Abstract:Several peptide dual agonists of the human glucagon receptor (GCGR) and the glucagon-like peptide-1 receptor (GLP-1R) are in development for the treatment of type 2 diabetes, obesity and their associated complications. Candidates must have high potency at both receptors, but it is unclear whether the limited experimental data available can be used to train models that accurately predict the activity at both receptors of new peptide variants. Here we use peptide sequence data labelled with in vitro potency at human GCGR and GLP-1R to train several models, including a deep multi-task neural-network model using multiple loss optimization. Model-guided sequence optimization was used to design three groups of peptide variants, with distinct ranges of predicted dual activity. We found that three of the model-designed sequences are potent dual agonists with superior biological activity. With our designs we were able to achieve up to sevenfold potency improvement at both receptors simultaneously compared to the best dual-agonist in the training set.

What problem does this paper attempt to address?

The problem this paper attempts to address is the design of novel human glucagon receptor (GCGR) and glucagon-like peptide-1 receptor (GLP-1R) dual agonists with enhanced bioactivity. Specifically, the researchers aim to use machine learning methods, particularly deep multi-task neural network models, to train models from known peptide sequence data to predict the activity of new peptide variants on both receptors. They also aim to optimize the sequences of these peptide variants through model-guided design to obtain new peptides with higher dual activity. ### Background - **Glucagon receptor (GCGR)** and **glucagon-like peptide-1 receptor (GLP-1R)** are important targets for the treatment of type 2 diabetes, obesity, and related complications. - **Peptide dual agonists** are a current research hotspot, but designing high-activity dual agonists is challenging because they need to exhibit high activity on both receptors simultaneously. - **Limited experimental data** makes it difficult to directly train models that accurately predict the activity of new peptides. ### Research Objectives - **Develop machine learning models**: Use existing peptide sequences and their activity data on GCGR and GLP-1R to train multi-task neural network models to predict the activity of new peptide variants on these two receptors. - **Design new peptide variants**: Through model-guided sequence optimization, design new peptide variants with different activity profiles, including: - High-activity dual agonists (perform well on both GCGR and GLP-1R) - Peptides that selectively activate GCGR - Peptides that selectively activate GLP-1R ### Methods - **Dataset**: Use a dataset of 125 experimentally characterized glucagon and GLP-1 peptide analogs. - **Model training**: Compare various regression model architectures, including single-task and multi-task convolutional neural network (CNN) models. - **Sequence optimization**: Design 15 peptides through a simple optimization strategy, with 5 peptides designed for each activity profile. - **Experimental validation**: Synthesize these designed peptides and measure their activity on GCGR and GLP-1R through cell-based assays. ### Results - **Model performance**: The multi-task convolutional neural network model significantly outperformed other models in predicting GLP-1R activity, while the performance difference on GCGR was not significant. - **Designed peptides**: Among the designed peptides, three (P1, P2, P3) exhibited high dual activity, with P3 showing more than 7 times the activity of the best dual agonist in the training set. - **Selective activation**: Most of the peptides designed to selectively activate GLP-1R (P11-P15) were successful, while the peptides designed to selectively activate GCGR (P6-P10) were less effective, possibly due to the smaller number of such peptides in the training data. ### Conclusion This study successfully demonstrates the potential of using machine learning methods to design high-activity peptide dual agonists, particularly achieving significant progress in predicting and optimizing GLP-1R activity. However, more experimental data is needed to improve the model's accuracy for peptides that selectively activate GCGR.

Machine learning designs new GCGR/GLP-1R dual agonists with enhanced biological potency

Molecular Insights into Ago-Allosteric Modulation of the Human Glucagon-Like Peptide-1 Receptor

Evaluation of biased agonism mediated by dual agonists of the GLP-1 and glucagon receptors

Structural Analysis of the Dual Agonism at GLP-1R and GCGR

Design of novel Xenopus GLP-1-based dual glucagon-like peptide 1 (GLP-1)/glucagon receptor agonists

De novo Design of G Protein-Coupled Receptor 40 Peptide Agonists for Type 2 Diabetes Mellitus Based on Artificial Intelligence and Site-Directed Mutagenesis

Recent Developments in Therapeutic Peptides for the Glucagon-like Peptide 1 and 2 Receptors

Design and Evaluation of 3-Phenyloxetane Derivative Agonists of the Glucagon-Like Peptide-1 Receptor

Molecular dynamics-guided optimization of BGM0504 enhances dual-target agonism for combating diabetes and obesity

Antiobesity therapeutics with complementary dual-agonist activities at glucagon and glucagon-like peptide 1 receptors

Glucagon-like peptide-1 receptor co-agonists for treating metabolic disease

Structural basis for the therapeutic advantage of dual and triple agonists at the human GIP, GLP-1 or GCG receptors

WITHDRAWN: Design and Evaluation of GLP-1 Receptor G-protein Biased Agonist with Prolonged Efficacy on Diabetes

Molecular modeling of the three-dimensional structure of GLP-1R and its interactions with several agonists

Dynamics of GLP-1R peptide agonist engagement are correlated with kinetics of G protein activation

Developing 2-Aminothiophene Derivatives as Positive Allosteric Modulators of Glucagon-Like Peptide 1 Receptor

Pharmacological characterization of mono-, dual- and tri-peptidic agonists at GIP and GLP-1 receptors

Polypharmacy through Phage Display: Selection of Glucagon and GLP-1 Receptor Co-agonists from a Phage-Displayed Peptide Library

Structural basis of peptidomimetic agonism revealed by small-molecule GLP-1R agonists Boc5 and WB4-24

Optimizing GLP-1R Agonist: A Computational Semaglutide Analogue with 112-Fold Enhanced Binding Affinity to GLP-1R

Computational approaches for clinical, genomic and proteomic markers of response to glucagon-like peptide-1 therapy in type-2 diabetes mellitus: An exploratory analysis with machine learning algorithms