Noble Gas Binding to Human Serum Albumin

Noble Gas Binding to Human Serum Albumin Using Docking Simulation: Nonimmobilizers and Anesthetics Bind to Different Sites Tomoyoshi Seto, MD, PhD*† Hideto Isogai, PhD† Masayuki Ozaki, MD* Shuichi Nosaka, MD*

BACKGROUND: Nonimmobilizers are structurally similar to anesthetics, but do not produce anesthesia at clinically relevant concentrations. Xenon, krypton, and argon are anesthetics, whereas neon and helium are nonimmobilizers. The structures of noble gases with anesthetics or nonimmobilizers are similar and their interactions are simple. Whether the binding site of anesthetics differs from that of nonimmobilizers has long been a question in molecular anesthesiology. METHODS: We investigated the binding sites and energies of anesthetic and nonimmobilizer noble gases in human serum albumin (HSA) because the 3D structure of HSA is well known and it has an anesthetic binding site. The computational docking simulation we used searches for binding sites and calculates the binding energy for small molecules and a template molecule. RESULTS: Xenon, krypton, and argon were found to bind to the enflurane binding site of HSA, whereas neon and helium were found to bind to sites different from the xenon binding site. Rare gas anesthetic binding was dominated by van der Waals energy, while nonimmobilizer binding was dominated by solvent-effect energy. Binding site preference was determined by the ratios of local binding energy (van der Waals energy) and nonspecific binding energy (solvent-effect energy) to the total binding energy. van der Waals energy dominance is necessary for anesthetic binding. CONCLUSIONS: This analysis of binding energy components provides a rationale for the binding site difference of anesthetics and nonimmobilizers, reveals the differences between the binding interactions of anesthetics and nonimmobilizers, may explain pharmacological differences between anesthetics and nonimmobilizers, and provide an understanding of anesthetic action at the atomic level. (Anesth Analg 2008;107:1223–8)

A

lthough some noble gases are anesthetics,1 the anesthetic action of helium and neon is so weak that these two gases have been used as media in experiments investigating the pressure reversal of anesthesia.2 Detailed reinvestigations of the minimum alveolar concentrations of noble gases have revealed that helium and neon do not have anesthetic effects and are nonimmobilizers,3 which are substances that are structurally similar to anesthetics but do not have the anesthetic effects predicted by the Overton-Meyer From the *Department of Anesthesiology, Shiga University of Medical Science, Otsu, Japan; and †Department of Applied Chemistry, Ritsumeikan University, Kusatsu, Japan. Accepted for publication April 29, 2008. Supported in part by Grant-in-Aid for Scientific Research Young Investigator (B) 19791062 from the Ministry of Education, Culture, Sports, Science and Technology (MEXT), in Japan. Hideto Isogai is currently at the Department of Molecular Life Science, Tokai University School of Medicine. Address correspondence and reprint requests to Tomoyoshi Seto, MD, PhD, Department of Anesthesiology, Shiga University of Medical Science, Seta Tsukinowa-cho, Otsu 520-2192, Japan. Address e-mail to [email protected]. Copyright © 2008 International Anesthesia Research Society DOI: 10.1213/ane.0b013e31817f1317

Vol. 107, No. 4, October 2008

rule. The anesthetic fluorine atom interacts strongly with the indole N atoms of tryptophan residues or amide hydrogen bonds in the backbone chain of gramicidin A (gA), a model of the peptide channels in lipid bilayers, whereas nonimmobilizers interact with gA only weakly.4 It is not even known whether anesthetics and nonimmobilizers bind to the same site or different sites. Human serum albumin (HSA) can be used as a suitable template for characterizing the structureactivity relationships of general anesthetics,5 and x-ray crystallography with 2.2-Å resolution has revealed that HSA has two binding sites for the IV anesthetic propofol. These binding sites, PR1 and PR2, are also binding sites for volatile anesthetics. Halothane, isoflurane, and enflurane (ENF) bind to PR1,6,7 which is a polar pocket, and ENF also binds to the less polar PR2.7 As the primary interactions of anesthetic binding in HSA are van der Waals interactions and hydrophobic interactions rather than electrostatic interactions,8 this ENF binding site PR2 is suitable for studying binding differences of anesthetics and nonimmobilizers in the molecular interaction level. In this study, we therefore used this ENF binding site to study affinity differences 1223

of anesthetics and nonimmobilizers. To simplify the analysis, we investigated only the binding of noble gases: the anesthetics xenon, krypton, and argon and the nonimmobilizers neon and helium. These gases have similar simple structures and full valence electron shells, so differences in binding will not be due to differences in molecular shape or electrostatic contributions. The progress of computational science has made it possible to predict the 3D-structure of the complex formed when a small molecule binds to a macromolecule and to calculate the binding location and binding mode of the small molecule in this complex. We predicted the locations around the ENF binding site (PR2) to which anesthetic and nonimmobilizer noble gases would bind in order to see whether both kinds of noble gases bind to the same site.

METHODS The structure of the propofol-HSA complex was obtained from the Protein Data Bank 1E7A (2.20-Å resolution, pH 7.0).6 The coordinates of the propofol molecule were deleted from the x-ray crystallography coordinates of the complex, and the remaining HSA structure was used as a docking simulation template. These crystallography data were amended to generate complete structural data in the following manner. After the template coordinates of the heavy atoms in HSA were fixed, hydrogen atoms absent in the crystallography data were added. To keep the added hydrogen atoms from overlapping the heavy atoms or other hydrogen atoms added, their positions were optimized to the minimum energy. The complete HSA structure (with the added hydrogen atoms) was used for docking simulations with noble gases. MOE-Dock 2002.3 Program in the Molecular Operating Environment 2002.3 (Chemical Computing Group, Montreal, Canada) was used to perform molecular docking for HSA and the noble gases xenon, krypton, argon, neon, and helium by using the Merck molecular force field 94s force field parameters.9,10 The albumin structure prepared as described in the previous paragraph was used, and the xenon binding position was sought within the 31 ⫻ 31 ⫻ 31-Å search box (docking box) set-up including four ␣-helical chains surrounding the bound propofol molecule. The simulated annealing method in MOE-Dock 2002.3 that was used to find the global minimum is based on the Monte Carlo method.11 It explores various states of a configuration space by generating small random changes in the current state and then accepting or rejecting each new state according to the Metropolis criterion.12 The xenon-docked structure was determined by minimizing the energy of the complex. The concept of simulated annealing is explained in the Appendix. The MOE-Dock 2002.3 calculates relative binding of free energies, electrostatic energy, van der 1224

Docking Simulation Study of Noble Gas

Table 1. Results from 25 Independent Docking Runs of Xenonlysozyme Complex Run number

U_total energy (kcal/mol)

#1 #2 #3 #4 #5 #6 #7 #8 #9 # 10 # 11 # 12 # 13 # 14 # 15 # 16 # 17 # 18 # 19 # 20 # 21 # 22 # 23 # 24 # 25

⫺ 7.23 ⫺ 7.23 ⫺ 7.23 ⫺ 7.23 ⫺ 7.23 ⫺ 4.56 ⫺ 4.24 ⫺ 4.24 ⫺ 4.24 ⫺ 4.24 ⫺ 4.23 ⫺ 4.23 ⫺ 4.23 22.02 22.03 32.71 48.55 48.55 72.55 78.29 78.29 78.29 78.29 91.02 138.61

Five runs in 25 trials showed a stable minimum energy and xenon position. Minimum energy and binding site can be regarded as global minimum in this docking box.

Waals energy,9,13 and solvation energy (i.e., solvent electrostatic correction). Solvation energies were calculated by the Poisson-Boltzmann equation implemented in MOE 2002.3.14 –16 The iteration limit of MOE-Dock 2002.3 was set to 8000, and the number of cycles was set to 8. The reproducibility of the minimum was checked by repeating the same search trial 25 times. The details of the calculation have been reported elsewhere.17 The validity of the MOE-Dock 2002.3 simulation was confirmed by using it, with Merck molecular force field 94s parameters, to calculate the structure of the complex calculated for xenon docked to the xenon binding site on lysozyme (Protein Data Bank entry 1C10) and then comparing that structure with the structure of the xenon-lysozyme complex determined from x-ray crystallography experiments.

RESULTS Docking precision was checked by comparing the simulated and x-ray crystallography determined results for xenon re-docking to the xenon binding site 1C10. The simulated xenon position differed by 2.9 Å from 1C10.18 Since the van der Waals radius of xenon is 2.16 Å, one sees that the docking simulation reproduced the complex structure determined in x-ray crystallography experiments (Table 1). Thus MOE-Dock 2002.3 proved to be a reliable tool for predicting the binding site of anesthetic gases in this study. ANESTHESIA & ANALGESIA

Figure 1. Xenon (dark blue) and interacting side chains of amino acid residues in the enflurane (ENF) (PR2) site of human serum albumin.

The amino residues constituting the ENF binding site, identified by extracting those within 4.5 Å of the position of the docked xenon, were Phe507, Phe509, Phe551, and Ala528 (Fig. 1). These are hydrophobic amino acids. The xenon atom interacted with the aromatic planes of two Phe residues and directly contacted the methyl group of Ala528. Xenon, krypton, and argon atoms docked to the ENF binding site are shown in dark blue in the space filling model in Figure 2. These three atoms overlap, so their binding positions, which correspond to the binding positions of the isopropyl groups of propofol, are essentially the same. Neon and helium, shown in light blue, are docked at positions away from the ENF binding site. Total binding energy and the components of this energy were determined from MOE results for each of the noble gases and are listed in Table 2. Both van der Waals interaction energy and solvation energy contribute to the binding energy. The binding energies for xenon, krypton, and argon decrease with the size of the atom. Local binding energy (in this study, van der Waals energy), which contributes to site specificity, is a larger proportion of the total binding energy for these 3 atoms (40%) than it is of the total binding energy for neon and helium (10%–26%). The total binding energies for the three anesthetic gases were correlated with their minimum alveolar concentrations (Fig. 3). It showed: U_total ⫽ 1.015 ⫻ log10 minimum alveolar concentration ⫺ 8.399. Confidence limits were [⫺2.842, 4.889] and [⫺12.121, ⫺4.677], respectively.

Figure 2. A: Human serum albumin structure: enflurane binding site ENF(PR2) and propofol binding site (PR1). Binding residues were shown in ENF site. B: Positions of noble gas atoms binding to the ENF(PR2) site. Dark blue: xenon, krypton, and argon. Light blue: neon and helium. Table 2. Noble Gas Binding Energies to the Enflurane (PR2) Site of Human Serum Albumin (Calculated by MOE-Dock 2002.3) Noble gas Helium Neon Argon Krypton Xenon

Total binding Van der Waals Solvation energy energy energy (kcal/mol) (kcal/mol) (kcal/mol) ⫺4.47 ⫺6.95 ⫺6.84 ⫺7.73 ⫺8.08

⫺1.19 ⫺0.71 ⫺2.66 ⫺2.70 ⫺3.59

⫺3.28 ⫺6.24 ⫺4.18 ⫺5.03 ⫺4.49

Total binding energy (at room temp) ⫽ Van der Waals energy ⫹ solvation energy.

DISCUSSION The present study, finding that anesthetic noble gases bind to the ENF binding site and that nonimmobilizer noble gases do not, suggests that the binding site preferences of anesthetics and nonimmobilizers could account for the pharmacological differences between them. Vol. 107, No. 4, October 2008

The total binding energy of a noble gas consists of van der Waals energy and solvent-effect energy, and the ENF binding site consists of three hydrophobic residues whose spatial configuration is fixed. The van der Waals interactions between the site and a noble gas will be too small and weak because the strength of © 2008 International Anesthesia Research Society

1225

Figure 4. Energy diagram for xenon and helium binding to the enflurane (ENF) (PR2) site of human serum albumin.

Figure 3. Relation between the binding energy and minimum alveolar concentration (MAC) of anesthetic noble gases. (MAC data from Ref.3).

those interactions is inversely proportional to the sixth power of distance. Snug fitting between an anesthetic and the binding site enables strong van der Waals interaction, is considered to produce local affinity with the site, and is thought to determine the binding site preferences of the noble gases. The entropy effect term due to solvent exclusion in solvent effect is large enough to drive the binding or coagulation of small molecules,19 but the MOEs energy calculation of solvent effects does not include the effect of solvent exclusion.14 –16 To calculate solvent effects accurately, solvent exclusion effect should be considered for energy calculation.19,20 A semiquantitative discussion can nonetheless be based on the MOEs calculation because the known structure of the xenon-lysozyme complex is reproduced by minimizing the MOEs calculated energy. Solvent-effect energy is comparable to van der Waals energy in our calculation. The transfer of an anesthetic from the water phase to the HSA site is thought to proceed in two steps. The first is partial dehydration of the anesthetic and binding site (hydrophobic dehydration) and the second is direct binding of the anesthetic and the site. The energy stabilization resulting from this hydrophobic dehydration is a solvent effect and is thought to make a large contribution to binding energy.21 We think it is important that both the local van der Waals energy and the non-site-specific solvent-effect energy due to the dehydration during the transfer from the water phase contribute to the binding energy of anesthetics, as well as that of nonimmobilizers. The ratio of van der Waals energy to total binding energy is larger for anesthetics than nonimmobilizers. This means that anesthetics have a relatively high local affinity, whereas nonimmobilizers have mostly nonselective affinity (solvent-effect energy). The energy diagrams for xenon and helium binding to the ENF binding site of HSA are shown in Figure 4. The 1226

Docking Simulation Study of Noble Gas

ratio of the energies explains the nature of anesthetic and nonimmobilizer binding. Whether anesthetics act on a specific or nonspecific site has long been a question. Many investigators hypothesizing that anesthetics act on a specific site think that the site could be a neuronal ion-channel protein. At clinically relevant concentrations, anesthetics enhance a neuronal ion-channel receptor, the ␥-aminobutyric acid type A (GABAA) receptor.22 At one time, the GABAA receptor was thought to be the specific site of anesthetic action, but the anesthetics cyclopropane and butane were found not to enhance the GABAA receptor.23 This means that the GABAA receptor does not completely explain the actions of anesthetics. It has been observed that no single ionchannel or receptor explains anesthetic immobility.24 Nonspecific sites of anesthetics’ actions have been noted, and it has been hypothesized that inhaled anesthetics produce anesthesia by modulating the global dynamics of one or more channel proteins.25 Using molecular simulation of a model of the water-hexane interface, Chipot et al.26 show that anesthetics bind to this but nonimmobilizers do not. The anesthetic binding process begins with transferring to the interface, and proceeds to form a cavity at the interface. Solute-solvent interactions of electrostatic and van der Waals then occur between the anesthetic and water/hexane molecule at the interface. They concluded that the affinity to the interface was originated from the energy balance between cavity formation and solute-solvent interactions. The water-hexane interface model and our ENF-binding-site-in-HSA model are quite different; one is organic solvent and the other is a protein. However, xenon binding processes to HSA in this study resemble that of the water-hexane interface because cavity formation (dehydration) and solute-solvent interactions between xenon and amino acid residues of the binding site correspond to binding processes to the water-hexane interface. The anesthetic binding processes are common to both models. Therefore, anesthetic binding to HSA can be understood as the binding to the waterprotein interface, because the ENF site is open to the protein surface. Noble gas binding to lysozyme has ANESTHESIA & ANALGESIA

Figure 5. Conceptual schema of simulated annealing. Simulated annealing is a searching method for global minimum by high-temperature thermal vibration which gets over energy barrier which cannot get over with lowtemperature thermal vibration.

recently been predicted by computational simulation, and the distribution of noble gases in the waterprotein interface has been reported.27 (Fig. 1(a) in Ref.27). They showed that xenon bound to the waterprotein interface of the surface, whereas neon bound and distributed less density to it, thus, xenon and neon showed different affinity to the protein interface. This study in HSA also suggests that noble gas anesthetics have an affinity to the protein-water interface. We could not search the whole HSA molecule because our computational resources were insufficient, but we did find that helium and neon did not bind to the ENF binding site. We speculate that helium and neon may bind to a different site in HSA, because anesthetics and nonimmobilizers have different interactions with solvent or the protein site. Chemical genomics has attracted attention in recent years, and a probabilistic model of relations between chemical compounds (drugs) and genes has been proposed.28 With this development, it becomes possible to identify genes, which are the origin of expressed proteins, interacting with pharmacologically active chemical compounds (i.e., drugs). When the relationship between anesthetics and genes becomes clear, it will be possible to determine whether anesthetics act on specific or nonspecific sites. It is surprising that the binding energy of rare gas anesthetics for the HSA model protein was correlated with the pharmacological anesthetic potency measure “minimum alveolar concentration.” The ENF binding site in this study is merely an anesthetic binding site, but it has a certain character which is common to the site of anesthetic action. The anesthetic target site is unknown, but at least the molecular interactions between that site and anesthetics are common to the binding model in this study. Our findings based on an ENF Vol. 107, No. 4, October 2008

binding site on HSA may contribute to the elucidation of anesthetic action at the molecular level.

CONCLUSIONS This study predicts that the anesthetic noble gases xenon, krypton, and argon bind to a part of the ENF binding site of HSA different from the part(s) to which the nonimmobilizers neon and helium bind. Noble gas anesthetics have a relatively high local affinity, whereas nonimmobilizers have nonsite-specific energy. The calculated total binding energy of anesthetics was correlated with their minimum alveolar concentrations. Our finding may explain the binding preference related to the pharmacological difference between anesthetics and nonimmobilizers and provide an understanding of anesthetic action at the atomic level. Xenon anesthesia has recently been put to practical use. This study results guarantee reversibility of xenon anesthesia following scientific ground of the molecular interactions. The safety of clinical anesthesia could be ensured by selecting an anesthetic that had been scientifically tested. APPENDIX MOE-Dock 2002.3 searches for favorable binding sites and conformations between a small ligand and a rigid macromolecular target. Using simulated annealing method and a molecular mechanics forcefield, MOE-Dock 2002.3 can search binding site within a specified 3D docking box. Finding a binding site corresponds to finding the complex structure with the minimum energy. Simulated annealing is a global optimization technique based on the Monte Carlo method. It explores various sites by generating small random changes in the current site and then accepting or rejecting each new site according to the © 2008 International Anesthesia Research Society

1227

Metropolis criterion.12 Each such change to the ligand is called a move. According to the Metropolis criterion, moves that decrease the energy of the system are always accepted, while moves that increase the energy of the system are accepted according to probability P ⫽ exp (-␦u/kt), where ␦u ⫽ u1-u0 (u0 is the energy of the current state and u1 is the energy of the new state), t is the “temperature” of the simulation, and k is the Boltzmann constant. Random search (move) was iterated, until either the number of accepted moves or the number of rejected moves reached 8000. The temperature, held constant during each search, was systematically reduced in steps to find the global minimum energy site. In this study the initial temperature, 1000 K, was decreased by 180 to 100 K in each step. The global minimum temperature does not depend on the search method including temperature steps of simulated annealing or the parameters. To confirm the global minimum, independent calculations with simulated annealing were repeated 25 times, then agreed minimum value was taken as the global minimum of this study. That is to find the binding site in room temperature. Conceptual Schema of simulated annealing was represented in Figure 5. ACKNOWLEDGMENTS We are grateful to the Central Research Laboratory of Shiga University of Medical Science for the MOE license. REFERENCES 1. Miller KW. The nature of sites of general anaesthetic action. Br J Anaesth 2002;89:17–31 2. Moss GW, Lieb WR, Franks NP. Anesthetic inhibition of firefly luciferase, a protein model for general anesthesia, does not exhibit pressure reversal. Biophys J 1991;60:1309 –14 3. Koblin DD, Fang Z, Eger EI II, Laster MJ, Gong D, Ionescu P, Halsey MJ, Trudell JR. Minimum alveolar concentrations of noble gases, nitrogen, and sulfur hexafluoride in rats: helium and neon as nonimmobilizers (nonanesthetics). Anesth Analg 1998;87:419 –24 4. Tang P, Hu J, Liachenko S, Xu Y. Distinctly different interactions of anesthetic and nonimmobilizer with transmembrane channel peptides. Biophys J 1999;77:739 – 46 5. Liu R, Meng Q, Xi J, Yang J, Ha CE, Bhagavan NV, Eckenhoff RG. Comparative binding character of two general anaesthetics for sites on human serum albumin. Biochem J 2004;380:147–52 6. Bhattacharya AA, Curry S, Franks NP. Binding of the general anesthetics propofol and halothane to human serum albumin. High resolution crystal structures. J Biol Chem 2000;275:38731– 8 7. Liu R, Eckenhoff RG. Weak polar interactions confer albumin binding site selectivity for haloether anesthetics. Anesthesiology 2005;102:799 – 805 8. Liu R, Pidikiti R, Ha CE, Petersen CE, Bhagavan NV, Eckenhoff RG. The role of electrostatic interactions in human serum albumin binding and stabilization by halothane. J Biol Chem 2002;277:36373–9

1228

Docking Simulation Study of Noble Gas

9. Halgren TA. Merck molecular force field. I. Basis, form, scope, parameterization, and performance of MMFF94. J Comput Chem 1996;17:490 –519 10. Halgren TA MMFF VI. MMFF94s option for energy minimization studies. J Comput Chem 1999;20:720 –9 11. Hart TN, Read RJ. A multiple-start Monte Carlo docking method. Proteins 1992;13:206 –22 12. Metropolis N, Rosenbluth A, Rosenbluth M, Teller A, Teller E. Equation of state calculations by fast computing machine. J Chem Phys 1953;21:1087–92 13. Halgren TA. Merck molecular force field. II. MMFF94 van der waals and electrostatic parameters for intermolecular interactions. J Comput Chem 1996;17:520 –52 14. Still WC, Tempczyk A, Hawley RC, Hendrickson T. Semianalytical treatment of solvation for molecular mechanics and dynamics. J Am Chem Soc 1990;112:6127–9 15. Schaefer M, Karplus M. A comprehensive analytical treatment of continuum electrostatics. J Phys Chem 1996;100:1578 –99 16. Qui D, Shenkin S, Hollinger FP, Still WC. The GB/SA continuum model for solvation. A fast analytical method for the calculation of approximate Born radii. J Phys Chem A 1997; 101:3005–14 17. Seto T, Nosaka S. Structural prediction of barbital-acetylcholine receptor complex from docking simulation. International Congress Series 2005;1283:151– 4 18. Isogai H, Seto T, Nosaka S. Validation of hypothesis of pressure reversal mechanisms: “Anesthetics binds to site, and squeezed out by pressure.” International Congress Series 2005;1283: 318 –19 19. Kinoshita M. Interaction between big bodies with high asphericity immersed in small aspheres. Chem Phys Lett 2004;387: 47–53 20. Trudell JR, Koblin DD, Eger EI 2nd. A molecular description of how noble gases and nitrogen bind to a model site of anesthetic action. Anesth Analg 1998;87:411–18 21. Sawas AH, Pentyala SN, Rebecchi MJ. Binding of volatile anesthetics to serum albumin: measurements of enthalpy and solvent contributions. Biochemistry 2004;43:12675– 85 22. Jones MV, Brooks PA, Harrison NL. Enhancement of gammaaminobutyric acid-activated Cl- currents in cultured rat hippocampal neurones by three volatile anaesthetics. J Physiol 1992;449:279 –93 23. Raines DE, Claycomb RJ, Scheller M, Forman SA. Nonhalogenated alkane anesthetics fail to potentiate agonist actions on two ligand-gated ion channels. Anesthesiology 2001;95:470 –7 24. Sonner JM, Antognini JF, Dutton RC, Flood P, Gray AT, Harris RA, Homanics GE, Kendig J, Orser B, Raines DE, Rampil IJ, Trudell J, Vissel B, Eger EI II. Inhaled anesthetics and immobility: mechanisms, mysteries, and minimum alveolar anesthetic concentration. Anesth Analg 2003;97:718 – 40 25. Tang P, Xu Y. Large-scale molecular dynamics simulations of general anesthetic effects on the ion channel in the fully hydrated membrane: the implication of molecular mechanisms of general anesthesia. Proc Natl Acad Sci USA 2002;99:16035– 40 26. Chipot C, Wilson M, Pohorille A. Interactions of anesthetics with the water-hexane interface. A molecular dynamics study. J Phys Chem B 1997;101:782–91 27. Imai T, Hiraoka R, Seto T, Kovalenko A, Hirata F. Threedimensional distribution function theory for the prediction of protein-ligand binding sites and affinities: application to the binding of noble gases to hen egg-white lysozyme in aqueous solution. J Phys Chem B 2007;111:11585–91 28. Zhu S, Okuno Y, Tsujimoto G, Mamitsuka H. A probabilistic model for mining implicit ‘chemical compound-gene’ relations from literature. Bioinformatics 2005;21(suppl 2):ii245–ii251

ANESTHESIA & ANALGESIA