1. Studies of activation mechanisms of G-proteincoupled receptors
Recent crystal structures of class A (rhodopsin-like) G-protein-coupled receptors (GPCRs), namely β1- and β2- adrenergic receptors and adenosine A2A receptors, showed nearly identical transmembrane domain structures but differences in the states of molecular switches compared with rhodopsin, which was the first GPCR crystallized. Based on experimental data, agonist binding and receptor activation were proposed to occur through a series of conformational intermediates. The transition between these intermediate states involves disruption, creation, or reorganization of intramolecular interactions that stabilize the basal state of a receptor. These changes are elicited by the action of molecular switches (also called microswitches).
The major switches proposed so far for different GPCRs, reflecting shared activation mechanisms, include the “rotamer toggle switch” that involves the CWxPx(F/H) sequence on transmembrane helix TM6, the switch based on the NPxxY(x)(5,6)F sequence that links helices TM7 and H8, and the “ionic lock” that links transmembrane helices TM3 and TM6 and employs the (E/D)RY motif on TM3. There are also switches not assigned to any particular sequence motifs, such as the 3-7 lock that involves the interaction between TM3 and TM7 and is present only in selected receptor types.
In our earlier papers, we investigated early activation steps that occur simultaneously to ligand binding in the MOR (mu), DOR (delta), and KOR (kappa) opioid receptors. The first switch that was broken by agonist binding was the “3-7 lock,” a hydrogen bond D3.32-Y7.43 that links transmembrane helices TM3 and TM7. It was the first activation event observed. We also detected the action of a second switch: a rotamer toggle switch that involves a simultaneous change of side chain conformations of W6.48 and adjacent residues and was therefore called the extended toggle switch. In the case of opioid receptors, the other residue in this extended switch was H6.52. This residue also participated in the agonist-antagonist sensor, determined by the propensity for creating a hydrogen bond with Y3.33 for antagonists and H6.52 for agonists.
All studied ligands, analogs of morphine with a common tyramine structural scaffold, created a salt bridge with D3.32 with their protonated nitrogen atom of the tyramine group. This sensor was studied by us for MOR and later DOR and KOR. The proposed mechanism of its action was later confirmed by molecular dynamics simulations of a closely related agonist-antagonist pair of KOR ligands: 5’-GNTI and 6’-GNTI.
Cannabinoid receptors, similarly to opioid receptors, belong to class A (similar to rhodopsin) GPCRs. The docking of agonists and antagonists to cannabinoid CB1 and CB2 receptors revealed the importance of a centrally located rotamer toggle switch and its possible participation in the mechanism of agonist/antagonist recognition. The switch is composed of two residues, F3.36 and W6.48, located on opposite transmembrane helices TM3 and TM6 (which is dissimilar to other GPCRs) in the central part of the membranous domain of cannabinoid receptors. The CB1 and CB2 receptor models were constructed based on the adenosine A2A receptor template. The two best-scored conformations of each receptor were used for the docking procedure. In all poses (ligand-receptor conformations) characterized by the lowest ligand-receptor intermolecular energy and free energy of binding, the ligand type matched the state of the rotamer toggle switch; antagonists maintained an inactive state of the switch, whereas agonists changed it.
2. Structure of presenilin, a component of γ-secretase
The integral membrane protein ensemble γ-secretase is responsible for the proteolytic processing of various type I transmembrane domains, among them the amyloid precursor protein, whose final cleavage results in the release of the amyloid β peptide which is the major component of senile plaques found in Alzheimer’s disease. The catalytic activity of γ-secretase requires the endoproteolytic cleavage of its presenilin subunit during the maturation of the complex. The cleavage results in the generation of natural N- and C-terminal presenilin fragments, each of which harbors one of the two active-site aspartates that contribute to the formation of a hydrophilic cavity where catalysis is believed to occur. The N-terminal fragment of presenilin-1 is believed to have a classical transmembrane topology, consistent with all published models. Biochemical studies and topology predictions performed on the C-terminal fragment (CTF), however, have thus far yielded a plethora of ambiguous and contradictory results, which may be difficult to reconcile in the absence of structural information. By employing NMR measurements on cell-free expressed protein in SDS micelles, determining the first structural model of the C-terminal fragment of human presenilin-1 was possible. The structure revealed a topology by which the membrane would likely be traversed three times, consistent with the more generally accepted model, but contain unique structural features that may be adapted to accommodate the unusual intramembrane catalytic process and could account for the lack of consensus observed in most studies.
To further evaluate the structure of CTF, we performed molecular dynamics (MD) simulations both in micelles and in a lipid bilayer. To achieve the sufficient simulation time needed for micelle formation and extensive sampling of the conformational space, we chose a coarse-grain approach using the MARTINI method, which was primarily developed to study the behavior of large biological membrane systems. Simulations were performed with 200 coarse-grain dodecyl-phosphocholine (DPC) molecules, which formed a micelle consisting of approximately 80 molecules around CTF during the first 50 ns, during which time the structure of CTF was frozen. In the ensuing 3 γs simulation, the coarse-grain structure of CTF was allowed to change. The comparison with the initial NMR structure revealed that the interhelical angle between helices 7 and 8 remained relatively stable (140° in the NMR structure and 120° after the 3 γs simulation), and helices 9a and 9b became more antiparallel (105° in the NMR structure and 130° after the simulation).
We also performed coarse-grained simulations in DLPC (dilauroylphosphatidylcholine) and DPPC (dipalmitoylphosphatidyl-choline) bilayers using the MARTINI method and MD simulations of allatom representation of CTF by employing continuous environments with the Implicit Membrane Model (IMM1) method in the CHARMM program. As expected, the resulting structures showed larger differences both with respect to the NMR input structure and to each other.
Although the angle between helices 7 and 8 remained relatively stable, the angle between helices 9a and 9b changed depending on the thickness of the membrane.
Additionally, the position of the catalytic helix also shifted from the center of the membrane toward its border, again depending on the membrane width. Notably, however, in the present case, the bilayer is not necessarily a better suited environment than a micelle because several elements of CTF are believed to reside in close proximity or even contribute to the formation of the water-filled cavity that exists within the γ-secretase complex. In such a case, the more accommodating micelle may indeed provide a better hydrophobic environment for CTF in the absence of the other γ-secretase components.
