Mechanisms of promiscuity among drug metabolizing enzymes and drug transporters

Abstract

Detoxication, or 'drug-metabolizing', enzymes and drug transporters exhibit remarkable substrate promiscuity and catalytic promiscuity. In contrast to substrate-specific enzymes that participate in defined metabolic pathways, individual detoxication enzymes must cope with substrates of vast structural diversity, including previously unencountered environmental toxins. Presumably, evolution selects for a balance of 'adequate' k_cat /K_M values for a wide range of substrates, rather than optimizing k_cat /K_M for any individual substrate. However, the structural, energetic, and metabolic properties that achieve this balance, and hence optimize detoxication, are not well understood. Two features of detoxication enzymes that are frequently cited as contributions to promiscuity include the exploitation of highly reactive versatile cofactors, or cosubstrates, and a high degree of flexibility within the protein structure. This review examines these intuitive mechanisms in detail and clarifies the contributions of the classic ligand binding models 'induced fit' (IF) and 'conformational selection' (CS) to substrate promiscuity. The available literature data for drug metabolizing enzymes and transporters suggest that IF is exploited by these promiscuous detoxication enzymes, as it is with substrate-specific enzymes, but the detoxication enzymes uniquely exploit 'IFs' to retain a wide range of substrates at their active sites. In contrast, whereas CS provides no catalytic advantage to substrate-specific enzymes, promiscuous enzymes may uniquely exploit it to recruit a wide range of substrates. The combination of CS and IF, for recruitment and retention of substrates, can potentially optimize the promiscuity of drug metabolizing enzymes and drug transporters.

Keywords: conformational selection; detoxication enzyme; enzyme promiscuity; induced fit; stochastic enzymes.

Figure 1

GST structural features. (A) Overlay of GSTA1‐1 and GSTA4‐4 apo structures in gray and blue, respectively (PDB http://www.rcsb.org/pdb/search/structidSearch.do?structureId=1GSD and http://www.rcsb.org/pdb/search/structidSearch.do?structureId=1GUM). GST forms a homodimer; the second subunit is in the background in lighter gray/blue. The C‐terminal helix is flexible and not resolved (not observed here) in most apo A1‐1 structures. Trp21 is conserved in both GSTA1‐1 and GSTA4‐4 and lies at the intrasubunit domain–domain interface remote from the active site (yellow/red, A1‐1/A4‐4). (B) Overlay of GSTA1‐1 with GS‐EA substrate conjugate bound and A1‐1 mutant with GSH bound in which the C‐terminal helix is resolved in one subunit in gray and red, respectively (PDB http://www.rcsb.org/pdb/search/structidSearch.do?structureId=1GSE and http://www.rcsb.org/pdb/search/structidSearch.do?structureId=1EV4). The C‐terminal helix can adopt very different locations when different ligands or substrates are bound.

Figure 2

Schematized free energy landscapes for substrate‐specific ligand‐free wild‐type GSTA4‐4 (a), a mutant GSTA1‐1 with intermediate promiscuity that exhibits pre‐steady‐state lags or bursts (b), and promiscuous wild‐type GSTA1‐1 (c). The conformational landscape of GSTA1‐1 has minimal energy barriers to rearrangement, so there is minimal kinetic cost for the conformational selection wherein different substrates select different conformations. After binding S1 or S2, further conformational change occurs (IF). The mutant GSTA1‐1 W21F;F222W (b) has modest energy barriers that add kinetic lags or burst in steady‐state catalytic experiments, but they are not sufficiently large to prevent conformational heterogeneity. In both b and c, different substrates select different conformations.

Figure 3

Kinetic models for substrate‐specific vs. substrate‐promiscuous enzymes. Top: Limiting case models for substrate‐specific enzymes are IF and CS. Most enzymes likely include a contribution of both, due to nominal conformational heterogeneity of the substrate‐free enzyme. Bottom: Proposed model for promiscuous drug metabolizing enzymes. Different conformations of enzyme are color‐coded. A wide range of conformational states of the substrate‐free enzyme (E1, E2… En) is presented to recruit structurally distinct substrates. Different substrates (S1, S2) bind to one or more conformations of the enzyme, indicated by the CS step, followed by conformational changes to the catalytically productive states that are retained at the active site (IF). Different substrates can induce different multiple conformations (E* vs. E** for S1 and E’ vs. E” for S2). Different complexes for the same substrate result in product promiscuity (P1 vs. P2 and P3 vs. P4) and multiple ES complexes for a single substrate. The model combines data from GSTs, CYPs, and other detoxication enzymes.

Figure 4

Superimposition of three substrate‐bound structures of CYP3A4: with midazolam (sky blue, PDB: http://www.rcsb.org/pdb/search/structidSearch.do?structureId=5TE8), ketoconazole (salmon, PDB: http://www.rcsb.org/pdb/search/structidSearch.do?structureId=2V0M), and ritonavir (gray, PDB: http://www.rcsb.org/pdb/search/structidSearch.do?structureId=5VC0). Each of the structures (chain A only) has been individually superimposed on the ligand‐free structure (not shown; PDB: http://www.rcsb.org/pdb/search/structidSearch.do?structureId=1TQN). Left: Three structures oriented to show the topology with respect to the membrane. The edge of the heme cofactor is observed (red). Substrates (not shown for clarity) sit below the heme toward the membrane and interact differentially with the D, E, F', F, G', G, H helices and the N‐term half of the I helix. Right: Rotation of the structures, viewed from the membrane. These elements occupy very different locations with different substrates resulting in large differences in active site volume when different ligands are bound. The most prominent differences occurred in the F’‐G’ and F‐G region (red arrows right panel). The active site and surroundings are highly plastic and adapt to different substrates.

Figure 5

Extensive conformational heterogeneity of P‐gp revealed by H/DX MS. P‐gp consists of two intracellular nucleotide‐binding domains (NBDs) and twelve transmembrane helices that comprise the transmembrane domains that form the drug‐binding site. Two macroscopic conformations are well characterized. The inward‐facing form allows binding of drugs from the membrane, but they cannot diffuse to the extracellular side until the enzyme binds ATP (red) and switches to an outward‐facing conformation. H/DX MS studies 101 indicate that many peptides (blue) exhibit slow conformational exchange on a wide range of timescales. Ligand‐free P‐gp populates a wide range of conformations that likely increase its promiscuity via conformational selection. Yellow regions are peptides analyzed that do not exhibit slow conformational exchange.

Figure 6

Schematized energy landscapes for P‐gp. The reaction coordinate for the nucleotide‐binding domain (NBD) moving together and apart is shown. Top: Simple two‐state model. The protein exists in inward‐facing (IF in the Figure, distinct from induced fit used elsewhere) and outward‐facing (OF) conformations, and nucleotide binding or ‘trapping’ shifts the ensemble partially to the OF state. Bottom: Several recent experimental methods suggest that there is significant conformational heterogeneity superimposed on IF and OF states. These additional conformations exchange on a wide range of timescales and present many conformations for substrates to select. The resulting energy landscape is rough, in contrast to that of GSTA1‐1. Both energy landscapes could be exploited to increase substrate promiscuity.

Figure 7

Comparative overview of the properties of substrate‐specific enzymes and promiscuous detoxication enzymes that are altered by evolution, and the differential exploitation of conformational selection.