HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Supplemental Research Data All Versions of this Article: ps.062374707v1 16/5/781    most recent Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Mouawad, L. Articles by Lavalette, D. Search for Related Content PubMed PubMed Citation Articles by Mouawad, L. Articles by Lavalette, D. Social Bookmarking                   What's this?

CO migration pathways in cytochrome P450_cam studied by molecular dynamics simulations

¹ Inserm U759, Institut Curie-Recherche, Bâtiment 112, Université Paris-Sud, 91405 Orsay cedex, France
² Institut Curie-Recherche, Bâtiment 112, Université Paris-Sud, 91405 Orsay cedex, France
³ Institut de Biochimie et Biophysique Moléculaire et Cellulaire, CNRS UMR8619, Université Paris-Sud, 91405 Orsay cedex, France

(RECEIVED May 31, 2006; FINAL REVISION January 24, 2007; ACCEPTED January 26, 2007)

	Abstract

TOP Abstract Introduction Materials and Methods Results Discussion Acknowledgments References

 
Previous laser flash photolysis investigations between 100 and 300 K have shown that the kinetics of CO rebinding with cytochrome P450_cam(camphor) consist of up to four different processes revealing a complex internal dynamics after ligand dissociation. In the present work, molecular dynamics simulations were undertaken on the ternary complex P450_cam(cam)(CO) to explore the CO migration pathways, monitor the internal cavities of the protein, and localize the CO docking sites. One trajectory of 1 nsec with the protein in a water box and 36 trajectories of 1 nsec in the vacuum were calculated. In each trajectory, the protein contained only one CO ligand on which no constraints were applied. The simulations were performed at 200, 300, and 320 K. The results indicate the presence of seven CO docking sites, mainly hydrophobic, located in the same moiety of the protein. Two of them coincide with xenon binding sites identified by crystallography. The protein matrix exhibits eight persistent internal cavities, four of which corresponding to the ligand docking sites. In addition, it was observed that water molecules entering the protein were mainly attracted into the polar pockets, far away from the CO docking sites. Finally, the identified CO migration pathways provide a consistent interpretation of the experimental rebinding kinetics.

Keywords: internal cavities; ligand migration; molecular dynamics; docking sites; ligand rebinding kinetics

	Introduction

TOP Abstract Introduction Materials and Methods Results Discussion Acknowledgments References

 
Ligand binding with hemoproteins has been investigated for more than 30 years (Austin et al. 1975). The picture emerging today is that the reaction dynamics are dominated by protein conformational relaxation and ligand migration among protein cavities (Tetreau and Lavalette 2005). Experimental and theoretical investigations have emphasized three functional roles of internal cavities (Brunori and Gibson 2001): (1) They define preferred migration pathways for ligands; (2) they open transient channels allowing ligand diffusion from the solvent toward otherwise inaccessible buried active sites; and (3) they cause delayed (re)binding by providing transient docking sites for the ligand. Ligand trapping inside internal cavities has been documented by time-resolved crystallography (Srajer et al. 2001; Bourgeois et al. 2003; Schotte et al. 2003; Hummer et al. 2004), cryocrystallography (Brunori et al. 2000; Chu et al. 2000; Ostermann et al. 2000; Schlichting 2000), spectroscopy (Engler et al. 2000; Lamb et al. 2002; Nienhaus et al. 2002; Kriegl et al. 2003), kinetic competition experiments with xenon (Scott and Gibson 1997; Scott et al. 2001; Nienhaus et al. 2003a,b; Tetreau et al. 2004, 2005), and molecular dynamics (MD) simulations (Elber and Karplus 1990; Carlson et al. 1996; Meller and Elber 1998; Amara et al. 2001; Mouawad et al. 2005).

These studies were essentially concerned with CO migration in oxygen storage hemoproteins, mainly myoglobin, but also other globins such as cytoglobin (de Sanctis et al. 2004a, b) and neuroglobin (Ascenzi et al. 2004). During the last few years we have investigated the reaction dynamics of a series of catalytic hemoproteins: The cytochromes P450 mono-oxygenases. The cytochrome P450 family derives its name from its ability to bind also CO to form a stable but enzymatically inactive ternary complex displaying a strong absorbance at 450 nm. Cytochrome P450_cam has become a paradigm for this, functionally more complex, hemoprotein family in which the ligand coexists with the substrate within a well-buried heme pocket. In its reduced Fe^II form, this protein catalyzes the stereospecific hydroxylation of camphor (the substrate) at the 5-exo position in Pseudomonas putida; camphor is maintained in the distal pocket not far from oxygen (the natural ligand), which is bound to the iron atom of the heme (Poulos et al. 1985, 1987).

Laser flash photolysis investigations of cytochrome P450_cam(cam)(CO) revealed a much greater kinetic complexity than for oxygen storage proteins (Tetreau et al. 2005). Four different geminate processes, i.e., recombination of CO from inside the protein, each of them depending on temperature, were taking place between 77 and 300 K. Near 200 K all processes occurred simultaneously as shown by the kinetic rate spectrum of Figure 1. The two faster processes (G^I and G^I_R bands) were attributed to CO rebinding from its primary docking site above the heme, respectively before and after the displacement of camphor closer to the iron atom; this motion of the substrate corresponded to a conformational relaxation of the heme pocket triggered by the removal of CO from the heme. Kinetic competition experiments with xenon indicated that the two slower recombination processes (G_Mf and G_Ms bands) were due to delayed rebinding of CO subsequent to its migration in secondary docking sites located in internal cavities of the protein matrix. A crystallographic study of Fe^III P450_cam under Xe pressure had revealed the presence of four Xe sites (Wade et al. 2004) which could constitute potential candidates for CO docking after photodissociation. However, in the absence of crystallographic data on trapped CO intermediates or of time-resolved crystallography experiments, the actual ligand docking sites and migration pathways remain undetermined.

View larger version (13K): [in this window] [in a new window]   Figure 1. Normalized rate spectrum describing the kinetics of CO rebinding with cytochrome P450cam(cam) at 200 K, obtained by the "Maximum Entropy Method" (Tetreau et al. 2005). P(log k) represents the probability of CO rebinding with a rate constant equal to k (s–1). Four different geminate rebinding processes are observed, the fastest being GI.

	Materials and Methods

TOP Abstract Introduction Materials and Methods Results Discussion Acknowledgments References

 
All calculations started using the coordinates of the Fe^II P450_cam(cam)(CO) complex taken from the Protein Data Bank (PDB entry 3CPP [PDB] ; Raag and Poulos 1989). The polypeptide chain, consisting of 405 residues (residue 10–414), did not contain coordinates for the first nine amino acids. An acetyl group was added to residue 10 of the N-terminal part to avoid the artificial introduction of a positive charge in this position. The tautomeric forms of the histidines were assigned in a way that favored the formation of hydrogen bonds. To simulate photodissociation of the CO complex, the Fe–CO bond of the starting structure was cut before any calculations. We included all the internal crystallographic water molecules that were located within a distance of 16.5 Å from the center of mass of the protein along its first and second inertia axes and of 9 Å along the third one. This resulted in a total of 21 water molecules, to which the TIP3P model was applied (Jorgensen et al. 1983).

The force field parameters
All simulations, including those in the water box, were performed using the parameter set QUANTA/CHARMm20 (Accelrys), in which only polar hydrogens were considered explicitly for the polypeptide chain and camphor, whereas all hydrogen atoms were introduced for the heme group (taken from CHARMM22; MacKerell et al. 1998). For camphor the parameters were obtained from the XPLO-2D program (Kleywegt and Jones 1997) and the partial charges of camphor were taken from Helms and Wade (1997). The simulations were performed using the CHARMM program (Brooks et al. 1983).

The procedures
Simulations with the protein in the vacuum
The same procedure was followed for the various trajectories in the vacuum. The structure was first subjected to a slight energy minimization, during which the conjugate gradient method was used for 600 iterations under mass-weighted harmonic constraints that were applied to the system in order to prevent abrupt deviations from the crystal structure. The harmonic force constant was decreased every 100 steps taking the successive values of 250, 100, 50, 25, 10, and 5 kcal/mol/Ų. Finally, the energy was minimized for 10 additional steps without harmonic constraints.

The minimized structure was then heated up to 200, 300, or 320 K, depending on the trajectories, by 25 K increments of temperature every 5000 steps. The SHAKE procedure (Ryckaert et al. 1977) was applied on bonds involving a hydrogen atom and the time step was of 1 fsec.

The system was further equilibrated for 100 psec at constant temperature following the Berendsen procedure (Berendsen et al. 1984) with a time coupling to the heat bath of 0.05 psec. This was followed by 900 psec of productive dynamics during which the system was coupled to the heat bath every 5 psec.

To truncate the electrostatic and van der Waals interactions the switch function (Brooks et al. 1983) was used within a cut-on/cut-off interval. Two different intervals were tested: One trajectory was calculated with cut-on and cut-off values of 11 and 15 Å, respectively, and a set of trajectories with values of 5 and 9 Å, respectively. The dielectric constant was taken equal to the distance separating the charges.

Simulations with the protein in a water box
One trajectory was calculated with the protein embedded in a water box at constant pressure (1 atm) and temperature (300 K), using the Berendsen method with a compressibility of 4.63 x 10^–5 atm^–1. The particle mesh Ewald method (Essmann et al. 1995) was applied, with the width of the Gaussian distribution = 0.34 and a cubic B-spline (the order of its interpolation was equal to 4). The dielectric constant was equal to 1. The switch function was applied between 8 and 12 Å for van der Waals energy. The periodic boundary conditions were used.

The water box.
A water box of 88 x 87 x 66 ų was constructed so that its dimensions exceeded those of the protein by 10 Šin the x, y, and z directions. Its energy was minimized by the steepest descent method (SD) for 500 steps; it was heated in 6 psec to 300 K, then equilibrated for 45 psec with a time coupling of 0.5 psec.

The protein in the water box.
The energy-minimized cytochrome P450_cam with all its crystallographic water molecules was immersed in the center of the equilibrated water box. All water molecules that overlapped either the protein or the crystallographic waters (i.e., with the distance between heavy atoms <2.8 Å) were removed. Nineteen sodium ions were added randomly in the water box far from the protein to neutralize its net charge. The final system consisted of 48,710 atoms, 3973 of which corresponded to the protein itself (including camphor and CO).

The energy of the system was minimized keeping fixed the protein for 500 steps of SD; then, the energy of the whole system including the protein was minimized for an additional 500 SD steps, followed by 10 psec of heating from 95 to 300 K, and equilibration during 140 psec. Finally, a production run of 900 psec was performed.

The total number of simulations
A total of 37 trajectories were calculated: 36 were in the vacuum including only internal crystallographic waters and one in the explicit solvent. The trajectories in the vacuum differ by the temperature and cut-on/cut-off distances.

The trajectories will be referred to as:

Set I: 300 K, cut-on/cut-off 11/15 Å (1 trajectory)

Set II: 300 K, cut-on/cut-off 5/9 Å (15 trajectories)

Set III: 200 K, cut-on/cut-off 5/9 Å (10 trajectories)

Set IV: 320 K, cut-on/cut-off 5/9 Å (10 trajectories)

Set V: 300 K, in a water box (1 trajectory).

In each set, the trajectories differ by the initial random assignments of velocities.

Calculations of protein cavities
The internal cavities of cytochrome P450_cam were calculated by the program Alpha Shapes (Liang et al. 1998a,b), with a probe size of 1.4 Å. A home-built software was used to visualize them; it consisted of filling the empty space of cavities with hydrogen atoms and drawing their contour surface.

	Results

TOP Abstract Introduction Materials and Methods Results Discussion Acknowledgments References

 
We performed preliminary calculations in order to define which procedure was the most appropriate. For this purpose we calculated trajectories of 1 nsec at 300 K, in the vacuum and in a water box, using two different cutoffs.

Choice of the parameters
All calculations yielded satisfactory results both for fluctuations (Fig. 2) and rms deviations (Fig. 3). This was the case for calculations in the water box (set V) and in the vacuum (sets I and II), even when smaller cut-on/cut-off distances were used (set II).

View larger version (24K): [in this window] [in a new window]   Figure 2. Fluctuations of the C atoms along representative trajectories at 300 K (gray lines) compared to the B-factors taken from the crystal structure (pdb file 3cpp, black lines). Calculations in the vacuum, (A) with cut-off interval 11/15 Å (set I), (B) with cut-off interval 5/9 Å (set II). (C) Simulations in a water box with Ewald summation for electrostatics (set V). In calculations in the vacuum the internal crystallographic water molecules were included.

View larger version (26K): [in this window] [in a new window]   Figure 3. C rms deviations from the crystal structure along the same trajectories as in Figure 2: set I (gray line), set II (black line), and set V (dashed line). (Inset) Main features concerning the conditions of the simulations and the cut-on/cut-off values of the switch function.

Thirty-five trajectories were calculated at three different temperatures; 200 K (10 trajectories, set III) were used to compare with experiments; 300 K (15 trajectories, set II) and 320 K (10 trajectories, set IV) were chosen to accelerate ligand migration and escape from the protein, because the length of the trajectories (1 nsec) was too short to provide a good description of the experimental events that occur on a time scale between 50 nsec and several seconds.

Simulations with the protein in the vaccum
CO migration pathways
The position of CO along the trajectories was monitored using the distance of its center of mass to the heme iron atom. In all trajectories CO moved from the iron during the heating phase and then either came back over the metal or migrated away toward other regions of the protein.

In the following, we will present only the analysis of the trajectories at 300 K (set II), unless otherwise stated. Figure 4 shows that, in five trajectories (# 3, 5, 6, 8, and 15), CO remained confined above the heme during almost the whole time course and that only in one simulation (#2) did it escape from the protein. In all other cases, CO jumped very rapidly (1 psec) between different regions, where it remained confined for several picoseconds.

View larger version (39K): [in this window] [in a new window]   Figure 4. Distance of the CO center-of-mass to the iron atom vs. time along the 15 trajectories of set II (at 300 K). The docking sites (S1 to S6), in which CO remains confined at a given time, are indicated.

View larger version (31K): [in this window] [in a new window]   Figure 5. Snapshots of CO at time intervals of 100 psec along all trajectories at 200 K (set III), 300 K (set II), and 320 K (set IV). The C atoms of the polypeptide chain from each frame were superimposed on those of the crystal structure and all views are in the same orientation. The CO molecules are colored differently according to the region into which they are localized (blue: S1, cyan: S2, green: S3, yellow: S4, orange: S5, red: S6, and black: S7); the heme and camphor are in gray and pink, respectively. In the bottom panel, the entire protein is shown in the same orientation at 320 K with the CO molecules presented as hard spheres, and the secondary structure elements denoted in blue. The views were obtained using the Insight II program (Accelrys).

View larger version (17K): [in this window] [in a new window]   Figure 6. Representation of the residues closely approached by CO (at a distance <4 Å) during the time course of the 15 simulations of set II at 300 K. Residue numbers 415 and 417 correspond to heme and camphor, respectively. The height represents the maximum percentage of time that CO spent in the vicinity of each residue of the protein in any trajectory; only percentages exceeding 2% are shown for clarity.

Similar sites were observed at 200 and 320 K. However, at 200 K only S1, S2, and S3 were populated; at 320 K, S4 was split in two parts, S5 was more extended, and a nearby site S7 could be distinguished (Fig. 5) between helices E, G, H, and I, at an average distance between the center of mass of CO and Fe of 12.7 Å. Sites S1–S7 are in an essentially non-polar environment, suggesting that they might correspond to hydrophobic cavities; this point will be discussed below.

The percentage of the CO occupancy of the different sites, calculated from the sum of occurrences in all trajectories at each temperature, is given in Table 1. It appears that S2 is depopulated in favor of the other sites upon increasing temperature. Site S7 is populated only at 320 K.

The sites being well defined, the migration pathways of CO versus time can be described from the simulations. The passage from S1 to S3 takes place necessarily via S2, as can be seen in trajectories (Traj.) 1, 7, 9, 10, 12, 13, and 14 of Figure 4; from S3, CO can migrate toward either S4 (Traj. 9) or S5 (Traj. 1, 12, and 13); CO can reach S6 starting from either S1 (Traj. 6), S2 (Traj. 11) or S5 (Traj. 12). Finally, S7 is only populated from S6 (observation made at 320 K, data not shown). The same migration pathways are observed at the three temperatures; they are summarized in Figure 7.

View larger version (15K): [in this window] [in a new window]   Figure 7. Schematic representation of the CO docking sites and routes of migration in the same orientation as Figure 5. The slanted letters A, B, C, and D designate the pyrrole rings of the heme. The ellipses represent sites S1 to S7; their sizes are roughly related to the relative CO occupation of the sites. Straight and curved arrows indicate the migration and escape pathways, respectively. The thickness of the arrows reflects the number of passages from site to site observed from all trajectories of sets II, III, and IV. It can be seen that CO migrates the most frequently from S1 to S2 then S3 and finally S5, and that it escapes from S5 along p1.

CO escape trajectories from the protein.
CO never left the protein in any of the 1-nsec trajectories at 200 K. At 300 K it escaped only in one of the 15 trajectories, starting from S1, first along the normal to the heme plane, then skirting around camphor and leaving between Ala95 of helix B' and Glu198 of helix G; this path will be denoted p2. At 320 K, CO escaped from the protein in four of the 10 trajectories, by following the same direction in three of them: In Traj. 1, 7, and 8, CO escaped along path p1, i.e., from S5 on the proximal side of the heme between helices C and L (more precisely, in Traj. 1, between Val118 on one side and Leu358 and Leu362 on the other side, in Traj. 7, between Met121 and His361, and in Traj. 8, between Asp116, Gln117, and His361); in Traj. 5, CO escaped along path p3, i.e., from S3 on the edge of the heme between residues Asp125 of helix C and Ile222 of helix H.

Contrary to CO, the crystallographic water molecules (not shown in Fig. 5 for the sake of clarity) remained always well localized at the three temperatures, indicating that they have a long (>1 nsec) residence time and are therefore likely to play a structural role.

Internal cavities
To address the question whether the docking sites are coinciding with pre-existing internal cavities, the latter were monitored at 50-psec time intervals along selected trajectories (1, 3, and 10 at 300 K and 1 and 2 at 200 K). This choice was based on the criterion of the site occupation by CO with the aim to investigate whether the ligand was influencing the presence of nearby cavities. At 300 K, in Traj. 3 and 10, CO remained confined mostly in S1 and S3, respectively, while in Traj. 1 it was jumping between S3 and S5. At 200 K, in Traj. 1 and 2, CO remained in S3 and S2, respectively. Moreover, cavities were also calculated along two trajectories at 320 K for comparison. Because the presence of CO prevents a sound evaluation of the size of the cavity into which it resides, the ligand was removed from the frames before performing the cavities calculations, contrary to water molecules which were maintained. In each selected frame of each trajectory, a large number of cavities (>20) were found; they were spread throughout the protein matrix, and their shape, size, and connections changed from frame to frame (Fig. 8). The volume of the largest cavity was >400 ų. At 300 K, eight cavities (A–H) with a noticeable size (>50 ų) were found recurrently in many frames from selected trajectories. Because the cavities frequently changed in size and shape, a precise definition cannot be given, but their location can be characterized by the residues that most frequently bordered their walls in the structures sampled along the trajectories. Table 2 shows that all cavities have a large nonpolar surface area. Three of them (D, F, and H; see Fig. 8) are located far from the CO pathways, whereas cavity G, located near camphor, is along one of the CO escape paths (p2) described above. The four remaining cavities, A, B, C, and E, correspond to the CO docking sites: Cavity A to S1 and S2, B to S3 and S5, C to S7, and E to S6. The coincidence between docking sites and internal cavities is clearly apparent in Figure 9A.

View larger version (43K): [in this window] [in a new window]   Figure 8. Fluctuations of cavities along trajectory 3 at 300 K. All cavities taken from only eight frames at 100-psec time intervals are presented. Their names are indicated in the last panel. The views have the same orientation as in Figure 5.

View larger version (52K): [in this window] [in a new window]   Figure 9. (A) Comparison between CO docking sites and internal cavities. The CO molecules taken every 100 psec from all 15 trajectories at 300 K (as in Fig. 5, panel 2, same color code) and all cavities (in gray) located in the same regions, calculated every 50 psec from Traj. 1 (top) or Traj. 3 (bottom) at the same temperature, are drawn after superposition of the polypeptide chain on that of the minimized structure (only the heme and camphor of this structure are shown). One can see that docking sites coincide with pre-existing internal cavities, except S4. However, cavity A (corresponding to site S1 in blue) is not observed in Traj. 1, contrary to Traj. 3, because in Traj. 1 CO does not reside in site S1, which is therefore occupied by camphor (see Fig. 10). It can be observed in the top panel that cavities extending from the bottom of site S5 correspond to the CO escape path p1. CO molecules are shown as hard spheres not to scale for clarity. (B) Water penetration in cytochrome P450cam in the simulation in the solvent (set V). A snapshot of the protein corresponding to the last frame of the trajectory is viewed from the distal side of the heme, i.e., rotated 90° from A and C. Only water molecules located inside the protein, within a distance of 16.5 Å from the center of mass of the protein along its first and second inertia axes (represented by the white circle) and of 9 Å along the third one (axis pointing toward the reader), are shown. They are colored in blue: dark blue for water that have entered from the bulk, and cyan for crystallographic waters. Helices are in yellow, the rest of the backbone and the heme in light gray, and camphor in dark gray. To visualize the position of the docking sites (S1–S7), CO from all trajectories calculated in the vacuum at 300 K (set II) and 320 K (set IV) have been superposed on the structure and drawn as red hard spheres. It can be seen that water is localized in the less-helical moiety of the protein opposite to the CO docking sites, although in the present trajectory (set V), these sites were free since CO remained confined over the heme (in S1 and S2). (C) Comparison between CO and xenon docking sites. The protein has the same orientation as in A. Xe atoms are shown as red hard spheres (not to scale for more clarity) and CO (taken from all trajectories at 300 and 320 K) are presented as sticks, with the same color code as in Figure 5, except for site S7, which is in white. It is clearly apparent in the figure that Xe1 is near S7 and Xe4 is hidden among the CO molecules that spread from site S5 at 320 K. Xe2 and Xe3 are in cavity D; the residues lining this cavity are drawn as gray sticks.

View larger version (52K): [in this window] [in a new window]   Figure 10. Distance of camphor (gray line) and CO (black line) to the heme normal passing through Fe along two representative trajectories at 300 K (set II): Traj. 8 (top) and Traj. 12 (bottom). It shows that camphor comes close to the heme plane's normal when CO moves away. S1 to S6 indicate the docking sites in which CO is located.

Remarkably, this water flux did not modify the characteristics of the internal cavities which were found again in almost the same locations and with the same sizes as in the vacuum. Once again cavities D, F, and H were located far from the CO pathways, G, along the escape path p2 and A, B, C, and E in the CO docking sites. In fact, water molecules were entering and leaving the protein by following completely different pathways from those described for CO in the vacuum (Fig. 9B). They were mainly attracted into the polar pockets already occupied by crystallographic waters, avoiding the moiety of the protein corresponding to the docking sites S1 to S7 (except S6, which is close to structural water). In this trajectory, it happened that the CO ligand moved only between sites S1 and S2 as it was the case for several trajectories calculated in the vacuum. However, the presence of the internal cavities A, B, C, and E, that include sites S1, S2, S3, S5, S6, and S7, indicates that the CO pathways should not be significantly modified by the presence of water around and inside the protein.

	Discussion

TOP Abstract Introduction Materials and Methods Results Discussion Acknowledgments References

 
Simulations of the protein were performed under various conditions, using different cut-off distances (11/15 Å or 5/9 Å) or procedures (in the vacuum with only internal crystallographic waters or in an explicit water box). In all these simulations, it was observed that the cavities were located in the same regions of P450_cam and that the routes of CO migration were similar. The robustness of these results suggests that both were mainly determined by the tertiary structure of the protein.

Internal cavities
Internal protein cavities have been studied both theoretically (Wade et al. 1991; Bakowies and van Gunsteren 2002; Radresa et al. 2002; Teeter 2004; Vaitheeswaran et al. 2004; Rashin and Rashin 2005) and experimentally (Tilton et al. 1984; Prangé et al. 1998; Rubin et al. 2002). An increasing number of crystallographic studies revealed that the cavities may be occupied by xenon atoms when they are hydrophobic or by more or less mobile water molecules in polar environments. A number of theoretical methods have been developed to calculate their distribution and localization inside proteins, as well as their hydration. Both crystallographic studies and MD simulations performed on wild-type or mutant myoglobins (Elber and Karplus 1990; Carlson et al. 1996; Hummer et al. 2004; Radding and Phillips 2004; Bossa et al. 2005), hemoglobin (Mouawad et al. 2005), and truncated Hb (Milani et al. 2004) suggested that internal cavities play a functional role and partly determine the protein reactivity by defining preferred diffusion pathways and providing transient docking sites for ligands.

The first argument in favor of the presence of a large number of atom-sized voids in cytochrome P450_cam came from Williams et al. (1994). More recently, crystal structure of P450_cam in the presence of xenon revealed the presence of four Xe binding sites (Wade et al. 2004). However, this information was not sufficient to identify CO migration pathways and docking sites unambiguously and to correlate them with the observed geminate rebinding processes.

In our simulations, a large number of small cavities were found to appear and disappear dynamically along the trajectories (Fig. 8). Only eight with a noticeable size were persistent; they were in an essentially nonpolar environment (Table 2). We observed that the opening/closure of the passage between adjacent cavities was not strictly correlated with the presence of nearby CO, contrary to what was described previously for myoglobin (Carlson et al. 1996; Bossa et al. 2004). Except for the distal cavity A, which was located above the heme, the seven other cavities were observed independently of the CO position; their formation was rather determined by the structure of the protein. For instance, the largest and most persistent cavities (including those corresponding to docking sites) were observed in trajectory 3 (at 300 K) in which CO remained confined in site S1 above the heme. However, the presence of CO inside a cavity maintained it open for a longer time although it did not influence the size of adjacent cavities.

Similar cavities were also observed in the simulation of the protein in a water box but amazingly no water molecules entered these cavities (except F) although the protein was highly hydrated during the trajectory. Interestingly, this does not seem to be due to the hydrophobic character of the cavities, since cavity F, which is as hydrophobic as the others, was hydrated. A possible explanation may be based on the observation of Williams et al. (1994), who reported that in -helical proteins the cavities were less hydrated than in other proteins and attributed this observation to the smaller potential number of hydrogen bonds that water could establish with the protein backbone already involved in intra-helical interactions. In our case, except A and F, all other cavities were mainly located between -helices. In cavity A, water entry was prevented by the permanent presence of CO in this trajectory, while cavity F, surrounded mainly by turn T2 and one strand of 5, was filled with water molecules.

CO docking sites and migration pathways
In the present simulations, the diffusion of the CO ligand was free of any bias such as those resulting from artificial restraints or forces, or multicopy sampling. Nevertheless, the ligand did not follow arbitrary paths inside the protein but moved toward preferred regions into which it remained transiently confined. Seven docking sites (S1 to S7) were identified. The residues lining them were mostly hydrophobic (Table 1).

The CO migration consisted in jumps from site to site through well-defined paths. The transitions between adjacent sites were almost instantaneous while CO remained confined for comparatively long times in each site. The transitions were sometimes accompanied by side-chains motions (e.g., Phe111 and Met241 during the passage of CO from S3 to S4), but always by a slight tilt of the adjacent helices. These observations are in good agreement with those reported in other studies of CO migration in Mb (Schotte et al. 2003; Hummer et al. 2004).

In P450_cam, not all sites were directly connected with each other; in particular, S4 could be reached only from S3, and S7 only from S6; in contrast, S6 could be accessed from S1, S2, or S5 (Fig. 7).

Ligand escape from the protein
As a rule, entry and escape of ligands are not directly observable in experiments because they do not give rise to any spectroscopic signal. However, escape can be quantified provided it competes with some rebinding process. Experimental evidence for the presence of two escape pathways was reported only recently in Mb, but this case was rather exceptional (Tetreau et al. 2004). MD simulations provide the adequate and complementary alternative to the description of these processes.

In P450_cam, CO followed three different pathways to escape from the protein, p1, p2, and p3 (Fig. 7). In p1, the CO located in S5 left the protein between helices C and L, in p2 it escaped from S1 through helices B' and G, and in p3, it left from S3 between helices C and H. Pathways p1 and p2 were comparable to those observed by Lüdemann et al. (2000a,b) for the camphor substrate, using a Random Expulsion Molecular Dynamics method, in which an artificial randomly oriented force was imposed to the substrate. These authors have identified three paths (pw1, pw2, and pw3). The main path for camphor with the lowest energy barrier, pw2, corresponded to the present p2, from which CO escaped rapidly (Traj. 2 in Fig. 4) without exploring docking sites, likely because the channel could widen easily enough to permit the passage of camphor; pw1, the camphor path with the highest energy barriers, corresponded to p1, the passage from which CO escaped most frequently and along which lay the most populated docking sites, S3 and S5. Finally, pw3, located either between helices F and G or near EF loop, was completely different from p3 but was close to the docking site S7. In the present simulations, camphor remained close to its initial position because the trajectories were not long enough to allow spontaneous important displacements of such a large-sized molecule.

Both recent experimental and theoretical studies that have monitored the migration routes of camphor and other large-size substrates in P450_cam have shown that the substrate escapes from the protein without exploring docking sites (Lüdemann et al. 2000a, b; Dunn et al. 2001; Pylypenko and Schlichting 2004). We can see here that CO adopts a similar behavior when it leaves the protein along the main substrate path, p2, whereas when following other routes, it spends time in docking sites.

Comparison between CO docking sites, cavities, and xenon binding sites
In previous MD simulations of Mb, it was reported that CO hopped rapidly between pre-existing cavities (Elber and Karplus 1990; Bossa et al. 2005) or that the opening/closure of the passage between adjacent cavities was strictly correlated with the presence of CO nearby (Carlson et al. 1996; Bossa et al. 2004). In cytochrome P450_cam, both situations were encountered. Most sites (S2, S3, S5, S6, and S7) corresponded to pre-existing cavities among which CO migrated, whereas in site S1 competition between CO and camphor prevented the formation of a cavity (this point will be discussed below) and site S4 was only formed in the presence of CO nearby.

On the other hand, the four xenon sites that were identified by crystallography (Wade et al. 2004) coincided with some hydrophobic cavities calculated here. Indeed, Xe1 corresponded to both cavity C and site S7, and Xe4 to cavity B and site S5; Xe2 and Xe3 were at the same location as cavity D, far from any CO docking site (Fig. 9C). Although cavity D constitutes a Xe site, CO did not pass through it along any trajectory, presumably because this cavity is close to the surface (accessible to Xe from the solvent) but far from the active site (the initial position of CO in our simulations).

Structural interpretation of the CO geminate rebinding dynamics
The present results provide a basis for a structural interpretation of our previous laser flash photolysis kinetic study of CO rebinding with P450_cam.

Assignment of the geminate rebinding processes G^I and G^I_R: Heme pocket relaxation
Both the comparison of the crystal structures of the Fe^II and Fe^II–CO species and energy minimization calculations previously carried out revealed that, upon removal of CO, camphor relaxes toward a position closer by 0.7 Å to the heme normal passing through Fe. This was attributed to the increase of the steric repulsion between camphor and one heme propionate resulting from the doming of the heme after CO dissociation (Tetreau et al. 2005). These observations led us to propose that the two fastest geminate processes (G^I and G^I_R bands of Fig. 1) were associated with CO rebinding from the primary docking site before and after the substrate relaxation motion, respectively.

This view is further supported by the present MD results. Examination of the CO and camphor distances to the heme normal in all trajectories (Fig. 10) showed that, when CO was at a short distance to the normal (2 Å), i.e., in S1, camphor was pushed away at an average distance of 4 Å, as in Traj. 8. In contrast, when CO jumped between different sites farther from the heme normal, as in Traj. 12, camphor relaxed to its position in the deliganded structure, at a mean distance of 2.6 Å. The latter is in good agreement with the distance observed in the crystal deoxy-structure (2.7 Å) (Poulos et al. 1987). Thus, the G^I and G^I_R bands can be assigned to CO rebinding from primary docking sites S1 and S2, respectively. Although both sites lie above the heme, rebinding must be slower from S2 than from S1 because of the increased steric hindrance of camphor. A further support to this assignment is provided by the temperature dependence of these band amplitudes and of the populations of the sites. Indeed, despite the difference in the temperature range (from 140 to 200 K for experiments and 200 to 320 K for simulations), similar variations were observed with increasing temperature: The intensity of bands G^I and G^I_R decreased significantly, although that of G^I_R more importantly, and, in parallel, the population of the docking sites S1 and S2 decreased but more appreciably for S2 (Table 1).

Assignment of the geminate rebinding processes G_Mf and G_Ms: Ligand migration
The two other geminate rebinding bands of the rate spectra (G_Mf and G_Ms) correspond to much slower processes than G^I and G^I_R, suggesting that rebinding is delayed by CO migration to secondary docking sites. This conclusion was supported by xenon competition experiments. At a xenon pressure of 7 atm, the band corresponding to the slowest rebinding process (G_Ms) almost entirely disappeared, indicating that the corresponding site was saturated with Xe. In contrast, the faster G_Mf process decreased only slightly at higher pressures (Tetreau et al. 2005). These results suggested that G_Ms and G_Mf might correspond to rebinding from two different docking sites, the one associated with G_Ms having a significantly higher affinity for Xe than the other. Further, these two sites must be occupied either in parallel or sequentially, in which case the site responsible for the slowest rebinding process, G_Ms, should be at the end of the sequence.

The analysis of the CO migration pathways from the trajectories (Fig. 7) showed that S3 and S6 are the only sites directly accessible from the primary docking sites S1 and S2. This makes them good potential candidates for the faster rebinding process (G_Mf) especially as their environments are not as hydrophobic as those of other sites. Indeed, the presence of Arg112 in S3 and of Glu366 and two water molecules in S6 could result in relatively low affinity for Xe.

On the other hand, sites S4, S5, and S7, which can only be reached through S3 and S6, should correspond to the slowest rebinding process, especially as they are lined with highly hydrophobic residues, giving them the character of high affinity for Xe. However, in the crystal structure resolved under a xenon pressure of 7 atm, only sites S5 and S7 were occupied by Xe atoms. This may be due to the fact that site S4 does not correspond to a pre-existing cavity and is opened only by the presence of CO. Anyway, the X-ray structure indicates that S4 could not be associated with the G_Ms band which disappears at 7 atm of Xe.

This leaves two possible passageways starting from the primary docking sites, either S3S5 or S6S7. A large number of trajectories are needed to obtain converged site occupancies to rigorously discriminate between these two paths. Nonetheless, in the simulations at multiple temperatures S6 and S7 are consistently found to be less populated than S3 and S5. Indeed in Table 1 it can be seen that, while S6 is only populated at T 300 K, S3 is already occupied at T = 200 K and that, while S7 is filled only at T = 320 K, S5 is populated at T 300 K. The fact that S5 is not occupied at 200 K may be due to the short length of the trajectories (1 nsec) compared to the experimental time scale (>50 nsec). From these considerations one can conclude that it is highly probable that the CO migration along the passageway S3S5 is responsible for the experimental observations of the rebinding processes, G_Mf and G_Ms. Crystallography experiments that would demonstrate the presence of Xe in site S3 at higher pressures or cryocrystallography or time-resolved crystallography visualizing the CO migration pathways after dissociation would be very useful to definitely support these assignments.

At present the most extensive data on CO migration are available for myoglobin and cytochrome P450_cam. Except for the heme, these two proteins have structurally little in common. Their identified cavities are not related in any way and they are disposed differently around the heme. However, remarkably, both proteins exhibit a common feature: The presence of a series of cavities connecting the distal pocket to the proximal side of the heme. These connections may be important for a ligand that enters the protein from the proximal side, because it gives it the ability to reach its binding site from this region too. Such a possibility might enhance the ligand binding rate.

Conclusions
In the present simulations, the CO migration pathways in cytochrome P450_cam were found to correspond to fast transitions between well-defined docking sites which coincided with pre-existing internal cavities. This emphasizes the functional role of the cavities, although not all of them were visited by the ligand during the 1-nsec trajectories. One of these unvisited cavities (D), however, corresponds to Xe docking sites, suggesting the possible existence of some ligand entries that do not lead directly to the active site. Moreover, the simulations in a water box showed that water molecules did not follow the same paths as CO despite the similar size of the two molecules and the presence of open internal cavities along the CO pathways. This shows that the polarity of the ligand plays a predominant role in its migration within the protein. Finally, the simulations provide a consistent interpretation of the complex kinetics previously reported for CO rebinding in P450_cam, not only by confirming the occurrence of heme pocket relaxation and of ligand migration, but also by attributing to each of the four bands observed experimentally a specific location of CO in the protein (S1 to G^I, S2 to G^I_R, S3 to G_Mf, and S5 to G_Ms).

	Footnotes

 
Supplemental material: see www.proteinscience.org

Reprint requests to: Liliane Mouawad, Inserm U759, Institut Curie-Recherche, Bâtiment 112, Université Paris-Sud, 91405 Orsay cedex, France; e-mail: liliane.mouawad{at}curie.u-psud.fr; fax: 33-1-69 07 53 27.

Article published online ahead of print. Article and publication date are at http://www.proteinscience.org/cgi/doi/10.1110/ps.062374707.

	Acknowledgments

TOP Abstract Introduction Materials and Methods Results Discussion Acknowledgments References

 
We thank Eric Quiniou for his valuable assistance. This work was supported by the Institut National de la Santé et de la Recherche Médicale (INSERM) and the Institut Curie.

	References

TOP Abstract Introduction Materials and Methods Results Discussion Acknowledgments References

 
Amara, P., Andreoletti, P., Jouve, H.M., and Field, M.J. 2001. Ligand diffusion in the catalase from Proteus mirabilis: A molecular dynamics study. Protein Sci. 10: 1927–1935.[Abstract/Free Full Text]

Ascenzi, P., Bocedi, A., de Sanctis, D., Pesce, A., Bolognesi, M., Marden, M.C., Dewilde, S., Moens, L., Hankeln, T., and Burmester, T. 2004. Neuroglobin and cytoglobin. Biochem. Mol. Biol. 32: 305–313.[CrossRef]

Austin, R.H., Beeson, K.W., Eisenstein, L., Frauenfelder, H., and Gunsalus, I.C. 1975. Dynamics of ligand binding to myoglobin. Biochemistry 14: 5355–5373.[CrossRef][Medline]

Bakowies, D. and van Gunsteren, W.F. 2002. Water in protein cavities: A procedure to identify internal water and exchange pathways and application to fatty-acid proteins. Proteins 47: 534–545.[CrossRef][Medline]

Berendsen, H.J.C., Postma, J.P.M., van Gunsteren, W.F., DiNola, A., and Haak, J.R. 1984. Molecular dynamics with coupling to an external bath. J. Chem. Phys. 81: 3684–3690.[CrossRef]

Bossa, C., Anselmi, M., Roccatano, D., Amadei, A., Vallone, B., Brunori, M., and Di Nola, A. 2004. Extended molecular dynamics simulation of the carbon monoxide migration in sperm whale myoglobin. Biophys. J. 86: 3855–3862.[CrossRef][Medline]

Bossa, C., Amadei, A., Daidone, I., Anselmi, M., Vallone, B., Brunori, M., and Di Nola, A. 2005. Molecular dynamics simulation of sperm whale myoglobin: Effects of mutations and trapped CO on the structure and dynamics of cavities. Biophys. J. 89: 465–474.[CrossRef][Medline]

Bourgeois, D., Vallone, B., Schotte, F., Arcovito, A., Miele, A.E., Sciara, G., Wulff, M., Anfinrud, P., and Brunori, M. 2003. Complex landscape of protein structural dynamics unveiled by nanosecond Laue crystallography. Proc. Natl. Acad. Sci. 100: 8704–8709.[Abstract/Free Full Text]

Brooks, B., Bruccoleri, R., Olafson, B., States, D., Swaminathan, S., and Karplus, M. 1983. CHARMM: A program for macromolecular energy, minimization and molecular dynamics calculations. J. Comput. Chem. 4: 187–217.[CrossRef]

Brunori, M., Vallone, B., Cutruzolla, F., Travaglini-Allocatelli, C., Berendzen, J., Chu, K., Sweet, R.M., and Schlichting, I. 2000. The role of cavities in protein dynamics: Crystal structure of a photolytic intermediate of a mutant myoglobin. Proc. Natl. Acad. Sci. 97: 2058–2063.[Abstract/Free Full Text]

Brunori, M. and Gibson, Q.H. 2001. Cavities and packing defects in the structural dynamics of myoglobin. EMBO Rep. 2: 674–679.[CrossRef][Medline]

Carlson, M.L., Regan, R.M., and Gibson, Q.H. 1996. Distal cavity fluctuations in myoglobin: Protein motion and ligand diffusion. Biochemistry 35: 1125–1136.[CrossRef][Medline]

Chu, K., Vojtechovsky, J., McMahon, B.H., Sweet, R.M., Berendzen, J., and Schlichting, I. 2000. Structure of a ligand-binding intermediate in wild-type carbonmonoxymyoglobin. Nature 403: 921–923.[CrossRef][Medline]

de Sanctis, D., Dewilde, S., Pesce, A., Moens, L., Ascenzi, P., Hankeln, T., Burmester, T., and Bolognesi, M. 2004a. Crystal structure of cytoglobin: The fourth globin type discovered in man displays heme hexa-coordination. J. Mol. Biol. 336: 917–927.[CrossRef][Medline]

de Sanctis, D., Dewilde, S., Pesce, A., Moens, L., Ascenzi, P., Hankeln, T., Burmester, T., and Bolognesi, M. 2004b. Mapping protein matrix cavities in human cytoglobin through Xe atom binding. Biochem. Biophys. Res. Commun. 316: 1217–1221.[CrossRef][Medline]

Dunn, A.R., Dmoichowski, I.J., Bilwes, A.M., Gray, H.B., and Crane, B.R. 2001. Probing the open state of cytochrome P450cam with ruthenium-linker substrates. Proc. Natl. Acad. Sci. 98: 12421–12425.

Elber, R. and Karplus, M. 1990. Enhanced sampling in molecular dynamics: Use of the time-dependent Hartree approximation for a simulation of carbon monoxide diffusion through myoglobin. J. Am. Chem. Soc. 112: 9161–9175.[CrossRef]

Engler, N., Ostermann, A., Gassmann, A., Lamb, D.C., Prusakov, V.E., Schott, J., Schweitzer-Stenner, R., and Parak, F.G. 2000. Protein dynamics in an intermediate state of myoglobin: Optical absorption, resonance raman spectroscopy, and X-ray structure analysis. Biophys. J. 78: 2081–2092.[Medline]

Essmann, U., Perera, L., Berkowitz, M.L., Darden, T., Lee, H., and Pedersen, L.G. 1995. A smooth particle mesh Ewald method. J. Chem. Phys. 103: 8577–8593.[CrossRef]

Helms, V. and Wade, R.C. 1997. Free energies of hydration from thermodynamic integration: Comparison of molecular mechanics force fields and evaluation of calculation accuracy. J. Comput. Chem. 18: 449–462.[CrossRef]

Hummer, G., Schotte, F., and Anfinrud, P.A. 2004. Unveiling functional protein motions with picosecond X-ray crystallography and molecular dynamics simulations. Proc. Natl. Acad. Sci. 101: 15330–15334.[Abstract/Free Full Text]

Jorgensen, W.L., Chandrasekhar, J., Madura, J.D., Impey, R.W., and Klein, M.L. 1983. Comparison of simple potential functions for simulating liquid water. J. Chem. Phys. 79: 926–935.[CrossRef]

Kleywegt, G.J. and Jones, T.A. 1997. Model-building and refinement practice. Methods Enzymol. 277: 208–230.[Medline]

Kriegl, J.M., Nienhaus, K., Deng, P., Fuchs, J., and Nienhaus, G.U. 2003. Ligand dynamics in a protein internal cavity. Proc. Natl. Acad. Sci. 100: 7063–7074.[Abstract/Free Full Text]

Lamb, D.C., Nienhaus, K., Arcovito, A., Draghi, F., Miele, A.E., Brunori, M., and Nienhaus, G.U. 2002. Structural dynamics of myoglobin. Ligand migration among protein cavities studied by FTI/TDS spectroscopy. J. Biol. Chem. 277: 11636–11644.[Abstract/Free Full Text]

Liang, J., Edelsbrunner, H., Fu, P., Sudhakar, P.V., and Subramaniam, S. 1998a. Analytical shape computation of macromolecules: I. Molecular area and volume through alpha shape. Proteins 33: 1–17.[Medline]

Liang, J., Edelsbrunner, H., Fu, P., Sudhakar, P.V., and Subramaniam, S. 1998b. Analytical shape computation of macromolecules: II. Inacessible cavities in proteins. Proteins 33: 18–29.[CrossRef][Medline]

Lüdemann, S.K., Lounnas, V., and Wade, R.C. 2000a. How do substrates enter and products exit the buried active site of cytochrome P450cam? 1. Random expulsion molecular dynamics investigation of ligand access channels and mechanisms. J. Mol. Biol. 303: 797–811.[CrossRef][Medline]

Lüdemann, S.K., Lounnas, V., and Wade, R.C. 2000b. How do substrates enter and products exit the buried active site of cytochrome P450cam? 2. Steered molecular dynamics and adiabatic mapping of substrate pathways. J. Mol. Biol. 303: 813–830.[CrossRef][Medline]

MacKerell Jr, A.D., Bashford, D., Bellott, M., Dunbrack Jr, R.L., Evanseck, J.D., Field, M.J., Fischer, S., Gao, J., Guo, H., Ha, S., et al. 1998. All-atom empirical potential for molecular modeling and dynamics studies of proteins. J. Phys. Chem. B 102: 3586–3616.

Meller, J. and Elber, R. 1998. Computer simulations of carbon monoxide photodissociation in myoglobin: Structural interpretation of the B states. Biophys. J. 74: 789–802.[Medline]

Milani, M., Pesce, A., Ouellet, Y., Dewilde, S., Friedman, J., Ascenzi, P., Guertin, M., and Bolognesi, M. 2004. Heme-ligand tunneling in group I truncated hemoglobins. J. Biol. Chem. 279: 21520–21525.[Abstract/Free Full Text]

Mouawad, L., Marechal, J.-D., and Perahia, D. 2005. Internal cavities and ligand passageways in human hemoglobin characterized by molecular dynamics simulations. Biochim. Biophys. Acta 1724: 385–393.[Medline]

Nienhaus, K., Lamb, D.C., Deng, P., and Nienhaus, G.U. 2002. The effect of ligand dynamics on heme electronic transition band III in myoglobin. Biophys. J. 82: 1059–1067.[Medline]

Nienhaus, K., Deng, P., Kriegl, J.M., and Nienhaus, G.U. 2003a. Structural dynamics of myoglobin: Effect of internal cavities on ligand migration and binding. Biochemistry 42: 9647–9658.[CrossRef][Medline]

Nienhaus, K., Deng, P., Kriegl, J.M., and Nienhaus, G.U. 2003b. Structural dynamics of myoglobin: Spectroscopic and structural characterization of ligand docking sites in myoglobin mutant L29W. Biochemistry 42: 9633–9646.[CrossRef][Medline]

Ostermann, A., Waschipky, R., Parak, F.G., and Nienhaus, G.U. 2000. Ligand binding and conformational motions in myoglobin. Nature 404: 205–208.[CrossRef][Medline]

Poulos, T.L., Finzel, B.C., Gunsalus, I.C., Wagner, G.C., and Kraut, J. 1985. The 2.6 Å crystal structure of Pseudomonas putida CytP450. J. Biol. Chem. 260: 16122–16130.[Abstract/Free Full Text]

Poulos, T.L., Finzel, B.C., and Howard, A.J. 1987. High-resolution crystal structure of cytochrome P450cam. J. Mol. Biol. 195: 687–700.[CrossRef][Medline]

Prangé, T., Schiltz, M., Pernot, L., Colloc'h, N., Longhi, S., Bourguet, W., and Fourme, R. 1998. Exploring hydrophobic sites in proteins with xenon or krypton. Proteins 30: 61–73.[CrossRef][Medline]

Pylypenko, O. and Schlichting, I. 2004. Structural aspects of ligand binding to and electron transfer in bacterial and fungal P450s. Annu. Rev. Biochem. 73: 991–1018.[CrossRef][Medline]

Raag, R. and Poulos, T.L. 1989. Crystal structure of the carbon-monoxide-substrate-cytochrome P-450CAM ternary complex. Biochemistry 28: 7586–7592.[CrossRef][Medline]

Radding, W. and Phillips Jr, G.N. 2004. Kinetic proofreading by the cavity system of myoglobin: Protection from poisoning. Bioessays 26: 422–433.[CrossRef][Medline]

Radresa, O., Ogata, K., Wodak, S., Ruysschaert, J.-M., and Goormaghtigh, E. 2002. Modeling the three-dimensional structure of H⁺-ATPase of Neurospora crassa. Eur. J. Biochem. 269: 5246–5258.[Medline]

Rashin, A.A. and Rashin, A.H. 2005. Lattice models, packing density, and Boltzmann-like distribution of cavities in proteins. Proteins 58: 547–559.[CrossRef][Medline]

Rubin, S.M., Lee, S.-Y., Ruiz, E.J., Pines, A., and Wemmer, D.E. 2002. Detection and characterization of xenon-binding sites in proteins by 129Xe NMR spectroscopy. J. Mol. Biol. 322: 425–440.[CrossRef][Medline]

Ryckaert, J.-P., Ciccotti, G., and Berendsen, H.J.C. 1977. Numerical integration of the Cartesian equations of motion for a system with constraints: Molecular dynamics of n-alkanes. J. Comput. Phys. 23: 327–341.[CrossRef]

Schlichting, I. 2000. Crystallographic structure determination of unstable species. Acc. Chem. Res. 33: 532–538.[CrossRef][Medline]

Schotte, F., Lim, M., Jackson, T.A., Smirnov, A.V., Soman, J., Olson, J.S., Phillips Jr, G.N., Wulff, M., and Anfinrud, P.A. 2003. Watching a protein as it functions with 150-ps time-resolved X-ray crystallography. Science 300: 1944–1947.[Abstract/Free Full Text]

Scott, E.E. and Gibson, Q.H. 1997. Ligand migration in sperm whale myoglobin. Biochemistry 36: 11909–11917.[CrossRef][Medline]

Scott, E.E., Gibson, Q.H., and Olson, J.S. 2001. Mapping the pathways for O₂ entry into and exit from myoglobin. J. Biol. Chem. 276: 5177–5188.[Abstract/Free Full Text]

Srajer, V., Ren, Z., Teng, T.-Y., Schmidt, M., Ursby, T., Bourgeois, D., Pradervand, C., Schildkamp, W., Wulff, M., and Moffat, K. 2001. Protein conformational relaxation and ligand migration in myoglobin: A nanosecond to millisecond movie from time-resolved Laue X-ray diffraction. Biochemistry 40: 13802–13815.[CrossRef][Medline]

Teeter, M.M. 2004. Myoglobin cavities provide interior ligand pathway. Protein Sci. 13: 313–318.[Abstract/Free Full Text]

Tetreau, C. and Lavalette, D. 2005. Dominant features of protein reaction dynamics: conformational relaxation and ligand migration. Biochim. Biophys. Acta 1724: 411–424.[Medline]

Tetreau, C., Blouquit, Y., Novikov, E., Quiniou, E., and Lavalette, D. 2004. Competition with xenon elicits ligand migration and escape pathways in myoglobin. Biophys. J. 86: 435–447.[Medline]