Inhaled Anesthetic Binding: Features and Location

University Of Pennsylvania, Philadelphia PA

Investigators

Linked publications & trials

Abstract

Molecular Targets. The favored neuronal targets of anesthetics, the ligand gated, cys-loop receptor/ion channels, have not yielded to high resolution structural characterization, nor are they sufficiently plentiful to conduct biochemical binding experiments (this is true of other proposed neuronal targets, such as the 2-pore K-channels). Nevertheless, some data for the muscle-type nicotinic acetylcholine receptor are available, owing to of the unusually high density in the Torpedo fish electric organ. These channels are inhibited by inhaled anesthetics (Forman et al., 1995), consistent with the well-known muscle relaxation produced by inhaled anesthetics. Although functionally different in direction from the GABAA receptor (which is potentiated by anesthetics, not inhibited), this may be more a function of receptor design than the mechanism of the anesthetic, because effects in both receptors appear to involve cooperativity with native agonist (Jenkins et al., 2001;Raines &Zachariah, 1999). Although an oversimplification, cooperativity in the setting of rapid desensitization (nAChR), produces a net inhibition, while for a more slowly desensitized receptor (GABAA, Glycine), the net effect is potentiation. Using the electric organ from Torpedo, we and others have found apparent KD values of about 1 mM, and multiple sites (Eckenhoff, 1996;Xu et al., 2000). Combining photolabeling with microsequencing in collaboration with Professor Jon Cohen (Harvard), we found an agonist-sensitive site in the first transmembrane segment of the 6 subunit (Chiara et al., 2003). A 4A resolution cryoEM structure indicated that this region of each subunit is a 4-ohelical bundle (Miyazawa et al., 2003), and that the implicated "anesthetic" site appears to be in an cavity formed between these transmembrane bundles of a helices. Similarly, a photolabeled site for etomidate has been reported at an (presumably) analogous position in the GABAA receptor, and this location is generally consistent with the results of site-directed mutagenesis (Mihic et al., 1997). Although binding affinity is difficult to measure in the presence of multiple different sites, there would appear to be a convergence of results from disparate approaches on an inter-domain cavity in an ct-helical bundle as representing a pharmacologically relevant anesthetic binding site. Models. All of the few high resolution complexes between anesthetics and proteins are model systems. High resolution structures of the FFL/bromoform complex revealed two bromoform molecules bound near (but not in) the predicted ligand site (Franks et al., 1998), and in a surprisingly polar environment. But because the FFL conformation was the low ATP form, which is also the low anesthetic affinity form (KD >5mM anesthetic) (Dickinson et al., 1993) (Eckenhoff et al., 2001), it was not surprising that few specific contacts were noted. Binding sites for halothane and propofol have found in human serum albumin (HSA) with crystallography (Bhattacharya et al, 2000). These structures are of only moderate resolution (~2.5A) and the atoms forming the binding site show high B-factors (thermal factors), so detailed structural inferences are difficult to draw. However, the sites are characterized as being largely lined by apolar or uncharged polar residues, and with a near-absence of specific interactions like hydrogen bonds (except for propofol). Again, this might be considered reasonable since the dissociation constants for HSA are greater than 1 mM. Until recently, these two proteins (FFL and HSA) were the only high resolution structures available for the clinically used inhaled anesthetics - and the binding energetics of neither protein approaches the clinical EC50 for these drugs of about 200 uM. Thus, using these structures as a template for drug design or for target discovery could be misleading. A structure-based approach is seriously hampered by the unavailability of high resolution complexes of high affinity interactions. In order to supply these data, we screened (with ITC) many helical bundle proteins with high resolution structures deposited in the PDB, and found that ferritin, a natural iron binding protein found in the cytosol and nucleus of almost every tissue in all species, binds anesthetics with KD in the low micromolar range.. We were able to crystallize apoferritin (the iron removed) and obtain high resolution structures of the anesthetic complex (co-crystallized, not soaked). These structures represent the highest resolution data for an anesthetic protein complex yet deposited in the PDB (1zx1), and the only dataset for isoflurane (1xz3). Ferritin is a 24-mer of two similar, 4-helix bundle subunits, H and L. It is arranged as a 12-mer of L:L and H:L dimers. Because H is less abundant (~15%) than L, and the asymmetric unit is the monomer, electron density due to H is overpowered by that of L, and H cannot be seen in the fitted structure. The anesthetic binding site is at an intersubunit, interhelical cavity. Thus, ferritin exhibits similarities to the superfamily of ligand-gated channels;an oligomer of 4-a-helix bundles, with the anesthetic site at the interhelical, subunit interface (see Figure 4 below). Most importantly, the ferritin data are of sufficient resolution to determine the structural basis for high affinity binding, stereoselectivity and exclusion of non-immobilizers. Our initial publication of this work (Liu et al, 2005) found better fits of the inhaled anesthetic enthalpograms (ITC) with two class site models, suggesting affinity differences at the H:L versus L:L interface. However, we have been unable to show that the H:L interface is any different from the L:L with respect to anesthetic binding; thus single class site fits are used from here on. The ability to discover or design a general anesthetic compound based on known target binding site detail is unprecedented, and is a highly sought after goal in most anesthetic mechanisms research. Previous structure activity relationships have focused only on the ligand (Krasowski et al, 2002, Sewell &Sear, 2006), therefore are heavily biased for finding similar molecules, and not to leap to novel scaffolds. Given that the necessary detail for the cys-loop ligand gated channels, or any other "relevant" target, is unlikely to be generated in the near term, this approach to use a surrogate as the initial screen is a reasonable start, and if successful, will suggest that the structural detail from GABAA receptors might not be required for further anesthetic development. And it is now clear that better anesthetics are required. Currently there is considerable concern that the inhaled anesthetics may produce long term CNS effects - especially in the extremes of age, and in other vulnerable populations.