*Department of Anaesthesia and Critical Care, Massachusetts General Hospital, 55 Fruit Street, Boston, MA 02114 and the Department of Biological Chemistry and Molecular Pharmacology, Harvard Medical School, Boston, MA 02115, USA*Address for correspondence
Abstract
Br J Anaesth 2002; 89: 1731
Keywords: anaesthesia, general; anaesthesia, binding sites
General anaesthetics gain access to all parts of the body and thus interact at the molecular level with every component in the biosphere including lipids, proteins and even oligosaccharides.43 In order for any of these interactions to cause the various physiological changes that result in general anaesthesia, or its many side-effects, they must occur at clinically relevant concentrations and they must cause a change in the function of the system. Thus, in order to approach the molecular mechanisms of action of general anaesthetics it is necessary to understand both the forces that govern how anaesthetics interact with their sites of action and the mechanisms available to transform these interactions into functional outcomes.
Physicochemical indicators of the nature of the anaesthetic site
Early workers rejected the lock and key model for ligands binding to proteins because the diverse structures of general anaesthetics could not possibly fit a single lock. Instead they adopted a physicochemical approach, attempting to define the nature of that site by seeking correlations between general anaesthetic potency in vivo and various physical properties, particularly those defining interactions between anaesthetics and some other substance. If a poor correlation is observed, as is the case for example with aqueous solubility, one concludes that the nature of the site of pharmacological action is different. The well-known correlation between anaesthetic potency and olive oil solubility reigned as the most powerful in pharmacology until the discovery of opiate receptor binding assays 70 yr later. Olive oil was chosen to mimic the lipoids of the cell,40 49 but after half a century a more systematic approach was introduced that examined solvents on the basis of their solubility parameter.47 This parameter increases with solvent polarity from a low value of 6 for hydrocarbons. General anaesthesia correlates best with solvents with solubility parameters of 1011,45 implying some polarity at the site of action. However, different pharmacological actions of anaesthetics were found to occur at sites of different solubility parameter.47 For example, the convulsant gases appear to interact at a site of lower solubility parameter than anaesthetic gases.11 If there are several sites for anaesthetic actions that differ in polarity, the theory would predict that general anaesthetics of varying polarity would distribute amongst them selectively; the more polar ones favouring the more polar sites, for example. This predicts that the spectrum of action of each anaesthetic will depend on its polarity, and that agents of similar polarity will have similar spectrums of action.
One of the assumptions underlying the physicochemical approach is that when half of a group of animals is anaesthetized with any volatile agent, they are in an isonarcotic state in which their physiology is identical. This is the unitary hypothesis and it is questioned by the above notion that there may be a number of anaesthetic sites exhibiting a spectrum of polarity.55
Recent evidence for multiple sites
Two lines of recent research emphasize these concerns. First, subanaesthetic concentrations of volatile anaesthetics suppress learning. This action may occur at a site distinct from that causing general anaesthesia because a group of non-anaesthetic agents (termed non-immobilizers) can suppress learning in animals at predicted concentrations but are incapable of producing full anaesthesia as measured by immobility in response to a painful stimulus. The interpretation is that suppression of learning and immobility are two effects of anaesthetics that are mediated by different sites. Non-immobilizers are apolar whereas most full general anaesthetics have polar character. Learning suppression then occurs at a less polar site than that producing anaesthesia or immobility.24 The non-immobilizers lack action at the GABAA receptor, which has long been regarded as a common target for all general anaesthetics,74 but they do act on acetylcholine receptors,27 as do most volatile general anaesthetics.
Second, it has now emerged that two clinically used apolar general anaesthetics, xenon28 and cyclopropane,55 do not interact with GABAA receptors. Whilst one must be cautious about drawing conclusions until a wider range of subunits has been examined, nonetheless these findings are a challenge to the established view that action on GABAA receptors is central to an agents ability to produce general anaesthesia.74 These gases do, however, inhibit acetylcholine and NMDA receptors. Might these latter receptors have sites of lower polarity than those on GABAA receptors? While the location of anaesthetic sites on these receptors is not unambiguously established, the outer end of the cation channel of the acetylcholine receptor is lined with hydrophobic residues, whereas that in the GABAA receptor exhibits more polarity.
One interpretation of the facts that are currently available to us is that general anaesthesia can be produced by agents acting at several targets to reduce the overall level of excitability. General anaesthetics certainly act on the superfamily of receptors that includes GABAA, glycine, nicotinic acetylcholine and serotonin 5HT3 receptors and on the structurally unrelated NMDA receptors. The first two are inhibitory and the other three excitatory. General anaesthetics tend to enhance the activity of inhibitory receptors and inhibit the activity of excitatory receptors. All the more polar anaesthetics act on the inhibitory receptors and on acetylcholine receptors, whereas the less polar ones act on acetylcholine and NMDA receptors. A working hypothesis might be that more polar general anaesthetics enhance inhibition at GABAA and glycine receptors and inhibit excitation at acetylcholine receptors, whereas the apolar general anaesthetics make up for their lack of action on GABAA receptors by further inhibiting excitation at NMDA receptors.
An alternative explanation is that none of these receptor types is central to general anaesthesia and that the true target has yet to be discovered. Certainly genetic studies on anaesthesia in Caenorhabditis elegans have not implicated any of these ligand-gated ion channels.14 61 Nonetheless, the ligand-gated ion channels are acted upon by clinical concentrations of general anaesthetics and these interactions must give rise to some physiological sequelae. Working out the molecular basis of general anaesthetic action on these channels is a necessary step in understanding the basis of selectivity between targets and is a start down the road to rational drug design.
The anaesthetic-sensitive ligand-gated superfamily of ion channels
Structure
The difficulties of applying x-ray crystallography and NMR to membrane proteins, let alone ones as large as the ligand-gated superfamily of ion channels, means that there is no direct and definitive evidence at the atomic level for anaesthetic binding sites on them. Such detailed evidence is only found on the readily crystallizable soluble proteins that we will consider later in this review.
The ligand-gated ion channels are thought to have five subunits, each contributing the second of four transmembrane helices (M2) to lining a pore along a central axis67 (Fig. 1A). This picture is based largely on the acetylcholine receptor (nAcChoR), which has a unique advantage over the other members of this superfamily. The subtype found in Torpedo electroplaque membranes has a specific activity 10 000 times higher than that of any other member. Because of its abundance, biochemical and structural studies have been performed on the nAcChoR and the results extrapolated by homology to design experiments on other members of the family. The overall structure of the nAcChoR in the resting, open and desensitized conformations has been studied by cryoelectron microscopy to a resolution sufficient to reveal some of the secondary structure. The ion-conducting pore can be seen to be lined by -helices but the secondary structure of the remaining transmembrane structures remains speculative (Fig. 1B).
|
Allosteric behaviour
Many proteins can coexist in a number of conformations of similar energy. The most highly populated conformation will be the one with lowest free energy. So finely poised are these conformations that binding of a small ligand selectively to a minor conformation may stabilize it sufficiently that it now becomes the predominant conformation. The ligand-gated ion channels of receptors are normally in a resting non-conducting state, which has low affinity for agonist, in equilibrium with a small proportion in the desensitized state, which has high affinity for agonist but does not conduct when stimulated.17 Upon binding two agonist molecules of molecular weight 102 Da, these
250 kDa receptors convert to the conducting conformation in less than a millisecond. In the continuing presence of agonist, a third conformation slowly accumulates. This is the desensitized conformation, which is non-conducting. Table 1 illustrates this for the acetylcholine receptor.9
|
Effects of general anaesthetics
Studies focused on these ion channels have led to significant advances in our understanding of how general anaesthetics act. However, whilst one might expect that individual ion channels would have a simpler pharmacology than intact animals, it turns out paradoxically that even on a single protein general anaesthetics exert multiple effects.
The major actions of general anaesthetics on the ligand-gated ion channel superfamily are as follows. First, on the cation channels most, but not all, general anaesthetics inhibit agonist-induced ion currents, depressing the maximum of the agonist concentrationresponse curve without shifting its midpoint, as expected for insurmountable action. Inhibition may also occur on the anion channels (GABAA, glycine) but only at much higher concentrations.50 Second, general anaesthetics cause a leftwards shift in the agonist concentrationresponse curve to lower concentrations. This occurs in the anion channels and with some of the smaller agents in the cation channels. Third, many general anaesthetics enhance agonist-induced desensitization in all members of the family. Fourthly, intravenous agents bind to discrete sites which are allosterically linked to the agonist site.48 Many of these actions are elaborated upon by Dilger17 and here we will emphasize only mechanistic information. Our working model will be that there are a number of different anaesthetic sites on this superfamily of receptors, and that the relative affinity of these sites for general anaesthetics varies both between members of the family and between conformational states of each receptor.
Evidence for allosteric anaesthetic binding sites
Binding studies
The high specific activity of the nAcChoR in Torpedo membranes made it possible to directly determine by pharmacological binding assays that there is a [14C]barbiturate binding site on the acetylcholine receptor. This remains the only demonstration of its type and it is worth considering in some detail because it illustrates the principle of allosteric action. The enantiomers of pentobarbital bind stereoselectively to nAcChoR membranes from Torpedo.57 About half the binding is displaceable. Of the various proteins available in these membranes, the barbiturate bound to the nAcChoR, addition of agonist reducing the amount of binding. This was neither a competitive nor a non-specific interaction because the highly specific competitive antagonist -bungarotoxin had no effect on barbiturate binding but prevented the effect of agonist.18 When [3H]acetylcholine was titrated causing progressively more of the receptors to be driven into the desensitized state, [14C]amobarbital binding decreased in parallel with the state change. Thus, amobarbital binds with higher affinity to the resting than to the desensitized state, whereas acetylcholine binds with higher affinity to the desensitized state (Fig. 2). It follows that the two binding sites are coupled by an allosteric interaction. Such an interaction could lead to either an increase or a decrease in binding of the allosteric ligand. In this case a decrease is seen, often called a negative heterotropic interaction. The strength, and indeed even the direction, of this allosteric interaction can depend on the ligands structure. For example, secobarbital bound with equal affinity to the resting as to the desensitized state, whereas thiopental favoured the desensitized state. This means that the structureactivity relationships governing interaction between the barbiturates and their site are different for each conformation. Furthermore, it can be inferred from their potency for inhibiting the open channel that a third set of structureactivity relationships exists for the open state.16
|
Kinetic studies
Evidence for sites on open channels must come both from kinetic information obtained from electrophysiological data and from biochemical data obtained by very rapid cation flux experiments conducted on a millisecond time scale. The former are summarized in Dilgers article herein.17 The biochemical data on Torpedo membranes also point to an inhibitory site.69 First, the Hill coefficients for inhibition by alkanols longer than hexanol are close to 1, suggesting that a single site or a set of sites of equal affinity mediate inhibition. Second, the dependence of apparent Ki on alkanol chain length is steeper than that of lipid solubility or lipid disordering. Third, inhibitory activity is constrained sterically; only alcohols with molecular volumes below 340 Å3 or above
100 Å3 are able to inhibit the nAcChoR.70
Whilst this evidence is persuasive, more definitive proof is provided by the strategy of exposing the open channel to two anaesthetic agents simultaneously and examining the interactions between them.71 Essentially, if one agent occupies half the inhibitory sites, the second will then have only half the probability of finding an inhibitory site. If, on the other hand, the two agents act on a large site or by dissolving in the lipid bilayer, the presence of one will not interfere with the action of the other. By titrating octanol and heptanol against each other it was possible to show that they acted at distinct sites on the open state of the nAcChoR. This strategy should work for studying the enhancement of inhibitory currents by general anaesthetics but it has rarely been employed. In one example, the interaction between pentobarbital and propofol was examined on the glycine receptor and evidence for mutually exclusive action reported.51
There are two other interesting features that suggest mechanisms of alcohol action on the acetylcholine receptor. First, as mentioned, the Hill coefficients for channel inhibition are 1, but for the smallest alcohols (below hexanol) they become 2 (Fig. 3). One might cautiously interpret this to mean that two molecules of the smaller agent might be accommodated in the channel inhibition site. This behaviour is reminiscent of the luciferase binding site, which accommodates two molecules of shorter alcohols but only one of longer ones. At higher concentrations where alcohols stabilize the desensitized state of the acetylcholine receptor, they do so with even higher Hill coefficients.
|
In the ligand-gated ion channel superfamily, the anaesthetic sites characterized by site-directed mutagenesis are all near the gating machinery. Mutations at many of these sites alter gating as well as anaesthetic action.26 37 This Cassandrian view is strengthened by recent studies showing that widespread mutations on the acetylcholine receptor unassociated with any binding site can affect gating.31
Photoaffinity labelling
Whilst the evidence for the sites given above is strong and often self-consistent, structural data are required both to definitively prove the existence and to define the location of the sites. Three-dimensional structures are unlikely to be available for some time. A technique that can provide information at the sequence, or primary structure, level is photoaffinity labelling. It is complementary to site-directed mutagenesis and has been successfully employed to identify the amino acids that form the agonist and local anaesthetic sites on the nAcChoR and the benzodiazepine sites on the GABAA receptor.19 33 62 Such an approach has been pioneered for general anaesthetics using [14C]halothane.10 21 However, this agent has low radioactive specific activity, poorly understood photochemistry and must be photoactivated at 250 nm, a wavelength likely to interact with the target protein. For example, of the seven halothane sites found by crystallography on human serum albumin (HSA),5 [14C]halothane photolabels only the one of low affinity.20 Aromatic diazirines have very good photochemical properties and have been widely applied in other fields.4 7 Recently, a number of alkanols containing diazirine have been characterized. They are unexceptional general anaesthetics that obey the MeyerOverton rule, enhance GABA-induced currents and inhibit acetylcholine-induced currents. One of them has been synthesized in a tritiated form and used to define sites on the
-subunit of the nAcChoR from Torpedo.34
Anaesthetic sites on the nicotinic acetylcholine receptor
[3H]azioctanol photolabels the nAcChoR selectively at the -subunit. This is rather reassuring because a priori one might expect a lot of random photoincorporation when the diazirine is activated by shining light at 360 nm onto the sample. In fact, the carbenes so formed are relatively short lived, and often react intramolecularly. Thus, it is only a small fraction of the agent that is incorporated into the protein. Edman degradation techniques were employed to locate the binding sites on the
-subunit. The location of these is indicated by squares on the structural diagram of the subunit structure in Figure 1. The main site of photoincorporation was at the top of the M2 channel-lining region. A second site was in the lipidprotein interface region of M4, and much lower levels were incorporated in the N-terminal region.53 We will consider each of these below in order of increasing interest.
Two of the three residues that were identified in M4 are also photolabelled by the extremely hydrophobic photolabel TID, placing them in the lipidprotein interface, but the third appears to occur before the predicted start of M4 and may be on a turn before that helix. This selective labelling of the lipidproteinwater interface region is reminiscent of the situation in the model channel gramicidin.63 Theoretical mechanisms exist for how such interactions might have functional effects on proteins.8 These are based on the assumption that a membrane proteins topology changes when its conformation changes and that actions of general anaesthetics on lipid bilayer lateral pressure can therefore change the relative free energy of conformational states. An alternative notion, possibly applicable to steroid anaesthetics, is that anaesthetics bind at specific lipid-binding sites in the lipidprotein interface.12 56
The residues in the N-terminus are known to be in the agonist binding pocket. They can be located on the recent crystal structure of the acetylcholine binding protein and are very close to the subunitsubunit interface, and not far from residues detected by photolabelling with halothane.10
Photoincorporation was increased 10 fold at the M2 site when agonist was used to desensitize the receptor. At the highest concentrations used, [3H]azioctanol induced desensitization itself, enhancing photoincorporation in the absence of agonist. Some agonist-sensitive binding was also observed on the ß-subunit. The residue labelled was Glu-262, which is on the C-terminal of M2 and 20 amino acids from the charged residue located at the N-terminal of the pore; hence it is referred to as being in the 20' position (Fig. 4). It may seem paradoxical that a charged amino acid is labelled but glutamate also contains two methylene groups, and crystallographic studies show that such interactions are not unusual.5
|
Photolabels as molecular rulers
In a recent study two geometric isomers of octanol bearing a diazirine group on either the third or seventh carbon (3- and 7-azioctanol, respectively) were used to locate and delineate an anaesthetic site on adenylate kinase.1 This enzyme was chosen for a number of reasons. First, halothane inhibits muscle adenylate kinase, an action once implicated in general-anaesthetic-induced malignant hyperthermia.58 Second, adenylate kinases allosteric states have been studied in incredible detail, the structures of many conformations having been determined at high resolution.68
Adenylate kinase, which photoincorporated both 3- and 7-azioctanol at a molar ratio of 1:1 as determined by mass spectrometry (MS), was subjected to tryptic digest and the fragments separated and sequenced by HPLC/MS/MS. 3-Azioctanol photolabelled His-36, whereas its isomer, 7-azioctanol, photolabelled Asp-41. Inspection of the known structure of adenylate kinase showed that the side chains of these residues are within 5 Å of each other. This distance matches the separation of the 3- and 7- positions of an extended aliphatic chain (Fig. 5). The interpretation is that azioctanol binds its site on adenylate kinase with its aliphatic chain in an extended conformation.1
|
ATP.Mg2+ + AMP Mg2+ + 2ADP.72
All adenylate kinases share a similar three-domain structure (Fig. 5). They have a central domain (CORE) consisting of a five-stranded parallel ß-sheet surrounded by -helices, an
-helical nucleotide monophosphate binding domain (NMPbind) and a region that covers the active site during catalysis (LID).68 The alkanol site spans two domains of adenylate kinase. His-36 is part of the CORE domain and Asp-41 belongs to the NMPbind domain. Upon binding the substrate-mimicking inhibitor adenosine-(5')-pentaphos pho-(5')-adenosine (Ap5A), the NMPbind domain rotates relative to the CORE domain, closing up the alkanol binding pocket and reducing the photoincorporation of [3H]3-azioctanol by 75%, consistent with the picture of allosteric action discussed above.
Future directions for photolabelling
Photolabelling techniques are currently providing information about the location of general anaesthetic sites on membrane proteins that cannot be obtained by any other technique. Unfortunately, Edman degradation techniques require milligrams of protein, particularly when the sequences under study are composed of hydrophobic amino acid residues, and the techniques cannot be applied to neuronal ion channels because these express poorly. The latter issue is an important roadblock. Most electrophysiologists are happy with a few channels, and so have not attempted high-level expression of receptors. Even if high-level expression is achieved, the issue of purification will remain. Thus, a good strategy for photolabelling will be to develop expression systems with higher yield in parallel with exploring more sensitive sequencing techniques. In the latter regard, the use of MS provides many advantages.
Because different classes of general anaesthetic may have different sites, it is important to synthesize diazirine derivatives of volatile and intravenous general anaesthetics. A number of groups are making progress on this, analogues of halothane,23 etomidate and steroids13 being in various stages of development.
Nature of anaesthetic binding sites
Nature of sites on proteins
The tools of pharmacology, biochemistry and site-directed mutagenesis have all yielded evidence that points strongly to the existence of general anaesthetic sites on excitable proteins. These proteins are mainly membrane proteins. It turns out to be extraordinarily difficult to crystallize membrane proteins, and even when crystallization is achieved the crystals often do not yield structures of the highest resolution. This resolution is necessary if the interactions of small anaesthetic molecules with the proteins structures are to be resolved in detail. Thus, the best information on general anaestheticprotein interactions comes from structural studies of soluble proteins, most of which are of no conceivable relevance to general anaesthesia. While one may regret the lack of pharmacological relevance, general principles derived from such model, or surrogate, systems should apply also to membrane proteins.
Protein sites for small molecules
The xenon complex with myoglobin was the first anaestheticprotein interaction to be characterized in detail.60 64 At a pressure of 7 bar, one major (almost fully occupied) and three secondary (half-occupied) sites were found. The first site is highly occupied at clinical concentrations. Xenon binds to it with very little perturbation of the surrounding molecular structure. Cyclopropane and dichloro methane have also been shown to bind to the major xenon-binding site in myoglobin but their greater size distorts the surrounding protein structure and causes rearrangement of some amino acid side chains.59
Crystallographers have long had an interest in xenon because, as a heavy atom, it helps them do the necessary phasing that is required to solve structures using x-rays. It has been claimed that as many as 40% of proteins have cavities suitable for binding xenon.39 There is thus a large body of crystallographic data barely mined by anaesthesia researchers. Xenon binds to intramolecular as well as to intermolecular sites, to inaccessible cavities, as well as to exposed pockets and even into channel pores.52 A cavity is a region in a protein that is not occupied by protein atoms and that is entirely closed off from the surrounding aqueous phase. How do anaesthetics gain access to such cavities deep in the interior of proteins? Analysis of x-ray data indicates that in the interior of proteins, atoms make only small (0.250.5 Å) excursions on the picosecond time scale from their mean positions, whereas on the proteins surface the amplitude of such fluctuations may be several angstroms and they occur on the micro- to millisecond time scale. These fluctuations allow rapid access of small molecules with a rate constant of 106107 M1 s1, comparable to those at which inhibitors act on neuronal ion channels. In contrast, small molecules access surface-exposed pockets at diffusion limited rates (108109 M1 s1).25 Proteins resist forming cavities because of the high free energy cost incurred, so these cavities are probably important for the conformational flexibility of proteins, a hypothesis that is supported by the observation of an overall reduction in thermal fluctuations (x-ray temperature factors) of the protein upon xenon binding, probably caused by a restriction in the number of conformational states.
Intermolecular forces between small molecules and proteins
Intermolecular forces govern the free energy of interaction of general anaesthetics with proteins. The forces experienced inside a protein are complex and vary from location to location on an atomic scale. For example, the pKs of charged amino acids can vary 3- to 4-fold depending on the environment at their location. Consequently, the forces experienced by an anaesthetic are quite complex and are very difficult to calculate. Nonetheless, the forces between two isolated uncharged molecules are fairly well understood and provide us with some good rules of thumb.35 39 65
The interaction energy between two apolar (for example xenon) or dipolar (for example chloroform) molecules decreases (attraction increases) as (1/r6) where r is the separation between them. Often called van der Waals interactions, they are all electrostatic, essentially dipolar, in origin and consist of dispersion (induced-dipoleinduced-dipole), dipoleinduced-dipole and dipoledipole interactions. Dispersion forces arise from the electron clouds orbiting an atom. Although their distribution is uniform when averaged over time, at any moment the non-uniform distribution leaves a small net dipole that will interact with similar dipoles in neighbouring atoms. Dispersion forces tend to predominate. Even in CH3Cl, which has a permanent dipole of 1.87 Debye, they provide two-thirds of the interaction energy. The repulsive interaction between two molecules is essentially quantum mechanical in origin. Empirically, it can be described as depending on (1/r12). This is a very steep function of separation approximating a hard sphere interactionthe molecules collide much like billiard or snooker balls.
The total intermolecular pair potential w(r), which is the sum of the attractive and repulsive interaction energies, depends on distance as:
w(r) [(
/r)12 (
/r)6]
where is the diameter of the molecules and r is the separation of their centres. This is illustrated in Figure 6. As the two approach, little attraction occurs until their hard sphere surfaces are separated by one diameter (
) or their centres are separated by 2
. The maximum interaction does not occur until the centres are separated by 1.12
. On closer approach, repulsion increases rapidly and the interaction becomes unfavourable as the hard spheres clash. The diameters of a xenon atom and a methane molecule are 4.3 and 4.0 Å, respectively, so the strength of the interaction is maximum when they approach to within
0.5 Å and is negligible beyond 5 Å. The actual size of the interaction will increase with the atoms polarizability, but no emphasis should be put on calculating its absolute magnitude because the situation in a protein pocket is so complex that it might be misleading.
|
Furthermore, transfer of a gaseous anaesthetic from the gas phase to a suitable pocket in a protein involves confining it in a small space. This always carries a significant entropic cost, roughly the same for all apolar gases, that partially offsets the interaction energy decrease shown in Figure 6. Dipolar gases may lose the ability to rotate freely, adding further to this entropic cost.
While the situation seems complex, some remarkably robust conclusions can be drawn. First, as Figure 6 shows, all interaction potentials exhibit the same repulsive interaction. When the centres are separated by one diameter (i.e. the spheres are touching), there is no net interaction energy and further approach can only occur at tremendous cost. This means that an anaesthetic will be unable to access a pocket that is too small for it unless that pocket has some flexibility. Second, the maximum attraction does not occur until very close approach is achieved. At this point the separation of the spheres surfaces are 12 and 20% of a diameter for a (1/r6) and a (1/r2) dependence, respectively. The corresponding half-maximum interactions occur at 40% and 95% of a diameter. Thus, strong interactions require close approach between an anaesthetic and the surrounding amino acid residues. Therefore, by assessing the size of a binding pocket and the polarity of the residues exposed on its surface, reasonable guesses about its ability to interact with anaesthetics can be made. This analysis also suggests why polarity is a useful property of general anaesthetics. Polar anaesthetics can exert favourable interactions over a longer range than can apolar ones and are therefore less dependent on the goodness of fit in a pocket.
Detailed examination of an internal cavity
The issue of what factors govern binding of small molecules to cavities inside proteins has been systematically examined by Matthews and colleagues. Their studies provide useful insights into the characteristics a site needs if the MeyerOverton relation is to govern its interaction with small molecules and into the effects of site-directed mutagenesis on binding pockets. The model employed for convenience is the bacteriophage T4 lysozyme. It contains a small cavity of volume 48 Å3 situated 7 Å from the surface, and two smaller cavities of 1415 Å3. 54 Argon, krypton and xenon form a homologous series with the van der Waals volumes of 29, 35 and 45 Å3, respectively. All three bind to the large site but not to the smaller sites because the free energy cost of distorting the proteins structure to enlarge the smaller cavities is not repaid by the additional anaestheticprotein interaction energy. This steric exclusion suggests that only those proteins with large enough pre-existing internal cavities will have the potential to bind general anaesthetics. A recent analysis of a randomly selected group of proteins of known structure shows that cavities of 4050 Å3 are relatively common, whereas cavities the size of more typical general anaesthetics (say 150 Å3) occur in some 1015% of cases.22 This is a surprisingly large number, and one might ask why general anaesthetics do not have even more side-effects than they do. One explanation is that in many cases the affinity for general anaesthetics is insufficient for significant occupancy to occur under clinical concentrations. Another explanation might be that when binding does occur it does not always result in a change in the proteins function.
In T4 lysozyme, mutation of a leucine residue in the large cavitys wall to the smaller amino acid alanine (L99A mutant) eliminates an isopropyl group and expands that cavity to a volume of 178 Å.3 The dissociation constants of the gases in this enlarged cavity were estimated from partial pressure titrations to be about twice their anaesthetic potencies: 2,
8 and
32 atm for xenon, krypton and argon, respectively. (The anaesthetic partial pressures are 0.95, 4.5 and 18 atm respectively.) The engineered cavity is large enough theoretically to accommodate four xenon atoms with a squeeze, or six argon atoms, but in fact the noble gases are not randomly distributed in the cavity. Instead, they tend to occupy three highly preferred sites that are numbered here in order of decreasing affinity. Site 1 is created by the mutation, site 2 is essentially the same as in the wild-type site and site 3 is found between sites 1 and 2, giving an approximately collinear arrangement of three equally spaced sites.54 Sites 1 and 2 are separated by 5.4 Å. Even the smallest gas argon has a radius of 1.9 Å, precluding binding at the central site when the peripheral sites are occupied. Thus, the average occupancy of site 3 decreases as overall occupancy increases. Simulations indicated that in addition to the favourable binding energy for site 1, the second xenon bound with positive cooperativity to site 2. This was not caused by xenonxenon interactions because the separation was too great, but it might have resulted from a deformation of site 1 during binding. Such double occupancy of cavities appears to be quite common, being observed for volatile agents in luciferase and HSA (see below).5 29
Of relevance to more conventional general anaesthetic agents, the entropic cost of confining xenon to the site was less unfavourable than in the case of benzene because the latter lost not just its lateral translational freedom but also its freedom to rotate.39 This suggests that the more polar anaesthetics, which might benefit from complimentarity in the binding pocket (e.g. alignment of dipoles or orientation of hydrogen bonds), will pay a small entropic penalty relative to less polar agents.
Although, as expected, the binding affinity increased with the polarizability of the inert gases in the order argon, krypton, xenon, exceptional behaviour was observed in one engineered pocket where krypton exhibited a higher affinity than xenon. The origin of this effect might be a better fit of krypton in the pocket and steric exclusion of xenon, but the investigators were unable to rule out long-distance allosteric effects caused by binding at other sites.54 Nonetheless, the example is salutary because it raises the possibility that not all actions of general anaesthetics need to scale as multiples of anaesthetic potency.
A good test of the ability of the MeyerOverton relationship to predict binding to such internal cavities was provided by a study of alkylbenzene binding to the L99A mutant of T4 lysozyme.46 The benzene moiety was found to bind in xenon site 1, whilst site 2 bound the alkylbenzenes side chain. In site 1 the amino acid side chains were rather rigid and therefore capable of discriminating between ligands of different shape. At this site, geometric isomers created by moving methyl and ethyl groups around the benzene ring bound with affinities that were uncorrelated with their octanol/water partition coefficients because the geometric constraints of the site were not modelled by the solvent interactions. In site 2 the amino acid side chains were more deformable, allowing some promiscuity in binding. In one series of compounds where successive methylenes were added to the aliphatic chain of toluene, extending it to n-butyl, binding affinity did correlate with octanol/water partition coefficients. However, even here steric factors were important, with tert-butyl side chains being excluded from the site. One may conclude that the best correlations with octanol partitioning are likely to be found in sites that can deform to accommodate a small change in a ligands size without incurring a high energy cost.
A further factor governing binding was the polarity of the ligand. The cavity in the L99A form of T4 lysozyme was lined with apolar amino acid residues incapable of hydrogen bonding to ligands. The site excluded water and other polar molecules, such as ethanol and even chloroform.
The effects of mutations designed to alter the dimensions of cavities on the proteins structure are of importance to those designing strategies for site-directed mutagenesis.26 42 One of the issues is whether the mutation alters just the binding properties of the general anaesthetic, as is usually assumed, or whether it also alters the proteins structure, perhaps altering the kinetics governing interconversion between conformational states (e.g. gating) that would have the potential to complicate interpretation. In T4 lysozyme, enlarging a cavity through large-to-small mutations of amino acid residues in the cavity wall caused the protein structure to contract slightly so that the gain in volume was not as large as expected. This implies a change in the overall flexibility of the protein, which might affect function. However, binding was little affected because this contraction in cavity volume was reversed upon binding a non-polar ligand.3 More significantly, making large cavities requires particular caution. In one of a series of double mutations, an -helix neighbouring the cavity rearranged into a non-helical structure, collapsing into the cavity and preventing ligand binding.
Clinical general anaesthetics bound to proteins
The number of high-resolution structures containing general anaesthetics of clinical relevance is very limited. So far luciferase has been crystallized only in the conformation that has low affinity for general anaesthetics, and so HSA provides the highest resolution structures.5 29 This heart-shaped protein contains 585 amino acids organized into three homologous domains (labelled IIII), and each domain consists of two subdomains (A and B) that share common structural elements (Fig. 7). Its structure was determined to high resolution with either halothane or propofol bound. The smaller halothane bound at eight sites, the larger propofol at only two. All anaesthetic binding sites were preformed pockets or clefts capable of binding fatty acids.
|
|
It has been suggested that propofol has a different site on the GABAA receptor from isoflurane, which itself shares a site with halothane.36 38 It is thus of interest that only one of the two propofol sites in HSA bound halothane. Presumably halothane finds more cavities because of its smaller size, but the reasons for this selectivity have not been laid out. Nor was it possible to rationalize the relative affinity of the halothane sites on the basis of their amino acid composition.5
A question of concentration
One of the philosophical issues surrounding the use of x-ray crystallography is the high concentrations of general anaesthetics employed, necessary in order to ensure full occupancy of binding pockets. They would seem to be entirely justified currently because the proteins available to be studied are not of direct pharmacological significance and they are studied to obtain an understanding of proteinanaesthetic interactions. A more difficult case will arise when a relevant target is under study. One would expect in such a case that the apparent affinity of the site will be established by kinetic and pharmacological studies, long before crystallization is successful. These studies will then define the permissible concentrations for crystallization, which will need to be at least 10 times the dissociation constant to approximate saturation. If the dissociation constant is comparable to clinical concentrations (probably a worst-case scenario), would 10 times higher concentrations harm the proteins structure? Probably not because, in the case of HSA, good crystals were obtained up to 2.5 mM halothane (>10 times clinical concentrations), and in the stabilizing presence of fatty acids up to 10.5 mM, a solution which is more than half saturated.
How do general anaesthetics act on proteins?
Binding to a site on a protein at clinical concentrations is not a sufficient condition for anaesthetic action. The bound agent must interfere with the proteins function in some way. Older theories postulated that anaesthetics unfolded, or denatured, protein structure, but such actions occur at such high concentrations as to be irrelevant.
Allosteric mechanisms were invoked above to account for actions on the ligand-gated ion channel superfamily. The structural basis for these actions is not currently known, although in the case of the channel inhibition site, the central ion pore located at the junction of all five subunits clearly provides a site whose topology must perforce change as the receptors conformation changes.9 A similar principle seems to be followed by adenylate kinase where the octanol site spans two domains of the protein that move relative to each other during catalysis.1 In the case of the putative site between the first, second and third transmembrane segments of the GABAA receptor, the situation is less clear.32
Anaesthetics inhibit luminescence in both bacterial and firefly luciferase by directly competing with 1-decanal15 41 and luciferin,30 respectively, both cofactors that bind to well-defined sites. These binding sites accommodate a remarkable range of volatile anaesthetics at clinical concentrations but tend to exclude intravenous agents. Thus, general anaesthetics here interact with rather well-defined protein clefts or pockets, displacing a ligand essential for the enzymes function. Although they may also interact allosterically, and in the case of the firefly luciferase do so,29 this is not the critical action.
In HSA, propofol binds to one of the more important drug binding sites, providing an example of competitive interaction that may be of clinical relevance.5
Inhibition by competition with a natural ligand has some interesting properties. The apparent potency of an anaesthetic will depend upon the concentration of that ligand, decreasing as the latter increases. Thus, if the concentration of the ligand in a cell depends on the cells activity, so will the sensitivity of the protein to anaesthetics. For example, at low, but not at high, ligand concentrations, the activity of the protein will be reduced by general anaesthetics competing effectively for the ligand. If the activity of the protein is under independent feedback control, it is possible that the anaesthetic-induced inhibition might be only transitory, being overcome with time as the concentration of ligand is increased. Well-characterized examples of such actions remain to be found however.
Summary
The molecular nature of the site of general anaesthesia has long been sought through the process of comparing the in vivo potencies of general anaesthetics with their physical properties, particularly their ability to dissolve in solvents of various polarities. This approach has led to the conclusion that the site of general anaesthesia is largely apolar but contains a strong polar component. However, there is growing evidence that several physiological targets underlie general anaesthesia, and that different agents may act selectively on subsets of these targets. Consequently research now focuses on the details of general-anaestheticprotein interactions. There are large amounts of structural data that identify cavities where anaesthetics bind on soluble proteins that are readily crystallizable. These proteins serve as models, having no role in anaesthesia. Two problems make studies of the more likely targetsexcitable membrane proteinsdifficult. One is that they rarely crystallize and the other is that the sites have their highest affinity for general anaesthetics when the channels are in the open state. Such states rarely exist for more than tens of milliseconds. Crystallographers are making progress with the first problem, whilst anaesthesia researchers have developed a number of strategies for addressing the second. Some of these (kinetic analysis, site-directed mutagenesis) provide indirect evidence for sites and their nature, whilst others seek direct identification of sites by employing newly developed general anaesthetics that are photoaffinity labels. Such studies on acetylcholine, glycine and GABA receptors point to the existence of sites located within the plane of the membrane either within the ion channel lumen (acetylcholine receptor), or on the outer side of the -helix lining that lumen (GABAA and glycine receptors). Bound anaesthetics generally exert their actions on ion channels by binding to allosteric sites whose topology varies from one conformation to another, but definitive proof for this mechanism remains elusive.
Acknowledgements
This work was supported in part by the Department of Anaesthesia and Critical Care, Massachusetts General Hospital and Harvard Medical School and in part by a grant from the National Institute for General Medical Sciences (GM5848). I interact with an excellent group of colleagues at Massachusetts General Hospital and Harvard Medical School. They include George Addona, David Chiara, Jon Cohen, Stuart Forman, Shaukat Husain, Robert Mihalek, Megan Pratt, Douglas Raines, Warren Sandberg and Thilo Stehl.
References
1 Addona GH, Husain SS, Stehle T, Miller KW. Geometric isomers of a photoactivatible general anaesthetic delineate a binding site on adenylate kinase. J Biol Chem 2002; in press
2 Arias HR, McCardy EA, Gallagher MJ, Blanton MP. Interaction of barbiturate analogs with the Torpedo californica nicotinic acetylcholine receptor ion channel. Mol Pharmacol 2001; 60: 497506
3 Baldwin E, Baase WA, Zhang X, Feher V, Matthews BW. Generation of ligand binding sites in T4 lysozyme by deficiency-creating substitutions. J Mol Biol 1998; 277: 46785[ISI][Medline]
4 Bayley H. Photogenerated Reagents in Biochemistry and Molecular Biology. Amsterdam, New York, Oxford: Elsevier, 1983
5 Bhattacharya AA, Curry S, Franks NP. Binding of the general anaesthetics propofol and halothane to human serum albumin. High resolution crystal structures. J Biol Chem 2000; 275: 387318
6 Brejc K. Crystal structure of an ACh-binding protein reveals the ligand-binding domain of nicotinic receptors. Nature 2001; 411: 2618[ISI][Medline]
7 Brunner J. New photolabeling and crosslinking methods. Annu Rev Biochem 1993; 62: 483514[ISI][Medline]
8 Cantor RS. Breaking the MeyerOverton rule: predicted effects of varying stiffness and interfacial activity on the intrinsic potency of anaesthetics. Biophys J 2001; 80: 228497
9 Changeux JP, Edelstein SJ. Allosteric receptors after 30 years. Neuron 1998; 21: 95980[ISI][Medline]
10 Chiara DC, Eckenhoff RG, Dangott LJ, Cohen JB. A binding site for the volatile anaesthetic halothane within the nicotnic acetylcholne receptor identified by photoaffinity labeling. Biophys J 2000; 78: 360a
11 Cohen S, Goldschmid A, Shtacher G, Srebrenik S, Gitter S. The inhalation convulsants: a pharmacodynamic approach. Mol Pharmacol 1975; 11: 37985[Abstract]
12 Corbin J, Wang HH, Blanton MP. Identifying the cholesterol binding domain in the nicotinic acetylcholine receptor with [125I]azido-cholesterol. Biochim Biophys Acta 1998; 1414: 6574[ISI][Medline]
13 Covey DF, Evers AS, Mennerick S, Zorumski CF, Purdy RH. Recent developments in structureactivity relationships for steroid modulators of GABA(A) receptors. Brain Res Brain Res Rev 2001; 37: 917
14 Crowder CM. Behavioral effects of volatile anaesthetics in Caenorhabditis elegans. Anesthesiology 1996; 85: 90112[ISI][Medline]
15 Curry S, Lieb WR, Franks NP. Effects of general anaesthetics on the bacterial luciferase enzyme from Vibrio harveyi: an anaesthetic target site with differential sensitivity. Biochemistry 1990; 29: 464152[ISI][Medline]
16 de-Armendi AJ, Tonner PH, Bugge B, Miller KW. Barbiturate action is dependent on the conformational state of the acetylcholine receptor. Anesthesiology 1993; 79: 103341[ISI][Medline]
17 Dilger JP. The effects of general anaesthetics on ligand-gated ion channels. Br J Anaesth 2002; 89: 4151
18 Dodson BA, Braswell LM, Miller KW. Barbiturates bind to an allosteric regulatory site on nicotinic acetylcholine receptor-rich membranes. Mol Pharmacol 1987; 32: 11926[Abstract]
19 Duncalfe LL, Dunn SM. Mapping of GABAA receptor sites that are photoaffinity-labelled by [3H]flunitrazepam and [3H]Ro 154513. Eur J Pharmacol 1996; 298: 31319[ISI][Medline]
20 Eckenhoff RG. Amino acid resolution of halothane binding sites in serum albumin. J Biol Chem 1996; 271: 155216
21 Eckenhoff RG. An inhalational anaesthetic binding domain in the nicotinic acetylcholine receptor. Proc Natl Acad Sci USA 1996; 93: 280710
22 Eckenhoff RG. Promiscuous ligands and attractive cavities. Mol Intervent 2001; 1: 25868
23 Eckenhoff RG, Knoll FJ, Greenblatt EP, Dailey WP. Halogenated diazirines as photolabel mimics of the inhaled haloalkane anaesthetics. J Med Chem 2002; 94552
24 Eger II E, Koblin D, Harris R, et al. Hypothesis: inhaled anaesthetics produce immobility and amnesia by different mechanisms at different sites. Anesth Analg 1997; 84: 91518[ISI][Medline]
25 Feher VA, Baldwin EP, Dahlquist FW. Access of ligands to cavities within the core of a protein is rapid. Nat Struct Biol 1996; 3: 51621[ISI]
26 Forman SA, Miller KW, Yellen G. A discrete site for general anaesthetics on a postsynaptic receptor. Mol Pharmacol 1995; 48: 18[Abstract]
27 Forman SA, Raines DE. Nonanaesthetic volatile drugs obey the Meyer-Overton correlation in two molecular protein site models. Anesthesiology 1998; 88: 153548[ISI][Medline]
28 Franks NP, Dickinson R, de Sousa SL, Hall AC, Lieb WR. How does xenon produce anaesthesia? Nature 1998; 396: 324[ISI][Medline]
29 Franks NP, Jenkins A, Conti E, Lieb WR, Brick P. Structural basis for the inhibition of firefly luciferase by a general anaesthetic. Biophys J 1998; 75: 220511
30 Franks NP, Lieb WR. Do general anaesthetics act by competitive binding to specific receptors? Nature 1984; 310: 599601[ISI][Medline]
31 Grosman C, Zhou M, Auerbach A. Mapping the conformational wave of acetylcholine receptor channel gating. Nature 2000; 403: 7736[ISI][Medline]
32 Horenstein J, Wagner DA, Czajkowski C, Akabas MH. Protein mobility and GABA-induced conformational changes in GABA(A) receptor pore-lining M2 segment. Nat Neurosci 2001; 4: 47785[ISI]
33 Hucho FL, Tsetlin V. Structural biology of key nervous system proteins. J Neurochem 1996; 66: 178192[ISI][Medline]
34 Husain SS, Forman SA, Kloczewiak MA, et al. Synthesis and properties of 3-(2-hydroxyethyl)-3-n-pentyldiazirine, a photo activable general anaesthetic. J Med Chem 1999; 42: 33007[ISI][Medline]
35 Israelachvili J. Intermolecular and Surface Forces. London: Academic Press, 1991
36 Jenkins A, Greenblatt EP, Faulkner HJ, et al. Evidence for a common binding cavity for three general anaesthetics within the GABAA receptor. J Neurosci 2001; 21: 13643
37 Koltchine VV, Finn SE, Jenkins A, et al. Agonist gating and isoflurane potentiation in the human gamma-aminobutyric acid type A receptor determined by the volume of a second transmembrane domain residue. Mol Pharmacol 1999; 56: 108793.
38 Krasowski MD, Koltchine VV, Rick CE, et al. Propofol and other intravenous anaesthetics have sites of action on the gamma-aminobutyric acid type A receptor distinct from that for isoflurane. Mol Pharmacol 1998; 53: 5308
39 Mann G, Hermans J. Modeling proteinsmall molecule interactions: structure and thermodynamics of noble gases binding in a cavity in mutant phage T4 lysozyme L99A. J Mol Biol 2000; 302: 97889
40 Meyer HH. Zur Theorie der Alkoholnarkose. Der Einfluss wechselnder Temperatur auf Wirkungsstärke und Theilungscoefficient der Narcotica. Arch Exp Path Pharmakol 1901; 42: 33846
41 Middleton AJ, Smith EB. General anaesthetics and bacterial luminescence. II. The effect of diethyl ether on the in vitro light emission of Vibrio fischeri. Proc R Soc Lon B Biol Sci 1975; 193: 17390[ISI][Medline]
42 Mihic SJ, Ye Q, Wick MJ, et al. Sites of alcohol and volatile anaesthetic action on GABA(A) and glycine receptors. Nature 1997; 389: 3859[ISI][Medline]
43 Miller KW. The nature of the site of general anaesthesia. Int Rev Neurobiol 1985; 27: 161
44 Miller KW, Addona GH, Kloczewiak MA. Approaches to proving there are general anaesthetic sites on ligand gated ion channels. Toxicol Lett 1998; 100101: 13947
45 Miller KW, Paton WDM, Smith EB, Smith RA. Physico-chemical approaches to the mode of action of general anaesthetics. Anesthesiology 1971; 36: 33951[ISI]
46 Morton A, Baase WA, Matthews BW. Energetic origins of specificity of ligand binding in an interior nonpolar cavity of T4 lysozyme. Biochemistry 1995; 34: 856475[ISI][Medline]
47 Mullins LJ. Some physical mechanisms of narcosis. Chem Rev 1954; 54: 289323
48 Olsen RW, Fischer JB, Dunwiddie TV. Barbiturate enhancement of -aminobutyric acid receptor binding and function as a mechanism of anaesthesia. In: Roth SH, Miller KW, eds. Molecular and Cellular Mechanisms of Anaesthetics. New York: Plenum, 1986; 16578
49 Overton CE. Studies of narcosis and a contribution to general pharmacology. In: Studies of Narcosis. Lipnick RL, ed. London: Chapman and Hall and the Wood Library-Museum of Anaesthesiology, 1991; 1203
50 Pearce RA. Barbiturates and the brain. Anesthesiology 1999; 91: 11902[ISI][Medline]
51 Pistis M, Belelli D, Peters JA, Lambert JJ. The interaction of general anaesthetics with recombinant GABAA and glycine receptors expressed in Xenopus laevis oocytes: a comparative study. Br J Pharmacol 1997; 122: 170719[Abstract]
52 Prange T, Schiltz M, Pernot L, et al. Exploring hydrophobic sites in proteins with xenon or krypton. Proteins 1998; 30: 6173[ISI][Medline]
53 Pratt MB, Husain SS, Miller KW, Cohen JB. Identification of sites of incorporation in the nicotinic acetylcholine receptor of a photoactivatible general anaesthetic. J Biol Chem 2000; 275: 2944151
54 Quillin ML, Breyer WA, Griswold IJ, Matthews BW. Size versus polarizability in proteinligand interactions: binding of noble gases within engineered cavities in phage T4 lysozyme. J Mol Biol 2000; 302: 95577[ISI][Medline]
55 Raines DE, Claycomb RJ, Scheller M, Forman SA. Nonhalogenated alkane anaesthetics fail to potentiate agonist actions on two ligand-gated ion channels. Anesthesiology 2001; 95: 4707[ISI][Medline]
56 Rankin SE, Addona GH, Kloczewiak MA, Bugge B, Miller KW. The cholesterol dependence of activation and fast desensitization of the nicotinic acetylcholine receptor. Biophys J 1997; 73: 244655[Abstract]
57 Roth SH, Forman SA, Braswell LM, Miller KW. Actions of pentobarbital enantiomers on nicotinic cholinergic receptors. Mol Pharmacol 1989; 36: 87480[Abstract]
58 Sachsenheimer W, Pai EF, Schulz GE, Schirmer RH. Halothane binds in the adenine-specific niche of crystalline adenylate kinase. FEBS Lett 1977; 79: 31012[ISI][Medline]
59 Schoenborn BP. Binding of cyclopropane to sperm whale myoglobin. Nature 1967; 214: 11202[ISI][Medline]
60 Schoenborn BP, Watson HC, Kendrew JC. Binding of xenon to sperm whale myoglobin. Nature 1965; 207: 2830[ISI][Medline]
61 Sedensky MM. Model organisms: new insights into ion channel and transporter function. Stomatin homologues interact in Caenorhabditis elegans. Am J Physiol Cell Physiol 2001; 280: C13408
62 Smith GB, Olsen RW. Functional domains of GABAA receptors. Trends Pharmacol Sci 1995; 16: 1628[ISI]
63 Tang P, Eckenhoff RG, Xu Y. General anaesthetic binding to gramicidin A: the structural requirements. Biophys J 2000; 78: 18049
64 Tilton RF, Jr, Kuntz ID, Jr, Petsko GA. Cavities in proteins: structure of a metmyoglobin-xenon complex solved to 1.9 A. Biochemistry 1984; 23: 284957[ISI][Medline]
65 Tilton RF, Singh UC, Kuntz ID, Kollman PA. Proteinligand dynamics: a picosecond simulation of a myoglobinxenon complex. J Mol Biol 1988; 199: 195211[ISI][Medline]
66 Trudell JR, Bertaccini E. Molecular modelling of specific and non-specific anaesthetic interactions. Br J Anaesth 2002; 89: 3240
67 Unwin N. The Croonian Lecture 2000. Nicotinic acetylcholine receptor and the structural basis of fast synaptic transmission. Philos Trans R Soc Lond B Biol Sci 2000; 355: 181329[ISI][Medline]
68 Vonrhein C, Schlauderer GJ, Schulz GE. Movie of the structural changes during a catalytic cycle of nucleoside monophosphate kinases. Structure 1995; 3: 48390[ISI][Medline]
69 Wood SC, Forman SA, Miller KW. Short chain and long chain alkanols have different sites of action on nicotinic acetylcholine receptor channels from Torpedo. Mol Pharmacol 1991; 39: 3328[Abstract]
70 Wood SC, Hill WA, Miller KW. Cycloalkanemethanols discriminate between volume- and length-dependent loss of activity of alkanols at the Torpedo nicotinic acetylcholine receptor. Mol Pharmacol 1993; 44: 121926[Abstract]
71 Wood SC, Tonner PH, de AA, Bugge B, Miller KW. Channel inhibition by alkanols occurs at a binding site on the nicotinic acetylcholine receptor. Mol Pharmacol 1995; 47: 12130[Abstract]
72 Yan H, Tsai MD. Nucleoside monophosphate kinases: structure, mechanism, and substrate specificity. Adv Enzymol Relat Areas Mol Biol 1999; 73: 10334[ISI][Medline]
73 Zhou QL, Zhou Q, Forman SA. The n-alcohol site in the nicotinic receptor pore is a hydrophobic patch. Biochemistry 2000; 39: 149206[ISI][Medline]
74 Zimmerman SA, Jones MV, Harrison NL. Potentiation of gamma-aminobutyric acidA receptor Cl- current correlates with in vivo anaesthetic potency. J Pharmacol Exp Ther 1994; 270: 98791[Abstract]