BackgroundTo understand the effect of the long intracellular loop 3 (ICL3) on the intrinsic dynamics of human β2-adrenergic receptor, molecular dynamics (MD) simulations were performed on two different models, both of which were based on the inactive crystal structure in complex with carazolol (after removal of carazolol and T4-lysozyme). In the so-called loop model, the ICL3 region that is missing in available crystal structures was modeled as an unstructured loop of 32-residues length, whereas in the clipped model, the two open ends were covalently bonded to each other. The latter model without ICL3 was taken as a reference, which has also been commonly used in recent computational studies. Each model was embedded into POPC bilayer membrane with explicit water and subjected to a 1 μs molecular dynamics (MD) simulation at 310 K.ResultsAfter around 600 ns, the loop model started a transition to a “very inactive” conformation, which is characterized by a further movement of the intracellular half of transmembrane helix 6 (TM6) towards the receptor core, and a close packing of ICL3 underneath the membrane completely blocking the G-protein’s binding site. Concurrently, the binding site at the extracellular part of the receptor expanded slightly with the Ser207-Asp113 distance increasing to 18 Å from 11 Å, which was further elaborated by docking studies.ConclusionsThe essential dynamics analysis indicated a strong coupling between the extracellular and intracellular parts of the intact receptor, implicating a functional relevance for allosteric regulation. In contrast, no such transition to the “very inactive” state, nor any structural correlation, was observed in the clipped model without ICL3. Furthermore, elastic network analysis using different conformers for the loop model indicated a consistent picture on the specific ICL3 conformational change being driven by global modes.

We performed a detailed analysis of conformational transition pathways for a set of 10 proteins, which undergo large hinge-bending-type motions with 4-12 Å RMSD (root mean-square distance) between open and closed crystal structures. Anisotropic network model-Monte Carlo (ANM-MC) algorithm generates a targeted pathway between two conformations, where the collective modes from the ANM are used for deformation at each iteration and the conformational energy of the deformed structure is minimized via an MC algorithm. The target structure was approached successfully with an RMSD of 0.9-4.1 Å when a relatively low cutoff radius of 10 Å was used in ANM. Even though one predominant mode (first or second) directed the open-to-closed conformational transition, changes in the dominant mode character were observed for most cases along the transition. By imposing radius of gyration constraint during mode selection, it was possible to predict the closed structure for eight out of 10 proteins (with initial 4.1-7.1 Å and final 1.7-2.9 Å RMSD to target). Deforming along a single mode leads to most successful predictions. Based on the previously reported free energy surface of adenylate kinase, deformations along the first mode produced an energetically favorable path, which was interestingly facilitated by a change in mode shape (resembling second and third modes) at key points. Pathway intermediates are provided in our database of conformational transitions (http://safir.prc.boun.edu.tr/anmmc/method/1).

No abstract

No abstract

Crystal structures of neurolysin, a zinc metallopeptidase, do not show a significant conformational change upon the binding of an allosteric inhibitor. Neurolysin has a deep channel where it hydrolyzes a short neuropeptide neurotensin to create inactive fragments and thus controls its level in the tissue. Neurolysin is of interest as a therapeutic target since changes in neurotensin level have been implicated in cardiovascular disorders, neurological disorders, and cancer, and inhibitors of neurolysin have been developed. An understanding of the dynamical and structural differences between apo and inhibitor-bound neurolysin will aid in further design of potent inhibitors and activators. For this purpose, we performed several molecular dynamics (MD) simulations for both apo and inhibitor-bound neurolysin. A machine learning method (Linear Discriminant Analysis) is applied to reveal differences between the apo and inhibitor-bound ensembles in an automated way, and large differences are observed on residues that are far from both the active site and the inhibitor binding site. The effects of inhibitor binding on the collective motions of neurolysin are extensively analyzed and compared using both Principal Component Analysis and Elastic Network Model calculations. We find that inhibitor binding induces additional low-frequency motions that are not observed in the apo form. ENM also reveals changes in inter- and intradomain communication upon binding. Furthermore, differences are observed in the inhibitor-bound neurolysin contact network that are far from the active site, revealing long-range allosteric behavior. This study also provides insight into the allosteric modulation of other neuropeptidases with similar folds.