Unlike many signaling proteins that function as binary switches between ‘on and off’ states, G protein-coupled receptors (GPCRs) exhibit basal activity that can be increased or decreased by numerous ligands. A given receptor can recognize multiple ligands, allosteric modulators, and transducers to create a complex free energy landscape. Many of the lowest energy states have been captured by static structural techniques while detailing the wells’ widths, metastable states, and the transition between them, is still in its infancy. Nuclear magnetic resonance (NMR) spectroscopy can monitor the structure and dynamics of GPCR ensembles across fifteen orders-of-magnitude, but technical challenges have limited its application to super-microsecond timescales. Focusing on a prototypical peptide-binding GPCR, the neurotensin receptor 1 (NTS 1), we employed NMR and density functional theory (DFT) to probe global sub-nanosecond motions. The near random coil chemical shifts of the apo receptor produced a poor correlation with theoretical predictions that may indicate a high degree of conformational averaging in solution, a crystallization artifact, or both. Whereas orthosteric agonists and antagonists both rigidified the receptor, but to varying degrees, which suggests conformational entropy differentially contributes to their respective pharmacology. The strong correlations of observed and theoretical chemical shifts lend confidence to interpreting spectra in terms of local structure, methyl dihedral angle geometry, and pico-second timescale transitions. Together, our results suggest a role for sub-nanosecond dynamics and conformational entropy in GPCR ligand discrimination.
G protein-coupled receptors (GPCRs) are the largest family of membrane proteins – comprising approximately 3% of the human genome. They recognize a diverse set of stimuli at the plasma membrane to regulate processes from vision, smell and taste, to immune, neurologic and reproductive functions. As fundamental components of all major systems it is no surprise that GPCRs dominate the therapeutic market – accounting for more than 30% of FDA-approved drugs. Activation is initiated by agonist binding on the extracellular face, which produces a conformational change on the intracellular side. Unlike many signaling proteins that function as binary switches between ‘on and off’ states, GPCRs feature a ligand-independent basal activity that is increased or decreased upon ligand binding, and then further regulated by allosteric modulators. Activated receptors signal intracellularly through G protein and arrestin transducers equally (balanced signaling) or selectively (biased signaling). A single receptor may specifically recognize several ligands and respond uniquely to each, creating a complex conformational landscape. Thus, there is immense therapeutic potential in the ability to tune receptor signaling using partial agonists, biased agonists, and allosteric modulators, that is only beginning to be tapped.
The principles of allostery, whereby ligand association at one site elicits altered activity at a remote location, have been critical to our understanding of GPCR signaling. A dramatic example from crystallography is that orthosteric ligand binding translates the intracellular portions of transmembrane (TM) helices 5 and 6, which are located > 40 Å away, outward to accommodate heterotrimeric G proteins. The Monod-Wyman-Changeux (MWC) model of allostery (i.e. conformational selection) posits the pre-existence of both inactive and active conformations whose relative populations are modulated by the ligand. Indeed, elegant nuclear magnetic resonance (NMR) spectroscopy studies using 19F-labeled TMs confirm a conformational ensemble that in most instances includes more than two stable, low-energy states. Yet, in many instances the MWC model is unsatisfactory for understanding the full pharmacological landscape of partial agonists, allosteric modulators, and biased agonists. There is a growing body of evidence that allostery doesn’t necessarily require conformational changes on a scale that can be observed by static structural techniques such as cryo-EM and X-ray crystallography. First theorized by Cooper and Dryden nearly 35 years ago, dynamically-driven (DD) allostery asserts that the frequency and amplitude of sub-nanosecond motions around the average conformation (i.e. conformational entropy) can effectively reduce the energy barrier between inactive-active modes without the need for structural change.
NMR spectroscopy is uniquely capable of reporting on both types of allostery at near atomic resolution but, to date, has primarily focused on the role of MWC allostery in GPCR systems. This is largely due to the technical challenges of selective isotope labeling in eukaryotic cells and the resonance assignment-by-mutagenesis approach. Although there are examples that demonstrate a role for sub-nanosecond backbone dynamics in protein function, the fast motions of methyl-bearing side chain are clearly more informative reporters on DD allostery. Quantification of local sidechain motions, in the form of NMR generalized order parameters, have been used to develop an empirical, model-independent proxy for conformational entropy. The comparatively high mobility of methionine side chains, and low sequence abundance, has generally diminished their inclusion in these motional analyses. Yet, for practical reasons, selective methionine labeling remains one of the most commonly applied approaches to studying GPCRs by NMR.
Determination of the generalized order parameter is a far more laborious process than that of other NMR observables such as the chemical shift and linewidths. Chemical shift values reflect a nucleus’ local electronic environment, which makes them extremely sensitive probes of both structure and dynamics; for example, the 13 C methyl chemical shift of branched-chain amino acids (e.g. Ile, Leu, Val) is strongly influenced by the side-chain rotameric state (trans, gauche). Chashmniam and colleagues recently used density functional theory (DFT) quantum calculations to demonstrate that the (de-)shielding effect of neighboring atoms on the methionine methyl chemical shift is comparable, or greater, than the one arising from local side-chain geometry. They showed that the relative 13 C chemical shift values of methionine methyl groups can be predicted from a static high-resolution crystal structure by only considering atoms within a 6 Å sphere. When a linear correlation is observed between the theoretical and experimental 13 C chemical shift values of methionines located far apart in the structure, it indicates a similar degree of side-chain conformational averaging throughout the protein. The (common) degree of local conformational averaging is protein dependent and can be extracted from the slope of the regression line. Thus, this slope can be interpreted as an order parameter for the global flexibility of the protein, as detected locally by multiple methionine side chains in distant parts of the structure. This methionine chemical shift global order parameter (S MCS) can theoretically range from one (completely rigid) to zero (completely averaged). In other words, the experimental chemical shifts are scaled from the rigid-structure calculated values toward those expected from a totally flexible environment.
Here, we focus on a prototypical peptide-binding receptor, neurotensin receptor 1 (NTS 1) to test if S MCS can sense ligand-dependent changes in global receptor dynamics. We functionally validated and assigned the 13 C ε H 3-methionine resonances of a minimal-methionine NTS 1 variant in the presence of several orthosteric and allosteric ligands. We then compared the experimental 13 C ε chemical shifts to DFT-calculated chemical shifts from high resolution apo, agonist and antagonist bound NTS 1 crystal structures. Whereas the basal ensemble of the apo receptor appears highly dynamic, agonist and antagonist binding tune global motions that differentially modulate receptor rigidity. The strong linear correlation between experimental chemical shifts and DFT calculations suggested that mechanistic hypotheses could be derived from detailed structural examination. Given the number of 13 C ε H 3-methionine NMR studies and corresponding high-resolution GPCR crystal structures, our results reveal a tractable approach to exploring global receptor dynamics on the sub-nanosecond timescale.
Our study focuses on a selectively 13 C ε H 3-methionine labelled NTS 1 construct, termed enNTS 1 ΔM4, which was derived from a previously thermostabilized rNTS 1 variant (enNTS 1) by removing four (M181L, M267L, M293L and M408V) of the ten endogenous methionine residues. Preliminary experiments showed that mutagenesis did not adversely affect structural integrity as it had little to no observable effect on the remaining resonances’ chemical shifts. M267 5.68, M293 ICL3 and M408 H8 (superscript refers to Ballesteros-Weinstein numbering) are solvent exposed and highly degenerate in two-dimensional (2D) 1 H-13 C heteronuclear multiple quantum correlation (HMQC) spectra while M181 ICL2 overlaps with M352 7.36 in some instances. Thus, the mutations simplify the analysis of 2D 1 H-13 C HMQC spectra, while preserving the structure of enNTS 1. enNTS 1 ΔM4 retains six endogenous methionine residues (M204 4.60, M208 4.64, M244 5.45, M250 5.51, M330 6.57 and M352 7.36) with all, except for M244 5.45 being retained across species. Four of these methionines (M204 4.60, M208 4.64, M330 6.57 and M352 7.36) are also conserved in all NTS 2 sequences, but M250 5.51 is the only probe significantly conserved among peptide GPCRs (19%, ranking 2nd after leucine).
Figure 1.Ligand-induced chemical shift changes observed for 13 CH 3-methionine labelled enNTS 1 ΔM4.
A) Cylindrical representation of thermostabilized rNTS 1 (PDB 4BWB) with labelled methionine methyl groups shown as yellow spheres (superscript - Ballesteros-Weinstein nomenclature) and NT8-13 shown as purple sticks. B) ML314 reduces efficacy of NT8-13 dependent, enNTS 1 ΔM4-mediated G protein activation in a TGFα shedding assay using HEK293A cells. Cells were stimulated with increasing concentrations of NT8-13 in the absence (magenta) or presence of 0.1 μM (light pink), 3.2 μM (light blue), and 10 μM (blue) ML314. Error bars represent SEM from three independent experiments. C) ML314 potentiates NT8-13 dependent βArr1 recruitment to enNTS 1 ΔM4. βArr1 recruitment was measured by a NanoBiT-based assay using HEK293A cells. Cells were stimulated with increasing concentrations of NT8-13 in the absence (magenta) or presence of 3.2 μM (light blue), and 10 μM (blue) ML314. Note that there was a modest β-arrestin suppressive effect with 10 μM ML314 alone (dark blue symbol at 0 μM NT8-13) yet showing response to NT8-13 in a concentration-dependent manner. Error bars represent SEM from three independent experiments.
