetkau et al., 1999).
DNA isolation and sequencing. Genomic DNAs from wild-type and NinaEpp100 mutant flies were isolated as described previously (Ashburner, 1989). Wild-type and mutant ninaE genes were PCR amplified using the 5′ AGG ATC CAA TGG AGA GGT ACG ATC GGT GAA TCC AC and 5′ GGT TGT GGA TCC AAA GAA TTT ATG CC primers that mapped to nucleotides -8 and 1518 of the ninaE gene, respectively (O'Tousa et al., 1985). Both PCR primers contain engineered BamHI sites at their 5′ ends (in bold), and the first primer also deletes a BamHI site (underlined) in the first ninaE intron. The PCR reactions contained 3 μg of genomic DNA, 1× native Pfu buffer, 2.5 mm of each dNTP, and 250 pmol of each primer. After incubating at 96°C for 7 min, 1 μl of native Pfu DNA polymerase (Stratagene, La Jolla, CA) was added and amplified through 30 cycles of 96°C for 1 min, 54°C for 1 min, and 72°C for 4 min, followed by a 10 min extension at 72°C. The PCR products were size fractionated on a Tris acetate-EDTA agarose gel, purified with the Wizard PCR prep resin (Promega, Madison, WI), and blunt-end ligated into the pZero 2.1 vector (Invitrogen, Carlsbad, CA). The plasmid DNA was purified with Strataclean Resin (Stratagene), precipitated, resuspended in H2O, and sequenced using the Sequenase Quick Denature Kit (US Biochemicals, Cleveland, OH), with primers generated from the published ninaE genomic sequence (O'Tousa et al., 1985).
In vitro mutagenesis. The Pfu-generated wild-type ninaE gene was digested from pZero 2.1 with BamHI, gel purified, and cloned into the BamH1 site of the PK- ATG- vector, which contains the 3.2 kb ninaE promoter and a 0.7 kb ninaE poly(A+) tail (Kurada and O'Tousa, 1995; Shetty et al., 1998). The NinaEpp100 mutation was introduced into the wild-type ninaE gene by in vitro mutagenesis in the PK- ATG- vector using the Quick Change Site-Directed Mutagenesis kit (Stratagene) and the complementary primer pairs 5′ TCA ACT GCA TGA GAC TGT TCA AGT 3′ and 5′ ACT TGA ACA GTC TCA TGC AGT TGA 3′ (the site of the nucleotide change is in bold). The PCR reaction contained 50 ng of template DNA, 125 ng of each primer, 1× buffer, 10 mm of each dNTP, and 1 μl of Pfu DNA polymerase. The PCR amplification protocol was 95°C for 30 sec and then 12 cycles of 95°C for 30 sec, 60°C for 1 min, 68°C for 2 min, and a final 15 min extension at 68°C.
Germ-line transformation. A 5.4 kb KpnI fragment containing the ninaE promoter, the in vitro mutagenized ninaE genomic DNA, and the ninaE poly(A+) tail was gel purified and cloned into the KpnI site of the pCaSpeR-4 transformation vector (Ashburner, 1989). The sequence of the entire ninaE coding region was confirmed in this vector. This in vitro mutagenized DNA was germ-line transformed with the Δ2-3 P-element (ratio of 6:1) using standard techniques (Spradling, 1986).
Electrophysiology. Electroretinograms (ERGs) were performed as described previously (Paetkau et al., 1999). ERG traces were recorded using a MacAdios II analog-to-digital converter using SuperScope II software on a Macintosh IIx computer (Apple Computers, Cupertino, CA).
To measure light-induced currents (LICs), orange light (OG 590 Ditric edge filter) from a xenon high-pressure lamp (75 W) was delivered to isolated ommatidia via the objective lens (60× Olympus Optical, Tokyo, Japan) and attenuated up to seven orders of magnitude by neutral density filters. The maximal luminous intensity of the orange light at the level of the ommatidia was ∼3.0 log units above the intensity required for a half-maximal response of the R1-R6 photoreceptors. The effective intensity of the white light without the orange filter was determined by comparing the light intensities with and without the orange filter, which were required to elicit a small response of the same amplitude in wild-type flies. In Figure 4, the intensity of the white light stimulus was converted to the effective relative intensity of orange light.
Dissociated ommatidia were prepared from either newly eclosed adult flies (<1 hr after eclosion) or late pupae as described previously (Hardie, 1991). Whole-cell patch-clamp recordings were performed as described previously (Hardie and Minke, 1992; Peretz et al., 1994). Recordings were made at 21°C using patch pipettes of 5-10 MΩ pulled from fiber-filled borosilicate glass capillaries. Series resistance of 7-14 MΩ was carefully compensated (>80%) during all experiments. Signals were amplified with an Axopatch-1D or 2B (Axon Instruments, Foster City, CA) patch-clamp amplifier, sampled at 2 kHz, and filtered below 1 kHz.
Immunoblots of NINAE, ARR2, and Gqα proteins. Immunoblots of rhodopsin expression in flies 4-6 hr old were performed as described previously (Milligan et al., 1997). Head protein homogenates were electrophoresed, electrotransferred to Hybond polyvinylidene difluoride (PVDF) membrane (Amersham Biosciences, Piscataway, NJ), and incubated with the 4C5 anti-rhodopsin monoclonal antibody (Developmental Studies Hybridoma Bank, University of Iowa, Iowa City, IA). The 4C5 monoclonal antibody was detected with HRP-conjugated goat anti-mouse antibodies (1:5000) and ECL reagents (Amersham Biosciences). Band intensity was quantified on a scanning laser densitometer (Bio-Rad, Hercules, CA) and normalized to a standard curve generated by a dilution series of head protein extract, with five independent blots analyzed.
The ARR2 protein was isolated under dim red light from white-eyed wild-type and mutant fly heads that were 4 hr after eclosion. The heads were placed in 20 μl of PBS, pH 7.4, illuminated at the desired wavelength for 2 min, homogenized in darkness, and centrifuged with 15,000 × g. The supernatant fraction was transferred to a new Eppendorf tube (Eppendorf Scientific, Westbury, NY), and the pellet was resuspended in 20 μl of PBS. SDS extraction buffer (20 μl) was added to both fractions and incubated at 45°C for 1 hr before electrophoresing one head equivalent of cytosolic and membrane protein extracts through a 7-15% gradient SDS-PAGE gel. After electrotransfer, the PVDF membranes were incubated with a mixture of the 4C5 anti-rhodopsin monoclonal antibody and anti-ARR2 rabbit polyclonal antisera (Alloway et al., 2000; Kiselev et al., 2000) and detected with a mixture of HRP-conjugated mouse anti-rabbit and rabbit anti-mouse antibodies and the ECL system (Amersham Biosciences), with five independent blots analyzed.
The Gqα-protein was quantitated in different subcellular fractions as described previously (Kosloff et al., 2003). Wild-type and NinaEpp100 flies were dark adapted for 24 hr and then subjected to either blue light illumination (490 nm) or continued darkness for 10 min at 22°C. The heads were homogenized at 4°C in the dark in hypotonic medium (20 mm HEPES buffer, pH 7.6, containing 15 mm N-ethylmaleimide, 20 μg/ml leupeptin, 1 μg/ml pepstatin A, and 0.35 mg/ml o-phenantroline). The samples were centrifuged at 15,800 × g for 15 min at 4°C to separate the membrane pellet from the soluble proteins. The pellet was washed and centrifuged, and the supernatants were combined. Ultracentrifugation at 150,000 × g for 30 min did not change the Gqα distribution between the fractions. The proteins were precipitated in 5% TCA, electrophoresed on a 10% SDS-PAGE, and electrotransferred to nitrocellulose using a semidry transfer apparatus (Bio-Rad). The nitrocellulose was blocked and incubated with a rabbit anti-Gqα polyclonal antibody raised against the decapeptide of the C terminus of the protein. The nitrocellulose was incubated with anti-rabbit polyclonal antiserum and detected with ECL Amersham Biosciences reagents according to the instructions of the manufacturer and quantitated with Fuji LAS-1000. The amount of Gqα in each fraction was calculated as a percentage of the total Gqα in the pellet and supernatant, with four independent blots analyzed.
GTPase assay. GTP hydrolysis of Drosophila head extracts under various light conditions were performed essentially as described previously (Blumenfeld et al., 1985).
The NinaEpp100 mutant exhibits a dominant retinal degeneration and a recessive electrophysiological defect
In a screen to identify Drosophila mutants that lacked a deep pseudopupil, we identified a dominant retinal degeneration mutation that mapped <1 cM from the ninaE gene, which encodes the rhodopsin protein expressed in photoreceptor cells R1-R6 (O'Tousa et al., 1985; Zuker et al., 1985). To determine whether NinaEpp100 was a ninaE allele, we examined the ERGs of flies that were ∼12 hr after eclosion (before any signs of degeneration). White-eyed wild-type flies exhibited a prolonged depolarizing afterpotential (PDA) after a blue light converts a large fraction of rhodopsin to metarhodopsin (Fig. 1A) (Hillman et al., 1983; Dolph et al., 1993). The white-eyed ninaEI17/ninaEI17 null mutant lacked a PDA and possessed small light response amplitudes (Fig. 1B), which were attributable to the opsins expressed in photoreceptor cells R7 and R8. White-eyed NinaEpp100/ninaE+ flies exhibited a wild-type ERG light response and a PDA (Fig. 1C), whereas NinaEpp100/NinaEpp100 flies lacked the PDA (Fig. 1E). Thus, the NinaEpp100 flies exhibited a recessive mutant ERG phenotype. White-eyed NinaEpp100/ninaEI17 flies also lacked the PDA (Fig. 1D), which confirmed that NinaEpp100 was a ninaE allele. Immunoblots revealed that the NinaEpp100 mutant gene expressed ∼20% of the rhodopsin relative to the ninaE+ gene (Fig. 1F). We sequenced the ninaE gene from the NinaEpp100 mutant and identified a G-to-A transition mutation, which changed the glycine at amino acid 299 to arginine (Fig. 1G).
We examined the retinal degeneration that underlies the deep pseudopupil loss in the NinaEpp100 mutant. A 2-d-old wild-type retina possessed a regular repeating pattern of ommatidia (Fig. 2A, left column), whereas the ultrastructure of a 10-d-old retina revealed large R1-R6 rhabdomeres surrounding the central R7 rhabdomere (Fig. 2A, right column). In contrast, the 2-d-old NinaEpp100/ninaE+ retina possessed a few holes and reduced or missing rhabdomeres (Fig. 2B). Electron microscopy of the 10-d-old NinaEpp100/ninaE+ retina revealed degenerating R1-R6 photoreceptors (darkly stained) and phagocytosed photoreceptors (Fig. 2B, asterisks). To prove that the G299R mutation caused the dominant NinaEpp100 mutant phenotype, we created three independent transgenic lines that contained a wild-type ninaE gene that was in vitro mutagenized to introduce the G299R mutation. At both the light and electron microscope level, all three transgenic lines (ninaE+, P[ninaE-G299R]/ninaEI17) (Fig. 2C) exhibited a dominant degeneration phenotype that was indistinguishable from the NinaEpp100/ninaE+ flies. Therefore, the G299R mutation accounted for the dominant NinaEpp100 degeneration phenotype.
Because the NinaEpp100 mutant exhibited a recessive ERG light response defect, we examined the histology of NinaEpp100 homozygotes. The 2-d-old NinaEpp100/NinaEpp100 retina exhibited an increased number of large intracellular vesicles, more ommatidial disorganization, and fewer rhabdomeres relative to NinaEpp100/ninaE+ (Fig. 2D,B, respectively). The ultrastructure of a 10-d-old NinaEpp100/NinaEpp100 ommatidium revealed that the R1-R6 rhabdomeres were more disorganized and smaller than the central R7 rhabdomere, and many large intracellular vesicles (Fig. 2D, arrows) were present within some of the R1-R6 photoreceptors. Some photoreceptors were being phagocytosed, and others exhibited a decreased cytoplasmic volume (Fig. 2D, asterisks, arrowhead, respectively). Thus, the NinaEpp100 mutation caused a dominant retinal degeneration that was enhanced in the NinaEpp100 homozygote.
The NinaEpp100 mutant exhibits a light-enhanced retinal degeneration
We examined the effect of light on the NinaEpp100 retinal degeneration to differentiate between various potential degeneration mechanisms. We used the deep pseudopupil as a rapid and noninvasive technique to assay the integrity of the retina and photoreceptor cells (Franceschini, 1972). The NinaEpp100 mutant exhibited a dominant deep pseudopupil loss that was slightly faster in constant light than in constant darkness (Fig. 3, NinaEpp100/ninaE+, open circles, filled circles, respectively). The absence of any wild-type NINAE+ protein in the NinaEpp100/NinaEpp100 flies dramatically increased the onset and rate of deep pseudopupil loss relative to NinaEpp100/ninaE+ (Fig. 3). Furthermore, the NinaEpp100/NinaEpp100 degeneration time course exhibited a significantly faster deep pseudopupil loss in constant light relative to constant darkness (Fig. 3, open squares, filled squares, respectively). The NinaEpp100/ninaEI17 flies exhibited a deep pseudopupil loss that mimicked the NinaEpp100/NinaEpp100 degeneration time course (data not shown). The ninaED1 mutation, which is a serine-to-phenylalanine change at position 137 in the third transmembrane domain, causes a slow dominant retinal degeneration by blocking rhodopsin maturation in the endoplasmic reticulum (Kurada and O'Tousa, 1995; Kurada et al., 1998). The ninaED1 mutation suppressed the rapid NinaEpp100-dependent deep pseudopupil loss (Fig. 3, NinaEpp100/ninaED1, triangles). Therefore, the NinaEpp100 mutation must cause its rapid degeneration through a mechanism subsequent to the ninaED1 block of rhodopsin maturation in the ER (Kurada and O'Tousa, 1995).
The NinaEpp100 mutant is strongly desensitized
We performed whole-cell recordings on the homozygous NinaEpp100 mutant before photoreceptor degeneration to elucidate the underlying physiological defect. The wild-type LIC to dim orange light stimulus showed current fluctuations that are attributable to the summation of the quantum bumps (Fig. 4A, top left panel), whereas a bright orange light stimulus invoked an initial transient phase that declined to a small steady-state phase attributable to light adaptation (Fig. 4A, top right panel). The LIC from one of the most sensitive NinaEpp100 mutant cells to intense orange light elicited only a small response that lacked large current fluctuations (Fig. 4A, middle left panel). Furthermore, the intense white light required to elicit a sizable response in the NinaEpp100 mutant (Fig. 4A, middle right panel) was approximately four orders of magnitude brighter than the stimulus eliciting a similar response amplitude in a wild-type cell. The need for a more intense light stimulus to induce a wild-type-sized LIC response in NinaEpp100 suggested that the NinaEpp100 mutant possessed a highly reduced sensitivity to light. The individual NinaEpp100 cells exhibited a large variability in their sensitivity to light, with only a fraction of the light-sensitive NinaEpp100 mutant cells possessing the maximal peak response amplitude (data not shown). Because this amplitude was similar to that of wild-type flies, the mutant cells must contain a sufficient number of light-sensitive rhodopsin molecules. It was shown previously that the high gain of the phototransduction cascade resulted in the ninaEP332 mutant exhibiting a shift of ∼3.5 orders of magnitude in the intensity-response function (V-log I curve) and a wild-type peak amplitude of the receptor potential, although it had approximately three orders of magnitude less rhodopsin than wild-type flies (Johnson and Pak, 1986).
To confirm the reduced light sensitivity of the NinaEpp100 cells, we plotted the intensity-response function (i-log I curve) for several different individual NinaEpp100 mutant and wild-type cells (Fig. 4B). There was a large difference in the light intensity range at which wild-type and the NinaEpp100 photoreceptors responded, with a considerable fraction of mutant cells either not responding to the maximal white light intensity or only exhibiting minimal response amplitudes. This data clearly demonstrated that a main electrophysiological phenotype of the NinaEpp100 mutant relative to wild type is a large decrease in the sensitivity to light.
The NinaEpp100 desensitization is partially rescued by elimination of arrestin
One mechanism that could underlie the NinaEpp100 reduced light sensitivity is persistent binding of arrestin (ARR2) to NINAEpp100. Persistent arrestin binding was shown previously to result in retinal degeneration of several Drosophila mutants (Alloway et al., 2000; Kiselev et al., 2000) and the transgenic mouse model expressing the K296E mutant human opsin (Li et al., 1995). Whole-cell recordings on the arr2Y20STOP NinaEpp100 double mutant (Fig. 4A, bottom left panel) revealed the bump noise that was also observed in wild-type flies (Fig. 4A, top left panel), but absent in NinaEpp100 (Fig. 4A, middle left panel). Although the arr2Y20STOP NinaEpp100 double mutant revealed a large increase in light sensitivity relative to NinaEpp100, it was still significantly less sensitive than wild type (Fig. 4A, bottom right panel). We compared the intensity-response function (i-log I curve) of the arr2Y20STOP mutant photoreceptors with wild-type and NinaEpp100 cells and found that the family of arr2Y20STOP responses was nearly identical to wild type (Fig. 4B). Thus, the ARR2 protein specifically interacted with the NINAEpp100 mutant rhodopsin to decrease the sensitivity to light. However, this interaction failed to account for the entire range of reduced sensitivity of the NinaEpp100 cells.
We extended this analysis by plotting the peak amplitude versus log of the relative light intensity from in vivo ERG recordings and confirmed that the arr2Y20STOP NinaEpp100 double mutant possessed an increased light sensitivity relative to the NinaEpp100 mutant (Fig. 5A). Although these observations strongly suggested that arrestin contributed to the reduced light sensitivity of the NinaEpp100 mutant, Figures 4A (bottom row) and 5A revealed that the arr2Y20STOP NinaEpp100 double mutant remained less sensitive to light relative to wild-type flies. This remaining desensitization in the arr2Y20STOP NinaEpp100 fly may be attributable to either the NINAEpp100 molecules being constitutively active or a reduced interaction between NINAEpp100 and the heterotrimeric G-protein.
To investigate whether the NINAEpp100 molecules were constitutively active and induced persistent opening of the light-activated channels, we recorded whole-cell membrane currents to voltage steps in the dark (Fig. 5C-F). Wild-type cells revealed no significant currents under these conditions (Fig. 5C). However, small but significant (Yoon et al., 2000) currents were recorded from both NinaEpp100 and arr2Y20STOP NinaEpp100 mutant cells (Fig. 5D,E, respectively). The characteristics of these currents and their ability to be blocked by a low concentration of La3+ (Fig. 5F) suggested that they originate from constitutively open transient receptor potential (TRP) channels (Hardie and Minke, 1994), which are the major light-activated channels (for review, see Minke and Hardie, 2000). These constitutive currents were recorded in 58% of the NinaEpp100 cells and 74% of the arr2Y20STOP NinaEpp100 double mutant photoreceptors (Fig. 5B). Because the arr2Y20STOP mutant cells failed to exhibit any significant whole-cell membrane currents to voltage steps in the dark (data not shown), the arr2Y20STOP mutation only contributed to the constitutive opening of the TRP channels in the presence of the NINAEpp100 mutant rhodopsin (Fig. 5B). Although arrestin partially masked the production of the constitutive current, possibly through a stable NINAEpp100-ARR2 complex, it is not clear why some of the mutant cells were not constitutively active. This may be related to variability in the level of rhodopsin among different NinaEpp100 cells, which is consistent with the large variability in the light sensitivity between different NinaEpp100 mutant cells (Fig. 4B).
The dominant NinaEpp100 mutant requires both Gqα and ARR2 proteins for rapid degeneration
It was shown previously that the norpA, rdgB, and rdgC mutations produced a persistent NINAE-ARR2 complex that induced photoreceptor cell apoptosis (Alloway et al., 2000; Kiselev et al., 2000). Because the electrophysiology suggested that a stable NINAEpp100-ARR2 complex was present, we examined whether loss of the ARR2 protein suppressed the rapid NinaEpp100-dependent degeneration. At 10 d after eclosion, wild-type photoreceptors possessed large rhabdomeres (Fig. 6A), whereas a 4-d-old NinaEpp100 mutant possessed significantly smaller and disorganized rhabdomeres and signs of photoreceptor cell death (Fig. 6D). The arr2Y20STOP null mutant possessed slightly smaller and irregular-shaped rhabdomeres relative to wild type (Fig. 6B), which is consistent with the light-dependent deep pseudopupil loss described previously (Alloway and Dolph, 1999). The 4-d-old arr2Y20STOP NinaEpp100 double mutant also exhibited small and abnormally shaped rhabdomeres (Fig. 6E), similar to the arr2Y20STOP mutant. However, neither arr2Y20STOP nor arr2Y20STOP NinaEpp100 exhibited the massive rhabdomere loss and photoreceptor cell death that occurred in the NinaEpp100 mutant. This suggested that a stable NINAEpp100-ARR2 complex induced both photoreceptor desensitization and cell death. Furthermore, the time course of deep pseudopupil loss revealed that the arr2Y20STOP mutation significantly slowed, but did not abolish, the NinaEpp100 degeneration in both constant light and constant darkness (data not shown). Thus, the NINAEpp100-ARR2 complex was not the sole cause of retinal degeneration, and an additional independent mechanism must participate in generating the NinaEpp100 mutant phenotypes.
Because the electrophysiology revealed that the NINAEpp100 mutant protein was constitutively active, we examined whether activation of the light-sensitive channels induced the retinal degeneration phenotype. It was demonstrated previously that the early-onset retinal degeneration in the rdgABS12 mutant, which possesses constitutively active light-sensitive channels, was significantly slowed by the trp301 mutation (Raghu et al., 2000). We examined whether mutants blocking three different points in the phototransduction cascade could suppress the rapid NinaEpp100-dependent degeneration. The mutations we tested corresponded to the dgq (which encodes the Gqα that is stimulated by light-activated metarhodopsin), norpA (which encodes the phospholipase C that is activated by Gqα), and trp genes (which encodes the major light-activated channel). The norpAp24 mutant exhibited no signs of retinal degeneration when raised for 6 d in constant darkness (Fig. 7B). In contrast, the NinaEpp100 mutant possessed the previously described degeneration phenotype after 6 d in constant darkness (Fig. 7C). Although norpAp24; NinaEpp100 exhibited reduced and misshapen rhabdomeres after 6 d in constant darkness (Fig. 7D), it lacked the severe degeneration that was present in the NinaEpp100 mutant. This suggests that norpAp24 slightly delayed the NinaEpp100 degeneration. Similarly, the trp301 mutation only slightly suppressed NinaEpp100 degeneration relative to the arr2Y20STOP mutation (Fig. 7E). Thus, constitutive activation of the phototransduction cascade is only a minor component of the NinaEpp100 degeneration mechanism.
Surprisingly, a mutation in the dgq gene, which encodes the Gqα subunit target for metarhodopsin (Lee et al., 1990, 1994; Scott et al., 1995), suppressed the NinaEpp100 deep pseudopupil loss substantially better than the trp301 mutation (Figs. 6K, 7E, respectively). At 10 d after eclosion, the dgq1 mutant possessed only a few missing rhabdomeres relative to the massive rhabdomere loss and photoreceptor cell death in NinaEpp100 (Fig. 6C,D, respectively). The ultrastructure of a 10-d-old dgq1; NinaEpp100 double mutant exhibited only minor photoreceptor abnormalities that were similar to the dgq1 mutant (Fig. 6F,C, respectively). Furthermore, the time course of deep pseudopupil loss in the dgq1; NinaEpp100 double mutant was similar to the dgq1 mutant in both dark and light-raised flies, which was significantly slower than the NinaEpp100 mutant (Fig. 6K). However, both the dgq1; NinaEpp100 and arr2Y20STOP NinaEpp100 double mutants still exhibited a slow retinal degeneration (Fig. 6E,F,K). After 20 d, the dgq1; NinaEpp100 and arr2Y20STOP NinaEpp100 double mutants exhibited massive photoreceptor cell death (Fig. 6I,H, respectively) relative to an equivalent aged wild-type retina (Fig. 6J). Surprisingly, the dgq1; arr2Y20STOP NinaEpp100 triple mutant exhibited only very minor rhabdomeric abnormalities after 40 d (Fig. 6G). The absence of significant cell death in this 40-d-old triple mutant relative to either of the 20-d-old double mutants and the minor suppression of NinaEpp100 degeneration by the norpAp24 and trp301 mutants revealed that both the ARR2 and Gqα proteins, but not activation of the light-sensitive channels, play major roles in the rapid NinaEpp100-dependent retinal degeneration.
The dominant NINAEpp100 protein causes a constitutive disassociation and translocation of the heterotrimeric G-protein
Whole-cell recordings revealed that the NinaEpp100 mutant photoreceptors were constitutively active (Fig. 5B,D), but the weaks uppression of the NinaEpp100-dependent retinal degeneration by norpA or trp suggested that constitutive activation of the light-sensitive channels was not required for the rapid NinaEpp100-dependent retinal degeneration. We examined the extent of the constitutive activation by comparing the GTPase activity in wild-type and NinaEpp100 mutant flies. Whereas we found high levels of light-dependent GTPase activity in wild-type flies, significantly lower light-dependent GTPase activity levels were measured in the NinaEpp100 mutant (Fig. 8A), which suggested that a large fraction of the Gqα-protein was not activated in the NinaEpp100 mutant. We demonstrated previously that the Gqα-protein translocates from the rhabdomeric membrane to the cytosol during blue light illumination (Kosloff et al., 2003). Therefore, we examined whether the Gqα-protein in the constitutively active NinaEpp100 mutant photoreceptors was predominantly in the cytosol. In wild-type flies, ∼80% of the Gqα was associated with the membrane fraction in the dark (Fig. 8B), which represented the GDP-bound Gqα in the heterotrimer. During illumination, ∼70% of the Gqα translocated into the cytosol (Fig. 8B). In contrast, ∼70% of the Gqα was present in the cytosol of the NinaEpp100 mutant, regardless of illumination (Fig. 8B). This persistent and abnormal distribution of the Gqα-protein in the cytosol could partially account for both the absence of light-dependent GTPase activity and the desensitization of the NinaEpp100 mutant photoreceptor cell.
The formation of a stable rhodopsin-arrestin complex is required for the rapid photoreceptor degeneration
The ability of the arr2Y20STOP mutation to significantly reduce the desensitization of the NinaEpp100 photoreceptor and suppress the NinaEpp100 degeneration suggested that the ARR2 protein persistently bound the NINAEpp100 mutant protein (Alloway et al., 2000; Kiselev et al., 2000). ARR2 is predominantly a soluble protein when rhodopsin is not photoactivated (such as during or after orange light illumination) (Fig. 9). However, during blue light conversion of rhodopsin to metarhodopsin, ∼80% of ARR2 protein binds metarhodopsin in the membrane (Fig. 9) (Byk et al., 1993). The ARR2 binding to the membrane is rhodopsin dependent, because only low levels of ARR2 are detected in ninaEI17 null mutant membrane fractions under either light condition (Fig. 9). In contrast, a significant percentage of ARR2 remained associated with the NinaEpp100 membrane, regardless of the light conditions (Fig. 9). The lower percentage of ARR2 in the newly enclosed NinaEpp100 membrane relative to the wild-type membrane after blue light illumination is consistent with the ∼80% reduction of NINAE protein in the NinaEpp100 mutant compared with wild type (Fig. 1F). Thus, the role of ARR2 in the rapid degeneration and strong desensitization in the NinaEpp100 mutant is through its persistent binding to NINAEpp100.
The NinaEpp100 mutant degenerates through two independent pathways that involve the abnormal distribution of the ARR2 and Gqα
Because both the dgq1 and arr2Y20STOP mutations significantly suppressed the rapid NinaEpp100-dependent retinal degeneration (Fig. 6), we examined whether the Gqα-protein was required for the persistent association of ARR2 with the NINAEpp100 protein. We found that the percentage of ARR2 protein present in the membrane fraction of dgq1; NinaEpp100 double mutants was essentially the same as in NinaEpp100 mutant flies (Fig. 9). Because the loss of the Gqα protein did not reduce the level of the persistent NINAEpp100-ARR2 complexes, the dgq1 suppression of NinaEpp100 degeneration must be arrestin independent. This demonstrates that Gqα is involved in a novel mechanism for the rapid NinaEpp100 degeneration, which may involve its predominant and persistent localization in the cytosol. Furthermore, the complete suppression observed in the 40-d-old dgq1; arr2Y20STOP NinaEpp100 triple mutant clearly demonstrates that both the Gqα and ARR2-dependent degeneration pathways are both necessary and sufficient to account for the entire NinaEpp100-dependent degeneration process.
The NinaEpp100 mutant exhibits a novel retinal degeneration phenotype
We isolated a novel dominant rhodopsin mutation, NinaEpp100, that requires the presence of both Gqα and arrestin to induce a rapid degeneration. Transgenic expression of an in vitro-generated glycine-to-arginine mutation at position 299 conclusively demonstrated that this is the molecular cause of the dominant NinaEpp100 mutant phenotype (Fig. 2C). Previously, several dominant rhodopsin mutations were isolated that blocked rhodopsin trafficking and maturation out of the endoplasmic reticulum and slowly degenerated (Colley et al., 1995; Kurada and O'Tousa, 1995; Kurada et al., 1998). Several lines of data demonstrated that the NinaEpp100 mutation did not affect rhodopsin trafficking and maturation. First, the ninaED1 mutation, which is one of the most severe of the previously isolated dominant rhodopsin mutations (Kurada and O'Tousa, 1995; Kurada et al., 1998), suppressed the rapid NinaEpp100 degeneration (Fig. 3). Therefore, the NINAEpp100 protein must exert its effect subsequent to the ninaED1 trafficking defect. Second, NinaEpp100/ninaE+ flies lacked immature forms of rhodopsin on immunoblots (Fig. 1F). Thus, both wild-type and NINAEpp100 mutant proteins must complete their maturation, although the NINAEpp100 protein was present at ∼20% of the NINAE+ protein level (Fig. 1F). Third, electron microscopy independently immunolocalized both epitope-tagged NINAE+ and epitope-tagged NINAEpp100 in the rhabdomeres of NinaEpp100/ninaE+ flies (data not shown), which confirmed that both forms reach the rhabdomere. Therefore, NinaEpp100 represents a new Drosophila model for dominant rhodopsin mutations.
The mechanisms underlying the highly reduced sensitivity to light in the NinaEpp100 mutant and its implications on congenital night blindness (CNB) in humans
The strongly desensitized NinaEpp100 photoreceptor was characterized by the following: (1) the intense light required to elicit a small-amplitude response that lacked the typical bump noise (Fig. 4A), (2) the increased light intensity necessary to reach the saturated amplitude (Fig. 4B), and (3) the shift of the intensity-response relationship toward higher light intensity levels (Figs. 4B, 5A). Although either the 80% decrease in the number of rhodopsin molecules (Fig. 1F) or the 80% decrease of Gqα-protein associated with the membrane (Fig. 8B) would reduce the NinaEpp100 light sensitivity, they are insufficient to account for the observed four orders of magnitude reduction. A more significant mechanism is the constant state of light adaptation that results from the persistent NINAEpp100-ARR2 complex (Fig. 9). Normally, rhodopsin is present at a 5:1 molar ratio to ARR2 (Dolph et al., 1993). However, the NINAEpp100 mutant protein is expressed at only 20% of the wild-type level of rhodopsin in wild-type photoreceptors (Fig. 1F), which yields an ∼1:1 molar ratio with ARR2. Thus, if 30-40% of the ARR2 is associated with the NinaEpp100 membrane, approximately an equivalent percentage of the NINAEpp100 mutant protein is likely in a NINAEpp100-ARR2 complex. This would correspond to ∼35% of the NINAEpp100 being present in the persistent complex, which corresponds to 6-8% of the rhodopsin level in wild-type photoreceptors. This complex could account for the shift in the intensity-response curve toward brighter light (Fig. 5A). The fact that the light response in a fraction of mutant cells reached a normal peak amplitude at intense lights (Fig. 4B) is consistent with a significant number of light-sensitive rhodopsin molecules (i.e., ∼103-105) (Johnson and Pak, 1986) not being bound by arrestin. The NINAEpp100-ARR2 complex would also account for the minimal light-stimulated GTPase activity in the NinaEpp100 mutant photoreceptor cell. However, genetic elimination of ARR2 only partially increased the light sensitivity of the mutant NinaEpp100 (Figs. 4A, 5A), which could be explained by two different mechanisms. First, constitutively active NINAEpp100 would continually open the light-sensitive channels (Fig. 5), which leads to a persistent Ca2+ influx and photoreceptor desensitization. Second, activated Gqα-protein may stimulate the internalization of specific G-protein-coupled receptors in the absence of arrestin binding (Rochdi and Parent, 2003). Thus, constitutively active NINAEpp100 would activate Gqα, which could drive internalization of both wild-type and NINAEpp100 rhodopsin molecules to desensitize the photoreceptor cell.
This strong desensitization response, with only a minor effect on retinal degeneration, is similar to the human CNB phenotype (Dryja, 2000). One form of autosomal dominant CNB results from constitutively active forms of rod opsin (Jin et al., 2003). Three dominant rod opsin mutations were shown to cause a significant shift in the intensity-response relationship of transgenic Xenopus rods (Jin et al., 2003). The constitutive activity and desensitization observed in the NinaEpp100 mutant suggests that the NINAEpp100 mutant rhodopsin may serve as an important model for elucidating the molecular details of one form of autosomal dominant CNB.
A novel mechanism for dominant rhodopsin-mediated retinal degeneration
The NinaEpp100 dominant mutation induces three different degeneration mechanisms. The first involves constitutive opening of the light-sensitive channels. The Drosophila rdgABS12 and trpP365 mutations were shown to constitutively depolarize the photoreceptor cell and cause retinal degeneration (Raghu et al., 2000; Yoon et al., 2000). Furthermore, three different trp alleles (including trp301) significantly slowed the rapid rdgABS12 retinal degeneration, which confirmed that the rdgABS12 degeneration was dependent on the constitutive opening of the light-activated channels (Raghu et al., 2000). However, the constitutive current observed in the NinaEpp100 mutant was approximately fivefold smaller than that found in the Drosophila rdgABS12 and trpP365 mutants (Raghu et al., 2000; Yoon et al., 2000) and was not found in all of the cells (Fig. 5B). Furthermore, the weak suppression by the norpAp24 and trp301 mutations on the rapid NinaEpp100 degeneration (Fig. 7D,E, respectively) confirmed that the constitutive opening of the light-sensitive channels is a minor NinaEpp100 degeneration mechanism.
The second degeneration mechanism involves the formation of a stable NINAEpp100-ARR2 complex. A variety of Drosophila mutations were shown previously to induce retinal degeneration by generating a stable NINAE-ARR2 complex that was endocytosed in a clathrin-dependent manner (Alloway et al., 2000; Kiselev et al., 2000). However, NinaEpp100 is the first Drosophila rhodopsin mutation that persistently targeted ARR2 to the membrane in all light conditions (Fig. 9). On the basis of the arr2Y20STOP mutation suppressing the NinaEpp100 degeneration (Fig. 6), the NINAEpp100-ARR2 complex accounts for a major NinaEpp100 degeneration mechanism. The K296E opsin mutation in transgenic mice exhibited an analogous dominant photoreceptor degeneration phenotype (Robinson et al., 1992; Li et al., 1995; Rim and Oprian, 1995). Although the K296E mutant opsin constitutively activated the heterotrimeric G-protein in cell culture, it failed to constitutively stimulate the phototransduction cascade in the mouse photoreceptor because it was persistently phosphorylated and bound by arrestin. This stable rhodopsin-arrestin complex may be a common stimulator of photoreceptor apoptosis for constitutively active rhodopsin mutations rather than persistent activation of the phototransduction cascade.
The third and novel Drosophila degeneration mechanism requires Gqα but not activation of the visual transduction cascade, because neither the norpAp24 nor trp301 mutations suppressed NinaEpp100 degeneration to the same extent as dgq1 (Figs. 7, 6, respectively). Furthermore, the inability of the dgq1 mutation to disrupt the stable NINAEpp100-ARR2 complex (Fig. 9) and the consistent presence of 70% of the Gqα-protein in the cytosolic fraction of both NinaEpp100 and arr2Y20STOP NinaEpp100 mutants (Fig. 8B and data not shown, respectively) demonstrated that the Gqα- and ARR2-dependent degeneration mechanisms were independent. The light-independent localization of most of the Gqα-protein to the NinaEpp100 cytosol is consistent with the constitutively active NINAEpp100 stimulating Gqα dissociation from Gβγ and the subsequent depalmitoylation of Gqα (Kosloff et al., 2003). The cytosolic localization and the markedly decreased light-dependent GTPase activity in the NinaEpp100 mutant (Fig. 8) suggests that the NINAEpp100 protein either diminishes the interaction between the cytosolic Gqα-GTP with the membrane-associated norpA-encoded phospholipase C effector molecule, which also functions as a GTPase activating protein (Cook et al., 2000), or suppresses the reassociation of the cytosolic Gqα with the membrane. Because the loss of Gqα-protein significantly slowed the NinaEpp100 degeneration, the persistent cytosolic localization of Gqα induces the degeneration rather than its absence from the membrane.
Whereas most previously characterized retinal degeneration models identified a single major pathway, the NinaEpp100 mutant identified two major degeneration mechanisms, the stable NINAEpp100-ARR2 complex and the Gqα-dependent pathway. The ability of ninaED1 to prevent the rapid NinaEpp100-dependent degeneration (Fig. 3) is consistent with the requirement of NINAEpp100 to reach the rhabdomere and interact with the ARR2 and Gqα proteins. Recently, knock-out mice revealed that light-induced degeneration also involved two mechanisms (Hao et al., 2002), one requiring the transducin α subunit and the other involving photoactivated metarhodopsin and/or its bleaching intermediates (Hao et al., 2002). Thus, the NinaEpp100 mutant may reveal not only the complex interactions between constitutive activation and desensitization but also the multiplicity of mechanisms that may be involved in some of the rhodopsin-dependent degeneration mutations.
This work was supported by National Eye Institute Grants R01-EY12426 (D.R.H.) and R01-EY 03529 (B.M., Z.S.), a United States-Israel Binational Science Foundation grant (B.M., D.R.H.), and Israel Science Foundation grants (B.M., Z.S.). We thank Dr. Alexander Kiselev and Dr. Patrick Dolph for sharing their anti-ARR2 antisera and Dr. Joseph O'Tousa for kindly providing the ninaED1 Drosophila mutant. We also thank Shahar Frechter for providing unpublished data.
Correspondence should be addressed to David R. Hyde, Department of Biological Sciences, Galvin Life Science Building, University of Notre Dame, Notre Dame, IN 46556-0369. E-mail:.
Copyright © 2004 Society for Neuroscience 0270-6474/04/242516-11$15.00/0
↵* R.I. and I.C.-O. contributed equally to this work.