Phosphorylation of the InaD gene product, a photoreceptor membrane protein required for recovery of visual excitation.

In an approach directed to isolate and characterize key proteins of the transduction cascade in photoreceptors using the phosphoinositide signaling pathway, we have isolated the Calliphora homolog of the Drosophila InaD gene product, which in Drosophila InaD mutants causes slow deactivation of the light response. By screening a retinal cDNA library with antibodies directed against photoreceptor membrane proteins, we have isolated a cDNA coding for an amino acid sequence of 665 residues (M = 73,349). The sequence displays 65.3% identity (77.3% similarity) with the Drosophila InaD gene product. Probing Western blots with monospecific antibodies directed against peptides comprising amino acids 272-542 (anti-InaD-(272-542)) or amino acids 643-655 (anti-InaD-(643-655)) of the InaD gene product revealed that the Calliphora InaD protein is specifically associated with the signal-transducing rhabdomeral photoreceptor membrane from which it can be extracted by high salt buffer containing 1.5 M NaCl. As five out of eight consensus sequences for protein kinase C phosphorylation reside within stretches of 10-16 amino acids that are identical in the Drosophila and Calliphora InaD protein, the InaD gene product is likely to be a target of protein kinase C. Phosphorylation studies with isolated rhabdomeral photoreceptor membranes followed by InaD immunoprecipitation revealed that the InaD protein is a phosphoprotein. In vitro phosphorylation is, at least to some extent, Ca-dependent and activated by phorbol 12-myristate 13-acetate. The inaC-encoded eye-specific form of a protein kinase C (eye-PKC) is co-precipitated by antibodies specific for the InaD protein from detergent extracts of rhabdomeral photoreceptor membranes, suggesting that the InaD protein and eye-PKC are interacting in these membranes. Co-precipitating with the InaD protein and eye-PKC are two other key components of the transduction pathway, namely the trp protein, which is proposed to form a Ca channel, and the norpA-encoded phospholipase C, the primary target enzyme of the transduction pathway. It is proposed that the rise of the intracellular Ca concentration upon visual excitation initiates the phosphorylation of the InaD protein by eye-PKC and thereby modulates its function in the control of the light response.

In an approach directed to isolate and characterize key proteins of the transduction cascade in photoreceptors using the phosphoinositide signaling pathway, we have isolated the Calliphora homolog of the Drosophila InaD gene product, which in Drosophila InaD mutants causes slow deactivation of the light response. By screening a retinal cDNA library with antibodies directed against photoreceptor membrane proteins, we have isolated a cDNA coding for an amino acid sequence of 665 residues (M r ‫؍‬ 73,349). The sequence displays 65.3% identity (77.3% similarity) with the Drosophila InaD gene product. Probing Western blots with monospecific antibodies directed against peptides comprising amino acids 272-542 (anti-InaD-(272-542)) or amino acids 643-655 (anti-InaD-(643-655)) of the InaD gene product revealed that the Calliphora InaD protein is specifically associated with the signal-transducing rhabdomeral photoreceptor membrane from which it can be extracted by high salt buffer containing 1.5 M NaCl. As five out of eight consensus sequences for protein kinase C phosphorylation reside within stretches of 10 -16 amino acids that are identical in the Drosophila and Calliphora InaD protein, the InaD gene product is likely to be a target of protein kinase C. Phosphorylation studies with isolated rhabdomeral photoreceptor membranes followed by InaD immunoprecipitation revealed that the InaD protein is a phosphoprotein. In vitro phosphorylation is, at least to some extent, Ca 2؉dependent and activated by phorbol 12-myristate 13acetate. The inaC-encoded eye-specific form of a protein kinase C (eye-PKC) is co-precipitated by antibodies specific for the InaD protein from detergent extracts of rhabdomeral photoreceptor membranes, suggesting that the InaD protein and eye-PKC are interacting in these membranes. Co-precipitating with the InaD protein and eye-PKC are two other key components of the transduction pathway, namely the trp protein, which is proposed to form a Ca 2؉ channel, and the norpA-encoded phospholipase C, the primary target enzyme of the transduction pathway. It is proposed that the rise of the intracellular Ca 2؉ concentration upon visual excitation initiates the phosphorylation of the InaD protein by eye-PKC and thereby modulates its function in the control of the light response.
Phototransduction by rhabdomeral photoreceptors, particularly of Drosophila compound eyes, has become an important model system for the ubiquitous phosphoinositide-mediated signal transduction. The progress achieved in this field is based on the powerful genetic and molecular biological techniques available for Drosophila, which have been successfully complemented by biochemical studies in other flies such as Calliphora and Musca. Despite the rapid progress that has been achieved in the understanding of sensory transduction mechanisms in recent years, the phototransduction cascade operating in this type of sensory cells has not yet been entirely resolved. In particular, the biochemical processes regulating the recovery and adaptation of the visual response in rhabdomeral photoreceptors are still obscure.
Extracellular Ca 2ϩ enters the photoreceptors through ion channels and is required for rapid recovery of visual excitation (1)(2)(3)(4)(5). A major portion of the Ca 2ϩ influx into the photoreceptor cell appears to be carried by a Ca 2ϩ -selective class of channels that depend on, or may indeed be formed by, the transient receptor potential (trp) 1 protein (6,7). The primary structure of this trp protein was identified simultaneously by Wong et al. (8) and Montell and Rubin (9), and the trp gene product was shown to be localized to the rhabdomeral photoreceptor membranes. Direct measurements of the extracellular Ca 2ϩ concentration revealed a decline of extracellular Ca 2ϩ upon illumination in the eyes of wild type flies, which is significantly reduced in trp mutants (4). Signal transduction is also impaired in two other Drosophila mutants, inaC and InaD, which were originally classified as inactivation-no afterpotential mutants by Pak (10). While the InaD gene product is an 80-kDa protein of unknown function (11), the inaC gene was shown to encode an eye-specific protein kinase C (eye-PKC) (12,13). Thus, it is tempting to assume that the Ca 2ϩ -dependent deactivation of the visual response is controlled by phosphorylation of photoreceptor-specific proteins associated with the rhabdomeral photoreceptor membrane. Identified proteins that undergo lightdependent phosphorylation are rhodopsin and arrestin 2 (14 -17). However, neither rhodopsin nor arrestin 2 were found to be phosphorylated by the inaC protein because phosphorylation of activated rhodopsin is not enhanced by Ca 2ϩ (14,15), and the Ca 2ϩ -stimulated phosphorylation of arrestin 2 has been shown to result from the activation of a Ca 2ϩ -calmodulin-dependent protein kinase (17). Accordingly, the target proteins of eyespecific protein kinase C have yet to be specified.
In an attempt to identify proteins that are part of the biochemical pathway in rhabdomeral photoreceptors, we have used an immunological approach to isolate Calliphora cDNA clones, which encode photoreceptor membrane proteins. Antibodies directed against purified rhabdomeral membranes (18) were generated and employed for the isolation of genes encoding rhabdomere-specific proteins. By this approach, we have cloned the Calliphora homologs of the Drosophila InaD, inaC, and trp genes. In the present paper we show that the InaD gene product is associated with the rhabdomeral photoreceptor membrane and that it is a putative substrate of eye-PKC. We also provide for the first time evidence for an interaction among eye-PKC, InaD protein, trp protein, and the norpA-(no receptor potential A) encoded phospholipase C.

EXPERIMENTAL PROCEDURES
Fly Stocks-Calliphora erythrocephala Meig., chalky mutant, was reared on bovine liver to maintain a high rhodopsin content in the eyes. Adult male flies were raised at 25°C in a 12 h light/12 h dark cycle and were used for the experiments at an age of 8 -10 days posteclosion.
Generation of Antibodies-Immunization of rabbits was performed according to standard protocols (19). Isolated rhabdomes of 700 Calliphora eyes were used for each of the four injections. Final blood sampling was 4 months after the first injection. The obtained antiserum was purified on protein A-agarose columns (Bio-Rad Life Technologies, Munich) as described (19).
Polyclonal anti-InaD antibodies were generated as follows: a DNA fragment encoding the 23 C-terminal amino acids of the Calliphora InaD protein was amplified by polymerase chain reaction from cloned cDNA using sequence-specific primers. The polymerase chain reaction product was cloned into the expression vector pQE40 (Qiagen, Hilden/ Germany) in frame with six His codons and the dihydrofolate reductase gene. Fusion proteins were expressed in Escherichia coli M15 (pREP4), extracted with urea and purified on Ni ϩ -agarose columns according to the manufacturer's instructions. Purified fusion proteins were dialyzed against phosphate-buffered saline (137 mM NaCl, 3 mM KCl, 8 mM Na 2 HPO 4 , 2 mM KH 2 PO 4 , pH 7.2) and used for the immunization of a rabbit (200 g of protein/injection). Antibodies were purified from the antiserum by affinity chromatography on HiTrap columns (Pharmacia, Freiburg/Germany), which had previously been coupled with 1 mg of the antigen as described by the manufacturer. The purified antibodies are hereafter referred to as anti-InaD-(643-665), according to the InaD peptide from which they were raised. A second anti-InaD antibody (anti-InaD-(272-542)) was generated and purified in the same way by using a recombinantly expressed peptide comprising amino acids 272-542 of the Calliphora InaD protein.
For the production of anti-Calliphora trp antibodies a partial Calliphora cDNA clone encoding the C-terminal half of the trp protein was expressed in E. coli M15 (pREP4), and the expression product was used as an antigen. Antibodies directed against the Calliphora inaC protein and the ␣-subunit of the eye-specific G-protein were raised against bovine serum albumin-coupled synthetic peptides (CYMNPEFITMI and QNALKEFNLG, respectively), which correspond to the C-terminal region if these proteins. Antibodies directed against the Drosophila norpA-encoded phospholipase C and against the ␤-subunit of the Gprotein, which also detect the corresponding Calliphora proteins, were a generous gift of R. Shortridge (20) and J. B. Hurley (21).
Construction of a Calliphora Retinal cDNA Library, Immunoscreening, and Sequencing-A Calliphora retinal cDNA library was prepared in the UniZAP XR vector (Stratagene, Heidelberg/Germany) according to the manufacturer's instructions, using poly(A) ϩ RNA isolated from 500 Calliphora retinae. Screening of the library with antibodies against rhabdomeral proteins was performed as described by Sambrook et al. (22). Expression of recombinant proteins was induced by applying nitrocellulose filters preincubated in 1 mM isopropyl-␤-D-thiogalactopyranoside 3 h after plating the phages. After an additional incubation for 4 h at 37°C the filters were removed from the plates, washed briefly in Tris-buffered saline (20 mM Tris/HCl, pH 7.5, 150 mM NaCl), blocked for 2 h at 25°C in 3% bovine serum albumin in Tris-buffered saline, and incubated overnight at 25°C in the same solution containing 1 ⁄1000 volume of antiserum. Binding of primary antibodies was detected using alkaline phosphatase-conjugated protein A and with nitro blue tetrazolium/X-phosphate as a chromogen. Positive clones were rescreened, and plasmid DNA was obtained by in vivo excision. The nucleotide sequence of the longest cDNA clone was determined for both strands by the dideoxy chain termination method (23) using templates generated by nested deletions.
Isolation of Photoreceptor Membranes, SDS-PAGE, and Western Blot Analysis-Isolation of total eye membranes and rhabdomeral photoreceptor membranes was performed as described previously (15,18). Low salt extractions were carried out in 3 mM EGTA, 1 mM dithiothreitol in 5 mM sodium phosphate buffer, pH 6.2, for 10 min on ice. 50 mM sodium phosphate buffer, pH 6.2, containing 1.5 M NaCl was used for high salt extractions of purified photoreceptor membranes. After complementing extracts with 5 ϫ SDS-PAGE buffer (1 ϫ SDS-PAGE buffer: 4% SDS, 1% 2-mercaptoethanol, 1 mM EDTA, 15% glycerol in 65 mM Tris/HCl, pH 6.8) or solubilizing membrane proteins in 1 ϫ SDS-PAGE buffer, the proteins were separated by SDS-PAGE according to Laemmli (24) on 8 -20% gradient gels (Pharmacia Midget System). Following SDS-PAGE, proteins were transferred to polyvinylidene difluoride membranes (Bio-Rad), and Western blot analysis was performed using standard protocols (19).
Immunoprecipitation of the InaD Protein-Proteins were extracted from purified rhabdomeral membranes of 30 Calliphora eyes in 30 l of Triton X-100 buffer (1% Triton X-100, 150 mM NaCl, 50 mM Tris/HCl, pH 8.0, and 1 mM phenylmethylsulfonyl fluoride) for 15 min at 4°C. The extract was added to 10 l of protein A/G-agarose beads (Pierce), which had previously been incubated with anti-InaD-(272-542) for 1 h. Immunoprecipitation was performed for 2 h at 4°C and was followed by four washes with 500 l of Triton X-100 buffer. Precipitated proteins were eluted from protein A/G-agarose beads with 15 l of 1 ϫ SDS-PAGE buffer for 10 min at 80°C and were subjected to SDS-PAGE and Western blot analysis.
Protein Phosphorylation and Dephosphorylation-The standard assay for protein phosphorylation was carried out in a buffer containing Hepes-buffered saline (115 mM NaCl, 2 mM KCl, 10 mM Hepes), pH 6.8, 2 mM MgCl 2 , 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, 125 M EGTA, 250 M CaCl 2 , and purified photoreceptor membranes from 10 fly retinae per sample. When indicated, phorbol 12-myristate 13-acetate or bisindolylmaleimide I (Calbiochem, Bad Soden/Germany) were added at a final concentration of 1 M or 0.2 M, respectively. The phosphorylation reactions were started by the addition of 2 mM ATP supplemented with 2 Ci [␥-32 P]ATP (Amersham Buchler, Braunschweig). The free Ca 2ϩ concentration in these assays, calculated according to Fabiato (25), was 60 M. Phosphorylation reactions that contained nominally zero Ca 2ϩ were supplemented with 2 mM EGTA to remove internal Ca 2ϩ , and no external Ca 2ϩ was added. The soluble fraction of retinal proteins used in recombination experiments was obtained by homogenizing retinae in a small volume (about 0.5 l/ retina) of 1 mM phenylmethylsulfonyl fluoride in water and subsequently separating the soluble and particulate fraction by centrifugation at 50,000 ϫ g for 10 min. Aliquots of soluble proteins of six retinae were added per sample in recombination experiments. Rhabdomeral photoreceptor membranes were prepared under dim red light. For activating light-dependent metarhodopsin phosphorylation, samples were illuminated with blue light for 2 min immediately before the reactions were started. In some cases blue light illumination was omitted as noted in the figure legends. If not indicated otherwise, the phosphorylation was carried out for 5 min at 20°C in the dark. Thereafter, membranes were sedimented at 13,000 ϫ g at 4°C for 10 min, and proteins were extracted with high salt buffer, Triton X-100 buffer, or SDS-PAGE buffer and were subjected to SDS-PAGE or were immunoprecipitated as described above. For measuring phosphorylation time courses, reactions were terminated by adding 5 ϫ SDS-PAGE buffer at the indicated times, and the whole sample was subjected to SDS-PAGE. The amount of protein loaded was visualized by staining the gels with Coomassie Blue, and protein phosphorylation was detected by autoradiography using Kodak BiomaxMR films. Quantification of the relative amount of radioactivity present in a protein band was performed with a phosphor imager (FUJIX BAS 1000, Fuji). For determining the stoichiometry of phosphorylation the radioactivity of cut-out protein bands was measured in a scintillation counter. The amount of InaD protein present in the InaD protein band was calculated by laser densitometry using bovine serum albumin as a standard.

Isolation and Characterization of Calliphora
InaD cDNAs-Antibodies directed against proteins of the fly photoreceptor membrane had been generated by immunizing rabbits with rhabdomes (i.e. a subcellular fraction composed of the rhabdomeral photoreceptor membranes and the intraommatidial matrix; see Ref. 18) isolated from 2800 Calliphora eyes. The resulting antiserum was used to immunoscreen a Calliphora retinal cDNA library. Out of 280,000 clones screened, 200 clones expressed polypeptides that reacted with the antiserum. Partial sequencing analysis revealed that the positive clones isolated so far encode at least six different proteins.
Work presented in this study focuses on clones that show homology to the recently published Drosophila InaD gene (11).
We obtained 14 cDNA clones coding for the Calliphora homolog of InaD and determined the nucleotide sequence of the longest cDNA for both strands. This clone contained a 201-base pair 5Ј-untranslated region, a 1995-base pair open reading frame encoding a polypeptide of 665 amino acids (M r ϭ 73,349), and a 194-base pair 3Ј-untranslated region. The translation initiation site was assigned arbitrarily to the first AUG of the open reading frame at nucleotide 202, which is preceded by a stop codon at nucleotide 196, and fits well with the consensus sequence for translation initiation sites in Drosophila, (C/ A)AA(A/C)AUG (26). Alignment of the deduced amino acid sequence of the Calliphora cDNA clone with the Drosophila InaD sequence (Fig. 1a) shows that the two proteins display 65.3% overall amino acid identity and 77.3% similarity if conservative substitutions are taken into account. Furthermore, both proteins share similar biophysical characteristics, i.e. the same predicted isoelectric point of 8.6, a high abundance of basic (Lys, Arg, His) and acidic (Asp, Glu) amino acids that together comprise more than 30% of the polypeptide, and similar hydrophilicity profiles (Fig. 1b) that reveal no stretches of hydrophobic sequences of 20 or more residues in length.
Two repeats of 40 amino acids (underlined in Fig. 1a) that were shown to share limited sequence homology with the Drosophila disc-large (Dlg), the rat post-synaptic density protein (PSD95), the vertebrate tight junction protein ZO-1, and the human ROS protein (see Ref. 11 and references therein), are highly conserved (90% similarity) between the Calliphora and Drosophila InaD proteins, implying a common functional role within the family of proteins that contain these repeats. On the other hand, repeats consisting of Gly-(Gln/Met), which are present in the Drosophila InaD sequence between amino acids 142 and 158, are not found in the Calliphora sequence. Indeed, the region between residues 106 and 183 is the least conserved part of the two proteins. The only common feature within this region is the relatively high abundance of glutamine residues. Another striking sequence motif, which is present in the Calliphora and the Drosophila InaD protein, is the highly hydrophilic cluster of lysine and glutamate residues between amino acids 454 and 473 (indicated by bars in Fig. 1b). While potential phosphorylation sites of cAMP-and cGMP-dependent protein kinase and tyrosine kinase present in the Drosophila sequence at Thr 169 and Tyr 203 , respectively, are not found in the Calliphora sequence, eight potential phosphorylation sites of protein kinase C are conserved (Fig. 1a). Finally, two potential glycosylation sites at Asn 168 and Asn 312 (Drosophila) or Asn 162 and Asn 303 (Calliphora) are found at similar positions in both sequences.
Identification and Localization of the InaD Protein by Monospecific Antibodies-In order to obtain more detailed information on the function of InaD in fly phototransduction, it is crucial to know whether the InaD gene product is a membrane protein, and if so whether it resides in the rhabdomeral membrane. Isolation of the InaD clone by means of an anti-rhabdom serum already suggests that the InaD protein is associated either with rhabdomeres or with the intraommatidial (extracellular) matrix of the photoreceptor cells. The InaD protein was identified on Western blots using monospecific anti-InaD-(643-665) and anti-InaD-(272-542) antibodies that were raised against peptides containing the 23 C-terminal amino acids and amino acids 274 -542 of the Calliphora InaD protein, respectively. Both antibodies bind to a single protein with an apparent molecular mass of 75 kDa (Fig. 2). The apparent molecular mass of 75 kDa is in line with the molecular mass deduced from the cDNA sequence (73,349 Da). This demon-strates that the InaD protein is present in total eye membranes and in purified photoreceptor membranes (Fig. 2, lanes 1 and  3). It is detected neither in the fraction containing soluble proteins obtained from whole retinas after extraction with low salt buffer nor in extracts containing proteins of the intraom- matidial matrix (Fig. 2, lanes 2 and 4). The latter extract was prepared by extraction of purified photoreceptor membranes with a low salt buffer containing EGTA. However, the InaD protein is extracted from the rhabdomeral photoreceptor membrane if a high salt buffer containing 1.5 M NaCl is used (Fig. 2,  lane 5). As is expected using this cloning procedure, the Western blot indicates that the InaD protein is enriched in the photoreceptor membrane preparation as compared with total eye membranes. Taken together, the hydrophilic character of the InaD protein predicted by the sequence data, and its solubilization by a high salt buffer reveals that InaD is a peripheral photoreceptor membrane protein.
Purification of the InaD Protein by Immunoprecipitation-In order to purify the InaD protein by immunoprecipitation rhabdomeral membranes were treated with a buffer containing 1% Triton X-100, which quantitatively extracted the InaD protein from nonsoluble material. Anti-InaD-(643-665) failed to immunoprecipitate the InaD protein. Thus, we generated an antiserum that was directed against a different part of the InaD protein (anti-InaD-(272-542)) and could successfully be used for immunoprecipitation (Fig. 3). Resolving the immunoprecipitates obtained with anti-InaD-(272-542) by SDS-PAGE revealed that, in addition to the InaD protein band, two other protein bands with apparent molecular masses of 140 and 80 kDa were immunoprecipitated (Fig. 3a, lane 3). None of these proteins was precipitated in control experiments in which protein A/G beads alone were used (Fig. 3a, lane 4). The 140-kDa protein band turned out to represent a double band when resolved on 8% polyacrylamide gels (data not shown). Western blot analysis (Fig. 3b) showed that this protein band reacted with antibodies specific for the Calliphora trp protein and for the norpA-encoded phospholipase C. The 80-kDa protein represents the eye-specific protein kinase C (inaC protein). The immunoprecipitates were also probed with antibodies specific for the ␣and ␤-subunit of an eye-specific G-protein. These G-protein subunits were not detected in the immunoprecipitates. Since anti-InaD-(272-542) does not cross-react with rhabdomeral proteins other than the InaD protein on Western blots, the co-precipitation of inaC, trp, and norpA proteins by anti-InaD-(272-542) suggests that these proteins are complexed permanently or transiently with the InaD protein in the photoreceptor membranes.
Phosphorylation of the Calliphora InaD Protein-The co-immunoprecipitation of the InaD protein with an eye-specific protein kinase C prompted us to investigate whether or not the InaD gene product is a phosphoprotein. In order to test this hypothesis, we made use of the ability to enrich the protein by high salt extraction of purified photoreceptor membranes. In the experiment depicted in Fig. 4, the InaD protein was extracted with high salt buffer after performing phosphorylation of photoreceptor membrane proteins under the standard conditions described under "Experimental Procedures." The extracted peripheral proteins, as well as integral membrane proteins, were subjected to SDS-PAGE and autoradiographed. Of the seven protein bands detected in the high salt extract after staining the gel with Coomassie Blue, four proteins are phosphorylated. The most prominent of these phosphoproteins shows an apparent molecular mass of about 75 kDa, a value corresponding to the apparent molecular mass of the InaD protein. Autoradiography of a duplicate blot and subsequent probing of the very same blot with anti-InaD-(643-665) demonstrated that the radioactively labeled protein band at 75 kDa represents the InaD protein band (Fig. 4b).
In order to rule out that a phosphoprotein other than the InaD protein is present in the high salt extracts and has the same electrophoretic mobility as the InaD protein upon separation by SDS-PAGE, phosphorylated InaD protein was also purified by immunoprecipitation (Fig. 5). The presence of radioactive phosphate in the 75-kDa protein band, which was obtained by resolving the anti-InaD-(272-542) immunoprecipitates by SDS-PAGE and which was identified as the InaD protein with anti-InaD-(643-665), clearly demonstrated that the InaD protein is a phosphoprotein. The stoichiometry of InaD protein phosphorylation, determined for the InaD protein present in high salt extracts as described under "Experimental Procedures," was 0.4 -0.5 mol of phosphate/mol of InaD protein. Hence, a substantial fraction of the InaD molecules was not phosphorylated in the in vitro assays, which may indicate that a fraction of InaD molecules is isolated in a phosphorylated form or is compartmentalized in membrane vesicles to which externally added ATP or activators of the protein kinase have no access.
Since eye-PKC co-immunoprecipitating with the InaD pro-  tein is assumed to be a Ca 2ϩ -dependent protein kinase, we tested the effect of Ca 2ϩ on the phosphorylation of the InaD protein and compared its phosphorylation with phosphorylation of rhodopsin and arrestin, which has been studied previously (14 -17) (Fig. 6). Lowering of internal Ca 2ϩ by the addition of 2 mM EGTA to the phosphorylation assay significantly reduced the incorporation of radioactive phosphate into the InaD protein as compared with standard phosphorylation assays performed at a calculated free Ca 2ϩ concentration of 60 M. Under these conditions, the phosphorylation of two other proteins involved in signal transduction is affected by lowering the free Ca 2ϩ concentration. First, the amount of phosphorylated arrestin 2 associated with the rhabdomeral photoreceptor membrane is decreased. This finding is consistent with earlier reports indicating that arrestin 2 is phosphorylated by a Ca 2ϩ -calmodulin-dependent protein kinase (17). Secondly, the phosphorylation of rhodopsin is affected by calcium concentration. There the amount of radioactive phosphate attached to metarhodopsin is reduced in the presence of Ca 2ϩ , presumably due to dephosphorylation by a Ca 2ϩ -dependent rhodopsin phosphatase (27,28). We also investigated whether the light conditions to which the photoreceptor membranes were subjected before the reaction was started, might affect the incorporation of phosphate into the InaD protein. Under the in vitro conditions used here, the already intensively studied light activation of  1, 3, 5, and 7) or 60 M calculated free Ca 2ϩ (lanes 2, 4, 6, and 8). In lanes 5-8 the membranes were reconstituted with a soluble extract obtained from six Calliphora retinae. Before starting phosphorylation by adding 2 mM ATP, samples were either kept under dim red light (r, lanes 1, 2, 5, and 6) or illuminated with blue light for 2 min in order to convert 70% rhodopsin to metarhodopsin (panel b, lanes 3, 4, 7, and 8). The phosphorylation reactions were carried out for 5 min in the dark. Migration of molecular weight standards is indicated on the right; arrows on the left show the position of the InaD protein (InaD), arrestin 2 (Arr2), and opsin. b, time course for the incorporation of radioactive phosphate into the InaD protein. For comparison, time courses of phosphorylation of arrestin 2 and opsin are shown. Phosphorylation reactions were performed in the absence (q), or presence (o) of Ca 2ϩ as described in panel a for lanes 3 and 4, respectively. At the indicated times, the reactions were terminated by adding 5 ϫ SDS-PAGE buffer. All samples were subjected to SDS-PAGE as described under "Experimental Procedures," and protein phosphorylation was detected by autoradiography or quantified by using a phosphor imager. rhodopsin phosphorylation (14,15) was reproduced, but light dependence of InaD protein phosphorylation was not revealed (Fig. 6a, lanes 1 and 2 and lanes 3 and 4). Thus, we can exclude the possibility that activated rhodopsin (metarhodopsin), present in the membranes, directly promotes (or suppresses) the phosphorylation of the InaD protein. Furthermore, the addition of soluble proteins did not enhance the incorporation of phosphate into the InaD protein, but rather suppressed its phosphorylation (Fig. 6a, lanes 5-8). This indicates (i) that the protein kinase that catalyzes InaD phosphorylation resides in the photoreceptive membrane, and (ii) that soluble cofactors are not required for InaD protein phosphorylation.
Time courses of the protein phosphorylation revealed similar phosphorylation kinetics for the InaD protein and for opsin with no further increase in net phosphate incorporation 10 min after the reactions were started (Fig. 6b). Arrestin 2 phosphorylation, described as the most rapid protein phosphorylation observed in Drosophila eyes (29), saturated 2 min after starting the reactions. The Ca 2ϩ dependence of the phosphate incorporation into the InaD protein, arrestin 2, and metarhodopsin is evident throughout the entire phosphorylation time course, except in the initial phase of metarhodopsin phosphorylation (Fig. 6b). The Ca 2ϩ -enhanced phosphorylation of the InaD gene product suggested that this reaction is catalyzed either by a protein kinase C or by a Ca 2ϩ -calmodulin-dependent protein kinase. In order to discriminate whether the InaD protein is phosphorylated by a protein kinase C or a Ca 2ϩ -calmodulin-dependent protein kinase, protein kinase C was hyperactivated using a phorbolester or specifically inhibited with bisindolylmaleimide I. As shown in Fig. 7, the addition of phorbol 12myristate 13-acetate to the phosphorylation reaction enhances the phosphate incorporation into the InaD protein by 25%. In the presence of bisindolylmaleimide I InaD protein phosphorylation is reduced by 25%. These effects are statistically significant (see legend of Fig. 7), and they are comparable with those observed in studies with other photoreceptor membrane proteins, for example the protein kinase C-dependent phosphorylation of bovine rhodopsin (30). The protein kinase C activator and inhibitor used here had no significant effect on the phosphorylation of arrestin 2 and opsin (Fig. 7), indicating that the addition of the phorbolester or of bisindolylmaleimide I modulated specifically the phosphorylation of the InaD protein but did not generally enhance or quench the phosphorylation of rhabdomeral proteins. DISCUSSION This study describes the molecular and biochemical characterization of the Calliphora InaD protein. The experiments have been performed to understand the function of this protein in the deactivation of light-triggered responses of photoreceptor cells. Cumulative evidence suggests that the biochemical reactions involved in phototransduction are identical in Drosophila and Calliphora. The eyes of both species have the same morphological architecture, they contain photoreceptors with identical absorbance characteristics, and the photoreceptor cells respond to light stimuli in the same way. Biochemical studies show that identified proteins of the phototransduction pathway, including Rh1 opsin (31)(32)(33), arrestin 2 (17,27,34), and phospholipase C (35,36), perform identical functions in both species. However, these functionally homologous proteins are less conserved than they are between photoreceptors currently used as model systems of vertebrate phototransduction, for instance bovine, rat, and mouse. These differences in the overall homology allow us to identify conserved regions as probable sites of functional importance within the protein.
The sequence alignment of the Drosophila and Calliphora InaD proteins (see Fig. 1) highlights the weakly, as well as the highly, conserved regions of the protein sequences. The Nterminal region (amino acids 1-14 of the Drosophila sequence) and the stretch between amino acids 106 and 183 show little if any sequence homology, suggesting that these regions are functionally less important and were, therefore, subject to extensive mutation during the evolution of both fly species.
Other sites of the InaD protein are well conserved. With respect to the phosphorylation of the InaD protein investigated in the present study it is particularly striking that five out of eight conserved potential protein kinase C phosphorylation sites (at positions 19,194,329,330, and 553) reside within stretches of 10 -16 amino acids that are identical in the Drosophila and Calliphora InaD protein.
Despite the fact that there are some poorly conserved regions in the Drosophila and Calliphora InaD protein, the overall biophysical characteristics (for example the isoelectric point at 8.6, the high abundance of acidic and basic amino acids, and the hydrophilicity profile of both proteins) are nearly identical. The apparent molecular mass of about 75 kDa of the Calliphora InaD protein, as estimated by SDS-PAGE, fits the molecular mass calculated from the sequence data (73.4 kDa). The discrepancy between the calculated and apparent molecular mass of the Drosophila InaD protein (80 and 90 kDa, respectively) reported by Shieh and Niemeyer (11) is not evident in Calliphora. Due to the hydrophilic nature of the InaD protein, it has been proposed that the Drosophila InaD gene product is not an integral membrane protein (11). Our results obtained with the Calliphora homolog of InaD are in agreement with this prediction.
In the present study we show that the InaD protein is associated with the rhabdomeral photoreceptor membrane, from which it is extracted by buffers of high ionic strength. The attachment to the photoreceptive membrane may be crucial for InaD function, because functional impairment of the Drosophila InaD mutant (InaD P215 ; Ref. 10) results from a single point mutation in which a methionine (Met 442 ), located within a small stretch of hydrophobic amino acids, is replaced by lysine (11). In the Calliphora InaD protein leucine is present at the corresponding position, indicating that Met 442 is not necessarily required for normal InaD function, and may be exchanged with another hydrophobic amino acid. Distortion of the hydrophobic character of the region by a highly polar amino acid, for example lysine of InaD P215 , however, might lead to the mutant phenotype, because the Met 442 to Lys mutation may render a soluble InaD protein that is nonfunctional. Alternatively, the nonpolar character of this region may be crucial for hydrophobic protein-protein interactions. A significant contribution to the hydrophilic character of the InaD protein results from a conserved stretch of lysine and glutamate residues (see bars in Fig. 1b). Interestingly, similar lysine/glutamate-rich clusters are found in the bovine and mouse rod photoreceptor cGMPgated channels (37,38). Analysis of the structure-function relationship of cGMP-gated channels has not yet established the function of this hydrophilic cluster.
The biochemical experiments of the present study were designed to investigate whether or not the function of the InaD gene product might be controlled by phosphorylation. The striking conservation of several putative protein kinase C phosphorylation sites between the Drosophila and Calliphora InaD sequence (Fig. 1), the localization of the InaD protein and the inaC encoded eye-PKC in the rhabdomeral photoreceptor membranes ( Fig. 2 and Ref. 13), and, most importantly, the coimmunoprecipitation of the InaD protein and eye-PKC (Fig. 3) suggest that the InaD protein is a likely candidate for phosphorylation by eye-PKC. Moreover, Drosophila InaD and inaC mutants show a similar phenotype, which is characterized by a defect in photoreceptor deactivation and by abnormal light adaptation (1,2,11,13,39), indicating that the respective gene products are acting, or even interacting, in closely related steps of the transduction cascade. The phosphorylation studies presented here reveal that the InaD protein is a phosphoprotein (Figs. 4 and 5). The Ca 2ϩ -dependence of the InaD protein phosphorylation (Fig. 6) and the findings that the incorporation of phosphate into the InaD protein is moderately enhanced in the presence of a phorbolester and quenched by the protein kinase C inhibitor bisindolylmaleimide (Fig. 7) are in line with the assumption that this phosphorylation is catalyzed by eye-PKC.
Despite this evidence, the data do not yet allow us to unequivocally rule out the possibility of phosphorylation of the InaD protein by other protein kinases. Ca 2ϩ -dependent phosphorylation of arrestin 2, reported to result from a Ca 2ϩ -calmodulin-dependent protein kinase (17), is observed in parallel to the Ca 2ϩ -dependent phosphorylation of the InaD protein, indicating that the corresponding protein kinase is present in the membrane preparation used in the assays. Also, Matsumoto and colleagues (17,40,41) reported on the phosphorylation of an 80-kDa protein present in the photoreceptor cell layer of Drosophila eyes. The molecular mass of this phosphoprotein suggests that it might represent the Drosophila InaD protein. However, the phosphorylation of this Drosophila 80-kDa protein was shown to be activated by cAMP but not by calcium (17). At least the Calliphora InaD protein lacks consensus sites for phosphorylation by a cAMP-dependent protein kinase. In dark-adapted Drosophila eyes, this 80-kDa protein is in the nonphosphorylated state, but it rapidly (within 3 s) becomes phosphorylated when the flies are exposed to a 1-ms light flash (41). Furthermore, the light-dependent phosphorylation of this protein is not observed in Drosophila norpA mutants (40), indicating that it depends on the activation of the phototransduction cascade and occurs downstream of the norpA-encoded phospholipase C.
Phosphorylation by protein kinase C is shown to be involved in the desensitization of a number of vertebrate G-proteinmediated transduction cascades, e.g. vertebrate phototransduction (30,42) and ␤-adrenergic receptor signaling (43). There desensitization is achieved by a protein kinase C-dependent phosphorylation of the respective receptor (rhodopsin or ␤-ad-renergic receptor), which in contrary to the phosphorylation by rhodopsin kinase or ␤-adrenergic receptor kinase, occurs in the activated and the nonactivated state of the receptor. Protein kinase C-mediated phosphorylation was shown to uncouple the receptor from its G-protein (44), thereby terminating the signal response. The proposed deactivation of the visual response via phosphorylation of the InaD protein by eye-PKC would act at a different site of the transduction cascade.
Towards a model for the Ca 2ϩ -dependent response inactivation in fly photoreceptor cells, we propose that the InaD protein is modulated via phosphorylation by eye-PKC, which itself should be activated by the transient rise of the intracellular Ca 2ϩ concentration upon visual excitation. Phosphorylated InaD protein in turn may be a subunit of, or act on, a third protein, e.g. an ion channel, in order to regulate Ca 2ϩ influx into the cytosol. In this respect it is important to note that the trp protein, which is proposed to represent a novel Ca 2ϩ channel responsible for light-dependent inositol trisphosphate-mediated Ca 2ϩ entry (6, 7), co-immunoprecipitates with the InaD protein. Alternatively, the activated InaD protein could be part of a feedback control mechanism that acts on upstream members of the transduction cascade. One of these may be the norpA-encoded phospholipase C (11). Our finding that key proteins of the phototransduction cascade investigated here coimmunoprecipitate with the InaD protein may indicate that proteins that provide a control mechanism of visual excitation are associated into a functional protein complex.
In conclusion, we have for the first time provided evidence that the Calliphora homolog of the InaD protein is phosphorylated by the inaC-encoded eye-PKC. InaD protein phosphorylation may be part of the mechanism that regulates the deactivation of the light response in invertebrate photoreceptors, in a way that is distinct from the protein kinase C-mediated desensitization of vertebrate phototransduction or ␤-adrenergic receptor signaling. However, a similar mechanism may operate in other vertebrate and invertebrate signaling pathways in which trp homologs are used as part of a store-operated Ca 2ϩ entry (45).