X-ray Crystallographic Studies for Revealing Binding Sites of General Anesthetics in Pentameric Ligand-Gated Ion Channels

Abstract

X-ray crystallography is a powerful tool in structural biology and can offer insight into structured-based understanding of general anesthetic action on various relevant molecular targets, including pentameric ligand-gated ion channels (pLGICs). In this chapter, we outline the procedures for expression and purification of pLGICs. Optimization of crystallization conditions, especially to achieve high-resolution structures of pLGICs bound with general anesthetics, is also presented. Case studies of pLGICs bound with the volatile general anesthetic isoflurane, 2-bromoethanol, and the intravenous general anesthetic ketamine are revisited.

1. Introduction

Pentameric ligand-gated ion channels (pLGICs) are targets of general anesthetics (Campagna, Miller, & Forman, 2003; Franks & Lieb, 1994; Hemmings et al., 2005). However, there is no consensus on the underlying molecular mechanisms of how general anesthetics modulate pLGICs. A number of experimental methods and structural tools have been developed and utilized to identify the binding sites of general anesthetics and understand their modulation on pLGICs, including site-directed mutagenesis in combination with electrophysiology measurements (Mihic et al., 1997), photoaffinity labeling (Z. W. Chen et al., 2012; Chiara, Dangott, Eckenhoff, & Cohen, 2003; Chiara et al., 2012; Chiara et al., 2014; Eckenhoff & Shuman, 1993; Eckenhoff et al., 2010; Husain et al., 2010; Jayakar, Dailey, Eckenhoff, & Cohen, 2013; Jayakar et al., 2014; Woll, Dailey, Brannigan, & Eckenhoff, 2016; Yip et al., 2013), nuclear magnetic resonance (NMR) (Bondarenko, Mowrey, Liu, Xu, & Tang, 2013; Bondarenko et al., 2014; Bondarenko, Yushmanov, Xu, & Tang, 2008; Kinde et al., 2016), and X-ray crystallography (Q. Chen et al., 2015; Q. Chen et al., 2017; Nury et al., 2011; Pan, Chen, Willenbring, Mowrey, et al., 2012; Sauguet, Howard, et al., 2013; Spurny et al., 2013). Among these methods, X-ray crystallography is particularly well suited to determine atomic-resolution structures of large channel proteins in complex with anesthetics.

In the last decade, remarkable milestones have been reached in solving crystal structures of pLGICs, beginning with the prokaryotic channels ELIC (Hilf & Dutzler, 2008) and GLIC (Bocquet et al., 2009; Hilf & Dutzler, 2009) and more recently with the eukaryotic channels GluCl (Hibbs & Gouaux, 2011), 5HT_3AR (Hassaine et al., 2014), β3GABA_AR (Miller & Aricescu, 2014), α3GlyR (Huang, Chen, Michelsen, Schneider, & Shaffer, 2015), and α4β2nAChR (Morales-Perez, Noviello, & Hibbs, 2016). The crystallographic approach has also successfully led to the discovery of multiple binding sites of general anesthetics in pLGICs (Q. Chen et al., 2015; Q. Chen et al., 2017; Nury et al., 2011; Pan, Chen, Willenbring, Mowrey, et al., 2012; Spurny et al., 2013). These studies provide not only structural insights into the molecular recognition of general anesthetics and the mechanisms underlying anesthetic modulation of pLGICs, but also a structural framework for the discovery of novel general anesthetics.

In spite of several recent successes (Q. Chen et al., 2015; Q. Chen et al., 2017; Nury et al., 2011; Pan, Chen, Willenbring, Mowrey, et al., 2012; Spurny et al., 2013), the application of X-ray crystallography to study the interactions between general anesthetics and pLGICs is by no means straightforward. pLGICs are ~200 kDa (prokaryotic) or ~300 kDa (eukaryotic) integral membrane proteins composed of five identical or homologous subunits. They assemble into pentamers under optimal purification conditions, but they also have a tendency to aggregate or to dissociate into smaller oligomers. Although most of the extracellular domain (ECD) and the transmembrane helices are well ordered, there are flexible regions that interfere with crystallization, including the flexible N- and C-termini and the highly variable intracellular domain of eukaryotic pLGICs (Hibbs & Gouaux, 2011; Huang et al., 2015; Miller & Aricescu, 2014; Morales-Perez et al., 2016). Post-translational modifications of proteins expressed in eukaryotic systems may also interfere with monodispersity and the ability to form high-quality crystals (Hassaine et al., 2013). Formation of high quality co-crystals with anesthetics is further complicated by the nature of these relative small and mostly hydrophobic molecules. They often bind to multiple sites on pLGICs with low affinities (μM - mM). All these factors may affect the crystal quality and structural resolution.

In this chapter, we provide protocols for crystallization of pLGICs in the absence and presence of general anesthetics, and for the identification and functional validation of general anesthetic binding sites in pLGICs. Although these protocols were developed for prokaryotic pLGICs, similar principles should apply to the eukaryotic channels.

2. Production of monodisperse pLGICs with high purity

X-ray crystallography requires an adequate amount of monodisperse protein with high purity, often on the order of milligrams. Successful protein production normally includes three major steps: making proper constructs, finding a suitable expression system, and optimizing purification protocols. Sometimes, this procedure may need to be iteratively optimized based on the crystal diffraction quality of the proteins. pLGICs for X-ray crystallographic studies have been obtained as recombinant proteins overexpressed in Escherichia coli (E. coli) cells for prokaryotic pLGICs (Bocquet et al., 2009; Bocquet et al., 2007; Hilf & Dutzler, 2008, 2009) and from insect (Du, Lu, Wu, Cheng, & Gouaux, 2015; Hibbs & Gouaux, 2011; Huang et al., 2015) or human embryotic kidney (HEK) (Hassaine et al., 2014; Hassaine et al., 2013; Miller & Aricescu, 2014; Morales-Perez et al., 2016) cells for eukaryotic pLGICs. The lower cost, shorter timeline and versatile strains make E. coli an attractive candidate for protein expression, not only for prokaryotic proteins, but also for eukaryotic membrane proteins. Recently, we have successfully produced full-length wild type human α7nAChR using E. coli and demonstrated that the purified α7nAChR retains functionality and pharmacological properties of native α7nAChR (Tillman et al., 2016). The success with α7nAChR paves a new path for the production of eukaryotic pLGICs using E. coli. Other examples of eukaryotic membrane protein expression in E. coli can be found in a recent review (Snijder & Hakulinen, 2016).

2.1. Expression of pLGICs

All pLGICs crystallized to date have been modified to remove flexible regions or post-translational modifications that may interfere with crystal formation. All eukaryotic pLGICs structures obtained so far have removed most or all residues in the intracellular domain because it contains flexible disordered regions (Hassaine et al., 2014; Hibbs & Gouaux, 2011; Huang et al., 2015; Miller & Aricescu, 2014; Morales-Perez et al., 2016). Other highly flexible regions, including the N- and C-termini, may also need to be modified for crystallizations (Hassaine et al., 2014; Hibbs & Gouaux, 2011; Huang et al., 2015; Miller & Aricescu, 2014; Morales-Perez et al., 2016). In principle, non-homogeneous glycosylation of pLGICs expressed from insect or HEK cells can interfere with crystallization and this heterogeneity can be reduced with de-glycosylation enzymes or by point mutations of the glycosylation site (Hassaine et al., 2013; Miller & Aricescu, 2014). However, this step can be skipped in some cases with few glycosylation sites (Du et al., 2015; Hibbs & Gouaux, 2011; Huang et al., 2015; Morales-Perez et al., 2016). A cleavable purification tag, such as polyhistidine (Bocquet et al., 2009; Hibbs & Gouaux, 2011; Hilf & Dutzler, 2008, 2009) and strep (Hassaine et al., 2013; Huang et al., 2015; Morales-Perez et al., 2016), or antibody recognition motif (Miller & Aricescu, 2014) can be added to the N- or C-terminus to facilitate purification. For expressions in E. coli, fusion with a soluble protein (e.g., maltose binding protein) may aid in the expression of well-folded channel proteins (Bocquet et al., 2007). Table 1 summarizes fusion proteins and expression tags that have been used for overexpression of pLGICs. Fluorescence-detection size-exclusion chromatography (FSEC) (Kawate & Gouaux, 2006) is an efficient tool for construct screening (Bird et al., 2015), which has been used in the design of crystal constructs of pLGICs, including GluCl (Hibbs & Gouaux, 2011), β3GABA_AR (Miller & Aricescu, 2014), and α4β2nAChR (Morales-Perez et al., 2016).

Table 1.

Host cells, tags, and detergents used for expression and purification of pLGICs

pLGICs	Host cells	Tags	Detergents	References
ELIC	E. coli BL21(DE3)	N-His10MBP	UDM	(Bertozzi, Zimmermann, Engeler, Hilf, & Dutzler, 2016; Hilf & Dutzler, 2008; Zimmermann & Dutzler, 2011; Zimmermann, Marabelli, Bertozzi, Sivilotti, & Dutzler, 2012)
	E. coli BL21-Gold(DE3)			(Gonzalez-Gutierrez et al., 2012)
	E. coli C43			(Spurny et al., 2013; Spurny et al., 2012; Ulens et al., 2014)
	E. coli Rosetta(DE3)pLysS			(Q. Chen et al., 2015; Q. Chen et al., 2017; Pan, Chen, Willenbring, Yoshida, et al., 2012)
GLIC	E. coli BL21(DE3)	N-His10MBP	DDM	(Bertozzi et al., 2016; Hilf et al., 2010; Hilf & Dutzler, 2009; Velisetty & Chakrapani, 2012; Velisetty, Chalamalasetti, & Chakrapani, 2012)
	E. coli BL21-Gold(DE3)			(Gonzalez-Gutierrez, Cuello, Nair, & Grosman, 2013)
	E. coli C43			(Basak, Schmandt, Gicheru, & Chakrapani, 2017; Bocquet et al., 2009; Fourati, Sauguet, & Delarue, 2015; Laurent et al., 2016; Menny et al., 2017; Nury et al., 2011; Prevost et al., 2012; Sauguet et al., 2016; Sauguet, Poitevin, et al., 2013; Sauguet et al., 2014)
	E. coli Rosetta(DE3)pLysS			(Mowrey et al., 2013; Pan, Chen, Willenbring, Mowrey, et al., 2012)
	Drosophilia Schneider (S2)	C-His10		(Sauguet et al., 2014)
GLIC-GlyR				(Moraga-Cid et al., 2015)
ELIC-GLIC	E. coli BL21(DE3)	N-His10MBP		(Schmandt et al., 2015)
α7nAChR	Rosetta2(DE3)pLysS	C-His8	DPC	(Tillman et al., 2016)
GluCl	Sf9		DDM	(Althoff, Hibbs, Banerjee, & Gouaux, 2014; Hibbs & Gouaux, 2011)
α1GlyR				(Du et al., 2015)
α3GlyR		C-Strep		(Huang et al., 2015; Huang, Chen, & Shaffer, 2017; Huang, Shaffer, et al., 2017)
α4β2nAChR	GnTI-HEK		DDM+CHS	(Morales-Perez et al., 2016)
β3GABAAR	HEK293S-GnTI−	TETSQVAPA	DMNG + CHS	(Miller & Aricescu, 2014)
5HT3AR	T-REx-293	N-Strep	C12E9	(Hassaine et al., 2014)

A brief generalized protocol for overexpression of pLGICs in E. coli is given as follows:

1. Transform the expression plasmid into the chosen E. coli strain.

2. Pick a few colonies into 10 mL rich media, such as Luria-Bertani (LB) broth supplemented with the selective antibiotics. Grow cells overnight at 37 °C.

3. Inoculate (1:100) overnight culture into 1 liter LB supplemented with antibiotics. Grow cells at 37 °C in an environmental shaker at 250 rpm until optical density (OD, measured at 600 nm) reaches 0.6–0.8.

4. Lower the environmental shaker temperature to 15 °C. After 1-hour temperature equilibration, induce the culture by adding 0.2 mM isopropyl β-D-1-thiogalactopyranoside (IPTG).

5. Harvest cells after expression for ~16 h. Store cell pellets at −80 °C.

Note 1: Expression media, OD₆₀₀ at induction, IPTG concentration, temperature, and expression time should be optimized for individual proteins.

Note 2: For the expression of pLGICs in insect or HEK cells, please refer to the similar protein expression protocols referenced in Table 1.

2.2. Purification of pLGICs

The process of protein purification isolates the desired protein from other cell components in the expression system. In general, the purification of pLGICs involves up to four steps: 1) isolating the cell membrane that contains the expressed protein, 2) solubilizing the membrane with a proper detergent, 3) purifying the expressed pLGICs from other membrane fractions, and 4) cleaving the fusion protein or tags whenever necessary. General guides on purifying membrane proteins have been recently reviewed (Roy, 2015). Specific purification protocols for pLGICs used in crystallizations can be found in the references for Table 1.

One of the key factors affecting pLGIC purification is the choice of detergent, which is crucial not only for solubilizing the membrane but also for maintaining the correct folding of pLGICs (Hovius et al., 1998; Miller & Aricescu, 2014). Table 1 lists the detergents used for successful purification of pLGICs. n-Dodecyl-β-D-maltopyranoside (DDM) is a mild detergent commonly used for solubilizing and purifying pLGICs. FSEC can be applied to identify suitable detergents for a given pLGIC (Miller & Aricescu, 2014). The purification is normally performed at 4 °C to prevent aggregation.

A brief purification protocol for pLGICs expressed in E. coli is given as follows:

1. Lyse cells and collect the membrane with ultracentrifugation at least 100,000 × g for ~1 h.

2. Solubilize the collected membrane using ~ 2% of the chosen detergent.

3. Purify the pLGIC with immobilized metal ion affinity chromatography (IMAC) or antibody resin based on the designed purification tag. Check protein purity with sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE).

4. Cleave and separate the fusion protein and/or tags from the pLGIC if necessary.

5. Collect the purified pentameric fraction of the protein by size exclusion chromatography (SEC).

6. Check protein purity with SDS-PAGE.

Note 3: The amount of detergent to solubilize the membrane is about 20% in weight, and should be optimized according to the quantity of collected membrane.

2.3. Functional validation of pLGIC constructs used for crystallizations

As pLGICs are often modified to improve their monodispersity and fit crystallization requirements, it is necessary to check if modifications to the recombinant protein construct have resulted in adverse effects on pharmacological features and channel functions, particularly the characteristic response to general anesthetics. Most studies have performed functional measurements in Xenopus oocytes or HEK cells heterologously expressing the pLGIC constructs (Bocquet et al., 2007; Du et al., 2015; Hibbs & Gouaux, 2011; Hilf & Dutzler, 2009; Huang et al., 2015; Miller & Aricescu, 2014; Morales-Perez et al., 2016). The purified pLGIC can also be directly tested for channel function, either directly measuring function of reconstituted pLGICs in the lipid bilayer (Hilf & Dutzler, 2008) or by injecting vesicles containing purified pLGICs into Xenopus oocytes for functional measurements (Labriola et al., 2013; Tillman et al., 2016).

3. Crystallization of pLGICs in the Absence and Presence of General Anesthetics

Protein crystallization can be achieved through vapor diffusion (Delmar, Bolla, Su, & Yu, 2015), microbatch (Chayen, 1998), or free-interface diffusion (Segelke, 2005). For a new soluble protein, the standard crystallization procedure involves screening the purified protein against sparse-matrix conditions (Luft, Newman, & Snell, 2014; Newman et al., 2005), including pH, temperature, and the choice of salt and type of polyethylene glycol (PEG) at varying concentrations. Crystallization of membrane proteins, such as pLGICs, involves additional variables due to the requirement of using detergents or other membrane mimetics (Bill et al., 2011; Landau & Rosenbusch, 1996). Co-crystallizing pLGICs with anesthetics offers a reliable way to reveal anesthetic binding sites and pLGIC conformational changes induced by the anesthetic binding. However, the inclusion of general anesthetics may add extra complications to crystallization. The choice of anesthetics and the method to introduce anesthetics to crystallization samples can make a profound difference in outcomes. A comprehensive consideration of multiple factors increases success in obtaining pLGIC crystal structures.

3.1. Crystallization of pLGICs

Crystallization of pLGICs begins with systematic screening of multiple variables, including pH, temperature, the type of buffer solution, and the type and concentration of precipitants and salts. The screening can use commercially available membrane protein crystallization screening kits, such as MemGold, MemGold2 (Molecular Dimensions) and MembFac (Hampton Research). Table 2 summarizes the crystallization conditions for pLGICs structures that have been determined so far, including protein concentration, reservoir solution, additives, crystallization method, and cryoprotectants. The same detergent used in purification (Table 1) is typically also used in crystallization. The concentration of pLGICs used for crystallization spans a wide range, from as low as 1–2 mg/mL for GluCl to as high as 8–12 mg/mL for GLIC. Thus, it requires optimization for individual pLGICs. Additives, such as agonists, antagonist, channel modulators, lipids, and nanobodies, can be used to improve the quality of pLGIC crystals. High-quality diffracting crystals of pLGICs have been obtained using both sitting drop and hanging drop methods at temperatures ranging from 4 to 20 °C. The crystals often need one additional week or even longer after the initial formation to reach their peak size (Sauguet et al., 2014). The final crystals are normally collected with cryoprotection using 20–30% glycerol, ethylene glycol, or lower molecular weight PEG, such as PEG350MME or PEG400 (Table 2). The harvested crystals are flash frozen and stored in liquid N₂.

Table 2.

Conditions of crystallization and cryoprotection for pLGICs

pLGIC	PDB code	Protein (mg/mL)	Reservoir solution	Additive	Crystallization method	Cryoprotectant
ELIC	2VL0	10	200 mM (NH4)2SO4, 50 mM ADA (pH 6.5), 10–15% PEG 4000	0.5 mg/ml E. coli lipids	Sitting drop (4 °C)	30% glycerol or ethylene glycol
	2YKS		200 mM NaLiSO4, 50 mM ADA (pH 6.5), 10% PEG 4000			30% glycerol
	2YN6		50 mM BaAc2, 50 mM ADA (pH 6.5), 10% PEG 4000			30% ethylene glycol
	2YOE, 4A97, 4A98, 3ZKR, 4TWD, 4TWF, 4TWH, 5LG3, 5LID		200 mM (NH4)2SO4, 50 mM ADA (pH 6.5), 9–12% PEG 4000			30% glycerol
	3RQU, 3RQW		200 mM (NH4)2SO4, 100 mM MES (pH 6.1–6.3), 10–12% PEG 4000			20% glycerol
	5HEJ, 5HEO, 5HEU, 5HEW		200 mM (NH4)2SO4, 50 mM ADA (pH 6.5), 10–13% PEG 4000			30% ethylene glycol
	4Z90, 4Z91, 5SXU, 5SXV	5–6	200 mM (NH4)2SO4, 100 mM MES (pH 6.1), 10–12% PEG 4000	0.01–0.02 mg/ml E. coli lipids		20% glycerol
	3UQ4, 3UQ5, 3UQ7	9–10	200 mM (NH4)2SO4, 50 mM ADA (pH 6.5–6.9), 10–12% PEG 4000	0.5 mg/ml E. coli lipids	Hanging drop (4 °C)	30% ethylene glycol
GLIC	3EHZ, 3EI0	10	225 mM (NH4)2SO4, 50 mM NaAc (pH 4.0), 9–12% PEG 4000	0.5 mg/ml E. coli lipids	Sitting drop (4 °C)	30% glycerol
	5HEG, 5HEH, 5HEJ, 5HEO, 5HEU, 5HEW					30% ethylene glycol
	2XQ3, 2XQ4, 2XQ5, 2XQ6, 2XQ7, 2XQ8, 2XQ9, 2XQA		200–500 mM (NH4)2SO4, 50 mM NaAc (pH 4.0), 9–12% PEG 4000			30% ethylene glycol
	4F8H		225 mM (NH4)2SO4, 50 mM NaAc (pH 3.9–4.1), 10–12% PEG 4000			20% glycerol
	5J0Z	9–10	225 mM (NH4)2SO4, 50 mM NaAc (pH 3.9–4.2), 10–12% PEG 4000			30% ethylene glycol
	4IRE	~10	225 mM (NH4)2SO4, 50 mM NaAc (pH 3.9–4.3), 10–12% PEG 4000		Hanging drop (4 °C)	20% glycerol
	4LMK, 4LMJ, 4LML	9–10	200–250 mM (NH4)2SO4, 50 mM NaAc (pH 3.9–4.3), 10–12% PEG 4000			30% ethylene glycol
	3EAM, 3TLS, 3TLT, 3TLU, 3TLV, 3YLW, 3UU3, 3UU4, 3UU5, 3UU6, 3UU8, 3UUB	8–12	400 mM NH4SCN, 100 mM NaAc (pH 4.6), 15% PEG 4000	-	Hanging drop (20 °C)	20% glycerol
	3P4W, 3P50, 4HFB, 4HFC, 4HFD, 4HFE, 4HFH		400 mM NaSCN, 100 mM NaAc (pH 4.0), 12-16% PEG 4000
	4HFI, 4IL4, 4IL9, 4ILA, 4ILB, 4ILC	-	400 mM NaSCN, 100 mM NaAc (pH 4.0), 12–15% PEG 4000
	5IUX					No further cryoprotection
	4NPP	6–8	50 mM NiCl2, 100 mM NaAc (pH 4.0), 15–20% PEG 2000MME		Hanging drop (20 °C), microseeding	18% glycerol
	4NPQ	10	200 mM KSCN, 10 mM CaCl2, 3% TMAO 100 mM NaHEPES (pH 7.5), 13–15% PEG 4000
	4QH1, 4QH5		400 mM NaSCN, 100 mM NaAc or NaH2/Na2H-PO4 (pH 4.0), 2% DMSO, 10% glycerol, 12–14.5% PEG 4000			No further cryoprotection
	5HCJ, 5HCM	6–8	400 mM NaSCN, 100 mM NaAc (pH 4.0), 17% glycerol, 12–16% PEG 4000,
	4ZZB, 4ZZC	-	400 mM NaSCN, 100 mM NaAc (pH 4.0), 12–15% PEG 4000		Hanging drop (18 °C), microseeding	20% glycerol
	5L47, 5L4E, 5L4H	-	400 mM NaSCN, 100 mM NaAc (pH 4.0), 3% DMSO, 16% glycerol, 12–14% PEG 4000			No further cryoprotection
ELIC-GLIC	4YEU	9-10	200 mM (NH4)2SO4, 50 mM ADA (pH 6.7–7.6), 7.5–10% PEG 4000	0.5 mg/ml E. coli lipids	Sitting drop (4 °C)	30% ethylene glycol
GLIC-GlyR	4X5T	6-8	50 mM NiCl2, 0.1 M NaAc (pH 3.0), 16–20% (vol/vol) PEG 2000MME, 4% (vol/vol) DMSO, 11% (vol/vol) ethylene glycol		Hanging drop (20 °C), microseeding
GluCl	3RHW, 3RI5, 3RIA, 3RIF	1–2	70 mM NaCl, 50 mM sodium citrate (pH 4.5), 21–23% PEG400	Fab antibody, POPC or DPPC	Hanging drop (4 °C)	30% PEG400
	4TNV, 4TNW	1–2	100 mM KCl, 50 mM sodium citrate (pH 5.5), 35–36% pentaerythritol propoxylate (5/4 PO/OH)	Fab antibody, POPC	Hanging drop (4 °C)	No further cryoprotection
β3GABAAR	4COF	3	100 mM NaCl, 100 mM Li2SO4, 100 mM ADA (pH6.5), 11.5% PEG 4000	2% (w/v) benzamidine	Sitting drop (6.5 °C)	20% glycerol
5HT3AR	4PIR	2.7	0.1 M Na2SO4, 0.1 M Tris (pH 8.5), 20–25% PEG 10000	1:2 VHH5 nanobody, 0.56mM Cymal-6	Hanging drop (4 °C)	20% glycerol or 40% PEG 10000
α3GlyR	5CFB	3	100 mM KCl, 200 mM MgCl2, 25 mM sodium citrate (pH 4.0), 30–33% PEG400	0.2 mM strychnine	Hanging drop (4 °C)	No further cryoprotection
	5VDH, 5VDI	~3	200 mM CaCl2, 100 mM MES (pH 6.5), 22.5–27.5% PEG350MME	0.5 mM glycine		30% PEG350MME
	5TIO, 5TIN	~3	200 mM CaCl2, 100 mM MES (pH 6.5), 22.5–27.5% PEG350MME	0.5 mM glycine
α4β2nAChR	5KXI	1.5–2.5	50 mM ADA (pH 6.8), 12.5% PEG 1500 and 10% PEG 1000	1 mM nicotine	Hanging drop (14 °C)	PEG 1000, PEG 1500 and ethylene glycol

The crystals of pLGICs bound with anesthetics can be obtained in two different ways: grown in the presence of anesthetics (co-crystallization) or soaked with an anesthetic-containing solution after crystals form. It appears that the co-crystallization technique has outperformed the soaking method based on the currently available crystal structures of pLGICs bound with anesthetics (Table 3). Once a favorable crystallization condition is identified for a given pLGIC, it can be used as a starting point to optimize further the pLGIC crystallization in the presence of anesthetics or other ligands. In the presence of anesthetics, the original optimized condition for apo pLGICs often needs to be adjusted by tweaking pH, salt compositions, and concentrations of precipitants. Additives, such as glycerol and DMSO, have been shown useful in co-crystallizing pLGICs with anesthetics (Fourati et al., 2017). Most general anesthetics have limited solubility in the aqueous phase; the anesthetic concentration chosen for co-crystallization must be balanced, as a low concentration may lead to low incorporation while a high concentration may degrade crystal diffraction. Likewise, soaking anesthetics into pLGIC crystals also requires a delicate balance between permeation of sufficient anesthetics into the crystals and potential disruption of the crystal lattice. Non-traditional soaking methods have proved successful for incorporating anesthetics into pLGICs, including adding saturated volatile desflurane to the reservoir well after GLIC crystals are observed (Nury et al., 2011) and pressuring xenon into GLIC crystals in a gas chamber (Sauguet, Fourati, Prange, Delarue, & Colloc’h, 2016).

Table 3.

Anesthetics identified in crystal structures of pLGICs

pLGICs	Anesthetics	PDB code (Resolution), Functional States	Reservoir solution	Crystallization Conditions
ELIC	Isofluorane	4Z91 (3.39 Å), resting 4Z90 (3.0 Å), desensitized	200 mM (NH4)2SO4, 100 mM MES (pH 6.1), 10–12% PEG 4000	Co-crystallization* Sitting drop (4 °C)
	2-Bromoethanol	5SXU (3.1 Å), desensitized 5SXV (3.4 Å), resting
	Bromoform	3ZKR (3.65 Å), resting	200 mM (NH4)2SO4, 50 mM ADA (pH 6.5), 9–12% PEG 4000
GLIC	Desflurane	3P4W (3.2 Å), open	400 mM NaSCN, 100 mM NaAc (pH 4.0), 12–16% PEG 4000	Saturating amount of desflurane added in the reservoir well after crystals were formed, hanging drop (20 °C)
	Propofol	3P50 (3.3 Å), open		Pure propofol added in crystallization mixture, hanging drop (20 °C)
	Ketamine	4F8H (2.99 Å), desensitized	225 mM (NH4)2SO4, 50 mM NaAc (pH 3.9–4.1), 10–12% PEG 4000	Co-crystallization*, sitting drop (4 °C)
	2-Bromoethanol	4HFC (3.05 Å), desensitized F14′A mutant	400 mM NaSCN, 100 mM NaAc (pH 4.0), 12–16% PEG 4000	Co-crystallization*, hanging drop (20 °C)
	Ethanol	4HFE (2.8 Å), desensitized F14′A mutant
	Bromoform	4HFD (3.1 Å), desensitized F14′A mutant 4HFH (2.65 Å), desensitized
		5HCM (3.15 Å), desensitized locally closed mutant	400 mM NaSCN, 100 mM NaAc (pH 4.0), 17% glycerol 2% bromoform, 12–16% PEG 4000
	Xenon	4ZZB (3.4 Å), desensitized locally closed mutant 4ZZC (3.1 Å), open	400 mM NaSCN, 100 mM NaAc (pH 4.0), 12–15% PEG 4000	Crystals pressured under xenon, hanging drop (18 °C)
	Seleniated barbiturate	5L47 (3.3 Å), desensitized locally closed mutant	400 mM NaSCN, 100 mM NaAc (pH 4.0), 3% DMSO, 16% glycerol, 12–14% PEG 4000	Co-crystallization, hanging drop (18 °C)
	Thiopental	5L4E (3.5 Å), desensitized locally closed mutant
	Brominated barbiturate	5L4H (3.3 Å), desensitized locally closed mutant

^*Anesthetics were co-crystallized with proteins

UDM: n-undecyl-β-D-maltopyranoside

DDM: n-dodecyl-β-D-maltopyranoside

DMNG: decyl maltose neopentyl glycol

C12E9: α-dodecyl-w-Hydroxy-Poly(oxy-1,2-Ethanediyl)

DPC: n-dodecylphosphocholine

POPC: 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine

Cryoprotection conditions for the crystallization of apo pLGICs normally also work for pLGICs bound with general anesthetics, but may need some modification. For cryoprotection, the cryoprotectant should contain a slightly higher concentration of anesthetics than that used for co-crystallization or soaking.

3.2. Data collection and processing

Synchrotron radiation facilities are usually used for collecting high-resolution diffraction data of pLGIC crystals. BioSync (http://biosync.sbkb.org/index.jsp) keeps up-to-date information about synchrotron facilities around the world. Crystals of pLGICs are often small, fragile, and prone to radiation damage, so a micro-focus beamline with a fast read-out detector is highly preferred. It enables data collection from small crystals or from the selected better-diffracted regions in heterogenous crystals (Riekel, Burghammer, & Schertler, 2005) and allows for complete data collection before severe radiation damage. Data collection for the anesthetic-containing crystals is similar to that for apo crystals, unless anesthetic molecules contain anomalous diffracting atoms. In such a case, data collection must be performed at or above the X-ray absorption edge so that the anomalous signal is maximized or close to the maximum. The collected diffraction data can be indexed, integrated, and scaled using software such as XDS (Kabsch, 2010a, 2010b), HKL2000 (Otwinowski & Minor, 1997), MOSFLM (Battye, Kontogiannis, Johnson, Powell, & Leslie, 2011), CCP4 (Winn et al., 2011), and d*TREK (Pflugrath, 1999). Initial structures are normally obtained by molecular replacement with the program PHASER (McCoy, 2007; Mccoy et al., 2007) using known structures of homologous channels as search models. Structure refinements are performed with PHENIX (Adams et al., 2010) or BUSTER (Blanc et al., 2004). COOT (Emsley & Cowtan, 2004) is used for manual examination and rebuilding of the coordinates based on the electron density maps of 2Fo-Fc and Fo-Fc. Structure analyses are performed using PHENIX (Adams et al., 2010) and stereochemistry is evaluated using MolProbity (V. B. Chen et al., 2010). PyMOL (DeLano, 2002) and VMD (Humphrey, Dalke, & Schulten, 1996) programs can be used for structural analysis and visualization.

3.3. Identification of anesthetics bound to pLGICs

Identification of anesthetics bound to pLGICs cannot begin until the structural model of a given pLGIC fits well to the X-ray electron density. This often requires an iterative building-refinement process. In addition to the pLGIC electron density, extra electron density will be observed if anesthetic molecules bound to the pLGIC and diffracted well. How well the shape and size of this extra electron density matches with the molecular features of the anesthetic determines the certainty of anesthetic assignment. Comparing the electron densities from apo and anesthetic-containing crystals obtained at otherwise the same conditions can boost the certainty, as demonstrated in the previous studies of ELIC and GLIC bound with ketamine (Pan, Chen, Willenbring, Mowrey, et al., 2012), isoflurane (Q. Chen et al., 2015), thiopental (Fourati et al., 2017), desflurane, or propofol (Nury et al., 2011). Anomalous signals from xenon, bromide or selenium atoms, if they exist in the chosen anesthetics or anesthetic analogs, will also boost the certainty of the assignment of anesthetics. Xenon (Sauguet et al., 2016), 2-bromoethanol (Q. Chen et al., 2017; Sauguet, Howard, et al., 2013), bromoform (Laurent et al., 2016; Spurny et al., 2013), and brominated or seleniated barbiturates (Fourati et al., 2017) have all been successfully identified based on anomalous signals.

4. Functional Validation of the Bound Anesthetics

Due to the low binding affinity (μM - mM) of general anesthetics, multiple binding sites may be observed in the crystal structures of pLGICs. However, the functional impact of anesthetic binding to each site may not be equal. Therefore, it is necessary to evaluate the roles of individual binding sites in triggering functional changes of pLGICs. Site-directed mutagenesis combined with functional measurements, such as two-electrode voltage clamp experiments, can fulfill this purpose (Q. Chen et al., 2015; Q. Chen et al., 2017; Nury et al., 2011; Pan, Chen, Willenbring, Mowrey, et al., 2012; Sauguet, Howard, et al., 2013).

5. Case Studies

5.1. Crystallization of ELIC with the volatile general anesthetic isofluorane or 2-bromoethanol (Q. Chen et al., 2015; Q. Chen et al., 2017)

ELIC, a cation-conducting pLGIC from Erwinia chrysanthemi, is functionally inhibited by the volatile anesthetic isoflurane and ethanol. Elucidating the molecular interactions between ELIC and isoflurane or ethanol aids in our understanding of inhibitory mechanisms of general anesthetics and alcohols on pLGICs.

As a prokaryotic pLGIC, ELIC expresses well in E. coli, especially when fused to the C-terminus of maltose binding protein (MBP). A Herpes simplex (HRV) 3C protease site was inserted between MBP and ELIC for their subsequent separation. The pelB signal sequence and His₁₀ purification tag were placed at the N-terminus of the MBP-ELIC fusion protein. The E. coli strain Rosetta(DE3)pLysS provided the best expression from the T7 promoter in the expression plasmid. The pLysS and expression plasmids were selected by chloramphenicol (35 μg/mL) and kanamycin (50 μg/mL), respectively. The expression began with an overnight preculture, which was diluted 1:100 into LB media containing antibiotics and grown at 37 °C. When optical density reached 0.9, the culture was harvested via centrifugation at 5000 × g for 20 min at 15 °C. The cell pellet was suspended in M9 media and incubated at 15 °C for 1-2 h before induction with 0.2 mM IPTG. After expression for 24 h, the cells were harvested and suspended in buffer A (150 mM NaCl, 50 mM sodium phosphate, pH 8) with a protease inhibitor cocktail. The cells were lysed with a M-110Y microfluidizer and the membrane fraction was collected at 4 °C by ultracentrifugation at 235,000 × g for ~1 h. The fusion protein His₁₀-MBP–ELIC was extracted from the membrane with 3.5% (wt/vol) n-undecyl-β-d-maltoside (Anatrace) in buffer A, clarified by centrifugation at 5000 × g for 20 min, and purified using Ni-NTA chromatography at 4 °C. The purified fusion protein was cleaved overnight with HRV3C protease and ELIC was separated from MBP using Ni-NTA chromatography. The pentameric fraction of ELIC was obtained by SEC using a Superdex 200 10/300GL column. The buffer was exchanged to 0.025% (wt/vol) n-dodecyl-β-d-maltoside (Anatrace) in 10 mM sodium phosphate (pH 8) and 150 mM NaCl at this step. ELIC was concentrated to ~5-6 mg/mL for crystallization.

ELIC crystals were obtained using the vapor-diffusion method in sitting drops at 4 °C. E. coli polar lipids (0.01–0.02 mg/mL) and 2-bromoethanol (100 mM) were mixed and equilibrated with ELIC for at least half an hour before setting up for crystallization trays. The equilibrated protein was mixed in a 1:1 ratio with the reservoir solution and resulted in a 2-bromoethanol concentration of 50 mM. The reservoir solution contained 10–12% PEG 4000, 200 mM ammonium sulfate, and 100 mM MES buffer at pH 6.1. The crystallization of ELIC with isoflurane followed a similar procedure except that isoflurane was not directly added to the protein and was instead added to the reservoir solution at its saturated concentration (14.4 mM) (Xu, Tang, Firestone, & Zhang, 1996). The amount of E. coli polar lipids, 2-bromoethanol, and isoflurane were adjusted and optimized for the best diffracting crystals. The crystals typically formed in one or two weeks and were harvested after briefly soaking in the reservoir solution supplemented with 20% glycerol and 100 mM 2-bromoethanol or a saturated isoflurane solution. The harvested crystals were flash-frozen in liquid nitrogen for short storage before shipping to a synchrotron facility for data collection.

X-ray crystal structures of ELIC bound with isoflurane were determined with a resolution up to 3.0 Å (Q. Chen et al., 2015). The high electron density in the Fo-Fc density map revealed double occupancies of isoflurane inside the pore of ELIC, near T237(6’) and A244(13’). A contraction of the pore radius was observed near the extracellular entrance upon isoflurane binding. Electrophysiology measurements of oocytes expressing ELIC with a single-point mutation at either the T237(6’) or A244(13’) position validated that binding to these sites is responsible for the inhibitory action of isoflurane. The study presents compelling evidence for a direct pore-binding mechanism of isoflurane inhibition.

The X-ray crystal structure of ELIC bound with 2-bromoethanol was solved at 3.1 Å (Q. Chen et al., 2017) and showed multiple 2-bromoethanol binding sites, including inside the pore near T237(6’) in the transmembrane domain (TMD) and at three different locations in the extracellular domain (ECD). The strong anomalous signal from the bromide atom of 2-bromoethanol facilitated the determination of these binding sites, which would be difficult, if not impossible, to directly determine for the small, low affinity ethanol molecule. Mutations to the three binding sites in the ECD did not alter the functional response of ELIC to 2-bromoethanol or ethanol, suggesting a negligible contribution of ethanol binding in the ECD to the functional modulation of ELIC. In contrast, both 2-bromoethanol and ethanol showed reduced potencies on the T237(6’)A mutant, indicating that the binding to the pore predominately contributes to the inhibition of ELIC by ethanol. This conclusion was further substantiated using an ELIC-α1β3GABA_AR chimera, which has the ECD of ELIC but the TMD of α1β3GABA_AR. While 2-bromoethanol and ethanol inhibit ELIC, both 2-bromoethanol and ethanol potentiate the ELIC-α1β3GABA_AR chimera. Replacing the TMD of ELIC reverses the functional effect of 2-bromoethanol and ethanol, supporting a pore-blocking mechanism of short-chain alcohols in ELIC, and likely also in other cation-conducting pLGICs.

5.2. Crystallization of GLIC with the intravenous general anesthetic ketamine (Pan, Chen, Willenbring, Yoshida, et al., 2012)

GLIC, a prokaryotic cation-conducting pLGIC from Gloeobacter violaceus, is inhibited by ketamine, an anesthetic commonly known as an antagonist of NMDA receptors. The crystal structure of ketamine bound to GLIC provides novel insight into the molecular recognition and allosteric anesthetic modulation of pLGICs.

The expression and purification of GLIC were conducted in a similar manner as done for ELIC, including the use of the fusion protein MBP and the E. coli strain Rosetta(DE3)pLysS. The purified GLIC was concentrated, equilibrated with 0.5 mg/mL E. coli polar lipids in the presence of ketamine (1 mM) for at least 1 h at 4 °C, and then mixed in a 1:1 ratio with the reservoir solution (10–12% PEG 4000, 225 mM ammonium sulfate, and 50 mM sodium acetate buffer at pH 3.9 – 4.1). Crystals were obtained using the sitting-drop method at 4 °C and usually appeared within one week. The crystals were cryoprotected by soaking in the reservoir solution supplemented with 20% glycerol and 10 mM ketamine for 30 min. The harvested crystals were flash frozen in liquid nitrogen for short storage before data collection at a synchrotron facility.

The X-ray structure of GLIC bound with ketamine was determined at 2.99 Å. Ketamine was identified in a single inter-subunit cavity in the ECD in each of five equivalent subunits of GLIC. The binding site partially overlaps with the antagonist-binding site in homologous pLGICs. The functional impact of ketamine binding to the site revealed in the crystal structure was further investigated by additional electrophysiology experiments. A point mutation at N152C inside the ketamine-binding site greatly altered the activation of GLIC by protons and decreased the potency of ketamine inhibition on GLIC. In addition, chemical labeling of N152C with 8-(chloromercuri)-2-dibenzofuransulfonic acid (CBFS) resulted in the inhibition of GLIC, mimicking the effect of ketamine binding at this site. Treatment with 10 mM dithiothreitol (DTT) removed the CBFS label at N152C and brought the channel back to a normal function. The success in identifying the ketamine binding site by crystallography and functional validation of the relevance of this site for ketamine inhibition of GLIC provides a useful template for future studies of anesthetic binding to pLGICs.