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