Address correspondence and reprint requests to Yoshitaka Fukada, Department of Biophysics and Biochemistry, Graduate School of Science, The University of Tokyo, Hongo 7-3-1, Bunkyo-ku, Tokyo 113–0033, Japan. E-mail: firstname.lastname@example.org
Retinal cone cells exhibit distinctive photoresponse with a more restrained sensitivity to light and a more rapid shutoff kinetics than those of rods. To understand the molecular basis for these characteristics of cone responses, we focused on the opsin deactivation process initiated by G protein-coupled receptor kinase (GRK) 1 and GRK7 in the zebrafish, an animal model suitable for studies on retinal physiology and biochemistry. Screening of the ocular cDNAs identified two homologs for each of GRK1 (1A and 1B) and GRK7 (7–1 and 7–2), and they were classified into three GRK subfamilies, 1 A, 1B and 7 by phylogenetic analysis. In situ hybridization and immunohistochemical studies localized both GRK1B and GRK7-1 in the cone outer segments and GRK1A in the rod outer segments. The opsin/GRKs molar ratio was estimated to be 569 in the rod and 153 in the cone. The recombinant GRKs phosphorylated light-activated rhodopsin, and the Vmax value of the major cone subtype, GRK7-1, was 32-fold higher than that of the rod kinase, GRK1A. The reinforced activity of the cone kinase should provide a strengthened shutoff mechanism of the light-signaling in the cone and contribute to the characteristics of the cone responses by reducing signal amplification efficiency.
The phototransduction mechanisms are well documented in rod photoreceptor cells (Baylor 1996; Nathans 1999; Arshavsky et al. 2002; Ridge et al. 2003): Photoexcited rhodopsin activates transducin, which in turn activates 3′, 5′-cyclic GMP (cGMP)-phosphodiesterase (PDE). PDE thus activated hydrolyzes cGMP, leading to the closure of cGMP-gated cation channels for plasma membrane hyperpolarization. As the regulation for the deactivation process, G-protein coupled receptor kinase 1 (GRK1) phosphorylates photoexcited rhodopsin at its carboxyl-terminal region, to which arrestin binds for shortening the lifetime of photoexcited rhodopsin (Maeda et al. 2003). On the other hand, the deactivation of transducin is accelerated by regulator of G-protein signaling 9–1 (RGS9-1) through regulation of the kinetics of GTP hydrolysis. The signaling proteins in rods have their counterparts in cones, and hence a similar phototransduction mechanism have been postulated in cones (Cobbs et al. 1985; Hurwitz et al. 1985; Lerea et al. 1986; Ong et al. 1995; Baylor 1996; Zhu et al. 2003; Kennedy et al. 2004). Nevertheless, rods and cones exhibit distinctive light responses: Rods have several hundred times higher sensitivity to light than cones, whereas cones recover from photoexcitation much faster than rods (Baylor 1996), and the molecular mechanism(s) underlying these differences largely remains to be elucidated. The activation kinetics of transducin by cone pigments was similar to that by rhodopsin (Fukada et al. 1989; Starace and Knox 1997), and the photoresponse recorded from the transgenic Xenopus rods expressing cone pigment was indistinguishable from that of the wild-type rods (Kefalov et al. 2003). In addition, cGMP-hydrolyzing activity of cone PDE was similar to that of rod PDE (Gillespie and Beavo 1988; Booth et al. 1991), implying less contribution of the molecular properties of these components to the distinctive photoresponses. A more recent study has enabled isolation of rod and cone lysates from the carp retina, and suggested remarkably faster deactivation kinetics of light-activated opsins in cones than in rods (Tachibanaki et al. 2001). It is conceivable that such a difference would characterize the light response properties of cones by accelerating the shutoff kinetics and by reducing the amplification efficiency of the light signal when compared to those in rods. For evaluating this hypothesis, it is important to compare qualitatively and quantitatively the signaling proteins in the deactivation process between rods and cones. One of the components to be investigated is GRK7, which is regarded as a cone opsin kinase particularly in diurnal vertebrates due to its localization in cone outer segments (Hisatomi et al. 1998; Weiss et al. 1998; Chen et al. 2001; Weiss et al. 2001) along with its ability to phosphorylate rhodopsin (Weiss et al. 1998; Chen et al. 2001) and M cone opsin (Liu et al. 2005) in vitro. A more recent report by Rinner et al. (2005) has shown that the knockdown of GRK7 by morpholino delayed the recovery of the light response of larval zebrafish cones, showing the importance of GRK7 for the normal photoresponse of the cone.
In order to understand the molecular basis for the characteristic photoresponse of cones, we focused on GRK1 and GRK7 in the zebrafish (Danio rerio), an animal model having a cone-enriched retina. The present findings provide molecular keys to understanding the vertebrate cone function, and exploit a way of genetic approaches to the issue in this animal model possessing typical photopic visual system.
Materials and methods
Cloning of zebrafish GRK cDNAs
Partial sequences of both GRK1A and GRK7-2 genes were found in the zebrafish whole genome shotgun database, while cDNA fragments for GRK1B and GRK7-1 were found in the zebrafish EST database (also see Results). Total RNA isolated from the zebrafish eyes by using TRIzol reagent (Invitrogen, Corp., Carlsbad, CA, USA) was reverse-transcribed by ThermoScript (Invitrogen, Corp., Carlsbad, CA, USA). Based on the partial sequences found in the databases, we prepared gene-specific primers and performed both 5′- and 3′-RACE using the zebrafish ocular cDNAs as the template. Subsequently, RT-PCRs were performed by using gene-specific primers for 5′- and 3′-flanking regions of each GRK-coding sequence, and the entire coding sequences were determined. More than five independent clones were sequenced for each GRK subtype to eliminate PCR errors. The nucleotide sequences for the zebrafish GRK cDNAs have been deposited in the DNA Data Bank of Japan under DDBJ Accession Numbers (GRK1A, AB212993; GRK1B, AB212994; GRK7-1, AB212995; GRK7-2, AB212996).
The GRK sequences used for the phylogenetic analysis were obtained from GenBank except for those from Fugu (Fugu rubripes) and zebrafish (see Results). A phylogenetic tree of the vertebrate GRKs was constructed by the Neighbor-joining method. Because the amino acid sequences of both the N- and C-terminal regions showed only limited homology among members across the GRK subfamilies (e.g. between GRK1A and GRK4), these regions were eliminated from the calculation to exclude possible artefacts caused by inappropriate alignment. Therefore, we used the amino acid sequences of the kinase domain corresponding to the amino acid positions from 189 to 459 of zebrafish GRK1A (Fig. 1a). Any gaps present in the kinase domain were also eliminated from the calculation. The confidence of the tree was evaluated statistically by bootstrap analysis, in which a thousand trees were generated.
In situ hybridization analysis
In situ hybridization was performed as described previously (Mano et al. 1999) by using digoxigenin-labeled cRNA probes with 10-µm frozen ocular sections prepared from light-adapted zebrafish. The cRNA probes for detecting the zebrafish GRK mRNAs were 325–383 bases in length, and the nucleotide sequences of the probes were less than 63% identical with those in the corresponding region of any other GRK subtypes.
The eyes were isolated from zebrafish kept under the room light (light-adapted eye) or in darkness (dark-adapted eye) for more than 6 h, and subjected to preparation of frozen sections. The ocular sections prepared in 10-µm thickness (Wada et al. 1998) were incubated for 30 min at room temperature in a blocking solution composed of 0.1% Triton X-100 and 1.5% normal goat serum in phosphate-buffered saline (PBS; 15 mm Na-phosphate; pH 7.4, 140 mm NaCl), and then incubated either with G8 mouse monoclonal antibody (Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA; 0.2 µg/mL) raised against human GRK1 or with anti-GRK7 rabbit polyclonal antibody (1 : 500 dilution) raised against a partial sequence of carp GRK7 (Tachibanaki et al. 2005) for 24 h at 4°C. The sections were washed with PBS and incubated for 24 h at 4°C with anti-mouse IgG (NEW ENGLAND BioLabs, Inc., Beverly, MA, USA) or anti-rabbit IgG antibody (Cell Signaling Technology, Inc., Beverly, MA, USA) conjugated with horseradish peroxidase. Positive signals were visualized by using a Vectastain Elite ABC kit (Vector Laboratories, Inc., Burlingame, CA, USA) with a substrate solution of 0.4 mg/mL diaminobenzidine and 0.003% H2O2 in 50 mm Tris-HCl (pH 7.5).
Quantitative analyses of ocular GRK proteins in the zebrafish eye
Each of the cDNAs for the four GRKs was modified so as to append both His-tag and HA-tag to the N-terminus of each recombinant GRK protein (His-HA-GRK). These proteins were expressed in Escherichia coli, purified by Ni2+-chelating chromatography, and used as standards for quantitative analyses of untagged GRKs in immunoblots, in which the immunopositive signals for G8 and anti-GRK7 antibodies were detected by CDP-Star (NEW ENGLAND BioLabs, Inc., Beverly, MA, USA). The X-ray films used for chemiluminescent detection were scanned and the densitometry measurements were performed with the aid of Image Gauge ver. 4.0 software (Fuji Photo Film, Co., Ltd, Tokyo, Japan). The immunoblot analysis by using G8 antibody showed a good linearity of the staining densities of the positive bands in a concentration range of 0.8–3.2 ng/lane, and the antibody exhibited equivalent reactivities to both His-HA-GRK1A and His-HA-GRK1B proteins within the range. Similarly, anti-GRK7 antibody showed a linearity in a range of 2.8–9.2 ng/lane, and exhibited equivalent reactivity to both His-HA-GRK7-1 and His-HA-GRK7-2 proteins within the range. The eyes were isolated from the adult zebrafish and homogenized on ice with an extraction buffer (50 mm Tris-HCl; pH 7.5, 1 mm dithiothreitol, 50 µg/mL leupeptin, 50 µg/mL aprotinin, 1 mm benzamidine), and the protein concentration was estimated by the method of Bradford (Bradford 1976). The ocular homogenate was mixed with 4-fold protein amount of the zebrafish brain homogenate showing undetectable levels of immunoreactivities to G8 and anti-GRK7 antibodies. Similarly, the purified recombinant proteins were mixed with the brain homogenate to eliminate any difference in masking effects by the other proteins between the samples.
Spectrophotometric measurements of photopigments in the zebrafish eye
Fifty eyes were isolated from the zebrafish kept in darkness for 4 h, homogenized on ice with 2 mL of buffer P (50 mm HEPES; pH 6.6, 140 mm NaCl, 1 mm dithiothreitol, 10 µg/mL leupeptin, 10 µg/mL aprotinin) and centrifuged at 110 000 × g for 20 min at 4°C. The precipitate was homogenized with 2 mL of buffer P, separated into two tubes and centrifuged. Then each of the resulting precipitate was extracted with 420 µL of buffer P containing 0.75% (w/v) 3-[(3-cholamidopropyl) dimethylammonio]-1-propanesulfonate (CHAPS) and centrifuged at 110 000 × g for 20 min at 4°C. This extraction procedure was repeated once more, and 200 µL of each extract was subjected to spectrophotometric measurements. The content of each visual pigment in the extract was estimated from the absorption spectra recorded by the partial bleaching method by using irradiation light from a 1-kW tungsten lamp with cutoff and bandpass filters (Okano et al. 1989). The red-sensitive (LWS-1 and LWS-2), green-sensitive (RH2-4 and rhodopsin), blue-sensitive (RH2-1, RH2-2 and RH2-3), violet-sensitive (SWS2) and ultraviolet-sensitive (SWS1) visual pigments were bleached by irradiations of light with wavelengths longer than 646 nm (> 646 nm), > 572 nm, > 524 nm, > 443 nm and at 324–351 nm, respectively. This partial bleaching experiment was performed in the presence of 10 mm hydroxylamine at 4°C. On the other hand, the rhodopsin content was estimated from the difference absorption spectrum before and after complete bleaching by light (> 524 nm) of the sample after incubation with 100 mm hydroxylamine for 1 h at 22°C. This hydroxylamine-treatment thermally bleached all of the cone visual pigments in the chicken retina without light irradiation (Okano et al. 1989). In the case of the zebrafish visual pigments, we confirmed that the red-sensitive visual pigments were completely bleached by the hydroxylamine-treatment, and hence we assumed that the same treatment bleached the green-sensitive cone pigment as well. The absorption spectrum of the retinal-oxime overlaps with those of both the violet-sensitive (SWS2) and ultraviolet-sensitive (SWS1) visual pigments at the wavelengths of their absorption maxima, 416 and 355 nm, respectively (Chinen et al. 2003). To estimate the accurate optical densities of these visual pigments at the absorption maxima, each of the difference absorption spectrum recorded by the partial bleaching was corrected by subtracting the retinal-oxime's absorption spectrum multiplied by a factor that yields the absorption maximum of the corrected spectrum at 416 or 355 nm. As a measure of the amount of a visual pigment, we use the unit ΔOD·ml, which is a maximal optical density (ΔOD) of the difference spectrum between the pigment and its photoproduct or a product after the hydroxylamine-bleaching (retinal oxime plus opsin) multiplied by the volume (ml).
RT-PCR-based semiquantitative measurement of GRK mRNA levels
The zebrafish ocular mRNAs were reverse-transcribed by using primers designed from the sequences conserved between GRK1A and GRK1B (GGCTTGAGGTCTCT) or between GRK7-1 and GRK7-2 (AGAAGAGTGCAAG). The resulting cDNA was amplified by PCRs with primer pairs designed for the conserved sequences that are located on the 5′-side of the primer-binding site for the reverse-transcription: A primer set for both GRK1A and GRK1B was GTGGCCAACTCTGC (forward primer) and CTGAGGTCTCCTCC (reverse), and a set for both GRK7-1 and GRK7-2 was TCAGTGGACTGGTGGGC (forward) and TGTCAATCATCTCCTGTTGC (reverse). The PCRs with these primers amplified a mixture of two cDNA fragments of 803 bp (GRK1A) plus 800 bp (GRK1B) or 463 bp (GRK7-1) plus 451 bp (GRK7-2). We verified that the amplification efficiencies were indistinguishable between the two GRK fragments in control experiments, in which the plasmid DNAs harboring each of the GRK cDNAs were amplified independently. In the quantitative experiments, the two amplified cDNAs were discriminated from each other by digestion with gene-specific restriction enzymes, EagI for digestion of GRK1A-derived fragment, EcoRI for GRK1B, EcoNI for GRK7-1, and NcoI for GRK7-2. For quantification, the cDNA mixture containing an amplified pair of GRK1A plus GRK1B or GRK7-1 plus GRK7-2 was digested by either of the paired enzymes separately, and the two digests were subjected to the polyacrylamide-gel electrophoresis, followed by staining with SYBR GREEN I (TaKaRa Bio, Inc., Shiga, Japan). The intensities of the undigested DNA bands in the paired samples were measured by FLA2000 (Fuji Photo Film, Co., Ltd, Tokyo, Japan). Stoichiometry and specificity of this analysis were validated in each experiment by good coincidence of the total amount of the original cDNA fragment (prior to digestion) with the summed amount of each undigested cDNA fragment remained after the independent digestion by the two enzymes.
In vitro kinase assay
HEK293S cells were maintained in DMEM/F12 medium (Invitrogen, Corp., Carlsbad, CA) containing 10% fetal bovine serum (Invitrogen, Corp., Carlsbad, CA) in culture plates of 10 cm in diameter. The cells were transiently transfected by using LipofectAMINE PLUS (Invitrogen, Corp., Carlsbad, CA) with 4.0 µg of a pcDNA3.1 plasmid vector (Invitrogen, Corp., Carlsbad, CA) harboring one of the GRK cDNAs. Twenty-four hours after the transfection, the cells were collected from 8 plates, washed with ice-cold PBS and homogenized on ice with 1 mL extraction buffer (50 mm Tris-HCl; pH 7.5, 1 mm dithiothreitol, 50 µg/mL leupeptin, 50 µg/mL aprotinin, 1 mm benzamidine). The cell lysate was then centrifuged at 48 000 × g for 30 min at 4°C, and the supernatant (termed soluble fraction) was subjected to in vitro kinase assay with a substrate of urea-stripped bovine ROS membranes prepared as described previously (Shichi and Somers 1978). A reaction mixture was composed of the urea-stripped ROS membranes at a final concentration of 20 µm rhodopsin and the HEK293S cell extract containing various concentrations of GRK protein, in 50 mm Tris-HCl (pH 7.5) with 6 mm MgCl2, 1 mm dithiothreitol, 1 mm ethylene glycol bis-N,N,N′,N′-tetraacetic acid, 50 µg/mL leupeptin, 50 µg/mL aprotinin and 1 mm benzamidine. The protein concentration of the extract from GRK-expressing HEK293S cells was adjusted at a final protein concentration of 3.75 mg/mL by adding extract of non-transfected HEK293S cell. Then the mixture was irradiated with an intense orange light (> 524 nm) from a 1-kW tungsten lamp for 1 min on ice. One minute after the light irradiation, phosphorylation reaction was initiated by adding [γ-32P] ATP at a final concentration of 100 µm (∼3.7 kBq/µL). The reaction was carried out at 25 or 30°C under dim red light, and terminated at various time points by mixing 20 µL aliquots of the reaction mixture with 5 µL SDS-sample buffer (50 mm Tris-HCl; pH 6.8, 30% glycerol, 10% SDS, 250 mm dithiothreitol, 10 mm EDTA and 0.1% Coomassie Brilliant Blue R-250). These samples were subjected to SDS-PAGE, and phosphorylation levels of light-activated rhodopsin were estimated from the band densities in autoradiogram of the gel after subtracting the band density of the corresponding area of a control sample using non-transfected HEK293S cell extract. For determination of Km and Vmax values, the concentrations of light-activated rhodopsin in the reaction mixtures were varied in a range from 1.25 to 10 µm by two distinct methods termed diluting method and limited bleaching method. In the diluting method, the reaction mixtures contained various concentrations of urea-stripped ROS membrane, in which rhodopsin was fully bleached by irradiation of the intense orange light (> 524 nm) as mentioned above. In the limited bleaching method, on the other hand, the ROS membrane content was kept constant at rhodopsin concentration of 6.4 µm in the reaction mixture, and the bleach level of rhodopsin was varied by changing the irradiation time (5–40 s) of an intense red light (> 572 nm) or by using a neutral density filter (20% transmittance). The resulting bleach levels were determined spectroscopic measurements of the aliquot of each reaction mixture. The phosphorylation reactions were terminated at 20, 40 and 60 s. Similar sets of experiments were performed with the recombinant GRKs at concentrations varied from 1.0 to 3.0 µg/mL (for GRK1A) or from 0.06 to 0.2 µg/mL (for GRK7-1). The Km and Vmax values of each GRK were determined by averaging the data from these experiments performed for each GRK.
Partial purification of GRK proteins
GRK7-1 protein in the membrane fraction was extracted with the extraction buffer containing 2% CHAPS and the ‘membrane extract’ was dialyzed against the extraction buffer containing 0.005% or 0.01% Tween 80. The soluble fractions of GRK1A and GRK7-1 were also dialyzed with the 0.005% or 0.01% Tween 80-containing buffer. Fractions dialyzed against 0.005% Tween 80-containing buffer were used for the kinase assay, whereas those dialyzed with 0.01% Tween 80-containing buffer were subjected to anion-exchanger MonoQ (5/50 GL, GE Healthcare Bio-sciences) column chromatography with 20 mm Tris–HCl (pH 8.0) buffer containing 0.01% Tween 80. GRK7-1 in the membrane extract was loaded on to the column, to which a linear gradient of 0–250 mm NaCl was applied in total of 5 mL of the column buffer. Then GRK7-1 was eluted with a step gradient of 450 mm NaCl in the buffer at a flow rate of 0.5 mL/min. On the other hand, GRKs in the soluble fractions were eluted with a step gradient of 600 mm NaCl in the column buffer after application of a linear gradient of 0–450 mm NaCl formed in 9 mL of the buffer. A peak fraction of each chromatography was dialyzed with the extraction buffer containing 0.005% Tween 80 and used for the kinase assay.
Novel classification of GRKs
To clone the zebrafish GRK cDNAs, we first performed TBLASTN search against the fugu genomic database by using cDNA sequences for medaka GRK1 (OlGRK-R) and GRK7 (OlGRK-C) as the query sequences (Hisatomi et al. 1998). We found three and two genes that showed high similarity to GRK1 and GRK7, respectively. These fugu sequences were used as the queries for further in silico screening of the zebrafish whole genome shotgun and EST databases, and we found that all the obtained DNA fragments should constitute four zebrafish GRK genes; two GRK1 and two GRK7 homologous genes. The full-length cDNA sequences were determined by a series of 5′- and 3′-RACE and subsequent RT-PCR analyses using the zebrafish ocular mRNAs as templates (Fig. 1a and 2a).
A phylogenetic analysis of the amino acid sequence in the kinase domain among vertebrate GRKs revealed that two fugu GRK1 homologs and one of the two zebrafish GRK1s were clustered with mammalian GRK1s, whereas the remaining GRK1 homologs of both zebrafish and fugu were clustered with chicken GRK1 (Fig. 1b). These phylogenetic relationships suggest that an ancestral gene of GRK1s duplicated once at the early stage of vertebrate evolution (Fig. 1b, node a) into common ancestors of two GRK1 subfamilies, GRK1A and GRK1B. On the other hand, the ancestral GRK7 genes for the two orthologs found in both zebrafish and fugu probably diverged from a common ancestor at a stage (Fig. 1b, node b) after the divergence of the teleost from the common ancestor of the other vertebrates (Fig. 1b, node c). Therefore these two GRK7 orthologs were designated GRK7-1 and GRK7-2 in a style that is different from that adopted for GRK1. Taken together, the four members were classified into three GRK subfamilies, 1 A, 1B and 7 (Fig. 1b). According to this nomenclature, the two fugu orthologs of GRK1A that we found were designated GRK1A-1 and GRK1A-2 (Fig. 1b).
Cellular distribution of zebrafish GRKs
In the zebrafish, the genes encoding the four subtypes of GRK (1A, 1B, 7–1 and 7–2) are transcribed not only in the eye but also in the pineal gland (Fig. 2a), which is a well-characterized photoreceptive organ expressing a photoreceptive molecule, exorhodopsin (Mano et al. 1999). The two GRK7 genes (GRK7-1 and GRK7-2), but not GRK1 genes, are also expressed in the brain devoid of the pineal gland (Fig. 2a), implying a role of GRK7-1 and GRK7-2 in regulating G protein-coupled receptors other than visual opsin. Cellular distribution of GRK mRNAs in the zebrafish eye was investigated by in situ hybridization analyses. The positive signals for GRK1A, GRK1B and GRK7-1 mRNAs were detected in the retinal outer nuclear layer (ONL) composed of the nuclei of the photoreceptor cells (Fig. 2d-f). The signal for GRK7-2 mRNA was not detected by in situ hybridization analysis, most probably due to the lower expression level in the eye (Fig. 4h). The nuclei of the rods and cones are distributed in two distinct layers within the ONL of the zebrafish retina (Branchek and Bremiller 1984; Mano et al. 1999), and in situ hybridization with cRNA probes for rhodopsin (rod opsin) and red (a long wavelength-sensitive cone opsin) demonstrates that the rod nuclei locate at the inner half (at the vitreous side) of the ONL in the light-adapted zebrafish retina, whereas those of cones are at the outer half (at the sclera side) of the ONL (Figs 2b,c). Taking advantage of this distinguishable distribution, we were able to identify a cell-type of the photoreceptors showing the positive signals: The mRNA for GRK1A was localized in the rods, whereas GRK1B and GRK7-1 mRNAs were in the cones (Figs 2d,f).
To determine the subcellular localizations of GRK proteins in the zebrafish rod and cone cells, we performed immunohistochemical analyses using G8 monoclonal antibody and anti-GRK7 polyclonal antibody. The specificities of these antibodies were studied by immunoblot analyses, in which G8 antibody recognized both GRK1A and GRK1B recombinant proteins and anti-GRK7 antibody reacted with both GRK7-1 and GRK7-2 (Supplemental Fig. 1). In the zebrafish retinal sections, G8 antibody strongly immunostained the rod and cone outer segments, which were located alongside in the outer segment layer in the dark-adapted retina (Fig. 3g), but were distinctly visible in two different layers at the sclera and vitreous side, respectively, in the light-adapted retina (Fig. 3b). On the other hand, anti-GRK7 antibody immunostained the cone outer segments with no detectable signals in the rod outer segments of the light-adapted (Fig. 3d) and dark-adapted retina (Fig. 3i). These observations together with the in situ hybridization data localized GRK1A protein expression in the rod outer segments (Fig. 2d), whereas GRK1B and GRK7-1 proteins were localized in the cone outer segments (Figs 2e,f). The anti-GRK7 antibody also immunostained nerve fiber-like structures in the inner plexiform layer (Fig. 3e), implying the expression of GRK7-1 and/or GRK7-2 protein in the non-rod/non-cone neurons in the zebrafish retina.
Expression levels of GRKs in the zebrafish eye
The protein levels of GRKs in the zebrafish eye were assessed by immunoblot analyses with G8 antibody (Fig. 4a–c) and the anti-GRK7 antibody (Fig. 4d–f). The estimated amount of GRK1A plus GRK1B proteins both immunoreactive to G8 antibody was 40.8 ± 2.1 ng/eye (mean ± SEM, Fig. 4g), and that of GRK7-1 plus GRK7-2 proteins both immunoreactive to anti-GRK7 antibody was 63.5 ± 1.6 ng/eye (Fig. 4g). Because neither of the GRK antibodies discriminate between GRK1A and GRK1B nor between GRK7-1 and GRK7-2, it is not feasible to compare the relative amounts of the two subtypes in the eyes at the protein level. Instead, we developed RT-PCR-based semiquantitation method (see Materials and methods) to determine the relative abundance of their mRNAs. Then we found that either of the two subtypes of GRKs is preferentially transcribed in a particular tissue: In the eye, the expression level of GRK1A mRNA was 5.3 times higher than that of GRK1B (Fig. 4h), while GRK7-1 mRNA level was 8.1 times higher than that of GRK7-2 (Fig. 4h). Intriguingly, the relative abundance between the two subtypes was reversed in the pineal gland, that is, GRK1B mRNA level was 13.3 times higher than that of GRK1A (Fig. 4h), whereas GRK7-2 mRNA level was 1.6 times higher than that of GRK7-1 (Fig. 4h). The brain was a characteristic tissue where a single subtype, GRK7-2, among the four GRKs was almost selectively transcribed (25.1 times higher than GRK7-1). Assuming that the relative abundance of their protein levels reflects that of their mRNA levels, we could conclude that GRK1A and GRK7-1 are the predominantly expressed subtypes of the rod and cone cells, respectively.
To estimate the molar ratios of the GRK proteins to their substrate opsins in both the rod and cone cells, we estimated the contents of the visual pigments in the zebrafish eye by the spectroscopic analysis. The zebrafish has eight cone visual pigments (LWS-1, LWS-2, RH2-1, RH2-2, RH2-3, RH2-4, SWS2 and SWS1) and a single rod pigment, rhodopsin (Chinen et al. 2003). Depending on the spectral sensitivities, they were classified into five groups (Fig. 5), the red-sensitive (LWS-1 and LWS-2), green-sensitive (rhodopsin and RH2-4), blue-sensitive (RH2-1, RH2-2 and RH2-3), violet-sensitive (SWS2) and ultraviolet-sensitive (SWS1) pigments. The cone pigment content of each group in the CHAPS-solubilized ocular extract was determined by a partial bleaching method (see Materials and methods) by adopting the molar extinction coefficient of chicken red-sensitive cone pigment (Okano et al. 1992). On the other hand, the rhodopsin content was estimated spectrometrically by adopting the molar extinction coefficient of chicken rhodopsin (Okano et al. 1992), after incubation of the extract for 1 h at 22°C with 100 mm hydroxylamine, which would bleach RH2-4 and the other cone pigments (see Materials and methods). The difference spectrum of the green-sensitive visual pigments (rhodopsin and RH2-4) determined by the partial bleaching (0.0061 ΔOD·ml/eye at the absorption maximum; ΔOD·ml is a unit of amounts of the visual pigments, see Materials and methods and ref. 26) was nearly identical to that of rhodopsin (0.0062 ΔOD·ml/eye) measured after the hydroxylamine-bleaching (100 mm at 22°C), suggesting a low protein level of RH2-4 (Table 1). The estimated value of rhodopsin content was 308 pmol/eye, whereas the summation of cone visual pigments was 154 pmol/eye (Table 1).
Table 1. The visual pigments in the zebrafish eye
Maximum wavelength of diff. abs. spectrum (nm)
Optical density at abs. max. (OD·ml/eye)
*The difference spectrum of the green-sensitive visual pigments (rhodopsin plus RH2-4) determined by the partial bleaching was nearly identical to that of rhodopsin measured after the hydroxylamine-bleaching, suggesting a low protein level of green-sensitive cone pigment. This speculation is based on the assumption that the green cone pigment is thermally bleached by incubation at 22°C for 1 h in the presence of 100 mm hydroxylamine, which completely bleached the red-sensitive cone pigments. If the green cone pigment is stable under the condition, the amount of rhodopsin might be lower than the estimated value (in this Table), to which the green cone pigment would contribute.
**The absorption spectrum of the retinal-oxime overlaps with those of both the violet-sensitive (SWS2) and ultraviolet-sensitive (SWS1) visual pigments at the wavelengths of their absorption maxima, 416 and 355 nm, respectively (Chinen et al. 2003). To estimate the accurate optical densities of these visual pigments at the absorption maxima, each of the difference absorption spectrum recorded by the partial bleaching was corrected by subtracting the retinal-oxime's absorption spectrum multiplied by a factor that yields the absorption maximum of the corrected spectrum at 416 or 355 nm.
Recombinant proteins of the four subtypes of zebrafish GRKs were expressed in HEK293S cells and their enzymatic properties were investigated by determining rhodopsin phosphorylation kinetics in vitro. Rhodopsin was employed as the common substrate of the in vitro phosphorylation reaction in order to highlight the molecular difference between the GRKs. It has been reported that the N-terminal region and C-terminal isoprenylation are both important for the full activity of the GRK1 (Inglese et al. 1992; Yu et al. 1999). Therefore, non-tagged recombinant GRK proteins were expressed in the HEK293S cells, in which every GRK proteins were composed of closely migrated multiple bands detected by immunoblot analyses. Each of the cell lysate was centrifuged to separate soluble and membrane fractions. GRK7-1 protein was detected in both the soluble and membrane fractions whereas most of GRK1A protein was recovered in the soluble fraction. These three fractions contained comparable composition of the closely migrated multiple bands of each GRK (see Supplemental Table 1 for their activities).
In the kinase assay, each of the four GRK subtypes in the soluble fraction phosphorylated bovine rhodopsin in the urea-stripped rod outer segment membranes in a light-dependent manner (Fig. 6a), displaying a characteristic property of opsin kinase. Light-dependent phosphorylation of rhodopsin was also detected in the control experiment with the non-transfected HEK293S cell extract (Fig. 6a, control). This could be due to the rhodopsin-phosphorylating activity endogenously present in the HEK293S cells because the light-activated rhodopsin could be phosphorylated by various kinases. The band density in the autoradiogram due to this rhodopsin-phosphorylating activity was subtracted from those obtained in experiments with the cell extract containing the recombinant GRK (see Materials and methods). The GRK protein concentrations in the HEK293S cell extracts were quantified by immunoblot analyses with G8 and anti-GRK7 antibodies (Fig. 6c-f), and specific activities of the four GRK subtypes for the rhodopsin phosphorylation reaction were determined (Table 2, Figs 6g,h). In these assays, the rhodopsin concentration was fixed at 20 µm, which is much higher than the Km values (4 µm for rhodopsin in the ROS membrane and 0.6 µm for the reconstituted rhodopsin) reported for GRK1 purified from bovine retina (Palczewski et al. 1988; Kelleher and Johnson 1990). We found that the specific activity of GRK7-1 (645 ± 85 nmol/min/mg GRK7-1 at 25°C) was remarkably higher than those of GRK1A (22.5 ± 1.5 nmol/min/mg GRK1A), GRK1B (12.5 ± 3.0 nmol/min/mg GRK1B) and GRK7-2 (9.5 ± 2.5 nmol/min/mg GRK7-2). We then estimated apparent Km and Vmax values of GRK1A and GRK7-1 that are the representative GRKs of rods and cones, respectively. We employed two different experimental methods, which were termed diluting method and limited bleaching method (see Materials and methods). In the diluting method, the reaction mixtures contained various amounts of the urea-stripped ROS membrane, of which rhodopsin molecules were fully light-activated. This is a conventional method that has been employed to determine the kinetics parameters of mammalian GRK1 (Palczewski et al. 1988). In the limited bleaching method, on the other hand, we varied the substrate concentration by changing the bleach level of rhodopsin with a fixed amount of the urea-stripped ROS membrane in the reaction mixtures. In other words, the membrane:GRK ratios are different among the reaction mixtures in the diluting method, whereas the ratio is constant in the limited bleaching method. Variations in GRK activity among the recombinant proteins were evaluated by experiments with three or four independent preparations of GRK1A and GRK7-1 which were expressed separately, but variations among the preparations were not significant, and we were able to obtain reliable kinetics data that are fitted with Michaelis-Menten relationships (Figs 6g,h). In Lineweaver-Burk plots, the Km values for rhodopsin were calculated to be 2.3 ± 0.3 µm for GRK1A and 4.4 ± 1.2 µm for GRK7-1 in the diluting methods, whereas the values determined by the limited bleaching method were 0.4 ± 0.1 µm for GRK1A and 0.8 ± 0.2 µm for GRK7-1, which were, respectively, about five times lower those estimated by the diluting method (Table 3). On the other hand, the Vmax values were almost insensitive to the methods; 24.0 ± 0.8 (the former method) and 26.3 ± 3.8 nmol/min/mg (the latter method) for GRK1A, whereas 773.6 ± 111.0 (the former method) and 831.1 ± 75.9 nmol/min/mg (the latter method) for GRK7-1 (Table 3). Based on these Vmax values, we estimated the catalytic rate constant, Kcat values, of GRK1A and GRK7-1 to be 1.5 ± 0.5 min−1 and 48.1 ± 6.9 min−1, respectively (the diluting method), or 1.7 ± 0.2 min−1 and 51.6 ± 4.7 min−1, respectively (the limited bleaching method).
Table 2. Specific activities of the four GRK subtypes for rhodopsin phosphorylation
Initial velocity (nmol phosphate incorporated into Rh/min/mg GRK protein)
The reactions were performed at 25°C or 30°C, and the values determined at 25°C are presented as the mean ± SEM from three independent experiments.
22.5 ± 1.5
12.5 ± 3.0
645 ± 85
9.5 ± 2.5
Table 3. Kinetics parameters of zebrafish GRK1A and GRK7-1 for the light-activated rhodopsin phophorylation
The kinetics parameters for rhodopsin phosphorylation reactions were determined by the diluting method and limited bleaching method, and the Km and Vmax values were calculated by the Lineweaver-Burk analyses. The values are shown as the mean ± SEM from four (GRK1A) or five (GRK7-1) independent experiments by the limited bleaching method or from five (GRK1A) or six (GRK7-1) independent experiments by the diluting method.
24.0 ± 0.8
26.3 ± 3.8
2.3 ± 0.3
0.4 ± 0.1
773.6 ± 111.0
831.1 ± 75.9
4.4 ± 1.2
0.8 ± 0.2
GRK7-1 protein was also detected in the membrane fraction as described above, and we estimated its kinase activity with the fixed concentration of light-activated rhodopsin (20 µm). The specific activity of GRK7-1 extracted from the membrane (935 ± 113 nmol/min/mg GRK protein, n = 3) was slightly higher than that in the soluble fraction (679 ± 82 nmol/min/mg GRK protein, n = 3). The specific activity of GRK7-1 in the membrane extract increased markedly after partial purification by anion-exchange chromatography on MonoQ column (3677 ± 440 nmol/min/mg GRK protein, n = 3). With respect to GRK7-1 in the soluble fraction, on the other hand, the specific activity after the similar chromatographic purification was 475 ± 28 nmol/min/mg GRK protein (n = 3).
Vertebrate GRKs had been classified into seven subfamilies from GRK1 to GRK7, and most of the members in each subfamily were shown to form a monophyletic group (Hisatomi et al. 1998; Zhao et al. 1999), suggesting that the ancestral gene of each GRK subfamily diverged from one another at the latest at the early stage of the vertebrate lineage. It has been pointed out, however, that chicken GRK1 shows a curious phylogenetic relationship among the members of GRK1 (Zhao et al. 1999). In the present study, the newly analyzed members of GRK1 (in the two species, zebrafish and fugu) were included in the phylogenetic analysis, which now supports strongly the novel classification of the GRK1 genes into two subclasses, GRK1A and GRK1B, each of which appears to form a monophyletic group (Fig. 1b). The topology of the phylogenetic tree suggested that the two GRK1 subtypes diverged at the early lineage of all the vertebrates at the latest before the teleost lineage diverged from a common ancestor of the others. Evidently, GRK1A and GRK1B should be considered as the equivalent rank of categories to the other GRK subfamily (i.e. GRK2-GRK7). In our hands, no GRK1B ortholog was found in the mammalian genomic database, and vice versa no GRK1A ortholog was found in the chicken genome, suggesting that either of these two GRK genes have been lost in the mammalian or avian lineage.
All the four GRK subtypes cloned in the present study have the C-terminal consensus sequence for isoprenylation, termed CaaX motif (Fig. 1a), where C, a and X designate cysteine, aliphatic, and any amino acid, respectively. The cysteine residue in the CaaX motif is modified with either the farnesyl or geranylgeranyl depending on the X residue (Casey et al. 1991; Kinsella et al. 1991; Reiss et al. 1991; Clarke 1992; Fu and Casey 1999). The farnesylation-signal sequence is conserved among the vertebrate members of GRK1A, whereas geranylgeranylation-signal sequence is conserved among GRK1Bs and GRK7s in the vertebrates (see also Fig. 1a). Interestingly, the two types of CaaX motif in GRK1A, GRK1B and GRK7-1 appear to correspond to the distributions of the protein; that is, GRK1A (to be farnesylated) was expressed in the rods whereas both GRK1B and GRK7-1 (to be geranylgeranylated) were in the cones of the zebrafish (Fig. 2d-f), although the cellular distribution of GRK 7–2 (to be geranylgeranylated) was not determined in the present study. A mutation in the CaaX motif altered localization of bovine GRK1A in vitro; the wild-type farnesylated GRK1A translocates from cytosol to ROS membranes in a light-dependent manner, while the geranylgeranylated GRK1A mutant mainly present in the membrane fraction regardless of light conditions (Inglese et al. 1992). In the other members of retinal proteins, the γ-subunit of rod transducin (Gγ1 and cone transducin (Gγ8) are farnesylated (Fukada et al. 1990; Lai et al. 1990; Ong et al. 1995), whereas most other subtypes of heterotrimeric G protein γ subunits expressed in non-photoreceptor cells are geranylgeranylated (Matsuda et al. 1998; Matsuda and Fukada 2000), and such a difference in the isoprenylation appears to affect the function of G-protein (Matsuda et al. 1994; Matsuda et al. 1998; Myung et al. 1999; Fogg et al. 2001; Kassai et al. 2005). Therefore, cell type-specific isoprenylation may contribute to characteristics of the cell function, and as well the different isoprenylation of GRK proteins in the zebrafish rods and cones could influence their characteristic photoresponse properties.
In a number of diurnal vertebrates such as dog, pig, ground squirrel and medaka, GRK1A and GRK7 are selectively expressed in the rods and cones, respectively (Hisatomi et al. 1998; Weiss et al. 1998; Weiss et al. 2001). Consistently, zebrafish GRK1A and GRK7-1 were found in the rods and cones, respectively (Figs 2d, f). On the other hand, GRK1B, a member of the new subclass of GRK1 was colocalized with GRK7-1 in the zebrafish cone outer segments (Figs 2e,f and 3b,d,g,i). Similar coexpression of two GRK subtypes in the cone outer segments was observed in human and rhesus monkey (Chen et al. 2001; Weiss et al. 2001), and we speculate that, in some species, the cone cells may be potentiated with the activity of phosphorylating opsin by expressing multiple GRK subtypes in a random additive manner. It is interesting to note that nocturnal rodents seem to have lost GRK7 gene (Weiss et al. 2001), when we consider the relationship between the GRK species and the animal habitat (see the related Discussion below).
The summation of the ocular protein levels of GRK7-1 plus GRK7-2 was 1.6-fold higher than that of GRK1A plus GRK1B (Fig. 4g). Further molecular quantification with discrimination of each subtype was difficult due to their structural similarity, and therefore we estimated the relative protein abundance from their mRNA levels instead (Fig. 4) on the assumption that the protein levels reflect the mRNA levels. This assumption is not generally true but is known to be applicable in some cases (Bernard et al. 1997; Yamakuni et al. 1998). No significant sequence signals for destruction of mRNA were found in any of the four cDNA sequences. Together with the results of protein quantification (Fig. 4g), we estimated that the protein level of the rod kinase (GRK1A) was 0.54 pmol/eye and that of the cone kinases (GRK7-1 plus GRK1B) was 1.01 pmol/eye. The rod:cone ratio in GRK protein levels per eye was not so much different from the ratio in the cell number, 1 : 1.1 (Supplemental Fig. 2). On the other hand, the protein amount of the substrate, rhodopsin, was two-fold higher than the summation of those of the cone opsins in the zebrafish eye (Table 1). Taking consideration of GRK contents per eye, we estimated the molar ratio of the opsins to the GRKs (opsins/GRKs) to be 569 in the rod and 153 in the cone. The ratio in the cone was 3.7 times lower than that in the rod, and the relative enrichment of GRK molecules in the cone may contribute to a more rapid time-to-peak for cone opsin phosphorylation observed in the intact zebrafish retina, when compared to that for rhodopsin phosphorylation (Kennedy et al. 2004). In addition to GRK molecules, the expression level of RGS9-1, a key component that regulates lifetime of the active form of transducin, is much higher in cone cells of some mammalian species, such as chipmunk, monkey and human, than in their rod cells (Cowan et al. 1998; Zhang et al. 1999; Zhang et al. 2003). The high expression levels of these key components, such as GRKs and RGS9-1 regulating the kinetics of the deactivation process, may be responsible in part for the characteristic photoresponse properties of the cone cells.
The difference in Km values between GRK1A and GRK7-1 for light-activated rhodopsin was rather small; they were within a range of two fold in the both experimental methods tested, but the values were notably down-shifted in a parallel manner by employing the limited bleaching method (Table 3). This parallel shift in Km value may be due to the difference in the membrane:GRK ratio between the two methods because the binding of GRK to the membrane probably influences the efficiency of the association (interaction) between the enzyme GRK and the substrate rhodopsin. On the other hand, the Vmax value for the phosphorylation reaction of GRK7-1 was 32-fold higher than that of GRK1A (Table 3) in the both methods, suggesting an important contribution of the difference in Vmax to the characteristic photoresponses from the rod and cone cells. Recently, Horner et al. (2005) showed that human GRK1 and GRK7 share similar kinetic properties to light-activated rhodopsin by using FLAG-tagged recombinant GRKs. Apparently, their data do not come into line with our finding of marked difference in the enzymatic activities between GRK1A and GRK7-1. One possible explanation may begin with the idea noticing the species difference between these studies, human and zebrafish. Because a common ancestor of mammalian species is suggested to acquire once a nocturnal habit during the evolution (Bowmaker 1998), the high shutoff activity of cones should have become unnecessary at that time, and consequently GRK activity might have been reduced in the cones. At the later stage when some of the mammalian ancestors were returned to diurnal, they may have employed another strategy such as reinforcement of RGS9-1 expression level in the cones (as observed in the primates) in order to potentiate the rapid shutoff process of the cone response without restoration of the high GRK activity in the cone. This is an explanation from the viewpoint of species difference of GRKs. Alternatively, the variation might be due to the difference in the experimental conditions such as the presence and absence of FLAG tag. Because it has been suggested that N-terminal tagging may change the enzymatic activity of GRK protein (Yu et al. 1999), we employed non-tagged GRK proteins in the present study to determine their intrinsic activities. In our experiments, however, it is difficult to exclude the possibility that the soluble fraction of HEK293S cell lysate contain a factor that may influence GRK1A activity. To test this idea, anion-exchange MonoQ column chromatography was performed for three GRK preparations as the starting materials, that is, the soluble fractions of GRK1A, the soluble fractions of GRK7-1, and GRK7-1 in the membrane extract. The specific activity of GRK7-1 partially purified from membrane fraction was remarkably high (3677 ± 440 nmol/min/mg GRK protein), while that purified from the soluble fraction was modest (475 ± 28 nmol/min/mg GRK protein). On the other hand, GRK1A partially purified from the soluble fraction showed extremely low specific activity, a level that was unmeasurable under the condition and far lower than that of GRK7-1 partially purified from the soluble fraction. Based on these results, we concluded that the intrinsic activity of GRK7-1 (in both the membrane and soluble fraction) is considerably higher than that of GRK1A in the soluble fraction. Since farnesylated protein is generally far less hydrophobic than geranylgeranylated protein (Kassai et al. 2005), the farnesylated GRK1A should be recovered in the soluble fraction, in which most of the expressed GRK1A protein was recovered. In fact, bovine GRK1 that is farnesylated is recovered in the soluble fraction of the retinal lysate (Shichi and Somers 1978; Palczewski et al. 1988; Kelleher and Johnson 1990). Even taking consideration of the isoprenyl modification, it is most probable that GRK7-1 shows significantly higher activity than GRK1A. Very recently, a marked difference in activity between carp GRK1 and GRK7 has been reported though their kinetics parameters were not defined (Tachibanaki et al. 2005). A remarkable difference in kinase activity between GRK1A and GRK7-1 determined in the present study, together with the difference in opsin/GRKs ratio between the rods and cones, illustrates that the cones are equipped with significantly higher activity for terminating the light signaling than the rods, at least in the zebrafish. This finding supports the hypothesis that the intense shutoff activity determines the characteristic light response properties of cones not only by accelerating the recovery kinetics of the response but also by reducing the amplification of the light signal.
The expressions of GRK mRNAs were also detected in the zebrafish pineal gland and brain (Fig. 2). In many vertebrates including zebrafish, both the pineal gland and the deep regions of the brain harbor the photoreceptor cells such as pinealocytes and deep brain photoreceptors, respectively, both expressing opsins (Wada et al. 1998, 2000; Mano et al. 1999; Kojima et al. 2000). The relative abundance of the GRK mRNAs levels varied among the tissues (Fig. 4); both GRK1B and GRK7-2 were the dominant subtypes expressed in the pineal gland, whereas GRK7-2 was expressed almost exclusively in the brain, suggesting a role of these kinases phosphorylating non-visual opsins in the particular extra-retinal tissues.
We thank Mr M. Sagara (University Tokyo) for his help in some experiments and Drs S. Kawamura and Y. Shimauchi-Matsukawa (Osaka University) for anti-GRK7 antibody. We also thank Drs F. Tokunaga and O. Hisatomi (Osaka University) for providing unpublished cDNA sequence data of medaka GRK. This work was supported in part by a Human Frontiers Science Program grant (to YF), by Grants-in-Aid for Scientific Research from the Japanese Ministry of Education, Sports, Science, and Technology (to YF and TO), and by a research fellowship of the Japan Society for the Promotion of Science for Young Scientists (to YW).