Consistent with this, alanine substitution of Y 1. In addition to these interactions, Gln3 of oxyntomodulin formed transient hydrogen bonds and hydrophobic contacts with T 7. While Ser2 of oxyntomodulin and Ala8 of GLP-1 both interacted with TM7 residues, these interactions were stronger with oxyntomodulin due to a persistent hydrogen bond between Ser2 and E 7. Accordingly, alanine mutagenesis of TM7 residues influenced both GLP-1 and oxyntomodulin affinity, but there was a larger effect on oxyntomodulin, consistent with its stronger interactions, however, interestingly these residues were more important for GLPmediated cAMP production Supplementary Figs 9 , In both peptides, the N-terminal histidine sits in an enclosed pocket-forming hydrophobic and hydrogen bond interactions with E 6.
In the MD studies, weaker interactions of oxyntomodulin with residues deep in the cavity, particularly R 2. Overall, these differing receptor interaction patterns of GLP-1 and oxyntomodulin were associated with larger conformational dynamics within the oxyntomodulin binding pocket during the course of the MD simulation, with the cavity opening and closing, likely due to the lack of stable interactions within the base of the TMD binding pocket, TM2 and ECL2, coupled with more persistent interactions with the upper regions of TM1 and TM7, relative to GLP-1 Fig.
For example, Arg17 and Arg18 of oxyntomodulin form interactions with the ECD and top of TM2 in the simulations, whereas the corresponding alanine residues in GLP-1 could not form these interactions.
The MD simulations revealed a large number of conserved interactions between GLP-1 and exendin-4, including the majority of peptide hydrogen bonding within the TMD. Of particular note was the lack of persistent receptor interactions for His1 of exendin-4, in contrast to the extensive interactions observed for GLP-1 His7 and oxyntomodulin His1 , as described above Fig.
His1 of exendin-4 formed only limited transient interactions with TM5, however more persistent interactions with E 6. In line with strong and stable interactions of His1-Ala2 of GLP-1 compared with transient interactions of His1-Gly2 in exendin-4, truncation of these two N-terminal residues reduces GLP-1 affinity by — fold, whereas for exendin-4, this has no effect, albeit, for both peptides, these residues are required for GLP-1R-mediated cAMP production 22 , 23 , 24 , The similarity in their TMD interaction patterns is consistent with the strong correlation in the impact of TMD mutagenesis for these two agonists Fig.
Moreover, the more transient interactions of exendin-4 parallel the smaller effects of the mutagenesis on cAMP signalling for this peptide. These interactions may also influence the conformation of TM1 accounting for the more outward conformation in the exendinbound structure Fig. Interestingly, single alanine amino acid substitutions of some interacting residues within the TMD had much larger effects on exendin-4 affinity than removal of the first 8 residues of the peptide exendin 9—39 22 , 31 , suggesting that peptide ECD and TMD interactions are correlated; non-optimal interactions of exendin-4 with the TMD elicited by receptor alanine mutations, likely promotes faster peptide dissociation from the ECD, compared to when the TMD interacting residues are not present in the peptide.
This is also supported by previous studies, whereby Gly2Ala in exendin-4 was tolerated, but the converse for GLP-1 Ala2Gly reduced affinity 31 , 32 , Nonetheless, the TMD conformation differed and, overall, the receptor was relatively stable, exhibiting less flexibility compared with GLP-1R bound to exendin-4 or oxyntomodulin Supplementary Fig.
The N-terminal sequence of exendin-P5 differs considerably from the other peptides and is also extended by one residue Fig. Nonetheless, there are some commonalities with the other peptides in their pattern of interaction with the TMD. Val3-Asp4 interact with similar residues at the base of the binding cavity to those observed for residues 2 and 3 of the other peptides, however, with the exception of Y 1.
Consequently, Asp4 could also form interactions with K 2. This provides a rationale for the effect of mutagenesis of R 2. Despite occupying a different location in static structures, Glu1 of exendin-P5 interacts with multiple residues that interact with His1 of the other peptides, including E 6. However, with the exception of R 5. Beyond these residues, there are very few interactions formed with the remainder of the N-terminal 9 residues of this peptide Supplementary Table 3.
These data are consistent with alanine substitution of TMD residues lining the binding cavity generally having limited impact on exendin-P5 affinity, suggesting that its affinity is largely driven by interactions with the ECD Figs.
In contrast, transient interactions between the exendin-P5 N-terminus and TMD are clearly important for agonism, with the majority of residues within this cavity being required for eliciting cAMP signalling Figs. The stark contrast in the requirement for stable TMD interactions for exendin-P5 affinity relative to GLP-1, oxyntomodulin and exendin-4, raises important questions regarding molecular mechanisms for peptide binding and receptor activation. In the wildtype cell line, GLP-1, oxyntomodulin and exendin-4 competition curves were clearly biphasic, with potencies for the high-affinity site correlating with those reported from whole cell-binding assays in the wildtype GLP-1R expressing ChoFlpIn cell line used in the mutagenesis study Fig.
In contrast, exendin-P5 exhibited monophasic binding curves with a lower pIC 50 than the other peptides that were consistent with the pIC 50 achieved in the ChoFlpIn whole-cell assay.
Interestingly, when G s was overexpressed, there was a larger influence on GLP-1 that exhibited the highest stability of interactions within the TMD binding cavity in the MD studies, compared to oxyntomodulin and exendin-4, which were more dynamic Fig. Exact P values are shown on the relevant figures. To assess if there is any potential for different GLP-1R agonists to display differences in G s turnover, we employed this same Nanobit G s complementation assay, to determine ligand-induced G protein dissociation in whole cells Fig.
In addition, we measured the kinetics of an earlier step in the G protein activation cycle using an assay to measure the G s conformational change upon coupling to the ligand-activated receptor Fig. Consistent with previous observations 13 , we demonstrated that exendin-P5 exhibited faster G s conformational transitions, relative to GLP-1 and exendin-4, and this was coupled with faster dissociation of the G protein heterotrimer in the Nanobit assay Fig.
Oxyntomodulin also displayed significantly faster kinetics relative to GLP-1 in both assays, whereas exendin-4 was more similar to GLP Overlay of the four consensus cryo-EM static structures revealed very similar backbone conformations of the intracellular face of the receptor, with the greatest divergence for the ICLs where modelled , and similar engagement with G s in all four peptides bound structures Fig. The MD simulations described above included the G s heterotrimer and revealed that the receptor—G s interactions were very similar regardless of the bound agonist, albeit the majority of interactions were transient Fig.
Nonetheless, when exendin-4, oxyntomodulin and exendin-P5 were bound, the receptor exhibited less persistent hydrogen bonding with the G protein when compared to GLP The first column shows the contacts between GLP-1R top and G s bottom during the MD simulations in the presence of GLP-1, with no contacts in cyan and increasing contacts heat mapped from white to dark pink. The other three columns report the contact differences relative to GLP-1 for each residue of the GLP-1R and G s during the MD performed in the presence of the other agonists with blue indicating fewer contacts, white similar contacts, and red enhanced contacts.
To analyse the allosteric transmission of a signal from the peptide to G protein binding site, the MD simulations for each complex were analysed using Network and Community Analysis Correlation analysis revealed similar patterns in the presence of all peptides with highly correlated motions between residues within the ECD and within the TM domain, however, anti-correlated motions were evident between these two domains Supplementary Fig.
However, the persistence of interactions differed suggesting subtle distinctions in how the different agonists engage these networks. In this analysis, edges connect non-consecutive nodes if the corresponding residues are within 4.
Variability in the connectivity of the networks enables the network to be subdivided into local communities according to the Girvan—Newman algorithm These communities contain groups of nodes that are densely interconnected and communicate to the rest of the network through a few largely conserved edges.
Accordingly, nodes within the same community communicate with each other easily through multiple routes, whereas communication between critical nodes that cross the edges form bottlenecks for information transfer within the network. Application of the Girvan—Newman algorithm to the GLP-1R data splits the network into 14—18 communities for each complex, depending on the bound peptide Supplementary Table 6 , which includes 4—5 communities within the TMD.
The analysis revealed differences in the number and the location of critical nodes required to communicate signals through the receptor by different peptides Fig. However, while these critical nodes for each peptide differ, all four peptides engage highly conserved class B1 GPCR residues, consistent with conserved activation mechanisms for this subclass of GPCRs.
Nodes and edges are in red. Peptides are not displayed as these were not considered during network analysis. Overall, more critical nodes are utilised by GLP-1 suggesting more effective communication between the G protein and the peptide-binding sites, relative to the other peptides, which is consistent with a greater allosteric influence of the G protein on GLP-1 affinity.
In comparison, fewer conserved nodes were identified for communication between communities for the other peptides, particularly for oxyntomodulin Supplementary Fig. Interestingly, while this peptide engages fewer conserved nodes, overall the correlated motions within the receptor interactions were stronger than GLP-1 Supplementary Fig. Fewer critical nodes, particularly below the peptide binding site may suggest that oxyntomodulin has a shorter, therefore quicker, path of communication between the peptide and G protein binding site, relative to GLP This is consistent with the faster G protein conformational change and activation exhibited by oxyntomodulin.
In contrast to the other peptides, exendin-P5 exhibits less correlated motions within the TM bundle Supplementary Fig. However, while still requiring fewer conserved nodes in the base of the bundle than GLP-1, relative to oxyntomodulin more conserved nodes are used by exendin-P5 below the peptide-binding pocket to enable transmission of signalling to the G protein-binding site.
This is again consistent with exendin-P5 exhibiting slower G protein activation kinetics than oxyntomodulin, yet faster than GLP This analysis highlights the complexity of transmission of information within the receptor, with differences in the persistence of peptide—receptor interactions and receptor interactions linking the peptide and G protein-binding sites resulting in different efficacy and bias of different peptides agonists.
Combining experimentally determined GLP-1R structures with structure—function studies and simulations of receptor dynamics provides unique insights into how distinct agonists engage and activate the receptor. GLP-1, oxyntomodulin, and the clinically used mimetic exendin-4 are among the most extensively studied GLP-1R peptide agonists in functional and structure—function studies. Here, we reveal that while the peptides form similar interactions in the fully active, G s -coupled state of the receptor, exendin-4 and oxyntomodulin-occupied receptors are more dynamic, which may, in part, be linked to the more transient nature of their interactions with the TMD observed in MD simulations.
Oxyntomodulin is a biased agonist relative to GLP-1 4 , and forms distinct and more dynamic interactions with the GLP-1R, particularly with residues at the base of the peptide binding cavity that is located above the conserved central polar network that is important for receptor activation. While the profile of exendinmediated signalling is more similar to that of GLP-1, there are differences in receptor engagement by the two peptides that can be rationalised by the structural and dynamic data.
Like oxyntomodulin, exendin-4 exhibits more transient interactions than GLP-1 with residues at the base of the peptide binding cavity, albeit it that the pattern of interactions with the polar core are relatively conserved. The peptide and G protein binding sites within GPCRs are allosterically linked to enable the transmission of information from peptide binding to G protein coupling. While there was also conservation in the TM bundle interactions when engaged by the four peptides, differences were identified in how each peptide uses different networks to facilitate G protein coupling.
This transmission of information across the TM bundle also enables G proteins to allosterically influence ligand affinity for GPCRs Consistent with this, we show the G protein can allosterically influence GLP-1R agonist affinity, but this occurs in a peptide-dependent manner. Enhanced affinity in the presence of G s is correlated with the degree of closure of the extracellular side of the TMD cavity around the peptide N-terminus and also the dynamics of interactions with residues in this domain.
GLP-1, oxyntomodulin and exendin-4, whose affinities are influenced by the presence of G s , promote a more closed bundle relative to exendin-P5, whose affinity is less sensitive to the presence of G s. N-terminally truncated GLP-1 and exendin-4 peptides lacking the first 8 amino acids GLP-1 15—36 and exendin 9—39 , exhibit similar affinities in published studies 31 , to their corresponding full-length peptides in the absence of G protein, providing further support that the influence of G s on peptide affinity is predominantly related to interactions of the N-terminus with the TMD.
While it is likely that G s has the potential to influence the affinity of all peptides that bind in the TMD cavity, the specific amino acid sequence of individual peptides impacts complementarity with receptor residues, influencing the stability of interactions and contributing to the degree of allosterically-facilitated TMD closure around the peptide. With the exception of interactions of Glu1 and Asp3 with polar residues deep in the bundle, the first 9 residues of exendin-P5 do not form stable interactions with the TMD.
Consequently, the TMD is more open and interactions with the peptide-N-terminus play only a minor role in its overall affinity. Nonetheless, these transient interactions are clearly essential for receptor activation, with mutation of the majority of residues within the TMD cavity decreasing exendin-P5 efficacy, but not affinity. Previously we revealed that the exendin-P5-bound GLP-1R induces faster G s conformational transitions that are linked to nucleotide exchange and induces faster cAMP production when compared to the GLPbound receptor In this study we demonstrate that both oxyntomodulin and exendin-P5 induce faster G s conformational transitions and G s heterotrimer dissociation, suggesting that these agonists may induce faster turnover of G s than GLP-1 and exendin Interestingly, this profile appears to be correlated to the strength and nature of interactions of these peptides with key polar residues at the base of the GLP-1R binding cavity.
While GLP-1, and to a slightly lesser degree exendin-4, form very stable interactions with R 2. This is predominantly due to the side chain chemistry of the residue at the position equivalent to Glu9 of GLP While glutamic acid is conserved in exendin-4, this is replaced by aspartic acid and glutamine in exendin-P5 and oxyntomodulin, respectively.
As polar residues at these receptor locations are conserved across class B1 GPCRs and play a role in receptor activation for all receptors where studied 40 , 41 , 42 , 43 , 44 , 45 , stable vs transient interactions formed by peptide residues may also be associated with peptide efficacy at other class B1 receptors.
Given cryo-EM structures of the GLP-1R in complex with G s are stabilised by trapping the nucleotide-free G protein on the ligand-activated receptor, it is not surprising that the G s interactions are similar across the four structures, even when bound by different agonists, nonetheless, the MD simulations revealed potential differences in the dynamics of these interactions.
Interactions of G proteins with these receptor domains contribute to the separation of the G protein RHD and AHD, disruption of the P loop and the nucleotide-binding pocket, all of which contribute to the release of GDP, one of the key rate-limiting steps in G protein activation However, how and if predicted differences in dynamics of receptor G protein interactions correlate to the differences in the effect of the allosteric coupling of the G protein on TMD binding sites, or the rates of G protein activation are unclear and additional studies will be required to address this.
Using the glucagon receptor as an exemplar, Hilger et al. This implies that the activated receptor is primed to activate multiple rounds of G protein coupling following dissociation of the initial interacting transducer protein, and this is proposed to contribute to the sustained cAMP signalling following activation of class B1 GPCRs.
Overall, given the long-lived active receptor conformation, this would promote greater turnover of G s and production of cAMP over time.
A similar phenomenon has been observed at the calcitonin receptor, another class B1 GPCR, where human calcitonin, with a fast off-rate, turns over G protein faster than salmon calcitonin, which has a slow off-rate, and this was related to differences in the residency time of the G protein on the receptor and to the sensitivity of the G protein to GTP Oxyntomodulin exhibited even faster kinetics for G s conformational transitions and subunit dissociation than those induced by exendin-P5, and this was also correlated with more transient interactions at the base of the binding cavity and TM2.
However, in contrast, this ligand does not exhibit higher efficacy relative to GLP-1, as determined by comparing transduction ratios from cAMP signalling and their determined affinity measures Supplementary Table 2. This highlights the complexity of GPCR activation, where downstream signalling is influenced by the interplay of multiple transducers that can interact with activated receptors pleiotropic coupling , and can also be influenced by different trafficking profiles that alter the location of the receptor in the cell.
Oxyntomodulin is a biased agonist relative to GLP-1 1 , 3 and exendin-P5 26 , with a bias towards arrestin recruitment over cAMP production, which would compete for G protein interactions and may contribute to the lack of correlation between the faster G s dissociation and enhanced efficacy when assessing cAMP accumulation in whole cells over an extended timeframe. While in the consensus cryo-EM structures, GLP-1, exendin-4 and oxyntomodulin exhibit similar TMD conformations, MD simulations revealed that both exendin and oxyntomodulin- bound GLP-1Rs, which exhibit biased agonism towards arrestin recruitment, are more dynamic in this region, than complexes with GLP-1 bound.
In summary, combining structural data from cryo-EM, receptor mutagenesis, pharmacological assays and MD simulations advances our understanding of peptide agonist engagement with the GLP-1R. As class B1 peptide agonists engage their receptors via a two-domain interaction, their efficacy for G protein-mediated signalling is influenced by multiple factors, including the nature of interactions with the ECD and TMD, contributing to both peptide affinity and how ligand—receptor interactions influence G protein binding, nucleotide exchange and G protein dissociation from the receptor.
This study provides insight into how differential dynamics of peptide—ligand engagement with the GLP-1R TMD can promote differences in G protein-mediated signalling, improving molecular understanding of mechanisms that contribute to ligand-dependent differential efficacy at the GLP-1R. The complex was solubilized from the membrane by 0. As the data collection was split between two different days, the data was split into 18 optics groups.
Exendin movies were collected and subjected to motion correction using motioncor2 CTF estimation was done using Gctf software 48 on the non-dose-weighted micrographs. The particles were picked from dose-weighted and low-pass filtered micrographs using crYOLO automated picking routine Further rounds of 2D and 3D classification yielded a final particle stack of Final consensus refinement produced a structure resolved to 3.
The cryo-EM data collection, refinement and validation statistics are reported in Supplementary Data Table 6. Oxyntomodulin: Data processing of oxyntomodulin: movies were collected and subjected to motion correction using motioncor2 Contrast transfer function CTF estimation was done using Gctf software on the non-dose-weighted micrographs The particles were picked using gautomatch developed by K.
An initial model was made using EMAN2 51 based on a few automatically picked micrographs and using the common-line approach. Picked particles 1,, were subjected to two rounds of 3D classification with three classes. The refined revealed the final structure at 3. Cells were then solubilised in 0.
Growth media was replaced with stimulation buffer [phenol-free DMEM containing 0. Cell surface receptor detection was then performed as previously described. The A K d concentration, 3. Data were corrected for baseline and vehicle probe only responses. Data were corrected for baseline and vehicle control. The concentration-response curves were then plotted using the total area under the curve during the time of measurement post ligand addition.
Data were corrected to baseline and vehicle-treated samples. Pharmacological data were analysed using Prism 8 GraphPad. Concentration-response binding and signalling data were analysed using the one-site binding inhibition and the three-parameter logistic equations in Graphpad prism, respectively.
Normalised AUC for the indicated ligand concentrations was plotted as a concentration-response curve and fitted with a three-parameter logistic curve. The missing loops in the cryo-EM structures were reconstructed using modeller or by molecular superposition as described elsewhere Hydrogen atoms were first added at a simulated pH of 7. Long-range Coulomb interactions were handled using the particle mesh Ewald summation method PME 68 by setting the mesh spacing to 1.
Atomic contacts were computed using VMD A contact was considered productive if the distance between two atoms was lower than 3. A network is formed by an ensemble of nodes interconnected by edges. Edges connect pairs of non-consecutive nodes if the corresponding residues are in contact within 4. The resulting dynamical network was weighted by considering the probability w ij of information transfer across the edge connecting two nodes i and j ; calculated using Eq.
The whole network was subdivided into local communities according to the Girvan—Newman algorithm The Girvan-Newman algorithm detects the communities within a network e. A general two-site solution for the chemical exchange produced dependence of T2 upon the Carr-Purcell pulse separation.
Watt, E. The mechanism of rate-limiting motions in enzyme function. Intrinsic motions along an enzymatic reaction trajectory. Boehr, D. The dynamic energy landscape of dihydrofolate reductase catalysis. Korzhnev, D. Millet, O. The static magnetic field dependence of chemical exchange linebroadening defines the NMR chemical shift time scale.
Download references. We are grateful to L. Kay for critical reading of the manuscript and valuable suggestions. We thank R. Ebright and Y. Ebright for providing the DNA fragment and N. Popovych for her help with the preparation of some CAP mutants. Author Contributions C. You can also search for this author in PubMed Google Scholar.
Correspondence to Charalampos G. PDF kb. Reprints and Permissions. Tzeng, SR. Dynamic activation of an allosteric regulatory protein. Download citation. Received : 29 June Accepted : 07 October Issue Date : 19 November Anyone you share the following link with will be able to read this content:. Sorry, a shareable link is not currently available for this article. Provided by the Springer Nature SharedIt content-sharing initiative.
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.
Advanced search. Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily. Skip to main content Thank you for visiting nature. Abstract Allosteric regulation is used as a very efficient mechanism to control protein activity in most biological processes, including signal transduction, metabolism, catalysis and gene regulation 1 , 2 , 3 , 4 , 5 , 6.
Access through your institution. Buy or subscribe. This is a preview of subscription content. Change institution. Buy article Get time limited or full article access on ReadCube. References 1 Kuriyan, J. Acknowledgements We are grateful to L. View author publications. Supplementary information. PowerPoint slides PowerPoint slide for Fig. PowerPoint slide for Fig. Rights and permissions Reprints and Permissions.
About this article Cite this article Tzeng, SR. Copy to clipboard. Further reading Achieving pure spin effects by artifact suppression in methyl adiabatic relaxation experiments Fa-An Chao Domarin Khago R. Niessen Mengyang Xu Andrea G. Damry Marc M. Mayer Roberto A. Chica Communications Biology Comments By submitting a comment you agree to abide by our Terms and Community Guidelines.
Search Search articles by subject, keyword or author. Kay for critical reading of the manuscript and valuable suggestions. We thank R. Ebright and Y. Ebright for providing the DNA fragment and N.
Popovych for her help with the preparation of some CAP mutants. N2 - Allosteric regulation is used as a very efficient mechanism to control protein activity in most biological processes, including signal transduction, metabolism, catalysis and gene regulation. AB - Allosteric regulation is used as a very efficient mechanism to control protein activity in most biological processes, including signal transduction, metabolism, catalysis and gene regulation.
Shiou Ru Tzeng, Charalampos G.
0コメント