Structural comparison of cytochromes P450 2A6, 2A13, and 2E1 with pilocarpine

Authors


E. E. Scott, Department of Medicinal Chemistry, University of Kansas, 1251 Wescoe Hall Dr., Lawrence, KS 66045, USA
Fax: +1 785 864 5326
Tel: +1 785 864 5559
E-mail: eescott@ku.edu

Abstract

Human xenobiotic-metabolizing cytochrome P450 (CYP) enzymes can each bind and monooxygenate a diverse set of substrates, including drugs, often producing a variety of metabolites. Additionally, a single ligand can interact with multiple CYP enzymes, but often the protein structural similarities and differences that mediate such overlapping selectivity are not well understood. Even though the CYP superfamily has a highly canonical global protein fold, there are large variations in the active site size, topology, and conformational flexibility. We have determined how a related set of three human CYP enzymes bind and interact with a common inhibitor, the muscarinic receptor agonist drug pilocarpine. Pilocarpine binds and inhibits the hepatic CYP2A6 and respiratory CYP2A13 enzymes much more efficiently than the hepatic CYP2E1 enzyme. To elucidate key residues involved in pilocarpine binding, crystal structures of CYP2A6 (2.4 Å), CYP2A13 (3.0 Å), CYP2E1 (2.35 Å), and the CYP2A6 mutant enzyme, CYP2A6 I208S/I300F/G301A/S369G (2.1 Å) have been determined with pilocarpine in the active site. In all four structures, pilocarpine coordinates to the heme iron, but comparisons reveal how individual residues lining the active sites of these three distinct human enzymes interact differently with the inhibitor pilocarpine.

Database
Structural data are available in the Protein Data Bank database under the accession numbers 3T3Q (CYP2A6 I208S/I300F/G301A/S369G with pilocarpine), 3T3R (CYP2A6 with pilocarpine), 3T3S (CYP2A13 with pilocarpine), and 3T3Z (CYP2E1 with pilocarpine)

Abbreviations
CYP

cytochrome P450

PDB

Protein Data Bank

rmsd

root mean square deviation

RPS

reconstituted protein system

TCA

trichloroacetic acid

Introduction

The cytochrome P450 (CYP) superfamily EC1.14.14.1 contains enzymes that catalyze oxidative metabolism of the myriad of lipophilic drugs and xenobiotics to which humans are exposed. Each different CYP enzyme can bind to, metabolize, and be inhibited by a distinctive set of chemically and structurally diverse compounds. However, a single compound can often interact with multiple CYP enzymes. Understanding the structural basis for this overlapping ligand selectivity is important in order to rationalize and predict the participation of various human CYP enzymes in drug and xenobiotic metabolism, as well as their respective susceptibilities to inhibitors.

In humans, the CYP2A and CYP2E subfamilies represent an excellent system for such structure–function investigations. In humans, the functional CYP2A enzymes consist of two enzymes with 94% sequence similarity: CYP2A6, which is largely hepatic [1], and CYP2A13, which is primarily located in the respiratory tract [2,3]. Humans also have a single CYP2E1 enzyme, which is primarily hepatic [1] and has < 40% sequence similarity with the CYP2A enzymes. Nonetheless, CYP2E1 oxidizes and is inhibited by many of the same compounds as CYP2A enzymes. For example, p-nitrophenol and chlorzoxazone have traditionally been used as substrates to indicate the presence of active CYP2E1, especially in microsomes, where mixtures of CYP enzymes are present, but CYP2A13 has a higher catalytic efficiency than CYP2E1 with both substrates [4]. In comparison, CYP2A6 has a lower catalytic efficiency than CYP2E1 for chlorzoxazone 6-hydroxylation, and equivalent catalytic efficiency for p-nitrophenol 2-hydroxylation [4]. Styrene and toluene are also metabolized by CYP2E1, CYP2A13, and CYP2A6 [4].

Although there are various structures available for CYP2A6, CYP2A13, and CYP2E1, none of them offer a direct comparison, because they all contain different ligands [5–9], and a single CYP enzyme can often bind different ligands with different active site topologies. This appears to be especially true for CYP2E1, which is known to have dramatically different active site volumes and topographies, depending on the ligand [5,6]. To probe the differences and similarities between the human CYP2A and CYP2E1 enzymes, we utilized the muscarinic receptor agonist pilocarpine [(3S,4R)-3-ethyl-4-[(1-methyl-1H-imidazol-5-yl)methyl]dihydrofuran-2(3H)-one; Fig. 1]. Pilocarpine was previously identified as a competitive inhibitor of human CYP2A6 and, to a lesser extent, of CYP2B6 [10,11]. We have determined herein that pilocarpine also inhibits CYP2A13 and CYP2E1.

Figure 1.

 Difference spectra upon titration of CYP enzymes with pilocarpine and the resulting binding constants. Black arrows indicate the direction of type II shifts. (A) CYP2A6. (B) CYP2A13. (C) CYP2A6 I208S/I300F/G301A/S369G. (D) CYP2E1. The inset in (A) shows the structure of pilocarpine. In (D), the red arrows indicate the type I spectral shifts that occur at lower pilocarpine concentrations, and the black arrows indicate the type II shifts observed at higher pilocarpine concentrations. Kd values are the average of duplicate titrations, except for CYP2A13, where three titrations were performed.

In the present study, functional analysis of CYP2A13, CYP2A6 and CYP2E1 with pilocarpine combined with X-ray structures of these complexes have allowed direct comparisons among these three different human CYP enzymes. In addition, wild-type CYP2A13 and CYP2A6 were compared with a previously identified CYP2A6 mutant, CYP2A6 I208S/I300F/G301A/S369G, that gained the ability to metabolize phenacetin, like CYP2A13 [12]. Previous studies suggested that these four residues were the primary determinants for CYP2A phenacetin metabolism, and might be important for differential recognition of other ligands by CYP2A enzymes. The functional and structural results presented herein allow for a direct comparison of all three enzymes, and provide insights into the observed differences in inhibition by pilocarpine.

Results

Pilocarpine binding

Titrations of pilocarpine with CYP2A6, CYP2A6 I208S/I300F/G301A/S369G and CYP2A13 resulted in increases in absorbance at 424 nm and decreases at 420 nm, consistent with a type II interaction, in which a nitrogen coordinates to the heme iron (Fig. 1A–C). Titrations of CYP2E1 with pilocarpine were more unusual (Fig. 1D). At low pilocarpine concentrations, absorbance increases were observed at 391 nm and decreases at 430 nm, indicative of a type I interaction, in which pilocarpine binding displaces a water molecule from the heme iron but does not interact directly with the heme itself. However, at higher pilocarpine concentrations, the type I shift disappeared and a type II shift was observed, as with the CYP2A enzymes. The dissociation constants for type II binding to all three CYP2A enzymes were very similar (1.5–3.6 μm; Fig. 1A–C), whereas the same parameter for CYP2E1 was 10-fold higher (26.5 μm; Fig. 1D).

Pilocarpine inhibition

Pilocarpine inhibition constants were determined for CYP2A6, CYP2A6 I208S/I300F/G301A/S369G, CYP2A13, and CYP2E1 with both common and selective substrates (Table 1). Coumarin was used as the selective substrate for all of the CYP2A enzymes, whereas chlorzoxazone was used as a substrate for CYP2E1. As CYP2E1, CYP2A13 and CYP2A6 all metabolize p-nitrophenol, pilocarpine inhibition was also evaluated for all three wild-type enzymes with p-nitrophenol as the substrate. These results revealed that pilocarpine is a competitive inhibitor of CYP2A13 for both coumarin and p-nitrophenol, although the Ki was 34-fold higher for coumarin. CYP2A6 inhibition by pilocarpine revealed mixed inhibition, with Ki values for both substrates two-fold higher than for CYP2A13. CYP2A6 I208S/I300F/G301A/S269G showed competitive inhibition, with a Ki for inhibition of coumarin 7′-hydroxylation nearly identical to that of wild-type CYP2A13 with the same substrate. Pilocarpine inhibited CYP2E1 least effectively, with noncompetitive inhibition with both chlorzoxazone and p-nitrophenol as substrates. By comparison, pilocarpine had a Ki value for CYP2E1 546-fold greater than that for CYP2A13 with the same substrate.

Table 1.   Inhibition constants for pilocarpine for CYP2A6, CYP2A6 I208S/I300F/G301A/S369G, CYP2A13, and CYP2E1. For Ki, all assays were performed in triplicate. Ki/Ki2A13, fold difference between the indicated Ki and the Ki for CYP2A13 with the corresponding substrate.
EnzymeSubstrateType of inhibitionKim)Ki/Ki2A13
  1. a Mixed inhibition with the larger α-value favors competitive inhibition, whereas the smaller α-value indicates more uncompetitive character.

CYP2A13CoumarinCompetitive48 ± 41
p-NitrophenolCompetitive1.4 ± 0.11
CYP2A6CoumarinMixed, α = 6.98a101 ± 172.1
p-NitrophenolMixed, α = 24.9a3.0 ± 0.52.1
CYP2A6 I208S/I300F/G301A/S369GCoumarinCompetitive49 ± 31
CYP2E1ChlorzoxazoneNoncompetitive360 ± 30 
p-NitrophenolNoncompetitive765 ± 30546

Structure of CYP2A6 with pilocarpine

The protein backbone of CYP2A6 with pilocarpine is very similar to other structures of CYP2A6. The average rmsd value of the CYP2A6–pilocarpine structure with available structures of CYP2A6 with coumarin, methoxsalen, N,N-dimethyl[5-(pyridin-3-yl)furan-2-yl]methanamine, N-methyl[5-(pyridin-3-yl)furan-2-yl]methanamine, [5-(pyridin-3-yl)furan-2-yl]methanamine and adrithiol is 0.24 Å [8,9]. In the present CYP2A6 structure, density for pilocarpine was located directly above the heme in all four CYP2A6 molecules. This density clearly oriented pilocarpine with the imidazole ring closest to the heme, with a distance of only 2.3 Å from the unsubstituted nitrogen in the imidazole ring to the heme iron (Fig. 2A). This is consistent with the type II spectral shift observed in the spectral ligand-binding assay (Fig. 1A). The furan ring packs against the I-helix with a weak hydrogen bond from the exocyclic keto oxygen of pilocarpine to the Asn297 NH2 donor (3.4 Å). The ethyl group of the furan ring is directed towards Leu370. Additional residues that pack against pilocarpine include Phe107, Phe209, Val117, Phe118, Ile300, Leu370, and Ile366.

Figure 2.

 Pilocarpine binding in the active site of CYP2A enzymes. (A) CYP2A6. (B) CYP2A13. (C) CYP2A6 I208S/I300F/G301A/S369G. (D) CYP2E1. In each panel, the 2|Fo| − |Fc| electron density is 1.0σ around the ligand and heme (blue mesh). In (B), the σA-weighted composite omit map is also shown in pink mesh.

Structure of CYP2A13 with pilocarpine

The structure of CYP2A13 with pilocarpine has a protein backbone very similar to that of a structure of CYP2A13 with indole [7] (rmsd 0.53 Å). Owing to the lower (3.0 Å) resolution, the density for pilocarpine was not as well defined in the CYP2A13 active site, but the ligand could be modeled with reasonable certainty in six of the eight molecules. The 2|Fo| − |Fc| map (Fig. 2B, blue mesh) has one portion of the pilocarpine with poor coverage – the ethyl side chain – but its placement was determined on the basis of the σA-weighted composite omit map (Fig. 2B, pink mesh). Similarly to the CYP2A6 structure, pilocarpine in the CYP2A13 active site is oriented with the imidazole ring closest to the heme, with a distance of only 2.5 Å from the unsubstituted nitrogen to the heme iron. This is also consistent with the type II spectral shift observed in the spectral ligand-binding assay for this enzyme (Fig. 1B). As in the case of CYP2A6, there appears to be a hydrogen bond from the furan ring keto oxygen to Asn297 (3.0 Å). The furan ring has a similar overall placement as in CYP2A6, except for the ethyl side chain, which is directed in the opposite direction – towards Phe300 rather than towards Leu370. Phe107, Ala117, Phe118, Phe209, Phe300, Leu370 and Leu366 all pack against pilocarpine in the CYP2A13 active site (Fig. 2B).

Structure of CYP2A6 I208S/I300F/G301A/S369G with pilocarpine

The protein backbone of CYP2A6 I208S/I300F/G301A/S369G is very similar to that of the same mutant with the substrate phenacetin bound (rmsd 0.27 Å). This CYP2A6 I208S/I300F/G301A/S369G structure also contained four copies of CYP2A6 in the asymmetric unit, and each copy contained density corresponding to pilocarpine clearly defined above the heme. As in the CYP2A13 and the CYP2A6 structures, the imidazole ring of pilocarpine is located close to the heme, with the unsubstituted nitrogen 2.5 Å from the heme iron (Fig. 2C), consistent with the type II spectral shift observed in the spectral ligand-binding assay (Fig. 1C). Although the plane of the furan ring is different, in that it stacks with Phe107 instead of against the I-helix as in CYP2A6 and CYP2A13, the hydrogen bond from the keto oxygen on the furan ring to Asn297 (3.0 Å) is maintained. Additional residues that pack against the pilocarpine include Phe107, Val117, Phe209, Phe300, Ala301, Ile366, and Leu370.

Structure of CYP2E1 with pilocarpine

The protein backbone of the CYP2E1 structure with pilocarpine is very similar to that of CYP2E1 with indazole, 4-methylprazole, and ω-imidazolyl fatty acid analogs [5,6], with an average rmsd of 0.31 Å. As in the CYP2A structures, the density in this 2.35-Å structure supported the clear placement of pilocarpine with the imidazole ring ∼ 2.18 Å from the heme iron (Fig. 2D), consistent with the type II spectral shift observed at high pilocarpine concentrations (Fig. 1D). In this orientation, no hydrogen bond is observed between pilocarpine and active site residues. Despite the presence of Asp295 in CYP2E1 at the position corresponding to Asn297 in CYP2A enzymes, in CYP2E1 Asp295 is shielded from pilocarpine by Ile115, and, as a result, the CYP2E1 active site is much more hydrophobic. Instead, pilocarpine is surrounded by several hydrophobic residues, including Ile115, Leu368, Val364, and phenylalanines (Phe116, Phe106, Phe207, Phe298, and Phe478), with Thr303 as the only polar residue.

Discussion

Comparison of CYP2A6 and CYP2A13

There are many similarities between the functional and structural characteristics of CYP2A6 and CYP2A13. CYP2A6 and CYP2A13 bind pilocarpine with similar, single-digit micromolar affinities, and have only a two-fold difference in the Ki value with two different substrates. The overall structural similarity is very high (rmsd 0.63 Å). The structural and functional evidence agree that pilocarpine binds with an imidazole nitrogen directly coordinated to the heme iron (Figs 1A,B and 3A). Additionally, both structures show the exocyclic oxygen of pilocarpine positioned within hydrogen-bonding distance to the conserved Asn297, one of only two polar residues lining the active site. The primary difference in pilocarpine binding to these two wild-type enzymes is in the orientation of the ethyl group of the furan ring (Fig. 3A). In CYP2A13, this ethyl group is directed towards residue 300 and away from Phe118 and Leu370, whereas in CYP2A6 the ethyl group is oriented in the opposite direction, towards Phe118 and Leu370 and away from residue 300. Residue 300 is a phenylalanine in CYP2A13 and an isoleucine in CYP2A6, whereas Phe118 and Leu370 are conserved. Although there are 11 first-shell and second-shell residue differences between the CYP2A6 and CYP2A13 active sites, the side chain present at position 300 may be one of the most significant differences between the two active sites. The identity of the residue at position 300 correlates not only with the ethyl orientation in pilocarpine in the CYP2A6 and CYP2A13 structures reported herein, but also with the ability to bind and monooxygenate phenacetin [12]. This is also a key residue in the binding of other ligands, including 2′-methoxyacetophenone, phenethyl isothiocyanate, and coumarin [13]. In addition to the role for the nonconserved residue at position 300, there are also several differences in the orientation of the three conserved phenylalanines, Phe118, Phe107, and Phe209, that line the active site. Overall, the sizes of the two active sites are similar, with the CYP2A6 volume (281.7 Å3) being slightly smaller than the CYP2A13 volume (309.4 Å3), but the proportions are different (Fig. 3B). The CYP2A13 active site has more space available for ligands near Phe300 and Phe209, owing to a combination of the phenylalanine at position 300 and positioning of Phe209 away from the active site in the CYP2A13 structure, whereas the CYP2A6 active site has more volume available for the ligand near Phe118 and above Leu370 (Fig. 3B).

Figure 3.

 Structural comparisons of CYP2A enzymes. Heme is shown as black sticks and iron as a red sphere. (A) Pilocarpine binds similarly in the CYP2A13 (yellow) and CYP2A6 (pink) active sites, with the imidazole nitrogen coordinated to the heme iron and the furan exocyclic oxygen hydrogen bonded to Asn297. (B) CYP2A13 and CYP2A6 active sites [colored as in (A)], with the corresponding mesh illustrating the cavity volumes. An increased active site volume is available near residue 300 in CYP2A13 and near Phe118 in CYP2A6. (C) Comparison of CYP2A13 and CYP2A6 active sites [colored as in (A)] with that of CYP2A6 I208S/I300F/G301A/S369G (green). Although the imidazole ring–Fe interaction and hydrogen bond to Asn297 are conserved, the furan ring of pilocarpine is positioned differently in CYP2A6 I208S/I300F/G301A/S369G.

Comparison of CYP2A6, CYP2A6 I208S/I300F/G301A/S369G, and CYP2A13

CYP2A6 I208S/I300F/G301A/S369G has a binding affinity for pilocarpine similar to that of both CYP2A wild-type enzymes, but a Ki for pilocarpine that is essentially identical to that of CYP2A13 and thus two-fold lower than that of CYP2A6 (Fig. 1; Table 1). On the basis of both this information and the ability of this CYP2A6 quadruple mutant to bind and de-ethylate phenacetin like wild-type CYP2A13 (as opposed to CYP2A6, which does not bind phenacetin [12]), we had proposed that CYP2A6 I208S/I300F/G301A/S369G might orient pilocarpine as observed in wild-type CYP2A13 rather than as observed in CYP2A6.

Comparison of the three pilocarpine CYP2A structures (Fig. 3C) reveals that two key features are shared in all three structures: (a) the placement of the imidazole ring and direct coordination of the nitrogen to the heme iron; and (b) a conserved hydrogen bond from the keto oxygen on the furan ring to Asn297. In CYP2A6 I208S/I300F/G301A/S369G, the ethyl side chain of pilocarpine is oriented towards residue 300 and away from Phe118 and Leu370, as in the CYP2A13 wild-type structure. Additionally, several residues are found in essentially identical positions as in CYP2A13, including the mutated residues Phe300, Ala301, Ser208, and Gly369. However, despite maintaining a hydrogen bond with Asn297, the pilocarpine furan ring is oriented very differently in the CYP2A6 I208S/I300F/G301A/S369G structure (Fig. 3C) than in either the CYP2A6 or CYP2A13 structure. Instead of packing against the I-helix, the furan ring rotates in CYP2A6 I208S/I300F/G301A/S369G, so that the furan face is more parallel with Phe107. The two nearby residues that differ significantly in their positions between CYP2A6 I208S/I300F/G301A/S369G and CYP2A13 are the conserved Leu370 and Phe118. In CYP2A6 I208S/I300F/G301A/S369G, the Leu370 side chain is deflected away from the central active site as compared with its position in CYP2A13 or CYP2A6, allowing sufficient volume for the furan ring of pilocarpine to rotate in a different manner than observed in CYP2A13 or CYP2A6 (Fig. 3C). Phe118 is found in the same position in CYP2A6 I208S/I300F/G301A/S369G as in wild-type CYP2A6, a position that also results in additional active site volume (Fig. 3C). In addition, there is one residue in this region that differs between CYP2A13 and CYP2A6 – residue 117, which is a valine in CYP2A6 and CYP2A6 I208S/I300F/G301A/S369G, and an alanine in CYP2A13. This residue, in addition to the position of the conserved Leu370, may play a role in the positioning of pilocarpine in the active site of CYP2A6 along with residues 208, 300, 301, and 369. Thus, although CYP2A6 I208S/I300F/G301A/S369G has nearly identical functional characteristics to CYP2A13 with both the substrate phenacetin and the inhibitor pilocarpine, its active site is not a complete structural mimic of that of CYP2A13 for all ligands.

Comparison of the inhibitor pilocarpine in CYP2E1 and CYP2A enzymes

Functional evaluation indicates that CYP2E1 has a type II affinity for pilocarpine that is approximately eight-fold lower than that of the wild-type CYP2A enzymes and has a 250-fold to 550-fold higher Ki (Fig. 1; Table 1). Pilocarpine binding to CYP2E1 is unusual in that, at low concentrations of pilocarpine, type I binding is observed, converting to type II binding at higher pilocarpine concentrations (Fig. 1D). This suggests that pilocarpine may bind in more than one mode. In addition, inhibition assays with pilocarpine indicate that pilocarpine is a noncompetitive inhibitor of CYP2E1. The noncompetitive nature of the inhibitor suggested that pilocarpine might bind in a second mode or second location. There is evidence for a second binding site in the sigmoidal kinetics of several human xenobiotic CYP enzymes, and some structural evidence as well. Most notable is a crystal structure of CYP3A4 with the antifungal ketoconazole that identified two molecules of ketoconazole in the active site, one located close to the heme and the second located above the first [14]. The structure of CYP2C9 with warfarin also has evidence of a secondary binding site, as warfarin is located too far from the heme for metabolism to occur [15]. However, the CYP2E1 structure with pilocarpine showed no evidence for pilocarpine binding anywhere other than immediately adjacent to the heme in the active site of CYP2E1. This is probably attributable to the very high concentration of pilocarpine (100 mm) present in the crystallization condition. The type I shift was only present at low concentrations of pilocarpine (< 10 μm).

CYP2A enzymes and CYP2E1 have active sites that are primarily hydrophobic, with a roof composed of phenylalanines. Several of the key hydrophobic residues in CYP2E1 have equivalents in the CYP2A enzymes (Fig. 4A). For example, CYP2E1 has a leucine at position 368 that is structurally equivalent to the leucine at position 370 in CYP2A enzymes. There are also several structurally equivalent phenylalanines that result in a very differently shaped active site in CYP2E1 than in the CYP2A enzymes (Phe207, Phe298 and Phe478 in CYP2E1 versus Phe209, Phe300 and Phe480 in CYP2A) (Fig. 4A). At the position corresponding to the hydrogen bond donor Asn297, the CYP2E1 active site has Asp295, but it does not hydrogen bond with pilocarpine, because Ile115 and Phe298 block the residue from actually forming part of the active site. Perhaps in part because of these differences, the pilocarpine orientation in the CYP2E1 active site is much more perpendicular to the heme than in the CYP2A enzymes, where this axis is more parallel to the I-helix (Fig. 4A). The hydrogen bond interaction with Asn297 in CYP2A enzymes, in addition to the imidazole ring interaction with the heme, may account for the higher binding affinity and lower Ki values for pilocarpine in the CYP2A enzymes.

Figure 4.

 Comparison of the CYP2E1–pilocarpine complex with CYP2A structures with pilocarpine and CYP2E1 structures with other ligands. Heme is shown as black sticks, and iron as a red sphere. (A) The active sites of CYP2E1 (blue), CY2A6 (pink) and CYP2A13 (yellow) are similar, in that all three contain several hydrophobic residues, including key phenylalanines, which form the top of the active site (numbering is CYP2A/CYP2E1). (B) CYP2E1, showing the active site voids (as mesh) for three different ligands: pilocarpine (blue), indazole (PDB 3E6I, cyan), and ω-imidazolyl-dodecanoic acid (PDB 3LC4, magenta). Changes in the active site topology are largely caused by reorientation of either the Phe478 or Phe298 side chain, with little alteration of the overall protein backbone. The figure is presented as a wall-eyed stereo figure.

In the first shell of residues forming the CYP2E1 active site, the absence of a residue corresponding to CYP2A Asn297 that can participate in a ligand-stabilizing hydrogen bond may also explain why pilocarpine binding to CYP2E1 can display both type I and type II interactions. In addition to the orientation observed in the structure with imidazole interacting with the heme (type II), pilocarpine may also bind in a noncoordinating orientation (type I). The available active site space suggests that pilocarpine may also be able to bind so that the dihydrofuran ring orients toward the heme, with the exocyclic oxygen interacting with the only hydrophilic residue, Thr303.

Comparison of the CYP2E1 active site with structures containing other ligands

The volume of the CYP2E1 active site is ∼ 190 Å3 with the low molecular mass ligands indazole or 4-methylpyrazole, and 420–470 Å3 with fatty acid substrate analogs of various chain lengths [5,6] (Fig. 4B). The active site volume of CYP2E1 with pilocarpine is 332.7 Å3, a volume slightly larger than those of the active sites of CYP2A6 and CYP2A13 with pilocarpine (280 and 310 Å3, respectively), and intermediate between those of the other reported CYP2E1 structures. However, the active site topology with pilocarpine is different yet again from the other two types of CYP2E1 structures reported previously. Differences in the CYP2E1 active site dimensions with fatty acid versus low molecular mass ligands arise through rotation of the Phe298 side chain without a shift in the backbone [6]. The CYP2E1–pilocarpine backbone also overlays with both of the other types of CYP2E1 structures, but in this case rotation of the Phe478 side chain opens the opposite wall of the active site to accommodate pilocarpine (Fig. 4B).

Conclusions

The current set of structural and functional data allows for a detailed comparison of binding of the same ligand across the human CYP2A and CYP2E subfamilies. In the current study, pilocarpine binding was different for both CYP2A6 and CYP2A13, as well as for CYP2E1, and these disparities can be correlated with detailed differences in the respective active sites. Overall, the binding of pilocarpine in the active site of CYP2A enzymes appears to be driven primarily by hydrogen bonding with the conserved Asn297 and steric interactions largely driven by both a small set of nonconserved residues (300, 301, 208, 369 and, perhaps, 117) and indirect effects of a few conserved residues such as Leu370 and Phe209. In other words, the pilocarpine ligand largely adapts its conformation to fit into the CYP2A active sites. Remarkably, the CYP2E1 active site appears to bind ligands in an entirely different manner. The binding of pilocarpine induces yet a third distinct conformation of the CYP2E1 active site, generated by changes in side chain conformations. In other words, CYP2E1 demonstrates much larger variation in the active site topology to accommodate various ligands. To date, none of the liganded structures of CYP2A enzymes have indicated this same type of flexibility. It remains to be seen whether structures of CYP2A enzymes with other ligands will reveal active sites that have essentially the same topology as those already known, or whether additional data will reveal more diversity in the CYP2A active site dimensions. This is especially true for CYP2A13, as only two structures have been reported to date, including the one presented herein. If the CYP2A structures are much less flexible than those observed for CYP2E1 and other xenobiotic CYP enzymes, this has positive implications for the accuracy of docking in predicting drug metabolism and procarcinogen bioactivation by these human CYP2A enzymes.

Experimental procedures

Protein design, expression, and purification

Truncated, His-tagged, fully functional versions of human CYP2A6, CYP2A6 I208S/G301A/I300F/S369G, CYP2A13 and CYP2E1 were designed, expressed and purified as described previously [5,12].

Spectral binding assay

Ligand-binding affinities were determined with a previously described spectral-ligand binding assay [13].

Enzyme assay

All metabolism and inhibition assays used a reconstituted protein system (RPS) consisting of 50 pmol of purified CYP2A or CYP2E1 incubated with 200 pmol of NADPH-cytochrome P450 reductase and 100 pmol of cytochrome b5 in a 1:4:2 ratio for 20 min at room temperature prior to use. This RPS mixture was added to assay buffer containing the desired substrate and inhibitor. The samples were preincubated at 37 °C for 3 min, and the reactions were initiated by the addition of 1 mm NADPH. Samples were incubated for 10 min at 37 °C, and the reaction was stopped with 300 μL of 20% trichloroacetic acid (TCA) on ice. All standards and zero samples had 300 μL of 20% TCA added prior to the addition of NADPH. Samples and standards were centrifuged at 4500 g for 10 min.

Coumarin 7-hydroxylation

RPS was added to buffer (50 mm Tris, pH 7.4, and 5 mm MgCl2) containing coumarin and pilocarpine. After completion of the assay as described above, all samples and standards were diluted by the addition of 1 mL (CYP2A6) or 200 μL (CYP2A13) of buffer (50 mm Tris, pH 7.4, and 5 mm MgCl2). The amount of 7-hydroxycoumarin present was determined by fluorescence following HPLC separation. The 7-hydroxycoumarin metabolite was detected by fluorescence with an excitation wavelength of 355 nm and an emission wavelength of 460 nm. The mobile phase consisted of 30% MeOH, 68% water, and 2% acetic acid, with a 1 mL·min−1 flow rate and a sample volume of 10 μL.

p-Nitrophenol 4-hydroxylation

RPS was added to 100 mm potassium phosphate (pH 6.8) with 2 mm ascorbic acid (for CYP2E1), p-nitrophenol, and pilocarpine. After assay completion as described above, the amount of 4-nitrocatechol was determined following HPLC separation. A mobile phase of 27% acetonitrile and 0.2% acetic acid was run at a rate of 1 mL·min−1. Monitoring of absorption at 345 nm allowed the detection of 4-nitrocatechol at ∼ 5 min, and p-nitrophenol at ∼ 8 min.

Chlorzoxazone 6-hydroxylation

RPS was added to 100 mm potassium phosphate buffer (pH 7.4) and varying concentrations of chlorzoxazone and pilocarpine. Sample reactions were stopped after 10 min at 37 °C by the addition of 25 μL of 60% perchloric acid, instead of TCA. Corresponding standards and controls had 25 μL of 60% perchloric acid, instead of TCA, added prior to the addition of NADPH. The 6-hydroxychlorzoxazone metabolite was detected with an absorption wavelength of 287 nm by HPLC with a mobile phase consisting of 20% acetonitrile, 78% water, and 2% acetic acid, with a 1 mL·min−1 flow rate.

Protein crystallization, data collection, and structure determination

CYP2A6, CYP2A6 I208S/I300F/G301A/S369G, CYP2A13 and CYP2E1 were cocrystallized with pilocarpine by hanging drop vapor diffusion. The CYP2A6–pilocarpine crystals were grown from 500 μm CYP2A6 with 100 mm pilocarpine in CM elution buffer (50 mm potassium phosphate buffer, pH 7.4, 20% glycerol, 1 mm EDTA, 0.5 m NaCl) with 2% Anapoe-35 in a 1:1 ratio with precipitant solution [30% poly(ethylene glycol) 3500, 0.175 m Tris, pH 8.5, 0.2 m ammonium sulfate]. The CYP2A6 I208S/I300F/G301A/S369G–pilocarpine crystal was grown similarly, with 100 μm CYP2A6 I208S/I300F/G301A/S369G and 100 mm pilocarpine in CM elution buffer and 2% Anapoe-35 in a 1:1 ratio with a slightly modified precipitant solution [30% poly(ethylene glycol) 3350, 0.100 m Tris, pH 8.5, 0.200 m ammonium sulfate]. The CYP2A13–pilocarpine crystal was grown from a solution of 200 μm CYP2A13, 100 mm pilocarpine and 2% Anapoe-35 in CM elution buffer in a 2:1 ratio with precipitant solution [30% poly(ethylene glycol) 3350, 0.175 m Tris, pH 8.5, 0.2 m ammonium sulfate]. The CYP2E1 crystal was grown at 455 μm in a buffer containing 100 mm pilocarpine, 120 mm potassium phosphate (pH 7.4), 0.5 m sucrose and 1 mm EDTA in a 1:1 ratio with precipitant solution [8% poly(ethylene glycol) MME 2000, 12% isopropanol, 0.1 m NaHEPES, pH 7.5].

Crystals were flash cooled in liquid nitrogen after being immersed in a cryoprotectant. For all CYP2A crystals, the cryoprotectant consisted of 700 μL of synthetic mother liquor and 300 μL of 100% ethylene glycol. The cryoprotectant for CYP2E1 was 0.1 m NaHEPES (pH 7.5), 5% isopropanol, and 1.4 m sucrose. X-ray diffraction data for CYP2A6, CYP2A6 I208S/I300F/G301A/S369G, and CYP2A13, each with pilocarpine, were collected at the Stanford Synchrotron Radiation Laboratory (Stanford, CA, USA) on Beamline 9-2 with a 0.98-Å wavelength and temperature of 100 K. Data were processed with mosflm and scala [16,17]. The CYP2E1–pilocarpine dataset was collected on Beamline 17-BM at the Advanced Photon Source and processed with hkl2000. All structures were solved by molecular replacement with phaser [17]. Model building and refinement of all structures were performed iteratively with coot [18] and refmac5 in the ccp4 suite [17]. Detailed collection and refinement statistics are given in Table 2.

Table 2.   Crystal data collection and refinement statistics of CYP enzymes binding pilocarpine.
 CYP2A6CYP2A6 I208S/I300F/G301A/S369GCYP2A13CYP2E1
  1. a Parentheses indicate highest-resolution shell.

Crystal data
 Space groupP21P21P1P4
 Unit cell
  a, b, c (Å)70.34, 158.00, 104.4570.95, 159.8, 103.971.50, 119.86, 154.87100.6, 100.6, 259.5
  α, β, γ (°)90.0, 92.1, 90.090.0, 91.8, 90.0101.0, 101.7, 93.690.0, 90.0, 90.0
 Molecules per AU4484
Data collection
 Resolution (Å)87.04–2.40 (2.46–2.40)87.04–2.10 (2.16–2.10)117.00–3.00 (3.16–3.00)37.0–2.35 (2.41–2.35)
 Total observationsa237 442 (17 456)301 406 (21 586)383 997 (55 808)385 690 (18 326)
 Unique observationsa85 624 (6420)126 622 (9537)96 716 (14 053)102 444 (10 635)
 Completenessa (%)96.5 (97.9)94.3 (96.0)98.3 (97.9)95.9 (84.3)
 Multiplicitya2.8 (2.7)2.4 (2.3)4.0 (4.0)3.76 (2.04)
 Rmergea (%)0.075 (0.400)0.075 (0.400)0.117 (0.634)0.105 (0.221)
 I/σIa13.2 (3.4)9.3 (1.4)9.4 (2.1)6.6 (2.6)
Refinement statistics
 Resolution (Å)79.00–2.4063.32–2.10102.88–3.0036.98–2.35
 No. of reflections81 313120 21891 88297 337
 R/Rfree (%)19.9/26.021.1/25.721.3/30.321.7/28.1
 Rmsd bond lengths (Å)0.0200.0090.0170.020
 Rmsd bond angles (°)1.8471.1821.7861.629
Number of atoms/average B-factor
 Protein15 040/31.815 101/31.530 058/55.415 166/35.4
 Ligand60/34.060/39.190/72.860/26.7
 Heme172/20.2172/21.3344/52.2172/22.1
 Sucrose00092/28.8
 Water217/25.9504/31.831/34.5772/29.6
 Coordinate error, Luzzati plot (Å)0.2770.2540.3880.321

The search model for CYP2A13 with pilocarpine was a 1.65-Å structure of CYP2A6 with N-methyl[5-(pyridin-3-yl)furan-2-yl]methanamine [Protein Data Bank (PDB) 2FDV]. A Matthews coefficient of 2.88 with 57.4% solvent suggested eight molecules in the asymmetric unit. Molecular replacement identified eight molecules in CYP2A13 with good packing. The 3.0-Å final model of CYP2A13 with pilocarpine contains residues 32–494 in molecules A–H, heme, six pilocarpine molecules, and 31 water molecules, and is deposited in the PDB as 3T3S. The crystallographic R-factor is 21.3% and the Rfree is 30.3%. In the Ramachandran plot, 86.2% of the residues are in the most favored region, 13.0% in the additionally allowed region, 0.5% in the generously allowed region, and 0.3% in the disallowed region.

The search model for both CYP2A6 structures was a 1.9-Å structure of CYP2A6 complexed with coumarin (PDB 1Z10, molecule B). The 2.4-Å structure of CYP2A6 with pilocarpine contains residues 32–494 and heme in all four copies of the asymmetric unit, four pilocarpine molecules, and 217 water molecules, and is deposited in the PDB as 3T3R. The crystallographic R-factor is 19.9% and the Rfree is 26.0%. In the Ramachandran plot, 91.5% of the residues are in the most favored region, 8.2% in the additionally allowed region, 0.4% in the generously allowed region, and 0.2% in the disallowed region.

The 2.1-Å final model for CYP2A6 I208S/I300F/G301A/S369G with pilocarpine contains residues 32–494 and heme in all four molecules, four pilocarpine molecules, and 504 water molecules, and is deposited in the PDB as 3T3Q. The crystallographic R-factor is 21.2% and the Rfree is 25.7%. In the Ramachandran plot, 91.2% of the residues are in the most favored region, 8.1% in the additionally allowed region, 0.3% in the generously allowed region, and 0.2% in the disallowed region.

The 2.35-Å final model of CYP2E1 with pilocarpine contains residues 31–494, heme and pilocarpine in all four molecules. There are also four sucrose molecules and 772 water molecules, and it is deposited in the PDB as 3T3Z. The crystallographic R-factor is 21.7% and the Rfree is 28.1%. In the Ramachandran plot, 89.6% of the residues are in the most favored region, 9.9% in the additionally allowed region, 0.2% in the generously allowed region, and 0.2% in the disallowed region.

In each structure, the single residue found in the disallowed region of the Ramachandran plot was well defined by electron density.

Protein figures and analysis

All protein figures were generated with pymol [19]. voidoo [20] was used to determine active site volumes with a 1.4-Å probe size. All figures were made with molecule A for CYP2A6, CYP2A6 I208S/I300F/G301A/S369G, and CYP2E1. All figures of CYP2A13 utilized molecule E, as this molecule contained the best ligand density.

Acknowledgements

This work was supported, in whole or in part, by National Institutes of Health Grant GM076343 (E. E. Scott). Crystals were grown by use of the facilities of the Protein Structure Laboratory at the University of Kansas, supported by National Institutes of Health Grant RR017708. The CYP2E1 data were collected at the Advanced Photon Source. Use of the IMCA-CAT beamline 17-BM at the Advanced Photon Source was supported by the companies of the Industrial Macromolecular Crystallography Association through a contract with Hauptman-Woodward Medical Research Institute. Use of the Advanced Photon Source was supported by the US Department of Energy, Office of Science, Office of Basic Energy Sciences, under Contract No. DE-AC01-06CH11357. The remainder of the X-ray data collection was carried out at the Stanford Synchrotron Radiation Laboratory, a national user facility operated by Stanford University on behalf of the United States Department of Energy Office of Basic Energy Sciences. The Stanford Synchrotron Radiation Laboratory Structural Molecular Biology Program is supported by the Department of Energy Office of Biological and Environmental Research, the National Institutes of Health National Center for Research Resources Biomedical Technology Program, and the National Institute of General Medical Sciences.

Ancillary