| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
* Department of Biological Sciences, Department of Chemistry and Biochemistry, University of Maryland, Baltimore, Maryland
Correspondence: Address reprint requests to Asst. Prof. Susan K. Gregurick, Chemistry and Biochemistry, University of Maryland, Baltimore County, 1000 Hilltop Circle, Baltimore, MD 21250. Tel.: 410-455-8698; E-mail: greguric{at}umbc.edu.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONEXPERIMENTAL MATERIALS AND...COMPUTATIONAL METHODSCOMPUTATIONAL RESULTSEXPERIMENTAL IN VITRO RESULTSCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL MATERIALS AND...COMPUTATIONAL METHODSCOMPUTATIONAL RESULTSEXPERIMENTAL IN VITRO RESULTSCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
; Jacobs, 1998; Sakmar et al., 2002). Once the signaling cascade has been activated, the receptor must then be quickly inactivated. Specifically for the photopigment rhodopsin, deactivation is initiated by a receptor-specific serine/threonine kinase, rhodopsin kinase (GRK1), which phosphorylates residues on the cytoplasmic tail and hence initiates partial deactivation. Complete deactivation of the receptor does not occur until a second regulatory protein, arrestin, binds. The interaction between receptor and arrestin results in the fast quenching of the signal transduction by decreasing the coupling of the receptor to the G-protein, transducin (Alloway and Dolph, 1999; Wilden et al., 1986; Lohse et al., 1990, 1992).
Phosphorhodopsin induces a large conformational change in arrestin which shifts this protein from an inactive, low-affinity binding state to an active high-affinity binding state, where there is a fivefold increase of the affinity of arrestin for phosphorhodopsin (Kuhn et al., 1986). Mutagenic studies have elucidated three regions of arrestin that respond to different states of rhodopsin (inactive, active, and phosphorylated; Gurevich, 1998). It is the interaction between phosphorylated rhodopsin residues and the arrestin protein that is of interest to our studies.
The cytoplasmic tail of rhodopsin contains seven possible phosphorylation sites. In vivo, the identity and number of phosphate groups incorporated into the tail is still under investigation. Several studies have found evidence that up to three phosphates are incorporated into the cytoplasmic tail of rhodopsin (Zhang et al., 1997; Kennedy et al., 2001a; Ablonczy et al., 2002; McDowell et al., 2001), and site-specific mutagenesis indicates that at least two rhodopsin phosphorylation sites are required for full desensitization (McDowell et al., 2001), whereas studies with transgenic mice indicate at least three phosphorylated resides are required (Mendez et al., 2000). Candidate phosphorylation sites include serine 334, serine 338, threonine 340, and serine 343 (Brannock et al., 1999; McDowell et al., 2001). The mechanism that these phosphorylated residues play in the desensitization process is still under investigation; however, it is believed that they will initially interact with arrestin via a charge-charge mechanism (Vishnivetskiy et al., 2000).
Therefore, our study aims to determine specific site-site interactions between the cytoplasmic tail of rhodopsin and arrestin using both computational methods and in vitro biochemical assays. For our approach, we draw on previous studies that have examined the regions necessary for this interaction. Specific crystallographic studies of bovine arrestin have elucidated regions that may respond to the phosphorylated residues on rhodopsin. The crystal structure of the inactive conformation of arrestin suggest that it is composed of two ß-sheets oriented in such a way as to create a hydrophobic core between them (Gray-Keller et al., 1997). Within this hydrophobic core there exist at least two salt bridges which act to stabilize the low-affinity binding state (Hirsch et al., 1999). During activation, the phosphorylated residues are believed to penetrate into the hydrophobic core and thereby disrupt these salt bridges which then induces the conformational change in arrestin to achieve an active high-affinity binding state (Vishnivetskiy et al., 2000, 1999; Gurevich, 1998; Palczewski et al., 2001).
The initial interaction between rhodopsin and arrestin, however, is not well understood. Vishnivetskiy et al. (2000) proposed that it might be a charge-charge interaction between two key positively charged lysine residues (Lys14 and Lys15) on arrestin, interacting with the negatively charged phosphorylated residues of rhodopsin. This charge-charge interaction is thought to guide the rhodopsin into the polar core of the arrestin protein and thus allow for arrestin's conformational change. At present, this mechanism is still hypothetical; however, it is based on site-specific mutation studies of arrestin (K14A and K15A) that resulted in a nearly eightfold reduction of rhodopsin binding (Vishnivetskiy et al., 2000).
It is the aim of this study to investigate the desensitization process in G-protein-coupled receptors, from both a computational and experimental prospective. In particular, we will study the rod-specific visual pigment, rhodopsin, in complex with visual arrestin as a model system for this process. We should point out that any simulation of phosphorylated rhodopsin must contend with the fact that a molecular force field to describe specific phosphorylated residues does not exist at this time. To address this issue, we will perform key mutation studies to model the negatively charged phosphorylated residues as either aspartic or glutamic acid. Therefore, our simulations are only directly comparable to the experimental work of McDowell et al. (2001) and the current experiments. We stress that although this computational work is of peptide binding to arrestin, it is believed that the larger rhodopsin-arrestin complex will interact in a similar fashion as the peptide-arrestin complex (McDowell et al., 2001; Puig et al., 1995). To the best of our knowledge, this investigations represent the first such simulation of the desensitization process. Recently, Orry and co-workers have begun a study of identifying ligand binding sites in G-protein-coupled receptors using a novel docking approach (Cavasotto et al., 2003; Orry and Wallace, 2000). Our approach is slightly different in that we are looking at specific protein-protein interactions, allowing for full conformational flexibility of both proteins. Therefore, our approach is not suitable for a full screening of drug assays, but is meant to simulate possible binding events in the desensitization process.
|
EXPERIMENTAL MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL MATERIALS AND...COMPUTATIONAL METHODSCOMPUTATIONAL RESULTSEXPERIMENTAL IN VITRO RESULTSCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
. Rhodopsin concentrations of membrane preparations were approximated by Western blot using mAb 4D2, an antibody against the N-terminus of bovine rhodopsin (antibody was a gift from R. Molday). Blots were quantitated using ImageQuant (Molecular Dynamics, Sunnyvale, CA) software. The concentration of rhodopsin in membrane preparations varied between 50 and 100 nM. The samples were stored at -80°C until ready for use. The 50-µL rhodopsin samples were reconstituted with 360 µM 11-cis retinal at 4°C for 1 h on a rotating plate.
Purification of arrestin from bovine retinas
Arrestin was purified from dark-adapted bovine retinas according to the method described by Hargrave (1986) and the samples were stored at -80°C.
Purification of transducin from bovine rod outer segments
Transducin was purified as described by Wessling-Resnick and Johnson (1987), and the protein was stored in 50% glycerol at -20°C.
Expression and purification of rhodopsin kinase
Rhodopsin kinase was expressed in High-Five insect cells (Invitrogen). A viral construct containing the rhodopsin kinase gene was a gift from K. Palczewski. High-Five cells were grown as a suspension culture to a cell density of 1 x 106 cells/mL. High-Five cells were infected with baculovirus containing RK gene at a multiplicity of infection = 5. The cells were incubated at 27°C for 72 h and were harvested by centrifugation (1000 x g) of 50 mL aliquots and subsequently stored at -80°C until ready for purification. For RK purification, a single aliquot was thawed in ice water for 30 min and resuspended into 3 mL of homogenization buffer (10 mM BTP, 0.04% Tween 20). The cells were homogenized with a Dounce homogenizer and the homogenate was mixed with 10 mL volume of regenerated DE52 resin (Whatman, Maidstone, Kent, UK). The kinase was eluted from the column with 100 mM NaCl in the homogenization buffer. Rhodopsin kinase positive fractions were verified using a filter-binding assay as described previously by Brannock et al. (1999).
GTP
S35-binding filter-binding assays
The ability of arrestin to bind to rhodopsin was indirectly measured through measurement of transducin activation by rhodopsin. A filter-binding assay was employed for this purpose. 2.5–5 picomoles of rhodopsin were reconstituted in the presence of 360 µM 11-cis retinal in a 50 µL volume for 1 h on a vertically rotating platform at 30 rpm. 12 µL of the reconstituted membranes were incubated in a buffer containing 30 mM BTP (pH 7.5), 3 mM MgCl2, 0.4 mM ATP, 0.3 mM DTT, and 2–4 µL kinase or kinase elution buffer (see Arrestin-Mediated Rhodopsin Deactivation Assays, below) to generate a final volume of 35 µL. The mixture was incubated at 30°C for 15 min under bright light. After the 15-min incubation, 15 µL of arrestin or arrestin buffer were added to generate 60% of the final arrestin concentration. The mixture was incubated for another 15 min at 30°C under bright light. The reaction was terminated by placing the tubes on ice for 2 min, at which time reaction volumes were increased to generate 100% of the final arrestin concentration of 5 µM, 2 µM transducin, 100 mM NaCl, 10 mM Tris-Cl (pH 7.5), 5 mM MgCl2, and 1 mM DTT to a final volume of 90 µL. The reactions were started by adding 10 µL of 30 µM GTPS35. 10 µL of the reaction mixture were added to nitrocellulose filters on a Millipore vacuum manifold at 30-s intervals. The filters were washed with 15 mL of reaction buffer. Then the filters were added to 5 mL of scintillation fluid (Amersham Biosciences, Piscataway, NJ) and allowed to shake vigorously overnight. Reactions rates were expressed as picomoles of GTPS35 bound per min. The linear regressions were calculated using SigmaPlot software.
Arrestin-mediated rhodopsin deactivation assays
Arrestin-mediated rhodopsin deactivation was indirectly measured by measuring transducin activation. Deactivation was expressed as percentage of inhibition of transducin activation. For experiments requiring phosphorylation of rhodopsin, three reactions were run simultaneously and the rates of transducin activation (v) were calculated for each. Reaction rates were as follows: v1 = (–) kinase, (–) arrestin; v2 = (+) kinase, (–) arrestin; and v3 = (+) kinase, (+) arrestin. Enough kinase was added to the reaction so v2 = 50% for Wt (v1). The decrease in rate of transducin activation due to arrestin-mediated deactivation was given as: % inhibition = [1 - (v3/v2)]100. For the experiments with rhodopsin mutants that had glutamic acid and aspartic acid substitutions, only two reactions were run. Reaction rates are as follows: v1 = (–) kinase, (–) arrestin, and v2 = (–) kinase, (+) arrestin. For these reactions, the decrease in rate of transducin activation due to arrestin-mediated deactivation is given as: % inhibition = [1 - (v2/v1)]100, and is reported as an arrestin effect.
|
COMPUTATIONAL METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL MATERIALS AND...COMPUTATIONAL METHODSCOMPUTATIONAL RESULTSEXPERIMENTAL IN VITRO RESULTSCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
; Ewing and Kuntz, 1997; Makino and Kuntz, 1997; Eisen et al., 1994). Recently Orry and Wallace utilized a docking approach to test their molecular model of the Endothelin G-protein-coupled receptor (Orry and Wallace, 2000; Cavasotto et al., 2003). Although this approach is highly successful for rigid-body (and semiflexible) ligands, it will not allow for gross conformational changes to occur in neither protein nor ligand. However, in our case, the entire complex must be allowed to change conformation during the docking procedure. This requires an effective search of all conformational space available to the complex, as a function of conformational changes in either, or both, docking proteins. In other words, both proteins must be able to refold upon interaction. Therefore, a rigid-body docking algorithm would not suffice for this particular study where there is a definite conformational change in the complex. Thus to effectively search the conformational space available to the rhodopsin-arrestin complex, and hence get a sense of the interactions between the two proteins, a torsion space, all-atom, Monte Carlo simulated annealing algorithm was developed. Our algorithm is based on the all-atom Monte Carlo algorithm of Avbelj and Moult developed for the folding of small peptide structures (independent folding units) of between 11 and 14 residues in length (Avbelj and Moult, 1995). Although we utilized the same free energy function (
G) and Monte Carlo sampling techniques as Avbelj and Moult, we have extended the original algorithm to include a Monte Carlo simulated annealing optimization procedure that will effectively handle the refolding of protein-protein complexes.
In accordance with the original Monte Carlo algorithm, all of the atoms were allowed to move in the simulation; however, only the bonding angles (
, 
, and 
) were allowed to vary. Thus, the length of all bonds was kept fixed throughout the simulation. As in many standard Monte Carlo techniques, an angle (backbone or side-chain) was selected and moved and then a Metropolis criterion (Metropolis et al., 1953) was used to evaluate the acceptance of the move as follows: if the free energy (G) decreased, the new conformation was accepted; if the free energy increased, then the move was accepted if
 |
G is the free energy of the complex, k is the Boltzmann constant, and T is the current simulation temperature (in K). Angles were randomly chosen for a given move and the following move set of Avbelj and Moult was employed throughout the simulations:
Of the angles chosen (backbone or side-chain) the following types of moves were allowed:
).
If any given move resulted in a steric clash, then that move was immediately disregarded and a new angle was chosen. The initial temperature of the simulation was set at 300 K, the final temperature was set to 50 K and a cooling rate of 0.988 was employed. Thus the temperature decreased 
0.1 K per every 10 simulation steps.
The free energy function (G)
The MC-SA procedure optimizes geometry by optimizing an all-atom implicit solvent free energy function. The free energy function used in the MC-SA algorithm, defined as G, was derived by Avbelj and Moult (1995). It is a potential of mean force based on a set of 114 nonhomologous protein crystal structures and has the form
 | (1) |
is the Coulomb energy of residue k, arising from interactions between the NH and C=0 groups with those of the flanking residues (k-1 and k+1). In general, is unfavorable for residues in a helical conformation and favorable for extended conformations. The balance of this term and that of main-chain hydrogen bonding will establish the secondary structure of the protein.
The second term in the free energy function is defined as the electrostatic energy for the other intramolecular interactions. The sums over k and l are over polar (k) and charged (l) residues of types i and j, for which the distance between the proton donor and acceptor is shorter than 6.5 Å. Ki(k)j(l) is a scaling factor that is dependent on the type of charged or polar groups involved. The equation for the Coulomb energy, 
, is given as
 | (2) |
The third term in the free energy function (Eq. 1) is the solvation free energy, and is based on the average local solvent accessibility of the atoms in the observed conformation. For each set of atoms of the same type in any configuration, 
can be expressed as in Avbelj and Moult (1995),
 | (3) |
The exponent n is equal to 1 for carbon and sulfur atoms and 3 for oxygen and nitrogen atoms. In Eq. 3, is the difference between the surface area for an estimated random coil conformation (At; Lee and Richards, 1971) and the exposed surface area for each atom, m, in the current conformation (Am).
Taken together, the three terms in Eq. 1 represent the potential of mean force, which is optimized for the lowest energy conformation using the Monte Carlo simulated annealing algorithm.
|
COMPUTATIONAL RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL MATERIALS AND...COMPUTATIONAL METHODSCOMPUTATIONAL RESULTSEXPERIMENTAL IN VITRO RESULTSCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
).
Modeling the cytoplasmic tail of rhodopsin
Due to the size of rhodopsin (348-residue Protein Data Bank entry: 1HZX.pdb), we are not yet able to include all of the residues of the rhodopsin-arrestin complex in our simulations. This is because the number of degrees of freedom, or the number of movable dihedral angles, is too large for our current simulation to handle. However, at present we are only interested in studying the initial mechanism of the desensitization process, which involves only the tail portion of rhodopsin (residues 329–348) associating with the full arrestin protein. Thus the remaining 328 residues of rhodopsin are relatively unimportant for this initial study. It must be stressed, however, that the full 353 residues of arrestin are important for this study to ascertain if a conformational change of this protein occurs when associating with rhodopsin.
Because the cytoplasmic tail of rhodopsin is disordered in solution and thus the x-ray crystal structure is not available for these residues, we have modeled the tail as follows:
G = -174.68 Kcal/mol, and has a two-turn helix turn structure, as illustrated in Fig. 1.
|
G = -3464.86 kcal/mol, with an RMS change of 0.539 Å as compared to the basal state crystal structure. This indicates that the crystal structure of arrestin is indeed at an energy minimum, hence any conformational changes observed during association with rhodopsin will be due to the interactions of the rhodopsin-arrestin complex and not an artifact of our simulation procedures.
The energy-minimized wild-type cytoplasmic tail of rhodopsin was manually docked into the binding pocket of arrestin using the program PSSHOW (Swanson, 1995). Care was taken that the distance between most residues of rhodopsin and arrestin were >9.0 Å apart, to begin our simulations from an unbiased starting configuration. Fig. 2 illustrates the wild-type complex, which is used as a starting structure for the simulations.
|
RMS (root mean-square deviation) between the initial structure (Fig. 2) and the final structure (not shown) is 0.56 Å. In fact, the wild-type cytoplasmic tail began to leave the active site during the simulation. Furthermore, there did not appear to be any interaction between the two structures. This is seen in Table 2, where the residue-residue distance (in Å) is never <8 Å. This confirms the experimental evidence that unphosphorylated (wild-type) rhodopsin will not interact with basal state arrestin. As a check of our simulation, we mutated the seven serine/threonine residues to alanine in the cytoplasmic tail of rhodopsin. We then performed the same MC-SA simulation on this mutant complex and again observed neither conformational changes (Table 1) nor binding interactions in the complex (Table 2), as expected.
|
|
|
|
as well as our current experimental work. In the investigations of McDowell and co-workers, the authors found evidence for an interaction between the all-aspartic and all-glutamic peptide tail analogous with arrestin. These interactions are not necessarily the same as in the phosphorylated case but it is exactly this interaction which we will elucidate as a foundation for future studies.
Simulations of the rhodopsin aspartic tail analog
In an attempt to elucidate the charge-charge interaction of rhodopsin with R-arrestin we have performed computational site-directed mutagenic studies of this system. Our mutation studies include both aspartic acid and glutamic acid mutations for the serine/threonine residues of the cytoplasmic tail of rhodopsin. We have also looked at different initial orientations of the rhodopsin tail interacting and refolding within the binding pocket of R-arrestin. We define one orientation as parallel to the ß-sheet of the Lys14/Lys15 pair (Fig. 2), and the other initial orientation of the rhodopsin is defined as perpendicular (Fig. 3). Based on our Monte Carlo simulated annealing simulations, we find that only the parallel orientation of the rhodopsin tail analog is capable of interacting with the Lys14/Lys15 residues on arrestin. This finding is best illustrated in Tables 1–3 and Fig. 4. In Table 1 we see that the CRMS distance between the initial and final structures of the aspartic acid mutant complex is 1.9 Å. The change in this distance is due mostly to the refolding of the aspartic mutant tail (CRMS 5.3 Å). In previous simulations of both the wild-type and the alanine mutant, the tail did not refold in the binding pocket of arrestin, indicating that the charge-charge interaction is necessary for this refolding process to occur.
|
|
and came as a surprise to us. The final structure of the all-aspartic acid tail analog interacting with arrestin (simulation 2) is illustrated in Fig. 4. The key residues are drawn in stick representation to highlight these strong interactions.
We also determined if different parallel conformations would interact in the same way with the R-arrestin protein. To address this issue, we created a new initial parallel conformation by refolding the all-aspartic tail analog separately into a new configuration. We then placed this new structure into the binding pocket of R-arrestin in a parallel conformation (Fig. 5). This new structure was also tested for all the constitutive mutation studies we performed. We find that during the course of the simulation the rhodopsin residues D338 and D340 (parallel structure 2) interact with the R-arrestin protein residues K14 and K15. However, this interaction must be somewhat weaker as the interaction distance is 6.2 Å and 8.7 Å, respectively (simulation 4, Table 2). In a recent article by Kumar and Nussinov on the relationship between ion pair geometry and electrostatic strengths in proteins, the authors find that if the ion pair is within 4 Å this tends to be a stabilizing interaction, whereas if the ion pair is >4 Å apart this tends to be a destabilizing interaction (Kumar and Nussinov, 2002). This work is done by exploring 11 nonhomologous protein NMR structures and is based on a continuum electrostatic calculation. Because we have not calculated the electrostatic strength of our ion pairs separately, we cannot say if the short-range 3.7 Å distance is stabilizing or destabilizing to the complex. We are now in the process of investigating this effect for the ion pairs formed in the complex.
|
Simulations of the rhodopsin glutamic tail analog
To address the issue of the effect of the side-chain size (Brannock et al., 1999) on the charge-charge interaction, we performed a glutamic acid mutation study. We begin with the same initial parallel structure as in the all-aspartic mutation study (Fig. 2). However, in this case, we mutated all-rhodopsin tail serine and threonine residues to glutamic acid. We then performed the same Monte Carlo simulated annealing algorithm as before. The results are presented in Tables 1–3. We see from Table 1 that the all-glutamic tail analog has refolded in the complex (C RMS is 5.58 Å) as in the case of the all-aspartic acid tail analog. However, in this particular case, we find that residue E338 on the rhodopsin tail analog will interact with the K14 residue on R-arrestin (Fig. 6). This interaction was not seen in the all-aspartic acid analog. Moreover, the all-glutamic acid analog did not show any evidence of a E334-K14 interaction or E340-K15 interaction, as in the all-aspartic acid case. We also created a two additional mutants in a parallel orientation: S334E:T340E and S338E:T340E. Although we know from experimental evidence that the S334E:T340E mutant should interact with the R-arrestin (as well as the S334D:T340D mutant), we could find no evidence of this interaction for the glutamic acid analogue. The bulky side chains of the glutamic acid mutant proved to be difficult to simulate. Thus we preformed only one simulation for each glutamic acid mutant. Based on our limited simulations, we find a somewhat weaker interaction between the glutamic acid tail analog and the R-arrestin protein, compared to the aspartic acid analog.
|
EXPERIMENTAL IN VITRO RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL MATERIALS AND...COMPUTATIONAL METHODSCOMPUTATIONAL RESULTSEXPERIMENTAL IN VITRO RESULTSCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
|
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL MATERIALS AND...COMPUTATIONAL METHODSCOMPUTATIONAL RESULTSEXPERIMENTAL IN VITRO RESULTSCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
-helix until residue V337, and thereafter the tail adopts a more random loop structure. This result is somewhat consistent with recent EPR spectroscopy that indicates the cytoplasmic tail is part of a short H-VII helix (residues 328–333; Altenback et al., 2001); however, solution NMR studies indicate that the tail is dynamically disordered beyond residue 327 (Langen et al., 1999).
Tables 1–3 summarize the conformational energetics, C root mean-square deviation (RMSD) (Table 1) and the residue-residue interaction distance (Table 2) between the initial and final structures of various mutated rhodopsin-arrestin complexes which we have simulated. We note that the alanine mutant is not represented in Tables 1 and 2, inasmuch as it is similar to the wild-type simulation. We find that both the glutamic acid and aspartic acid mutated tail analogs produced a slight conformational change in R-arrestin (0.56 and 0.133 Å). However, both of the negatively charged tail mutants refolded into completely new conformations when introduced in complex with R-arrestin (5.58 and 5.269 Å). Moreover, both the aspartic and glutamic acid tail analogs showed potential charge-charge interactions with R-arrestin, although at different sites. For example, in the aspartic tail analog we see specific interactions of S334D (tail) with K15, K300, H301, and R29 (R-arrestin) and also T340D (tail) with K15 (R-arrestin) (Fig. 4), whereas the glutamic tail analog mainly interacts at S338E (tail) with K14 (R-arrestin) (Fig. 6).
What about the role of the serine 343 residue in the rhodopsin cytoplasmic tail? There is compelling in vivo and in vitro work to suggest that this serine residue is indeed phosphorylated, and as such, the question is, does it play a role in desensitization (Ohguro et al., 1995, 1996; Zhang et al., 1997; Kennedy et al., 2001b). To address this issue, we have analyzed our simulation results for the interaction of residue 343 with the arrestin lysine residues (K14, K15) of interest (Table 3). We find, in one simulation (the all-glutamic acid mutant, conformation 1), an interaction between E343 (rhodopsin) and K15 (arrestin). Moreover, this is an obvious interaction as the interaction distance changes from 17.22 Å to 7.35 Å. The question is how to interpret these results, since we find only one simulation with an obvious rhodopsin S343 interaction with arrestin. In all other simulations studied, we consistently see interactions of rhodopsin residues S338, T340, and strikingly S334 with the arrestin protein. One possibility could be that we have simply not run enough simulations to sample all possible interactions of rhodopsin with arrestin. We are currently running multiple mutant glutamic acid simulations to test this hypothesis.
Another possibility is, if phosphorylated, residue S343 will have a weak, but viable interaction with the arrestin protein. We tested this hypothesis with our in vitro assay (Fig. 7). We find that single mutants of aspartic acid on any phosphorylated rhodopsin residue produces a relatively insignificant arrestin effect, with the possible exception of residue D334. Moreover, the double mutant D335/D343 also produces an insignificant arrestin effect. This suggest that the role of residue S343 in desensitization is secondary at best. However, a polar (negative) charge at this position may facilitate the formation/stabilization of the rhodopsin-arrestin complex. We stress that we have not tested all possible mutant combinations involving residue S343; we merely suggest that, if one relates the distance between the ion pair residues to the strength of this interaction, that the order of affinity would be S334 > T340 > S338 > S343. This would imply that mutation of the S334 residue could have a significant effect on the binding of the rhodopsin-arrestin complex. Our results are consistent with the dephosphorylation results of Palczewski and co-workers, who find that dephosphorylation occurs in the following order (slowest to fastest): S334 < S338 < S343 (Ohguro et al., 1996).
We believe that our findings may be significant, inasmuch as it is thought that the rhodopsin residues Ser334, Ser338, and Thr340 are key phosphorylation sites (McDowell et al., 2001). Our results also indicate the relative importance of the R-arrestin Lys14 and Lys15 as a possible charge-charge interaction site (Vishnivetskiy et al., 2000). Moreover, we find that the orientation of the cytoplasmic tail of rhodopsin in the binding pocket of arrestin is important. That is to say, only a parallel tail orientation facilitates binding, whereas a perpendicular orientation did not seem to show any interaction between the two proteins. We also note that our simulations do not show any evidence of a significant role of the other serine/threonine residues in this binding process. We must stress that this is the first computational study of such a system.
Based on our computational studies, we predicted that the S334D/T340D aspartic acid mutant analog of rhodopsin would bind to arrestin in a comparable way to phosphorylated rhodopsin. This hypothesis was tested with our in vitro assays (Fig. 7) and results indicate this to be the case. We are now in the process of studying the effect of this ionic interaction, both experimentally and computationally.
|
ACKNOWLEDGEMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL MATERIALS AND...COMPUTATIONAL METHODSCOMPUTATIONAL RESULTSEXPERIMENTAL IN VITRO RESULTSCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
S.K.G. acknowledges funding from the University of Maryland, Baltimore County in the form of a Special Research Initiative Award, and P.R.R. acknowledges the National Science Foundation for continued support (grant #0119102).
Submitted on August 8, 2003; accepted for publication December 12, 2003.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL MATERIALS AND...COMPUTATIONAL METHODSCOMPUTATIONAL RESULTSEXPERIMENTAL IN VITRO RESULTSCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
Alloway, P. G., and P. J. Dolph. 1999. A role for the light-dependent phosphorylation of visual arrestin. Proc. Natl. Acad. Sci. USA. 96:6072–6077.
Altenback, C., K. Cai, J. Klein-Seetharaman, H. G. Khorana, and W. L. Hubbell. 2001. Structure and function in rhodopsin: mapping light-dependent changes in distance between residue 65 in helix tm1 and residues in the sequence 306–319 at the cytoplasmic end of helix tm7 and in helix h8. Biochemistry. 40:15483–15492.[Medline]
Avbelj, F., and J. Moult. 1995. The role of electrostatic screening in determining protein main chain conformational preferences. Biochemistry. 34:755–764.[Medline]
Brannock, M., K. Weng, and P. R. Robinson. 1999. Rhodopsin's carboxyl terminal threonines are required for the arrestin mediated quench of transducin activation in vitro. Biochemistry. 38:3770–3777.[Medline]
Cavasotto, C. N., A. J. W. Orry, and R. A. Abagyan. 2003. Structure-based identification of binding sites, native ligands and potential inhibitors for G-protein-coupled receptors. Protein Struc. Func. Gen. 51:423–433.
Eisen, M. E., D. C. Wiley, M. Karplus, and R. Hubbard. 1994. HOOK: a program for finding novel molecular architectures that satisfy the chemical and steric requirements for a macromolecule binding site. Protein Struc. Func. Gen. 19:199–221.
Ewing, T., and I. Kuntz. 1997. Critical evaluation of search algorithms for automated molecular docking and database screening. J. Comp. Chem. 18:1175–1189.
Gray-Keller, M. P., P. B. Detwiler, J. L. Benovic, and V. V. Gurevich. 1997. Arrestin with a single amino acid substitution quenches light-activated rhodopsin in a phosphorylation-independent fashion. Biochemistry. 36:7058–7063.[Medline]
Gurevich, V. V. 1998. The selectivity of visual arrestin for light-activated phosphorhodopsin is controlled by multiple nonredundant mechanisms. J. Biol. Chem. 273:15501–15506.
Hargrave, P. A. 1986. Molecular Dynamics of the Rod Cell, Vol. 1. Academic Press, Orlando, FL.
Hirsch, J. A., C. Schubert, V. V. Gurevich, and P. B. Sigler. 1999. A model for arrestin's regulation: the 2.8 Å crystal structure of visual arrestin. Cell. 97:257–269.[Medline]
Holm, L., and C. Sander. 1994. Protein structure comparison by alignment of distance matrices. J. Mol. Biol. 233:123–138.
Jacobs, G. H. 1998. Photopigments and seeing: lessons from natural experiments. Invest. Opth. Vis. Sci. 39:2204–2216.
Kennedy, M., K. A. Lee, G. A. Niemi, D. B. Craven, G. G. Garwin, J. C. Saari, and J. B. Hurley. 2001a. Multiple phosphorylation of rhodopsin and the in vivo chemistry underlying rod photoreceptor dark adaptation. Neuron. 31:87–101.[Medline]
Kennedy, M. J., K. A. Lee, G. A. Niemi, K. B. Craven, G. G. Garwin, J. C. Saari, and J. B. Hurley. 2001b. Rapid and reproducible deactivation of rhodopsin requires multiple phosphorylation sites. Neuron. 32:451–461.[Medline]
Kuhn, H., S. W. Hall, and U. Wilden. 1986. Light-induced binding of 48-kDa protein to photoreceptor membranes is highly enhanced by phosphorylation of rhodopsin. FEBS Lett. 176:473–478.
Kumar, S., and R. Nussinov. 2002. Relationship between ion pair geometries and electrostatic strengths in proteins. Biophys. J. 83:1595–1612.
Langen, R., K. Cai, C. Altenbach, H. G. Khorana, and W. L. Hubbell. 1999. Structural features of the C-terminal domain of bovine rhodopsin: a site-directed spin-labeling study. Biochemistry. 38:7918–7924.[Medline]
Lee, B., and F. M. Richards. 1971. The interpretation of protein structures: estimation of static accessibility. J. Mol. Biol. 55:379–400.[Medline]
Lohse, M. J., S. Andexinger, J. Pitcher, S. Trukawinski, J. Codina, J. Faure, M. G. Caron, and R. J. Lefkowitz. 1992. Receptor-specific desensitization with purified proteins. Kinase dependence and receptor specificity of ß-arrestin and arrestin in the ß2-adrenergic receptor and rhodopsin systems. J. Biol. Chem. 267:8558–8564.
Lohse, M. J., J. L. Benovic, J. Codina, M. G. Caron, and R. J. Lefkowitz. 1990. ß-arrestin: a protein that regulates ß-adrenergic receptor function. Science. 248:1547–1550.
Makino, S., and I. Kuntz. 1997. Automated flexible ligand docking method and its application for database search. J. Comp. Chem. 18:1812–1825.
McDowell, J. H., P. R. Robinson, R. L. Miller, M. T. Branock, A. Arendt, W. C. Smith, and P. A. Hargrave. 2001. Activation of arrestin: requirement of phosphorylation as the negative charge on residues in synthetic peptides from the carboxyl-terminal region of rhodopsin. Invest. Opth. Vis. Sci. 42:1439–1443.
Mendez, A., M. E. Burns, A. Roca, J. Lem, L. W. Wu, M. I. Simon, D. A. Baylor, and J. Chen. 2000. Rapid and reproducible deactivation of rhodopsin requires multiple phosphorylation sites. Neuron. 28:153–164.[Medline]
Metropolis, N., A. W. Rosenbluth, M. N. Rosenbluth, A. Teller, and E. Teller. 1953. Equations of state calculations by fast computing machine. J. Chem. Phys. 21:1087–1091.
Molday, R. S. 1998. Photoreceptor outer segment proteins, phototransduction and retinal degenerative diseases: the Friedenwald Lecture. Invest. Opth. Vis. Sci. 39:2493–2513.
Ohguro, H., J. P. V. Hooser, A. H. Milam, and K. Palczewski. 1995. Rhodopsin phosphorylation and dephosphorylation in vivo. J. Biol. Chem. 270:14259–14264.
Ohguro, H., M. Rudnicka Nawrot, J. Buczylko, X. Zhao, J. A. Taylor, K. A. Walsh, and K. Palczewski. 1996. Structural and enzymatic aspects of rhodopsin phosphorylation. J. Biol. Chem. 271:5215–5224.
Orry, A. J. W., and B. A. Wallace. 2000. Modeling and docking the endothelin G-protein-coupled receptor. Biophys. J. 79:3083–3094.
Palczewski, K., Z. Pulvermuller, J. Buczylko, and K. P. Hofmann. 1991. Phosphorylated rhodopsin and heparin induce similar conformational changes in arrestin. J. Biol. Chem. 266:18649–18654.
Puig, J., A. Arendt, F. L. Tomson, G. Abdulaeva, R. Miller, P. A. Hargrave, and J. H. McDowell. 1995. Synthetic phosphopeptide from rhodopsin sequence induces retinal arrestin binding to photoactivated unphosphorylated rhodopsin. FEBS Lett. 362:185–188.[Medline]
Sakmar, T. P., S. T. Menton, E. P. Marin, and E. W. Awad. 2002. Rhodopsin: insights from recent structural studies. Annu. Rev. Biophys. Biomol. Struct. 31:443–484.
Sayle, R. A., and E. J. Milner-White. 1995. RASMOL: biomolecular graphics for all. Trends Biochem. Sci. 20:374–376.[Medline]
Swanson, E. 1995. PSSHOW, Vers. 1.9. University of Washington, Seattle, WA.
Trosset, J. Y., and H. Scheraga. 1998. Reaching the global minimum in docking simulations: a Monte Carlo energy minimization approach using Bezier splines. Proc. Natl. Acad. Sci. USA. 95:8011–8015.
Vishnivetskiy, S. A., C. L. Paz, C. Schubert, J. A. Hirsch, P. B. Sigler, and V. V. Gurevich. 1999. How does arrestin respond to the phosphorylated state of rhodopsin? J. Biol. Chem. 274:11451–11454.
Vishnivetskiy, S. A., C. Schubert, G. C. Climaco, Y. V. Gurevich, M. G. Velez, and V. V. Gurevich. 2000. An additional phosphate-binding element in arrestin molecule. Implications for the mechanism of arrestin activation. J. Biol. Chem. 275:41049–41057.
Weiss, E. R., S. Osawa, W. Shi, and C. D. Dickerson. 1994. Effects of carboxy-terminal truncation on the stability and G-protein-coupling activity of bovine rhodopsin. Biochemistry. 33:7587–7593.[Medline]
Wessling-Resnick, M., and G. L. Johnson. 1987. Allosteric behavior in transducin activation mediated by rhodopsin. Initial rate analysis of guanine nucleotide exchange. J. Biol. Chem. 262:3697–3705.
Wilden, U., S. Hall, and H. Kuuhn. 1986. Phosphodiesterase activation by photoexcited rhodopsin is quenched when rhodopsin is phosphorylated and binds the intrinsic 48 kDa protein of rod outer segments. Proc. Natl. Acad. Sci. USA. 83:1174–1178.
Zhang, L., C. D. Sports, S. Osawa, and E. R. Weiss. 1997. Rhodopsin phosphorylation sites and their role in arrestin binding. J. Biol. Chem. 272:14762–14768.
This article has been cited by other articles:
 |
|
 |
|
  M. E. Sommer, W. C. Smith, and D. L. Farrens Dynamics of Arrestin-Rhodopsin Interactions: ARRESTIN AND RETINAL RELEASE ARE DIRECTLY LINKED EVENTS J. Biol. Chem., February 25, 2005; 280(8): 6861 - 6871. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
) in association with R-arrestin.
