Eur. J. Biochem. 208,659-667 (1992) 0FEBS 1992

Mechanistic studies on rhodopsin kinase Light-dependent phosphorylation of C-terminal peptides of rhodopsin Neil G. BROWN, Charles FOWLES, Ram SHARMA and Muhammad AKHTAR Department of Biochemistry, University of Southampton, England (Received May 25,1992)

-

EJB 92 0731

The phosphorylation of a synthetic peptide, corresponding to the C-terminal 11 amino acids of bovine rhodopsin (MI, residues 338 - 348), was studied under different conditions. The peptide was only phosphorylated in the presence of photoactivated rhodopsin. Using the same protocol, 12 other peptides, mapping in the rhodopsin C-terminal, were screened for their effectiveness as substrates for rhodopsin kinase. It was found that the peptides became poorer substrates with increasing length, and the best substrates comprised the most C-terminal9 - 12 amino acids as opposed to other parts of the C-terminus. It was noted that the absence of the two-terminal residues Pro347 and Ala348 impaired peptide phosphorylation. The effect of the decay of metarhodopsin I1 on the phosphorylation of rhodopsin and the peptides was determined, and it was found that the rhodopsin and peptide phosphorylations decayed with half times of approximately 33 min and 28 min, respectively. The sites of phosphorylation on the peptides were determined and in all cases the phosphorylation was found to be predominantly on serine residues. Only the 11-residue peptide (MI, residues 338348) contained significant threonine phosphorylation, which was about 25% that on serine residues. Cumulatively, the results suggest that Ser343 is the preferred site of phosphorylation in vitro. The reason for the poor substrate effectiveness of the larger peptides was examined by competitive experiments in which it was shown that a poorly phosphorylated larger peptide successfully inhibited the phosphorylation of a ‘good’ peptide substrate. The studies above support a mechanism for rhodopsin kinase that we have termed the ‘kinase-activation hypothesis’. This requires that the kinase exists in an inactive form and is activated only after binding to photoactivated rhodopsin.

Photoactivation of bovine rhodopsin leads to the production of a relatively long-lived intermediate know as metarhodopsin I1 [l -31. This conversion also results in dramatic protein conformational changes, which are indicated by altered proteolytic susceptibility [4, 51 and the exposure of the Schiff-base linkage to external reactions [6, 71. A metarhodopsin-11-like species (regarded as the ‘activated’ form and symbolised as Rho*) initiates the process of visual transduction [8 - 111 through interaction with transducin (guanosinenucleotide-binding regulatory protein) [12, 131. This culminates, via a cascade mechanism [14, 151, in the hyperpolarisation of the rod cell [16, 171. This process is terminated by a phosphorylation mechanism involving Rho* [18, 191 and rhodopsin kinase [20-221. This kinase, in vitro, acts specifically on several serine and threonine residues in the C-terminal domain of bleached rhodopsin [23-261. This dual role of Rho* in the transduction and termination processes is illustrated in Scheme 1. With respect to the mechanism of light-dependent phosphorylation, it was originally assumed that the C-terminal domain of rhodopsin is inaccessible to the kinase but becomes Correspondence to M. Akhtar, Dept. of Biochemistry, Biomedical Sciences Building, University of Southampton, Basset Crescent East, Southampton, England SO9 3TU Abbreviations. ROS, rod outer segments; Rho*, photoexcited rhodopsin.

li’

GDP GTP + (GDP)T T T

Transducin

f j j j

-4 \

+

TpT7 + ( G W T a

Signal-Transmission

.

a hv P Y

Rhodopsin

Rho’

_L

Rho

Kinase

(Phosphorylated)n-Rho’

Regeneration ..........................................................

Signal-Termination

:

Scheme 1. The involvement of photoactivated rhodopsin (Rho* = metarhodopsin 11) in signal transmission and signal terminationprocesses. The transformation shown by the broken arrow occurs in three steps involving the release of nPi through the action of a phosphatase, hydrolysis of the Schiff-base linkages and reaction with 1I-cis-retinal (for details see [35]). T,T,T, represents the three subunits of transducin.

available following the conformational rearrangement accompanying the formation of metarhodopsin I1 [18, 221. Recently however, using synthetic peptides corresponding to the C-terminal region of the protein, we made the surprising observation that, in our hands, these peptides were not handled as substrates by the kinase alone, but in the presence of rhodopsin and light, dramatic phosphorylation of the peptides

660

Optiphase 'Hisafe' 3 was obtained from LKB and [y32P]ATP and NCS tissue solubiliser from Amersham. All other chemicals, except those used in peptide synthesis, were obtained from Sigma, BDH or Interchem. Bovine retinae were collected from a local slaughterhouse (FMC Ltd.).

of a rhodopsin kinase extract (40 pl). In control incubations, one constituent was replaced by phosphate buffer. After mixing under dim-light conditions, the samples were sonicated for 1 rnin in a Decon FXlOO sonication bath, then incubated at 30 "C for 5 rnin in a thermostat-controlled water bath to equilibrate. The kinase reactions were initiated by illumination with continuous white light (150-W bulb in a photoflood lamp at a distance of 30 cm). Under these conditions, 100% bleaching, as determined by the decrease in the absorbance at 500 nm, was achieved within 10 min. ROS phosphorylation was determined using a modification of the method described by Kiihn and Wilden [32]. At the appropriate times, 5 p1 aliquots were placed in 200 p1 20% (mass/ vol.) trichloracetic acid at 0°C. After 10 min the ROS were recovered by pelleting by centrifugation in an Eppendorf centrifuge at 10000 x g for 2 rnin and the supernatant discarded. The pellets were washed twice in 500 p1 10% (mass/vol.) trichloroacetic acid containing 10 mM KH2P04 at 0°C for 5 min by the same method and the final pellet resuspended in 100 p1 NCS tissue solubiliser and incubated for 10 min at room temperature. Incorporated 32Pwas determined by liquid scintillation counting in 900 ~1 optiphase 'Hisafe' 3 using a Phillips PW470 liquid scintillation counter.

Methods

Phosphorylation of peptides by rhodopsin kinase

Retinae from freshly killed cattle (less than 15-min old) were collected and immediately stored in a light-proof flask at 0°C for transportation to the laboratory. They were always processed the same day and all procedures, unless otherwise stated, were performed under red safelight conditions in a cold room at 4°C.

Peptide incubations were performed as the rhodopsin phosphorylation above, except that some of the phosphate buffer was replaced with the appropriate peptide ( 5 -25 p1 30 mM stock) to obtain a final concentration of 3 - 15 mM. At the appropriate times, aliquots (10 - 40 pl) were centrifuged in an Eppendorf centrifuge and 5 pl of the supernatants were applied to a TLC plate and subjected to high-voltage electrophoresis, as described below, for 2.5 h. The plates were oven dried and autoradiographed at - 70 "C for 15- 20 h using Fuji NIF RX X-ray film. After the film was developed, enabling bands of phosphopeptide to be seen, these bands of cellulose were scraped off the plate and, after addition of 5 ml Optiphase 'Hisafe' 3 and mixing, were analysed by scintillation counting. The phosphopeptides, like their unphosphorylated counterparts, migrated from the positive to the negative terminal.

occurred [27]. This observation was interpreted in terms of an alternative mechanism which envisaged that the kinase normally exists in an inactive form and is activated following interaction with bleached rhodopsin. After an initial scepticism [28, 291, this observation has been confirmed by others [30,31] and a general acceptance of our mechanism has begun to emerge. In this paper, we describe experiments which provide further support for the involvement of a metarhodopsin11-like species in the activation of the kinase, as well as defining the optimal structures of the peptides which may serve as substrates for the kinase and identifying the principal site that is phosphorylated. EXPERIMENTAL PROCEDURES Materials

Preparation of rod outer segments, rhodopsin kinase and kinase free rod outer segments

Rod outer segments (ROS) from 100-140 retinae were purified by sucrose density gradient as described previously [27]. The ROS for use in phosphorylation experiments were resuspended in 100 mM sodium phosphate, pH 7.0, 2 mM MgC12,0.1 mM EDTA, 0.1% (by vol.) 2-mercaptoethanol at 4- 8 mg/ml rhodopsin (0.1 -0.2 mM) calculated from c500= 40000 M-' cm-'. Rhodopsin kinase extracts were obtained from ROS by resuspension and homogenisation as described previously [27]. Rhodopsin kinase activity was typically 1-2 nmol 32Pincorporated min-' mg protein-' in extracts containing 2 - 3 mg/ ml protein. ROS recovered after kinase extraction were washed with 5 M urea as described before [27], then resuspended in 100 mM sodium phosphate, pH 7.0, as above. These ROS were found to be free of the kinase activity and were used as a source of rhodopsin in the experiments using the kinase extracts described below. Phosphorylation of rhodopsin by rhodopsin kinase

In all experiments either non-extracted or urea-extracted ROS ( 5 pl containing 0.1 -0.4 mM rhodopsin) and 2430 mM ATP (5 pl) were used. The final incubation volume of 50 p1 contained 10-40 pM rhodospin and 2.4-3 mM [ Y ~ ~ - P ] A(50000TP 150000 cpm/nmol) in 100 mM sodium phosphate, pH 7.0. With the unextracted ROS, the rest of the incubation (40 pl) consisted of phosphate buffer, whereas with the urea-extracted ROS, the rest of the incubation consisted

Peptide synthesis and purification

The peptides were assembled by the manual stepwise solidphase methodology developed by Merrifield [33]. At the end of the synthesis, the protected peptide was cleaved and deprotected from the resin in a single step by the low/high hydrogen fluoride procedure [34]. After the removal of hydrogen fluoride, the residue was washed with ethyl acetate and the peptide was extracted with 50% (by vol.) acetic acid (100 ml). The peptide solution was lyophilised and yielded crude peptide as a white solid. This was dissolved in a minimum volume of 0.1% (by vol.) trifluoroacetic acid and was desalted on a PDlO G25 gel-filtration column (Pharmacia) and eluted with 10% (by vol.) acetic acid. The peptide-containing fractions were pooled and lyophilised. The peptides were purified to a single peak by reversephase HPLC on Dynamax C4 (30 nm 4.6 mm x 250 mm) and Zorbax PEP-RPIC8 (6.2 mm x 80 mm) columns at a flow rate of 1 ml/min using a gradient over 0 - 80% (by vol.) acetonitrile using solution A [O.l% (by vol.) trifluoroacetic acid/H20]and solution B [0.1% (by vol.) trifluoroacetic acid/80% (by vol.) acetonitrile/H20].

661 Peptide purity was further determined by high-voltage TLC using a pyridine/acetic acid/water (1 : 10:89) buffer at pH 3.5, electrophoresised at 1000 V for 90 min at 7°C. A single band was observed when the cellulose plates were visualised by ninhydrin. Under our electophoretic conditions, the peptides migrated from the positive to the negative terminus. The compositions of the purified peptides were confirmed by their amino acid analyses and by fast-ion-bombardment spectrometry. Analysis of the phosphoamino acid content of phosphopeptides At the end of the peptide incubations, aliquots (30-40 ~ 1 were microcentrifuged and all the supernatants subjected to electrophoresis as described above. The plates were dried in air, the phosphopeptides were visualised by autoradiography and the phosphopeptide bands scraped off the plate. The phosphopeptides were then eluted with 5 ml distilled water onto CI 8 reverse-phase sep-pac columns that had been successively washed with 30%, 5% and 0% (by vol.) aqueous acetonitrile ( 5 ml each). The phosphopeptides were eluted from the columns with a 070,5% and 30% (by vol.) aqueous acetonitrile stepwise gradient (3 ml each). The phosphopeptide-containing fractions [usually 5% (by vol.) acetonitrile] were lyophilised and subjected to acid hydrolysis in 6 M HCI (100 pl) at 110°C for 2 h. Acid was removed by NaOH/P20S under vacuum for 3-5 h and the samples were then resuspended in 1 ml distilled water and lyophilised. The samples were resuspended in 11 pl of a solution of 1 mg/ml each of phosphotyrosine, phosphothreonine, and phosphoserine solution, with 1% (mass/vol.) xylene cyanol as a marker, and subjected to electrophoresis as described above for 90 min. These samples migrated toward the positive terminal, in contrast to the unhydrolysed peptides and their phosphorylated counterparts. Phosphoamino acid standards were visualised using ninhydrin stain and the plates then autoradiographed for 15 h to 2 weeks, depending on the peptide concentration, the level of 32Pincorporation and the specific activity of the ATP used in the kinase assays. On developing the films, the radioactive phosphoaminoacid bands visualised were matched with those bands of phosphoaminoacid standards visualised with ninhydrin, on the electrophoresis plates. RESULTS The phosphorylation of C-terminal peptides by rhodopsinkinase in the presence of bleached rhodopsin The basic protocol used to monitor the phosphorylation of synthetic peptides was based on high-voltage cellulose TLC electrophoresis at pH 3.5, where the parent peptides were found to migrate further towards the cathode than their phosphorylated counterparts. The time of electrophoresis was adjusted to ensure that the separation between the non-phosphorylated and phosphorylated peptides was maximal. The experiment performed with the 11-residue peptide (VII, residues 338 - 348) comprising the eleven C-terminal residues of rhodopsin (Fig. 1a) shows that its phosphorylation from [y32P]ATPwas strictly dependent on the kinase, rhodopsin and light (Fig. la, lane 1). When the incubation containing all the components of the reaction was performed in the dark, the level of phosphorylation was only 6% of that in the light. No significant phosphorylation was observed when either rhodopsin or kinase were omitted, even though the incubation was performed in the light. In these experiments, the incorpo-

ration of 32Pinto opsin was also measured and, in proportionate terms, the level of protein phosphorylation under different conditions mirrored that of the peptide. Thus, the phosphorylation of the protein only occurred in the complete system containing [ Y - ~~ P ] A Tkinase P , and rhodopsin in the light. The main requirements for the phosphorylation of the 11residue peptide (VII, residues 338 - 348) found here are similar to those reported by us for the 10-residue peptide (VIII, residues 339 - 348) and subsequently confirmed by Palczewski et al. [29, 301 and Onorato et al. [31] with other synthetic peptides. Using the protocol of Fig. l b , 12 other peptides were ) screened for phosphorylation by the kinase (Table 1) and it was found that the 11-residue peptide (VII, residues 338348) was in fact the best substrate; closely following this were the 10-residue peptide (VIII, residues 339 - 348) and 9-residue peptide (IX, residues 340 - 348). There was a gradual impairment of phosphorylation as the peptide chain length increased. It is particularly interesting to note that the 20-residue peptide (11, residues 329 -348) and 31-residue peptide (I, residues 318 - 348) which contained all the potential phosphorylation sites of rhodopsin and were designed to mimic its phosphorylation domain most closely, were in fact the poorest substrates, being phosphorylated at a level which was barely 2% that of the 11-residue peptide (VII, residues 338- 348). Attention is also drawn to the fact that, compared to the 10-residue peptide (VIII, residues 339 - 348), the absence of Lys339 in the 9-residue peptide (IX, residues 340 - 348) had only a small affect on the level of phosphorylation. The family of peptides lacking the two C-terminal amino acids of rhodopsin, Pro347 and Ala348, showed considerably reduced activities. For example, the 10-residue peptide (X, residues 337 - 346) and the 8-residue peptide (XI, residues 339 - 346) were phosphorylated at about 10- 18% the rate obtained with the optimal 11-residue peptide (VII, residues 338 -348). In these experiments, each peptide was used at 3 mM, since this was found to be the highest convenient concentration for comparative evaluations. The use of higher concentrations was impractical in some cases because of the insolubility of the substrates and, in other cases, because of the possibility that at high concentrations the peptides, due to their amphipathic nature, may act as detergents. Fig. 2 shows a comparison of the time courses of the simultaneous phosphorylation of rhodopsin and the best peptide substrate, the 11-residue peptide (VII, residues 338 -348). After 2 h incubation, the amount 0f32Pincorporated into the peptide (0.29 mol) approached 10% of that incorporated into the rhodopsin (3.2 nmol). This demonstrates the high level of peptide phosphorylation under the conditions used in this work. However, the native rhodopsin was still by far the preferred substrate of rhodopsin kinase. Fig. 2 also shows that, in the presence of 3 mM 11-residue peptide (VII, residues 338 - 348), the rhosopsin phosphorylation was unimpaired. This is not suprising since the affinity of synthetic peptides for the kinase was about three orders of magnitude lower than that for the natural substrate, bleached rhodopsin. The K , values for the 11-residue peptide (VII, residues 338 - 348) and the 10-residue peptide (VIII, residues 339 - 348) were found in this study to be 6 - 7 mM compared to about 3 pM reported for rhodopsin [28]. The effect of larger peptides on the phosphorylation of good substrates The poor phosphorylation of larger peptides could be because of several reasons, including the possibility that their

662

B

A Anode(+)

Anode + I

origin

Lane

1

2

3

4

Origin

5 Lane l a

lb

2a

2b

3a

3b

4a

4b

Cathode ( - )

Cathode(-)

Fig. 1. The phosphorylation of the 11-residue peptide (VII) containing the C-terminal residues 338 - 348 of bovine rhodopsin, and other related peptides, under different conditions. (A) The phosphorylation of the 11-residue peptide (VII, residues 234 -248). Incubations were performed as described in the Experimental Procedures for 2 h at 30°C under white light or in darkness in a total volume of 50 pl containing 3 mM peptide, 15.2 pM rhodopsin in urea-extracted ROS, 35 pL1 kinase extract and 2.5 mM [p3’P]ATP. After 2 h, the incubations were microcentrifuged and a 5-pl sample of the supernatant analysed by cellulose TLC electrophoresis followed by autoradiography as described in Experimental Procedures. Lane 1, complete system incubated in light; lane 2, as in lane 1 but incubated in dark; lane 3, as in lane 1 but without peptide; lane 4, as in lane 1 but without rhodopsin; lane 5, as in lane 1 but without kinase extract. Results shown are from one of four experiments where peptide phosphorylation under each condition varied by less than 6%. (B) The general screening of other peptides for phosphorylation. The incubations were performed as described in Experimental Procedures in a total volume of 50 pl containing 3 mM of the appropriate peptide, 35.6 pM rhodopsin in unextracted ROS (which contain the endogenous kinase) and 2.7 mM [Y-~’P]ATP.The other details are the same as in (A). Lanes la, 2a, 3a, and 4a are incubations with the 10-residue peptide (VIII, residues 339-348), 10-residue peptide (X, residues 337 - 346), 1I-residue peptide (VII, residues 338 - 348) and 12-residue peptide (VI, residues 337 - 348) respectively in light, while lanes lb, 2b, 3b and 4b are the corresponding peptides incubated in the dark. The results shown are from one of four experiments where peptide phosphorylation varied by less than 5% (see also Table 1).

Table 1. The relative efficiency of the phosphorylationof various synthetic peptides. The incubations were performed using 3 mM of each peptide as described in Fig. l b . Following autoradiography, the phosphopeptide zones were scraped from the plates for scintillation counting. The potential phosphorylation sites are underlined in the sequences below. The results shown are an average of four experiments where peptide phosphorylation varied by less than 5%. The residue numbers of the peptides relate to the corresponding sequences of bovine rhodopsin. Actual molecular masses were found by m/z 1 from FIBS (If: I), except for peptide I which was characterised by amino acid sequencing. Theoretical molecular masses (M H’) were calculated from monoisotopic values and are shown in brackets.

+

+

Peptide (residue number)

Sequence

Molecular mass

1

VTTLCCGKNPLGDDEASTTVSKTETSQVAPA

Relative peptide phosphor ylation

kDa 31-mer (318-348) I1 20-mer (329 - 348) Ill 16-mer (333-348) IV 14-mer (335 - 348) V 11-mer (329 - 339) VI 12-mer (337-348) VII 11-mer (338-348) VIII 10-mer (339 - 348) IX 9-mcr (340 - 348) X 10-mer (337-346) XI 8-mcr (339-346) XI1 6-mer (341 - 346) XI11 4-mer (343 - 346)

n.d. (3123.5) GDDEA~T~VZKTETSQVAPA 1993 (1993.9) A~EVSKTETSQVAPA1577 (1577.8) TTVSKTETSQVAP A 1419 (1419.7) G D D E A ~ V ~ K1109 (1109.5) 1217 (1217.6) V~KTEEQVAPA 1118 (1118.6) SKTESQVAPA - _ KTETSQVAPA 1031 (1031.5) TET SQVAPA 903 ( 903.4) V~KTET~QVA 1049 (1049.5) 863 ( 863.4) VTTLCCGKNPLGDDEASTTVSKTETSQVA 634 ( 634.3) 404 ( 404.2) SQVA

conformations in solution favoured the formation of ‘hairpin’ structures which were not recognised by the kinase. This hypothesis was examined by competitive experiments in which the phosphorylation of a good peptide substrate was performed in the presence of a poorly phosphorylated larger peptide. In order to facilitate unambiguous analysis, the two peptides were chosen so that their phosphorylated derivatives were readily separable by electrophoresis. This requirement was satisfied by the 11-residue peptide (VII, residues 338 348) and the 20-residue peptide (11, residues 329 - 348) whose

K~EEQVA

2 3 5 5 11 36 100 76 75 18 11 0 0

phosphorylated products separated maximally under our electrophoretic conditions. Fig. 3 shows that 20-residue peptide (11, residues 329 - 348) was a good inhibitor of the phosphorylation of the 11-residue peptide (VII, residues 338 348) and when the latter was present at a concentration of 3 mM, its phosphorylation was almost completely abolished by 1 mM 20-residue peptide (11, residues 329 - 348). The 20-residue peptide (11, residues 329 - 348) does, therefore, compete with the small peptides for binding to the kinase but the interaction leads to the formation of a catalytically

663

--E 0

c .-0

0

:

3.5

I

0.30 3.0

c

m L 0

P

2.5

2 0.20

2.0

0

c

a

0.15

1.5

m

1.o 0.5

0.05

:: 0

A=

0

a

2 m

0.10

u)

c

:

c -

r

c

c n

-

.-

0.25

:

0

c

Carnode

o

LT

0

I

I

I

I

30

60

90

120

c-,

I

a L

0

0.00

f

150

Time bin) Fig. 2. The time course of the phosphorylation of the 11-residue peptide (VII, residues 338 - 348) and rhodopsin. Incubations were performed as described in Experimental Procedures for 2 h at 30°C under whitelight illumination in 50 pl volumes with the following final concentrations: 3 mM 11-residue peptide (VII, residues 338-348), 15.7 pM rhodopsin in unextracted ROS and 2.4 mM [ Y - ~ ~ P I A TAt P .various times, 5-pl samples were removed and processed either for rhodopsin phosphorylation or peptide phosphorylation as described in Experimental Procedures. The upper rhodopsin line (V) represents the rhodopsin phosphorylation in the absence of any peptide while the lower rhodopsin line ( W ) represents the rhodopsin phosphorylation in the presence of 3 mM 1I-residue peptide (VII, residues 338 - 348). The points on the peptide graph correspond to the lanes on the inset autoradiograph showing the 32Pincorporation in the 11-residue peptide (VII, residues 338- 348) at the various times. Lane 1,lO rnin of incubation; lane 2, 30 rnin of incubation; lane 3, 60 rnin of incubation; lane 4, 120 min of incubation. The results shown are from one of four experiments where rhodopsin and peptide phosphorylation varied by less than 4%.

c

0 ._ I

100

m L

O

f0? 0 -c

80

60

8

I

m c u) n

40

0

.c

a

:

.-I

20

-a m

K

incompetent complex. It should be noted that, in the experiments of Fig. 3, the phosphorylation of bleached rhodopsin was also studied and this was found to be inhibited only by less than lo%, which is within the range of variation of the analytical procedures used.

O

0.0

0.2

0.4

0.6

0.8

1 .o

Inhibiting Peptide Concentration (mM)

Fig. 3. The effect of the 20-residue peptide (11, residues 329 - 348) on the phosphorylationof the 11-residue peptide (VII, residues 338- 348). Incubations were performed as described in Experimental Procedures for 2 h at 30°C under white-light illumination in volumes of 50 pl. Final concentrations were: 3 mM 1I-residue peptide (VII, residues 338 - 348), 0 - 1 .O mM 20-residue peptide (11, residues 329 - 348), 13.7 pM rhodopsin in unextracted ROS and 2.6 mM [Y-~’P]ATP. After 2 h, the incubations were microcentrifuged and a 5-pl sample of the supernatants analysed by cellulose TLC electrophoresis, autoradiography and scintillation counting as described in Experimental Procedures. The results shown are an average of four experiments where peptide phosphorylation varied by less than 5%.

The nature of the phosphorylated residue in the peptide substrates Next, experiments were performed to identify the site(s) of peptide phosphorylation and, for this purpose, critical studies were performed on the 10-residue peptide (VIII, residues 339 - 348) which contains three potential phosphorylation sites at its two threonyl residues and its single seryl residue. Following large-scale phosphorylation, the phosphopeptide was first separated by electrophoresis, then further purified by HPLC when the material containing 32Peluted just ahead of the position of the parent 10-residue peptide (VIII, residues 339 - 348). Edman degradation of the phosphorylated peptide by pulse-liquid-phase sequencing gave positive identification of the first eight amino acids, showing the expected sequence. However, during none of the cycles did the thiazolinone extract contain sufficient 32P radioactivity above the background to justify the identification of the phosphoamino acid in the sequence. This finding is not surprising since, under the conditions of Edman degradation, phosphoseryl and phosphothreonyl residues undergo p-elimination, releasing inorganic phosphate that is retained on the membrane-polybrine phase and not eluted by mild solvents used in liquid-pulse sequencing. Protocols to overcome this difficulty have been described but have not yet been used successfully in our laboratory. As an alternative, we undertook a phosphoamino acid analysis of the peptide and, following a partial hydrolysis of the 32P-labelled 10-residue peptide (VIII, residues 339 - 348) with 6 M HC1 for 2 h, the hydrolysate, along with standards of phosphoserine, threonine and tyrosine were subjected to electrophoresis at pH 3.5. Fig. 4 shows that in the phosphoamino acid region, phosphoserine was the most highly radiolabelled species, while the radiolabelling of phosphothreonine was only about 5% of the former. A similar bias in favour of serine phosphorylation was found when the phosphopeptides obtained from the incubations of the 8residue and 9-residue peptides (XI, residues 339 - 346; IX, residues 340-348) were analysed. However, in the case of the 11-residue peptide (VII, residues 338 - 348) a significant radiolabelling of the phosphothreonine band was observed, though this was still less than 25% of the phosphoserine band. Since the latter peptide, 11-residue peptide (VII, residues 338 - 348), contains two seryl residues, its predominant site of phosphorylation cannot be defined; however, the results obtained for the 8-residue, 9-residue and 10-residue peptide (XI, residues 339 - 346; IX, residues 340 - 348; VIII, residues 339 - 348, respectively), each of which has a single serine residue, clearly show that, in vitro, it is the residue corresponding to Ser343 that is modified. The absence of phosphorylation at the Thr340 and Thr342 is also noteworthy. The effect of the decay of metarhodopsin I1 on the phosphorylation of peptides Past studies [20 -261 have clearly established, as is evident from the present work, that rhodopsin becomes a substrate for the kinase only following its bleaching by light. That metarhodopsin I1 is the species phosphorylated is, however,

664 Cathode

(-)

Origin

r:

100

.-0 m Y

80

60

40 0 J=

a Phosphothreonine Phosphoserine

al

.-c>. m 0 CT

Lane

1

2

3

20

0 1 0

I

I

I

I

I

10

20

30

40

50

60

4

Anode

(+)

Photoexcited Rhodopsin Decay (rnin)

Fig.4. The nature of the phosphorylated residue in the peptide sub- Fig. 5. The effect of the decay of metarhodopsin I1 on the phosphorylastrates. The phosphorylations were performed and peptide phos- tion of the 11-residue peptide (VII, residues 338 348) and rhodopsin. phorylation analysed as described in Experimental Procedures. Final Incubations were performed as described in Experimental Procedures concentrations in the 50-p1 incubations were: 15 mM peptide, at 30°C under white-light illumination in 50-pI volumes with the 15.7 pM rhodopsin in unextracted ROS, and 2.5 mM [ Y - ~ ~ P I A T P following . final concentrations: 3 mM 1I-residue peptide (VII, resiThe resultant phosphopeptide bands were collected, hydrolysed and dues 338-348), 15.2 pM rhodopsin in urea-extracted ROS, 35 pl analysed by cellulose TLC electrophoresis as described in Experimen- kinase extract and 2.7 mM [ Y - ~ ~ P I A TAll P . constituents except the tal Procedures. It should be noted that the radioactivity in the hydroATP were incubated for 5 rnin in the light. After a further incubation lysed samples migrate towards the anode under our electrophoretic for 0 - 60 rnin in the dark, the ATP was added and a final incubation conditions, unlik,e the phosphopeptides or their unphosphorylated performed for 30 min, again in the dark. 5-pl samples were used to counterparts which migrate toward the cathode. Lane 1, 10-residue analyse the rhodopsin phosphorylation (A) and following microcenpeptide (VIII, residues 339- 348); lane 2, 8-residue peptide (XI, resi- trifugation, 5-pl samples of the supernatants were analysed for peptide dues 339- 346); lane 3, 9-residue peptide (IX, residues 340- 348); phosphorylation (B) as described in Experimental Procedures. The lane 4, 11-residue peptide (VII, residues 338 - 348). The results shown points on the peptide graph (B) correspond to the lanes on the inset are from one of four experiments where peptide phosphorylation autoradiograph showing the 32P incorporation in the 1I-residue varied by less than 7%. peptide (VII, residues 338-348) after varying times of decay of photoexcited rhodopsin. Lane 1, 5-min bleaching prior to ATP addition and 30-min dark incubation; lane 2, as 1 but with 20-min dark not unambiguously proven. The possible role of metarho- incubation between bleaching and ATP addition; lane 3, as 1 but with dopsin I1 in protein, as well as peptide, phosphorylation was 40-min dark incubation between bleaching and ATP addition: lane 4, studied as follows. Samples of ROS containing the kinase, as 1 but with 60-min dark incubation between bleaching and ATP together with the 11-residue peptide (VII, residues 338 - 348) addition. Results shown are an average of four experiments where were exposed to light for 5 min, then placed in the dark. The rhodopsin and peptide phosphorylation varied by less than 4%.

-

phosphorylation reaction was initiated by the addition of [y32P]ATP at 0, 20, 40 and 60min post-bleaching and the samples allowed to incubate for a further 30 min, also in the dark. The incubation mixtures were then processed to quantify the incorporation of 32P into opsin and the peptide. Fig. 5 shows a progressive decrease in the phosphorylation of both the substrates as the interval between illumination and the initiation of phosphorylation increases. In broad terms, the profile of protein phosphorylation obtained here is identical to that described previously [I81 when the half-time of decay was found to be 30-40 min. This compares favourably with the value of approximately 33 rnin found in the present study. In addition, Fig. 5 gives important information that protein and peptide phosphorylation activities had similar decay halftimes of approximately 33 rnin and approximately 28 min, respectively. An apparently puzzling feature of the experiments of the type, shown in Fig. 5 , is that even after 60 rnin post-bleaching, when spectrophotometric studies show that metarhodopsin I1 has largely decayed, there is still a significant phosphorylation activity present. Similar observations have been made previously when experiments were conducted to correlate either phosphorylation [18] or transduction activity [ l l ] with the decay of metarhodopsin 11. The simplest explanation of such observations is that species intervening between metarhodopsin I1 and opsin (for example metarhodopsin HI), but

not opsin itself, have metarhodopsin-11-like conformational characteristics and hence mimic its biological properties. Alternatively, a more sophisticated rationale for the residual Rho* activity has been offered by Kibelbek et al. [ll].It should be borne in mind that the protocol of the type employed in Fig. 5, or for that matter in previous attempts to define the chemical nature of Rho*, in which metarhodopsin I1 is allowed to decay in the absence of a vital component (such as ATP or transducin) is completely artificial. In vivo, under normal illumination conditions, such a situation does not exist and metarhodopsin I1 produced in minute amounts is rapidly removed by the pathways of Scheme 1. Notwithstanding this, the in-vitro experiments, showing that the decay products of metarhodopsin I1 are phosphorylated by the kinase, have an important physiological significance. It is known that a ROS protein phosphatase is involved in the regeneration of rhodopsin from its phosphorylated derivatives [35 - 371. The phosphatase acts on phosphorhodopsin as well as phosphoopsin and hence, by implication, on the phosphorylated derivatives of other decay products. In view of this property of the phosphatase, should in-vivo dephosphorylation generate metarhodopsin 111, this species is likely to participate in false signalling. It therefore makes physiological sense to endow the kinase with a broader specificity, so that all the

665 decay products of rhodopsin are kept in a phosphorylated form, and thus prevented from interaction with transducin, the first enzyme of the transduction cascade.

ranged from 60% inhibition to 10% stimulation! [39]. These findings appear to be in conflict with the result reported in this paper, but several important differences in the experimental conditions used here and by Kelleher and Johnson [39] should be noted. In our case, rhodopsin contained in ROS was used in the presence of 3 mM ATP when up to 4.5 phosphates/ DISCUSSION rhodopsin were incorporated. Kelleher and Johnson used vesContinuing studies on the role of the phosphorylation/ icle-reconstituted rhodopsin with 100 pM ATP and achieved dephosphorylation in the termination of the visual signal and the incorporation of 0.3 phosphates/rhodopsin. Attention is regeneration of rhodopsin, we have now further explored the also drawn to another feature. In our initial experiments we mechanism of action of rhodopsin kinase using synthetic sub- had noticed a concentration-dependent inhibition of rhodopstrates. The requirements of light-activated rhodopsin for the sin phosphorylation by certain peptides, but the results were phosphorylation of a decapeptide, the 10-residue peptide not reproducable and depended on the batch of the peptide (VIII, residues 339 - 348) by the kinase, originally demon- used. It was subsequently found that peptides which appeared strated by us [27] and subsequently validated using other homogeneous as judged by HPLC, were often contaminated peptides [30,31], is further substantiated by the present study. with reagents which inhibited the activity of rhodopsin kinase. In the experiments of Fig. 1, we have found a 15 - 30-fold In the light of this experience, the purity of peptides were difference in the phosphorylation of peptides between exper- checked by three methods (HPLC, electrophoresis and mass iments performed in the light and those in the dark. Further- spectromety) and the only peptides studied in detail were those more, the omission of the kinase or rhodopsin produced negli- which gave, at 3 mM, less than 15% inhibition of rhodopsin gible peptide phosphorylation. The obligatory requirement of phosphorylation. Rho* for peptide phosphorylation, shown by the experiments In the peptide substrates, the residue corresponding to of Fig. 1, is further strengthened by the metarhodopsin-I1 Ser343 was preferentially phosphorylated. Furthermore, the decay study of Fig. 5 and supports the notion that the enzyme peptides were only monophosphorylated. This is not surpristhat acts on the synthetic peptides is rhodopsin kinase rather ing since the peptides have poor affinity (in the millimolar than another kinase present in our preparation. The other range) for the enzyme and the level of monophosphorylated kinases reported to be present in ROS include protein kinase peptide produced during the incubation was low for these C, types I and I1 A kinases, insulin or insulin-like-growth- species to compete for multiple phosphorylation with the origfactor-stimulated kinase and phosphatidylinositol-stimulated inal substrate present in large amounts. This conclusion is kinase. Of these, only protein kinase C has been shown to supported by our recent work (Pullen and Akhtar, unpubphosphorylate rhodopsin [38]. However, this phosphorylation lished results) in which we have prepared monophosdid not show light dependence and occurred with unbleached phorylated and disphosphorylated derivatives of the 10-resias well as bleached rhodopsin. Furthermore, protein kinase C due peptide (VII, residues 339 - 348) and the 12-residue requires, for its activity, 0.1 -0.2 mM calcium, which was not peptide, (VI, residues 337 - 348) by unambiguous chemical included in our incubations, and 0.1 mM EDTA was added synthesis. As expected, it was found that the electrophoretic mobilities of the two types of derivatives were different. In to remove any contaminating calcium. Table 1 also shows that the best peptide substrates were every case, the enzymatically produced phosphopeptides cothose which corresponded to the entire 10 - 12 C-terminal migrated with the authentic samples of monophosphorylated amino acids of rhodopsin. The peptides of similar sizes, but peptide. It should be borne in mind that, since the electrolacking the last two residues of rhodopsin, Pro347 and Ala348, phoretic separation of peptides is based on net charge, this had considerably reduced activities. The results highlight that technique cannot separate various isomers of monophosthe phosphorylation phenomenon studied here has a prefer- phorylated peptides from each other. If the results obtained with the peptides are extended to ence for the C-terminal sequences of the visual receptor, though these amino acids are not absolutely essential since the phosphorylation of the receptor itself, then Ser343 should the 11-residue peptide (V, residues 329 - 339), lacking the last be the most favoured site of modification. The absence of significant phosphorylation on the threonine residues in the 10 amino acids, was phosphorylated, albeit poorly. The gradual deterioration of phosphorylation with the 10-residue peptide (VIII, residues 339 - 348) is broadly conincreasing size of the peptides is also note worthy. Two possi- sistent with the findings of Thompson and Findlay [26]. These ble explanations for this observation were considered. Either, workers, during work on the identification of residues T Povine rhodopsin, were able in solution, the larger peptides adopted secondary structures radiolabelled with [ Y - ~ ~ P I Ain which prevented their recognition by the kinase, or these to detect radioactivity at all the potential serine sites (positions peptides interacted with the kinase but in a ‘non-productive’ 334, 338 and 343) but only in two (positions 335 and 336) of mode. The competition experiments in Fig. 3 show that the the possible six threonine sites. Fig. 5 shows that the decay of Rho* affects its own phos‘poor’ peptide substrates were, in fact, effective inhibitors of the phosphorylation of good peptide substrates, thus phorylation and that of the peptide (11-residue peptide VII, suggesting that longer peptides did bind to the kinase but the residues 338 - 348) similarly; both these processes were found interaction led, presumably, to the formation of catalytically to have half-times of phosphorylation decay of about 30 min. There was, however, a difference in that 1 h after the initiation incompetent complexes. In the present study, the peptides were found not to inhibit of decay, when rhodopsin phosphorylation could still be dethe phosphorylation of Rho* to any significant extent. This tected, phosphorylation of the peptide had nearly ceased. This is explicable by the greatly differing affinities of the peptides small divergence in the two phosphorylation reactions is not and Rho* for the kinase. In a previous study, the phosphoryla- surprising since, as we shall see later, the phosphorylation of tion or Rho* was studied in the presence of a number of the receptor and of peptides involve similar, but not identical, peptides corresponding to various domains of rhodopsin and catalytic complexes. Attention has already been drawn to the it was found that their effects on the phosphorylation process fact that the half-time for the decay of phosphorylation of

666 loop between helices

5&6

nJ

A “ZH

Kinase

-

+ Phospho-Rho’

CH3C0. Rhodopsin

Rho’

active

B Phospho-Rho

COZH

ATp

+ Active Kinase

CH3C0. Rhodopsin

Rho’

inactive

B2

I

CI

(Peptide)

-

B, I z r P e p t i d e i

AT P

Phosphopeptide

COZH CH3C0.

Scheme 2. Alternative mechanisms for the Light-dependent phosphorylation of rhodopsin. (A) Light excitation exposes the C-terminus of rhodopsin to the kinase; (B) kinase exists in an inactive form but is activated following binding to a domain in Rho*, presumably in the loop between transmembrane segments 5 and 6 [30]. In the illustration above, this loop is shown to undergo a conformational change. The peptide phosphorylation may then occur either following the dissociation of the active form of the kinase (Path B,) or within a Rho*.active kinase complex (Path B2). The arrangement of transmembrane helices and connecting loops shown above is purely illustrative and a more realistic view of this is shown in the inset provided by Professor J. B. C. Findlay, Leeds University.

Rho* does not mimic the half-time of decay of metarhodopsin 11, which has been estimated to be 10 - 15 min. The lack of this correlation has been observed by other investigators and particularly noteworthy are the experiments of Kibelbek et al. [ l l ] who found that the decay products of metarhodopsin I1 (e.g. metarhodopsin 111) participated in productive interaction with transducin. In the Results, we have pointed out that there is physiological merit in rhodopsin kinase displaying a somewhat broad substrate specificity so that all those intermediates of bleached rhodopsin which possess tranduction activity are kept in a phosphorylated, hence desensitised, state. The mechanism

The fact that dark-adapted rhodopsin is resistant to the action of the kinase but becomes an effective substrate following photoactivation may be rationalised by at least two alternative hypotheses (Scheme 2). The first (Scheme 2A) and the classical view which we have named ‘site accessibility hypothesis’, assumes that the C-terminus of rhodopsin, the site for the action of the kinase, is inaccessible to the enzyme when rhodopsin is in the dark-adapted state, and it is only after photoactivation that a new conformation of the visual protein is obtained in which the C-terminus becomes available for modification [18, 221. If this mechanism operated, then a straight-forward phosphorylation of the synthetic peptides by the kinase would have been expected. The finding that exogenously added peptides become effective substrates for the kinase only in the presence of bleached rhodopsin is, however, most readily explained by the second hypothesis (Scheme 2 B), termed the ‘kinase-activation hypothesis’, which requires that the kinase normally exists in an inactive form and is activated following binding to Rho*. It is interesting to recall that such a possibility was originally highlighted by the observation that isolated frog retina, when incubated with [y32P]ATPunder low-bleaching conditions, incorporated up to 20 phosphate residues/bleached rhodopsin molecule (40, 371.

Since rhodopsin does not contain 20 phosphorylation sites in a single molecule, one of the possibilities considered was that the high phosphate incorporation may be due to the phosphorylation of unbleached rhodopsin molecules [41, 421 by the kinase rendered ‘active’ through interaction with Rho*. Recently a more dramatic version of such an experiment has been reported in which the photoactivation of only 1000 or fewer of the 3 x lo9 rhodopsins present/rod cell resulted in the incorporation of 1400 phosphates into the total rhodopsini each activated rhodopsin (Rho*) [43]. The region of bleached rhodopsin which may interact with the kinase has been suggested to be the cytoplasmic loop corresponding to residues Lys231- Arg252. This inference is based on the observation that peptides corresponding to this sequence activate the kinase, albeit only marginally, and more importantly that rhodopsin cleaved at this loop is not, or only poorly, phosphorylated [30]. We now come to consider the precise nature of the activation process and the intimate details of the interactions through which the peptide substrates are phosphorylated by the kinase in the presence of bleached rhodopsin. In principle, there are two possibilities. First, that the binding of inactive kinase to Rho* results in the formation of a complex from which the active kinase eventually dissociates according to the kinetic scheme Rho* + inactive kinase t--f Rho* . inactive kinase H Rho* , active kinase ++ Rho* active kinase,

+

and participates in peptide phosphorylation. The second possibility is that the activated form of the kinase only exists within the binary complex, Rho* . activated kinase, and that the phosphorylation of peptides is due to their ability to ‘smuggle’ into the complex where they act as surrogate substrates and compete for phosphorylation with the C-terminal tail of the receptor (Scheme 2B). We have shown that although larger peptides are not phosphorylated themselves, they inhibit the phosphorylation of

667 smaller peptides, thus these (larger peptides) must interact 14. with the kinase but without producing a catalytically com- 15. petent complex. This behaviour may be explained by the 16. model B2 (Scheme 2) in which the difficulty of correctly po17. sitioning the larger peptides within a crowded crevice at the active site is intuitively obvious. Not withstanding this, further 18. work is required to distinguish between the two mechanisms represented by Paths B1 and B2 of Scheme 2. 19. There is now overwhelming evidence that rhodopsin is 20. prototypical of other guanosine-nucleotide-binding-regulatory-protein-linked receptor systems [44]. Thus, the most ex- 21. tensively studied member of this class, j-adrenergic receptor, is desensitised by a mechanism in which an agonist-occupied 22. form of the receptor is phosphorylated by a j-adrenergic- 23. receptor kinase. The latter displays several enzymological similarities to rhodopsin kinase and, recent, dramatic progress 24. in the cloning and sequencing of the two kinases has revealed a high degree of sequence similarity [45]. In view of this, the detailed mechanism proposed for rhodopsin kinase may also 25. be applicable to P-adrenergic-receptor kinase and other re- 26. lated enzymes. 27.

We thank the Ulverscroft Foundation and Wellcome Trust for financial support of our work in the field of vision. N. G. B. and C. F. were the recipients of postgraduate studentships from the Science and Engineering Research Council (SERC). We thank Dr M. G. Gore and Mr L. Hunt of the Protein Sequencing Unit of SERC Centre for Molecular Recognition, Southampton University, for peptide sequencing. We thank Dr D. L. Corina for mass spectral analysis which was performed using a VG TS-250 mass spectrometer, purchased by a grant from the Wellcome Trust. Lastly, we thank Mr Bill Boxhall of the Fresh Meat Company, Ltd, Salisbury, whose cooperation in the obtaining of retinae made this research possible.

REFERENCES 1 . Matthews, R. G., Hubbard, R., Brown, P. K. & Wald, G. (1963) J. Gen. Physiol. 47, 21 5 - 240. 2. Yoshizawa, T. & Wald, G. (1963) Nature 197, 1279-1286. 3. Wald, G. (1968) Science 162, 230-239. 4. Kiihn, H., Mommertz, 0. & Hargrave, P. A. (1982) Biochem. Biophys. Acta 679, 95 - 100. 5. Pellicone, C., Nullans, G., Cook, N. J. & Virmaux, N. (1985) FEBS Lett. 181, 179-183. 6. Bownds, D. & Wald, G. (1965) Nature 204,254-257. 7. Akhtar, M., Blosse, P. T. & Dewhurst, P. B. (1965) Life Sci. 4, 1221- 1226. 8. Emeis, D., Kiihn, H., Reichert, J. & Hofmann, K. P. (1982) FEBS Lett. 143, 29 - 34. 9. Bennett, N. Michel-Villaz, M . & Kiihn, H (1982) Eur. J . Biochem. 127,97 - 103. 10. Longstaff, C., Calhoon, R. D. & Rands, R. R. (1986) Proc. Natl Acad. Sci. USA 83,4209-4213. 11. Kibelbek, J. Mitchell, D. C., Beach, J. M. & Litman, B. J . (1991) Biochemistry 30, 6761 -6768. 12. Wheeler, G. L. & Bitensky, H. W. (1977) Proc. Nut1 Acad. Sci. USA 74,4238 -4242. 13. Fung, B. K.-K., Hurley, J. B. & Stryer, L. (1981) Proc. Natl Acad. Sci. USA 78, 152-156.

Note added in proof. Phosphorylation of the synthetic peptides at Ser343 has now been confirmed by direct sequencing.

28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45.

Chabre, M. & Deterre, P. (1989) Eur. J . Biochem. 179, 255-266. Stryer, L. (1991) J . Biol. Chem. 266, 10711-10714. Fesenko, E. E., Kolesnikov, S. S. & Lyubarsky, A. L. (1985) Nature 313, 310-313. Zimmerman, A. L., Yamanaka, G., Eckstein, F., Baylor, D. A. & Stryer, L. (1985) Proc. Nut1 Acad. Sci. USA 82,8813-8817. McDowell, J. H. & Kiihn, H. (1977) Biochemistry 16, 40544060. Kiihn, H. (1978) Biochemistry 17, 4389-4395. Bownds, D., Dawes, J., Miller, J. & Stahlman, M. (1972) Nature 237,125-127. Kiihn, H., Cook, J. H. & Dreyer, W. J. (1973) Biochemistry 12, 2459 - 2502. Frank, R. N. & Buzney, S. M (1975) Biochemistry 14, 51105117. Sale, G. J., Towner, P. & Akhtar, M. (1978) Biochem. J . 175, 421 -430. Hargrave, P. A., Fong, S.-L., McDowell, J. H., Mas, M. T. Curtis, D. R., Wang, J. K., Juszczak, F. & Smith, D. P. (1980) Neurochem. 1,231 -244. Wilden, U. & Kiihn, H. (1982) Biochemistry 21, 3014-3022. Thompson, P. & Findlay, J. B. C. (1984) Biochem. J. 220, 773780. Fowles, C., Sharma, R. & Akhtar, M. (1988) FEBS Lett. 238, 56 -60. Palczewski, K., McDowell, J. H. & Hargrave, P. A. (1988) Biochemistry 27, 2306-2313. Palczewski, K., Arendt, A., McDowell, J. H. & Hargrave, P. A. (1989) Biochemistry 28, 8764- 8770. Palczewski, K., Buczylko, J., Kaplan, M. N., Polans, A. S. & Crabb, J. W. (1991) J. Biol. Chem. 266, 12949-12955. Onorato, J. J., Palczewski, K., Regan, J. W., Caron, M. G., Lefkowitz, R. J. & Benovic, J. L. (1991) Biochemistry 30,5118 5125. Kiihn, H. & Wilden, U. (1982) Methods Enzymol. 81,490-491. Barany, G. & Merrifield, R. B. (1979) in The peptides (Gross, E. & Meienhoffer, J., eds) pp. 1-248, Academic Press, New York. Tam, J. P., Heath, W. F. & Merrifield, R. B. (1983) J. Am. Chem. SOC.105, 6442-6455. Fowles, C., Akhtar, M. & Cohen, P. (1989) Biochemistry 28, 9385-9391. Palczewski, K., Hargrave, P. A., McDowell, J. H. & Ingebritsen, T. S. (1989) Biochemistry 28,415-419. Miller, J. A. & Paulsen, R. (1975) J . Biol. Chem. 250,4427-4432. Kelleher, D. J. &Johnson, G. L. (1986) J . Biol. Chem. 261,47494757. Kelleher, D. J. &Johnson, G. L. (1990) J . Biol. Chem. 265,26322639. Bownds, D., Dawes, J. & Miller, J. (1973) in Biochemistry & physiology of visual proteins (Langer, H., ed.) pp. 267-273, Springer-Verlag, New York. Miller, J. A,, Paulsen, R. & Bownds, D. (1977) Biochemistry 16, 2633 - 2639. Aton, B. R. (1986) Biochemistry 25, 677-680. Binder, B. M., Biernbaum, M. S. & Bownds, D. (1990) J. Biol. Chem. 265, 15333-15340. Palczewski, K. & Benovic, J. L. (1991) Trend~yBiochem. Sci. 16, 387 - 391. Lorenz, W., Inglese, J., Palczewski, K., Onorato, J. J., Caron, M. G. & Lefkowitz, R. J. (1991) Proc. Nut1 Acad. Sci USA 88, 8715- 8719.