Examining how enzymes self-organize in a membrane.

It is becoming clear that enzymes have dynamic structures that respond to changes in their environment (1–3). The presence of substrate and cofactors, the identities and oxidation states of ligands and metal centers, all can have a dramatic effect on the compactness, conformation, and oligomerization state of an enzyme. For membrane-bound or membrane-associated enzymes, how the enzyme interacts with the lipid bilayer is yet another variable that affects overall protein structure and stability. There is a paucity of structures for intact membrane proteins in the Protein Data Bank (PDB), making it difficult to determine to what extent membrane association modifies protein conformations. Although many structures have been obtained using versions of membrane proteins with their membranebinding domains removed, these are not particularly helpful for understanding protein– bilayer interactions. In PNAS, Monk et al. pull back the curtain a bit on this important question (4). In their paper, Monk et al. describe several crystallographic structures of lanosterol 14α demethylase (CYP51) from Saccharomyces cerevisiae. The enzyme consists of three domains: a catalytic C-terminal cytochrome P450 (CYP) linked via an intermediate transmembrane helix (TMH) to an N-terminal amphipathic (AH) helix that resides in vivo on the inner side of the ER membrane. A remarkable feature of the described structure is the clear segregation of the catalytic domains from the TMH domains within the crystal lattice. The TMH domains are well ordered, and there is a high degree of interaction between the AH domains and symmetry-related CYP domains on the opposite sides of the TMH region. Specific interactions between residues on the TMH domain and the CYP domain constrain their relative orientations, resulting in a spacing of 33 Å between CYP domains separated by the TMH domains, a typical value for lipid bilayer thickness (Fig. 1). Although no ordered bilayer components are detected (the protein crystallizations were performed using a nonionic detergent), the inference is obvious that the crystallographic structures accurately capture the relative arrangement of the three subunits of the bitopic (single TMH) structure. These results stand out for several reasons. First, this is (surprisingly!) the only complete structure of a bitopic membrane protein deposited in the PDB, and it provides some important general insights into how enzymes are constrained at membrane surfaces. The degree to which the specific interactions between the TMH and the CYP domains appear to orient and control the interface between the lipid bilayer and the CYP is unexpected: rather than a loose anchor, the TMH provides a relatively rigid dock for the catalytic domain in the bilayer (Fig. 2). Like a boat tied to a single stanchion, a series of hydrogen bonding and polar interactions between the TMH and the CYP fixes the position of the active site entrance in the lipid bilayer by holding the “prow” of the CYP in place. At the same time, the arrangement should be entropically favorable, as it permits a fair amount of lateral motion on the part of the CYP in the plane of the bilayer. That the constrained region of the CYP is indeed the entrance (vestibule) to the active site is confirmed by a structure with a bound extended conformation inhibitor: The inhibitor extends from the heme iron (to which it is linked by a dative covalent bond) to a helical bend close to the TMH– CYP interaction site. Furthermore, electron density consistent with a lanosterol-sizedmolecule is observed in a secondary vestibule site located on the other side of the TMH–CYP interaction locus, in a position that is expected to be submerged in the lipid bilayer. As such, the structures described by Monk et al. provide evidence for substrate access to the CYP active site directly from the lipid bilayer. A single data point, perhaps, but it opens up a different way of considering such interactions in other bitopic membrane proteins. For those of us whowork with cytochromes P450, the results ofMonk et al. are exciting for other reasons as well. The cytochrome P450 monooxygenases comprise one of the oldest and most widespread enzyme superfamilies. More than 100,000 putative P450 genes have been deposited in GenBank to date, and TMH CYP