Tài liệu Báo cáo khoa học: The lipid ⁄ protein interface as xenobiotic target site Kinetic analysis of tadpole narcosis ppt

The lipid ⁄ protein interface as xenobiotic target site
Kinetic analysis of tadpole narcosis
Joachim Altschuh
1
, Sebastian Walcher
2
and Heinrich Sandermann, Jr
3,4
1 Institute of Biomathematics and Biometry, GSF – National Research Center for Environment and Health, Neuherberg, Germany
2 Lehrstuhl A fu
¨
r Mathematik, RWTH Aachen, Germany
3 Institute of Biochemical Plant Pathology, GSF – National Research Center for Environment and Health, Neuherberg, Germany
4 ecotox.freiburg, Germany
High-resolution X-ray diffraction structures of integral
membrane proteins have revealed structurally diverse
annular, nonannular, single leaflet and laterally con-
nected lipid-binding sites [1–4]. Each of the multiple
lipid-binding sites of integral membrane proteins may
be, to a certain extent, structurally unique, as best
documented for the tightly bound phospholipid, cardio-
lipin [1–4]. In spite of their analytical power, spectro-
scopic methods have so far failed to resolve the
multiple protein-bound lipids. Both ESR [5,6] and
fluorescence spectroscopy [2,7] use overall relative
macroscopic lipid-binding constants. The similarity of
these binding constants among different membrane
proteins has led to the conclusion that binding to the
annular lipid shell shows relatively little structural
specificity [2]. The overall microscopic lipid-binding
constants of four integral membrane enzymes also

were quite uniform [8]. By contrast, there is evidence
for special lipid-binding sites having high structural
specificity, or dominating the lipid dependence of func-
tional proteins. For example, the mitochondrial b-hyd-
roxybutyrate dehydrogenase needs the specific lipid,
phosphatidylcholine, in a background of other phos-
pholipids [9,10], and a specific requirement for cardio-
lipin in a background of other lipids is documented for
cytochrome oxidase [11] and processes of oxidative
phosphorylation [12]. In order to kinetically resolve
protein-bound lipids, an extended Adair-type analysis
of tadpole narcosis was performed employing micro-
scopic lipid-binding constants.
More than a century ago, studies of tadpole narcosis
led to the Meyer–Overton rule describing a positive
Keywords
Adair theory; kinetic modelling; lipid ⁄ protein
interface; nicotinic acetylcholine receptor;
tadpole narcosis
Correspondence
H. Sandermann, GSF – National Research
Center for Environment and Health, Institute
of Biochemical Plant Pathology, Ingolsta
¨
dter
Landstraße 1, D-85764 Neuherberg,
Germany
Fax: +49 89 3187 3383
Tel: +49 89 3187 2285
E-mail:

(Received 16 December 2004, revised
4 March 2005, accepted 10 March 2005)
doi:10.1111/j.1742-4658.2005.04657.x
High-resolution X-ray diffraction structures of integral membrane proteins
have revealed various binding modes of lipids, but current spectroscopic
studies still use uniform macroscopic binding constants to describe lipid
binding. The Adair approach employing microscopic lipid-binding con-
stants has previously been taken to explain the enhancement of agonist
binding to the nicotinic acetylcholine receptor by general anaesthetics in
terms of the competitive displacement of essential lipid activator molecules
[Walcher S, Altschuh J & Sandermann H (2001) J. Biol. Chem. 276,
42191–42195]. This approach was extended to tadpole narcosis induced by
alcohols. A single class, or two different classes of lipid activator binding
sites, are considered. Microscopic lipid and inhibitor binding constants are
derived and allow a close fit to dose–response curves of tadpole narcosis
on the basis of a preferential displacement of more loosely bound essential
lipid activator molecules. This study illustrates the potential of the Adair
approach to resolve protein-bound lipid populations.
Abbreviations
nAChR, nicotinic acetylcholine receptor; SSD, sum of squared deviations.
FEBS Journal 272 (2005) 2399–2406 ª 2005 FEBS 2399
correlation between the lipophilicity of chemicals and
their anaesthetic potency [13,14]. The underlying
molecular mechanism has remained controversial.
General anaesthetics were proposed to act primarily by
perturbing the membrane lipid phase [13,14], or by
attacking hydrophobic protein pockets [15,16]. The
latter hypothesis is favoured by site-directed amino
acid exchanges [17–19] and by photoaffinity labelling
[19,20]. However, many of the findings supporting lipid

or protein target sites can also be interpreted in terms
of the lipid ⁄ protein interface as a third candidate
target site. We have previously developed a kinetic
framework for anaesthetics acting by competitive
displacement of essential lipid activators from the
lipid ⁄ protein interface [21–23]. This mechanism has
been tested with synaptosomal Ca
2+
-ATPase [24] and
the Torpedo nicotinic acetylcholine receptor (nAChR)
[23]. Subsequent to our report [23], a competitive dis-
placement of lipids bound to the nAChR has also been
reported for free fatty acids [25] and local anaesthetics
[26]. The nAChR is the by far best-characterized mem-
ber of the superfamily of ligand-gated ion channel
proteins [27], which is currently thought to harbour
the true target site(s) of general anaesthesia [15–20].
The lipid ⁄ protein interface of the nAChR and its
microscopic binding and inhibition constants are used
here as a prototype that allows a close fit to dose–
response data for tadpole narcosis.
Results
Experimental system
This study was based on the fact that the reported Hill
coefficients of tadpole narcosis are in a diagnostic win-
dow that is typical for the activation [28] and inhibi-
tion [21] of lipid-dependent functional membrane
proteins, namely Hill coefficients in the range 2.5–4.0.
Tadpole narcosis has been characterized by Hill coeffi-
cients of around 4.0 when 14 different aliphatic alco-

hols were tested [29] and of around 3.4 with four
different cycloalcohols [30]. The Hill coefficients of the
previously analyzed enhanced agonist binding to the
nAChR were between 2.3 and 4.0 [23,31], and were
thus in the same diagnostic window. The multiple-site
kinetic formalism for lipid-dependent proteins has
therefore previously been applied to the nAChR [23]
and is now applied to tadpole narcosis. General anaes-
thetics are well known to also interact with the lipid-
free pore region of the nAChR, but this inhibitory
effect has much lower kinetic cooperativity (Hill coeffi-
cients 1.0–1.3) [32,33]. In order to employ the pre-
viously characterized lipid ⁄ protein interface of the
nAChR as a prototype target site of tadpole narcosis,
the microscopic inhibition constants of narcotic chemi-
cals need to be known. The K
I
-value of 1-hexanol has
previously been determined [23]. The additional K
I
val-
ues needed for the present study were derived from
other previously determined K
I
values as described in
Experimental procedures and summarized in Table 1.
Kinetic modelling
A basic task of our kinetic modelling was to explain
the ‘leftward shift’ of the dose–response curves of tad-
pole narcosis relative to those of nAChR agonist bind-

ing enhancement. It is known that the members of
ligand-gated channel superfamilies [18–20], and mem-
brane proteins generally [1–4], have sequence-specific
rough surfaces that are likely to lead to nonuniform
binding constants for the multiple essential lipid acti-
vators. Therefore, it seemed appropriate to develop a
general Adair-type formalism for the case of different
binding sites (see Appendix). The more loosely bound
lipid ligands have higher microscopic lipid dissociation
binding constants, K
L
, and will more easily undergo
displacement by general anaesthetics. The K
L
-value
previously derived for the nAChR [23] is therefore
allowed to increase in order to simulate the unknown
narcotic receptor(s) of tadpoles, all other parameters
characterizing the nAChR lipid ⁄ protein interface
remaining constant. The initial analysis assumes a uni-
form increase in the microscopic dissociation constant,
K
L
, of all 40 lipid activator sites of the nAChR. This
is illustrated for 1-hexanol in Fig. 1. An approximation
to the data points for tadpole narcosis and the ‘left-
ward shift’ is achieved by increasing K
L
8.2-fold to
387 nm. This overall increase is within the $ 10-fold

range of average macroscopic lipid-binding constants
of various membrane proteins as documented by ESR
[5] and fluorescence spectroscopy [2]. A 10-fold range
of overall macroscopic binding constants for different
lipids has recently also been documented by ESR-
spectroscopy for the nAChR [26]. In the case of
b-hydroxybutyrate dehydrogenase there was a 15-fold
difference in macroscopic binding constants between
the specific lipid activator, phophatidylcholine, and the
nonactivating lipid, phosphatidylethanolamine [10].
The curve fits to the data points for tadpole narcosis
by the additional alcohols of Table 1 are shown in
Fig. 2 along with the theoretical curves for the
enhancement of agonist binding to the nAChR. The
latter curves were calculated with the K
I
values of
Table 1 and Eqn (1) of Walcher et al. [23].
As discussed above it is unlikely that all lipid-bind-
ing sites of a sensitive target protein differ by the same
Lipid ⁄ protein interface J. Altschuh et al.
2400 FEBS Journal 272 (2005) 2399–2406 ª 2005 FEBS
factor from closely related nontarget members of the
same superfamily. We therefore applied our general-
ized Adair-type formalism to differentiate between two
lipid-binding classes (Appendix). A fixed number, m,
of the total of 40 lipid-binding sites of the receptor is
allowed to have increased K
L
values. For example, if

m ¼ 4 lipids are assumed to be more loosely bound, a
factor of 270 is required to achieve a fit to the data
points for 1-hexanol. The results of a more extensive
analysis of assuming two lipid-binding classes are sum-
marized in Table 1. The corresponding dose–response
curves are shown in Fig. 1 for 1-hexanol and in Fig. 2
for the other tested alcohols.
Statistical quality
Although the focus of the mathematical approach is on
the ‘leftward shift’, illustrated in Figs 1 and 2, a statisti-
cal analysis was performed for additional justification.
The sums of squared deviations (SSD) for a uniform
increase in K
L
were between 0.022 and 0.38 but were
higher for small values of m. All SSD values are listed
in Table 1. It should be noted that the SSD values of
tadpole narcosis are of poorer quality than of agonist
binding enhancement [23]. However, a survey of the lit-
erature on tadpole narcosis showed that the two studies
used [29,30] were among the best documented. The
curve fits of Figs 1 and 2 were obtained by varying only
Fig. 1. Fitted dose–response curves for 1-hexanol. The principle
(upper) is to mathematically transfer the dose–response curve for
the nAChR [23] to the data set for tadpole anaesthesia [29]. This
‘leftward shift’ of the whole-animal response is achieved by
increasing the microscopic lipid dissociation constant, K
L
, of the
nAChR over all binding sites: m ¼ 40 (solid line), or over part of

the binding sites: m ¼ 10 (dotted line), and m ¼ 4 (dashed line).
The data points for enhancement of agonist binding (s) were taken
from Miller et al. [31] and those for tadpole narcosis (d) from
Alifimoff et al. [29].
Table 1. Results of regression analyses. Values of log K
OW
, K
I
, K
L
¢ and SSD for all alcohols analysed. In addition to the fitting procedure des-
cribed in the Results, the data were also fitted using the standard two-parameter fits of the Gauss and logistic procedures; the correspond-
ing SSD values are shown in the last two columns.
Chemical log K
OW
a
K
I
b
(mM)
No. receptors
m with K
L
¢
c
K
L
¢
c
(lM) SSD

d,e
SSD
d,f
SSD
d,g
1-Hexanol 2.03
h
0.069
j
40 0.387 0.22 0.21 0.21
10 1.85 0.25
4 12.7 0.37
1-Octanol 3.00
h
0.011
k
40 0.642 0.38 0.23 0.22
10 3.16 0.46
4 26.6 0.70
Cyclohexanol 1.23
h
0.25
k
40 0.182 0.022 0.0092 0.011
10 0.723 0.031
4 3.36 0.062
Cycloheptanol 1.88
i
0.080
k

40 0.214 0.022 0.028 0.029
10 0.899 0.024
4 4.74 0.046
Cyclooctanol 2.53
i
0.026
k
40 0.244 0.038 0.012 0.011
10 1.04 0.049
4 5.60 0.092
a
Logarithms of octanol ⁄ water partition coefficient.
b
Microscopic inhibition constant.
c
Microscopic lipid dissociation binding constant of a
class of more loosely bound activator molecules binding to a number m of receptor binding sites.
d
Sum of squared deviations.
e
For nonlinear
regression with the present kinetic model.
f
Gauss fit.
g
Logistic fit.
h
From Hansch et al. [43].
i
Estimated with ALOGP 2.1 [45].

j
From
Walcher et al. [23].
k
From regression analysis, see Experimental procedures.
J. Altschuh et al. Lipid ⁄ protein interface
FEBS Journal 272 (2005) 2399–2406 ª 2005 FEBS 2401
a single parameter, K
L
. Curve fits with comparable
SSD values were obtained when the data sets were sub-
jected to the standard two-parameter fits of the Gauss
and logistic procedures [34]. As expected, two variable
parameters generally produced a better fit, but the
mechanism-related model was, at least for m ¼ 40, of
comparable quality. Overall, statistical quality was
determined by the quality of the data sets used.
Discussion
Animal narcosis
Narcosis is assessed by some all-or-none quantal
response of individuals in an animal test population.
Theoretically, if all members of the population were
equally responsive, the dose–response curve would be
a sharp step-function with infinite steepness [18,35].
Hill coefficients for anaesthesia of mice, rats, dogs and
humans have been derived and were found to be
between 10 and 22 [15,18,36]. Similarly, high slope val-
ues were reported in a recent study with several mouse
strains when either the loss of righting reflex or a tail
clamp assay was employed [37]. High Hill coefficients

of around 13 were also reported for narcosis of brine
shrimps exposed in artificial sea water [38]. Uniquely
low slope values of narcosis have been reported for
tadpoles. The two independent studies [29,30] re-ana-
lysed here used different tadpole species and experi-
mental conditions. Loss of the righting reflex was
assessed in one study [29], whereas the second study
[30] measured the lack of sustained swimming. Hill
coefficients of around 4.0 [29] and 3.4 [30] were in a
diagnostic window for lipid-dependent enzymes and
receptors [21–23,28] so that our analysis became
possible. The quantal responses of tadpoles seem to
reflect the titration of some as yet unidentified lipid-
embedded target site(s). The much higher slope values
of narcosis obtained with other animal species and
man are attributed to population heterogeneity [35]
and ⁄ or synaptic-block [36] or threshold sensing [18]
mechanisms. It seems unlikely that the low Hill coeffi-
cients for tadpoles can be explained by population
heterogeneity alone, because this would imply an extre-
mely high heterogeneity for tadpoles in contrast to all
other animal species. A search for a particular mech-
anism therefore seems reasonable.
Role of loosely bound lipid activators
The above analysis shows that loosely bound essential
lipid activator molecules could act as an anaesthetic
Fig. 2. Fitted dose–response curves for the
straight-chain and cyclic alcohols of Table 1.
Only the microscopic lipid-binding constant
K

L
was allowed to vary, all other parameters
describing the lipid ⁄ protein interface [23]
remaining constant. The curves for the
nAChR shown on the right were calculated
as described in the Results. The following
fitted curves to the data points (d)for
tadpole narcosis [29,30] are shown on the
left: m ¼ 40; (solid line), m ¼ 10 (dotted
line), and m ¼ 4 (dashed line).
Lipid ⁄ protein interface J. Altschuh et al.
2402 FEBS Journal 272 (2005) 2399–2406 ª 2005 FEBS
target site as an alternative to lipid-free hydrophobic
protein pockets which are currently favoured in the lit-
erature [18,20]. So far, there is only indirect experimen-
tal evidence for such loosely bound lipid activator
molecules. X-Ray and electron diffraction studies have
well documented the opposite situation, namely the
existence of particular tightly bound lipid molecules
such as cardiolipin [1–4]. The lipid-binding sites of the
nAChR or other ligand-gated channel proteins have
not been resolved into subclasses by direct measure-
ment although the sum of the available evidence is
strongly in favour of heterogeneous lipid-binding sites
[19,26,27]. Indirect support comes from the case of a
K
+
-channel-associated peptide, in which mutation of a
single amino acid has been shown by ESR-spectro-
scopy to change lipid specificity [39].

The lipid ⁄ protein interface of the nAChR was used
here as a prototype to explain tadpole narcosis by assu-
ming a preferential displacement of loosely bound lipid
activator molecules. This analysis employed Adair-type
kinetics and microscopic lipid and inhibitor binding
constants. Recent reviews on high-resolution X-ray and
electron diffraction structures agree with a picture of
high diversity in lipid-binding modes [1–4]. The func-
tional importance of individual lipid ⁄ protein binding
sites is illustrated by site-directed amino acid exchan-
ges along transmembrane helices [40]. Amino acid
exchanges near the interface region, or in the lipid
hydrocarbon region, have been shown to influence the
orientation and function of transmembrane proteins
[19,40]. The existence of multiple lipid ⁄ protein binding
sites has led to different conclusions. On the one hand,
it was stated that a full description of binding to such a
heterogeneous surface is an impossibly difficult task [2].
On the other hand, functional assays of reconstituted
proteins are proposed to be most crucial to yield infor-
mation on the specificity of binding of particular vs.
bulk lipids, the number of their binding sites and the
occupancies and the cooperativity of their potential
interactions [1]. In these proposed functional assays,
Adair-type kinetics and the use of microscopic binding
constants appear to be one of the tools to achieve the
required resolution. Methods to directly determine
microscopic lipid-binding constants still remain to be
developed. However, an indirect support of the present
analysis is provided by molecular dynamics simulation

that has recently identified the lipid ⁄ protein interface as
anaesthetic target site of the gramicidin ion channel [41].
Experimental procedures
This study is based on two independent data sets for tad-
pole narcosis by various alcohols [29,30]. The previously
published characterization of the lipid ⁄ protein interface of
the nAChR [23] including the total number of n ¼ 40 lipid-
binding sites of the nAChR [19,42] was adopted, as were
the other kinetic constants previously derived [23]. This
concerns in particular the microscopic lipid dissociation
constant, K
L
, of the nAChR under the first-order assump-
tion of identity of lipid-binding sites. In terms of ‘two-
dimensional’ kinetics, K
L
amounted to 3.75 lipid molecules
per receptor molecule. This corresponded to a bulk lipid
concentration of 46.9 nm [31]. The previously determined
K
I
value of 1-hexanol (5500 molecules per receptor mole-
cule, corresponding to 68.7 lm) [23] is adopted here, while
the K
I
values of the other straight-chain and cyclic alcohols
studied were calculated from a regression equation of the
logarithms of K
I
values [23] vs. the logarithms of octa-

nol ⁄ water partition coefficients, log K
OW
[43]. The regres-
sion is based on data of ethanol, ether, 1-hexanol,
isoflurane, and methoxyflurane: log K
I
¼ )0.759Ælog K
OW
)2.67 (N ¼ 5, r
2
¼ 0.76, SD ¼ 0.44). The higher alcohols
studied in tadpole narcosis [29,30] had log K
OW
values out-
side the range of applicability of the regression and were
therefore not included in the analysis. mathematica 4.1
software was employed for all calculations.
References
1 Pebay-Peyroula E & Rosenbusch JP (2001) High-resolu-
tion structures and dynamics of membrane protein–
lipid complexes: a critique. Curr Opin Struct Biol 11,
427–432.
2 Lee AG (2003) Lipid–protein interactions in biological
membranes: a structural perspective. Biochim Biophys
Acta 1612, 1–40.
3 Bracey MH, Cravatt BF & Stevens RC (2004) Struc-
tural commonalities among integral membrane enzymes.
FEBS Lett 567, 159–165.
4 Palsdottir H & Hunte C (2004) Lipids in membrane
protein structures. Biochim Biophys Acta 1666, 2–18.

5 Marsh D & Horvath LI (1998) Structure, dynamics and
composition of the lipid–protein interface. Perspectives
from spin-labelling. Biochim Biophys Acta 1376, 267–
296.
6 Marsh D & Pa
´
li T (2004) The protein–lipid interface:
perspectives from magnetic resonance and crystal struc-
tures. Biochim Biophys Acta 1666, 118–141.
7 Alvis SJ, Williamson IM, East JM & Lee AG (2002)
Interactions of anionic phospholipids and phospatidylen-
ethanolamine with the potassium channel KcsA.
Biophys J 85, 3828–3838.
8 Sandermann H (2002) High free energy of lipid ⁄ protein
interaction in biological membranes. FEBS Lett 514,
340–342.
9 Sandermann H, McIntyre JO & Fleischer S (1986)
Site–site interaction in the phospholipid activation of
J. Altschuh et al. Lipid ⁄ protein interface
FEBS Journal 272 (2005) 2399–2406 ª 2005 FEBS 2403
d-b-hydroxybutyrate dehydrogenase. J Biol Chem 261,
6201–6208.
10 Loeb-Hennard C & McIntyre JO (2000) (R)-3-Hydroxy-
butyrate dehydrogenase: selective phosphatidylcholine
binding by the C-terminal domain. Biochemistry 39,
11928–11938.
11 Carr HS & Winge DR (2003) Assembly of cytochrome
c oxidase within the mitochondrion. Acc Chem Res 36,
309–316.
12 Haines TH & Dencher NA (2002) Cardiolipin: a proton

trap for oxidative phosphorylation. FEBS Lett 528,
35–39.
13 Seeman P (1972) The membrane actions of anesthetics
and tranquilizers. Pharmacol Rev 24, 583–655.
14 Miller KW (1985) The nature of the site of general
anesthesia. Int Rev Neurobiol 27, 1–61.
15 Franks NP & Lieb WR (1994) Molecular and cellular
mechanisms of general anaesthesia. Nature 367, 607–
614.
16 Franks NP & Lieb WR (1998) Which molecular targets
are most relevant to general anaesthesia? Toxicol Lett
100–101, 1–8.
17 Mihic SJ, Ye Q, Wick MJ et al. (1997) Sites of alcohol
and volatile anaesthetic action on GABA
A
and glycine
receptors. Nature 389, 385–389.
18 Yamakura T, Bertaccini E, Trudell JR & Harris RA
(2001) Anesthetics and ion channels: molecular models
and sites of action. Annu Rev Pharmacol Toxicol 41, 23–
51.
19 Barrantes FJ (2003) Modulation of the nicotinic acetyl-
choline receptor function through the outer and middle
rings of transmembrane domains. Curr Opin Drug
Discov Devel 6, 620–632.
20 Miller KW (2002) The nature of sites of general anaes-
thetic action. Br J Anaesth 89, 17–31.
21 Sandermann H (1993) Induction of lipid–protein mis-
match by xenobiotics with general membrane targets.
Biochim Biophys Acta 1150, 130–133.

22 Sandermann H & Pflugmacher D (1996) Induction of
lipid–protein mismatch by xenobiotics: kinetic coopera-
tivity. Biochim Biophys Acta 1300, 219–225.
23 Walcher S, Altschuh J & Sandermann H (2001) The
lipid ⁄ protein interface as xenobiotic target site: kinetic
analysis of the nicotinic acetylcholine receptor. J Biol
Chem 276, 42191–42195.
24 Pflugmacher D & Sandermann H (1998) The lipid ⁄ pro-
tein interface as a target site for general anesthetics: a
multiple-site kinetic analysis of synaptosomal Ca
2+
-
ATPase. Biochim Biophys Acta 1415, 174–180.
25 Antollini SS & Barrantes FJ (2002) Unique effects of
different fatty acid species on the physical properties of
the Torpedo acetylcholine receptor membrane. J Biol
Chem 277, 1249–1254.
26 Mantipragada SB, Horva
´
th LI, Arias HR, Schwarz-
mann G, Sandhoff K, Barrantes FJ & Marsh D (2003)
Lipid–protein interactions and effect of local anesthet-
ics in acetylcholine receptor-rich membranes from
Torpedo marmorata electric organ. Biochemistry 42,
9167–9175.
27 Miyazawa A, Fujiyoshi Y & Unwin N (2003) Structure
and gating mechanism of the acetylcholine receptor
pore. Nature 423, 949–955.
28 Sandermann H (1982) Lipid-dependent membrane
enzymes. A kinetic model for cooperative activation in

the absence of cooperativity in lipid binding. Eur J Bio-
chem 127, 123–128.
29 Alifimoff JK, Firestone LL & Miller KW (1986) Anaes-
thetic potencies of primary alcohols: implications for
the molecular dimensions of the anaesthetic site. Br J
Pharmacol 96, 9–16.
30 Curry S, Moss GWJ, Dickinson R, Lieb WR & Franks
NP (1991) Probing the molecular dimensions of general
anaesthetic target sites in tadpoles (Xenopus laevis) and
model systems using cycloalcohols. Br J Pharmacol 102,
167–173.
31 Miller KW, Braswell LM, Firestone LL, Dodson BA &
Forman SA (1986) General anesthetics act both specifi-
cally and nonspecifically on acetylcholine receptors. In
Molecular and Cellular Mechanisms of Anaesthetics
(Roth SH & Miller KW, eds), pp. 125–137. Plenum
Press, New York.
32 Forman SA, Miller KW & Yellen G (1995) A discrete
site for general anesthetics on a postsynaptic receptor.
Mol Pharmacol 48, 574–581.
33 Forman SA (1998) Direct interactions of anesthetics
and nonanesthetics with the nicotinic acetylcholine
receptor pore. Toxicol Lett 100–101, 169–178.
34 Pruscha H (1996) Angewandte Methoden der Mathemat-
ischen Statistik. Teubner, Stuttgart.
35 Munson PL, Mueller RA & Breese GR (1996) Principles
of pharmacology. Basic Concepts and Clinical Applica-
tions, pp. 10–11. Chapman & Hall, New York.
36 Woodbury JW, D’Arrigo JS & Eyring H (1975) Physio-
logical mechanism of general anesthesia: synaptic block-

ade. In Molecular Mechanisms of Anesthesia, Progress in
Anesthesiology, Vol. 1 (Fink BR, ed.), pp. 53–91. Raven
Press, New York.
37 Quinlan JJ, Ferguson C, Jester K, Firestone LL &
Homanics GE (2002) Mice with glycine receptor subunit
mutations are both sensitive and resistant to volatile
anesthetics. Anesth Analges 95, 578–582.
38 Takasaki M, Tatara T, Suezaki Y, Shirahama K,
Kamaya H, Ueda I & Totoki T (1991) Effect of inhala-
tion anesthetics on swimming activity of Artermia
salina. J Anesth 5, 287–293.
39 Horvath LI, Knowles PF, Kovachev P, Findlay JBC &
Marsh D (1997) A single-residue deletion alters the lipid
selectivity of a K
+
channel-associated peptide in the
b-conformation: spin label electron spin resonance
studies. Biophys J 73, 2588–2594.
Lipid ⁄ protein interface J. Altschuh et al.
2404 FEBS Journal 272 (2005) 2399–2406 ª 2005 FEBS
40 Miller AS & Falke JJ (2004) Side chains at the mem-
brane–water interface modulate the signaling state of a
transmembrane receptor. Biochemistry 43, 1763–1770.
41 Tang P & Xu Y (2002) 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 99, 16035–16040.
42 Ellena JF, Blazing MA & McNamee MG (1983) Lipid–
protein interactions in reconstituted membranes contain-

ing acetylcholine receptor. Biochemistry 22, 5523–5535.
43 Hansch C, Leo A & Hoekman D (1995) Exploring
QSAR. Hydrophobic, Electronic, and Steric Constants,
Vol. 2. American Chemical Society, Washington, DC.
44 Lang S (1965) Algebra. Addison-Wesley, Reading, MA.
45 Tetko IV, Tanchuk VY, Kasheva TN & Villa AEP
(2001) Internet software for the calculation of the lipo-
philicity and aqueous solubility of chemical compounds.
J Chem Inf Comput Sci 41, 246–252.
Appendix
The Adair approach employing microscopic lipid-bind-
ing constants has previously been taken to explain the
enhanced agonist binding to the nAChR in terms of
the competitive displacement of essential lipid activator
molecules [23].
We generalize and extend this approach by allowing
for different lipid-binding constants in this study. Thus
we consider a macromolecule P with a number n of bind-
ing sites for lipid L as well as for inhibitor I. Because the
sites may no longer be identical for lipid, we have micro-
scopic lipid dissociation constants K
L1
, K
L2
, , K
Ln
.
However, we assume that the sites are identical for small
inhibitor molecules that are assigned a uniform micro-
scopic lipid dissociation constant K

I
.
According to the basic model of lipid dependence
[28], there is an integer a < n with the following prop-
erty: a macromolecule is not functional if fewer than
n ) a binding sites are occupied by lipid, and fully
functional if at least n ) a sites are occupied by lipid.
We derive a formula for the fraction r of functional
macromolecules.
To begin, we give an expression for the total concen-
trations. Consider a fixed binding site (say, of number
j) and denoted by [P
j0
] the concentration of macromol-
ecules with empty site j. According to the mass-action
law, the concentration of molecules with lipid bound
to site j is given by
P
j0
ÂÃ
Á
½L
K
Lj
and the concentration of molecules with inhibitor
bound to site j is given by
P
j0
ÂÃ
Á

½I
K
I
Therefore the total concentration is given by
P
j0
ÂÃ
Á 1 þ
½L
K
Lj
þ
½I
K
I

Because the sites are independent, the total concentra-
tion is equal to the product
P
0
½Á1 þ
L½
K
L1
þ
I½
K
I

Á 1 þ

L½
K
L2
þ
I½
K
I

::: 1 þ
L½
K
Ln
þ
I½
K
I

with [P
0
] the concentration of ‘empty’ molecules (with
no binding site occupied). This expression may be rear-
ranged as
½P
0
Á
X
n
k¼0
Q
k

Á L½
k
with
Q
k
¼ 1 þ
½I
K
I

nÀk
Á S
k
1
K
L1
;
1
K
L
2
:::
1
K
Ln

and S
k
denoting the k-th elementary symmetric poly-
nomial in n variables, thus

S
0
ðx
1
; x
2
; ; x
n
¼ 1;
S
1
ðx
1
; x
2
; ; x
n
Þ¼x
1
þ x
2
þÁÁÁx
n
.
.
.
S
n
ðx
1

; x
2
; ; x
n
Þ¼x
1
Á x
2
ÁÁÁÁÁx
n
Generally, S
k
(x
1
, , x
n
) is formed by taking all
possible products of k distinct factors among the
x
1
, , x
n
, and then summing these up. For more
information see Lang [44].
To verify this expression, note
1 þ
I½
K
I
þ

L½
K
Lj
¼
1
K
Lj
L½þK
Lj
Á 1 þ
I½
K
I

and use the algebraic identity
L½þx
1
ðÞL½þx
2
ðÞÁÁÁL½þx
n
ðÞ
¼ L½
n
þ
X
nÀ1
k¼0
S
nÀk

x
1
; ; x
n
ðÞL½
k
The biochemical interpretation of this rearrangement is
the following: The term
P
0
½ÁQ
k
Á L½
k
is equal to the concentration of those macromolecules
to which exactly k lipid molecules are bound. Thus the
fraction of functional molecules is given by
J. Altschuh et al. Lipid ⁄ protein interface
FEBS Journal 272 (2005) 2399–2406 ª 2005 FEBS 2405
r ¼
P
n
k¼nÀa
Q
k
½L
k
Q
n
j¼1

1 þ
½L
K
Lj
þ
½I
K
I

because [P
0
] cancels in numerator and denominator.
For practical computations (with a usually quite small
compared with n) another transformation is useful:
Divide numerator and denominator by [L]
n
, and set
S ¼ 1 ⁄ [L]. Then
r ¼
P
a
k¼0
Q
nÀk
S
k
Q
n
j¼1
1

K
Lj
þ S þ S Á
½I
K
I

Here the numerator is just the degree a Taylor poly-
nomial of the denominator, which can be deter-
mined using built-in functions of symbolic computation
systems.
If all the K
Lj
¼ K
L
are equal, we recover the for-
mula from Walcher et al. [23] (with q ¼ 1).
In this study we discuss the scenario when there are
two classes of lipid-binding sites, i.e. there is a number
m,0<m £ n, such that m sites have dissociation con-
stant K
0
L
, and the remaining n ) m sites have dissoci-
ation constant K
L
for lipid.
Lipid ⁄ protein interface J. Altschuh et al.
2406 FEBS Journal 272 (2005) 2399–2406 ª 2005 FEBS