Meyer and Overton Revisited

The fact that certain lipophilic compounds depress voluntary activity and responses in numerous phyla is difficult to ascribe to coincidence, but also hard to consider due to natural selection. The natural selection force is curious to determine (if an organism is absolutely still does that make it more or less likely to die, or if cell metabolism is depressed is that protective or harmful), whereas the wide variety of molecular and cellular mechanisms altered by anesthetics seems curiously large (if you are beneficially immobilized, why is amnesia necessary?). However, this question raised by Sonner1 in this issue of the journal justifies a reexamination of the possible processes involved. In thinking about how molecules’ function and behaviors are altered by anesthetics, there is the obvious temptation to concentrate on specific proteins and changes in primary structure (DNA-defined amino acid substitutions), and to pay less heed to the protein environment (e.g., lipid bilayer). Yet, even the membrane bilayer itself is defined indirectly by the genetic code. If one is trying to rationalize the fact that unicellular organisms and complex animals can be quickly and reversibly anesthetized, we must examine the relevant cellular features that are common to both: the membrane proteins responsible for signaling. This widely diverse group of proteins share common features, the most obvious being transmembrane spanning α-helical amino acid sequences. A number of studies have documented anesthetic binding sites on a variety of natural2–5 and synthetic proteins.6–8 Effects on cell regulation and signaling caused by anesthetics in variably mutated membrane signaling proteins, with subsequent expression in live animals (knock-in and knock-out mice), particularly the γ-aminobutyric acid type A (GABAA) channel, has provided considerable insight into certain aspects of anesthetic effects.9–13 Beginning with the description of anesthetic-induced depression of the function of a pure protein system over two decades ago,14 and an earlier lack of evidence for changes in membrane lipid structure,15 the membrane bilayer has been increasingly discounted as a site for anesthetic action. In addition, the failure of the Meyer16 and Overton17 correlation (of anesthetic potency with lipid solubility) to incorporate a variety of lipid soluble compounds18 also led to focus on anesthetic binding by proteins. Understanding is usually gained only in those subjects that are most readily studied, in this case membrane signaling proteins. However, one problem that has plagued interpretation of anesthetic mechanism studies has been the Principal of Parsimony (Occam’s razor), that is, if one explanation (mechanism) can explain an effect, that mechanism (and that alone, e.g., protein binding) is applied universally. But in multicellular organisms, the anesthetic state has multiple components (loss of consciousness, amnesia, immobilization, analgesia) that appear to be mediated by actions at multiple molecular sites (pre- and postsynaptic, various ion channels, various receptors). Consequently, attribution to a single molecular mechanism (binding to a protein cavity) is perhaps too parsimonious. Although recent publications have emphasized protein binding by anesthetics, it may be premature to exclude bilayer actions from contributing to altered actions. Over two decades ago, a variety of observations were made regarding the effects of cholesterol in the membrane and its potential to increase order and decrease “fluidity.”19,20 Fluidity is the somewhat ambiguous term that suggested greater mobility and decreased stability of the membrane constituents, and its increase could perturb membrane function and result in the anesthetic state. However, measurable changes in membrane structure were small as measure did not appear to explain anesthetic action. The bilayer lipids came to be seen primarily as a “solvent” for the membrane proteins, albeit in a planar distribution. In that context, the addition of anesthetics to a mere solvent might not be expected to have a great effect. However, general anesthetics (incompletely halogenated hydrocarbons and ethers, heavier noble gasses, aromatic molecules, alcohols) have a fixed or inducible polarity and therefore occupy a specific location at the interfacial region of the bilayer, at the region between the polar head group and the hydrocarbon chains that form the interior of the membrane.21–24 In contrast, the nonanesthetic lipid soluble compounds occupy the nonpolar region of the hydrocarbon tails of the membrane interior. It is also noteworthy that in the transmembrane α-helices, aromatic amino acid residues seem to specifically present in the helix where it is near this interfacial region.25 If anesthetics specifically localize in the interfacial region, it seems likely that they could modify the lipid interfacial interaction with the amphipathic amino acids that are typically present in this region of the helix26 and thereby alter function. The interaction between the amino acid sequence of membrane proteins and their surrounding annular lipids is only beginning to be explored in detail. The lipids that surround the protein (annular lipids) are less mobile and diffuse within the membrane surface more slowly than the nonannular lipids.27 Altering the lipid constituents of the membrane can influence the microstructure and function of ion channels (e.g., nicotinic acetylcholine receptors28) and of G-protein linked receptors.29,30 The requirement for the binding of specific lipids in annular clefts or interactions with certain phospholipids headgroups (clefts) means that displacement of these lipids by anesthetics could alter function. In addition, while the stereospecificity of amino acids and proteins has been emphasized as explaining the stereospecific aspects on anesthestic action, the inherent stereospecificity of bilayer components (cholesterol, phospholipids, sphingomyelin) and their importance has not been fully explored. As opposed to specific sites that require certain membrane lipids, Cantor has suggested lateral pressure may have a regulating role in membrane proteins.31,32 Such a model incorporates the disruptions due to unsaturated acyl chains in the phospholipids, as well as the presence of cholesterol in altering the tension place upon a protein within the bilayer. Increased pressure at in the central region (surrounded by the bilayer interior) or at the interfacial region (near the phospholipid head groups) could have divergent actions function. Instead of calculations of altered internal pressure, Tang et al. have used molecular dynamic simulations to predict the location and behavior of anesthetics on the model channel gramicidin A.33,34 Their results demonstrate the anesthetic-induced changes in global dynamics of the protein that thereby alter its function. Both investigations emphasize the anesthetic action being mediated by a continuous but nonstationary interaction with the protein molecule to alter its function. Several lines of experimental evidence suggest that some lipophilic or amphipathic molecules bind to hydrophobic, intramembranous receptor sites via the membrane bilayer.35 Cantor has suggested that neurotransmitters themselves would interact not only with their specific receptor, but also infiltrate the membrane bilayer to induce changes in lateral pressure and protein function.36 Although speculative, it is clear that a number of drug and toxin effects are mediated by bilayer interactions that allow drug interaction with the membrane protein. It seems therefore particularly noteworthy, and especially in the context of evolutionary development, that a number of polypeptide toxins from a variety of species (tarantula, scorpion, sea anemone) have a prominent lipophilic region that binds to the bilayer. These molecules target ion channels by first being bound to the lipid membrane. Tarantula toxins that bind in lipid sites may serve as a model of anesthetics, working in the boundary lipids.37,38 The chili pepper-derived drug, capsaicin, is most noteworthy for its activation of the vanilloid receptor (TRPV1) that modulates pain transmission. Yet capsaicin, a relatively simple amphiphilic molecule (MW approximately 305 Da), interacts with membranes and modulates a variety of voltage-gated channels, as well as the acetylcholine receptor.39 The isoflavonoid genistein (MW approximately 270 Da), often used as a tyrosine kinase inhibitor, is able to modulate a wide variety of voltage-gated and ligand regulated channels.40 Suchyna et al.37 suggest that the actions of both molecules are mediated by altering the lipid bilayer thickness, so that channel function is inhibited or enhanced. In addition, nicotine has effects beyond the acetylcholine receptor, also depressing the voltage gated sodium channel, effects that could be mediated in a similar fashion.41 It is not inconceivable that for some membrane proteins, anesthetics behave like peptide toxins, residing in lipid phase but altering the function ion channel. For voltage-gated channels, which have perhaps been prematurely discarded as anesthetic sites of action, the interaction of their voltage-sensing region with the membrane with the bilayer (“the paddle in oil”42–44) could be particularly susceptible to alterations in the interfacial region of the bilayer. In addition, a more complex and sophisticated understanding of bilayer structure is emerging that has great relevance as to how anesthetics might influence evolving membrane protein function.25 Certain areas of the membrane incorporate cholesterol and sphingomyelin to form distinct “rafts” that incorporate and are bounded by certain membrane proteins.45 The ordered-liquid phase (Lo) induced by cholesterol and sphingomyelin existing only in specific regions (e.g., caveolae) are likely to be more resistant to anesthetic alteration since their more ordered structure tends to exclude various other lipid components.46 It is likely that these Lo phases are less likely to be susceptible to changes in lateral pressure. It remains to be verified if increased cholesterol incorporation into the membrane was a means of counteracting the effects of continuous halothane exposure in the prokaryote Acholeplasma laidlawii.47 Whereas from an evolutionary perspective, it seems most obvious to examine the protein functions that are altered by anesthetics, it may also be appropriate to think about proteins whose functions are not altered but rather “protected.” Are certain signaling molecules imbedded in the Lo rafts because mutations that moved them to the disordered liquid phase (Ld) subjected them to anesthetic-like agents that altered function in a manner to decrease survival of the organism? It seems possible that those transmembrane signaling proteins whose functions are necessary for cell survival might be sequestered in regions of cholesterol- and sphingomyelin-rich Lo membrane. Although the argument for a lipid site of action (not necessarily exclusive) might now be revived based on recent developments and research, it is interesting to make the case from an evolutionary perspective. More than 100 years ago, Meyer and Overton described the association between lipid solubility and anesthetic potency at a time when membrane bilayers had yet to be described. A much more complicated picture has emerged. Not only is the lipid bilayer structure better defined, but over the last decade, the emerging picture suggests that membrane proteins (which are of interest) do not merely float around in a sea of any old fatty oil (boring stuff!). Rather, the lipid membrane is a structured entity with differing domains possessing differing constituents: more structured lipid rafts of cholesterol and sphingomyelin are anchored by specific proteins in the region of a cell surface, whereas certain protein functions are moderated by specific lipids. Regions of the bilayer that are less affected by lipophilic agents may preserve the function of certain necessary pathways. Alterations in the interfacial region of other parts of the bilayer may interact with specific regions of membrane proteins in a phenomenon that over millions of years has resulted in selection of agent-induced changes in protein functions that are protective (timely immobilization or depressed metabolism?). Therefore, emergence of the susceptibility to polar-lipid alterations in the membrane protein signaling (which we call anesthesia) might involve far more than mere mutations in hydrophobic cavities of membrane proteins.