Computationally-guided tuning of ligand sensitivity in a GPCR-based sensor

Dániel Kéri, Reto B. Cola, Zacharoula Kagiampaki, Tommaso Patriarchi, Patrick BarthbioRxiv 2021 · DOI

TL;DR

This paper presents a computational approach to enhance the sensitivity of GPCR-based fluorescent biosensors. By combining computational design with traditional protein engineering, the researchers successfully engineered variants of the dopamine sensor dLight1 that exhibit significantly improved potency for dopamine and norepinephrine, paving the way for more sensitive neuromodulator detection in neuroscience.

What Problem Does This Paper Address?

Genetically encoded fluorescent sensors are powerful tools for studying neuromodulators in neuroscience, offering high spatiotemporal resolution and molecular specificity. However, their utility is often limited by the inherent sensitivity of the protein scaffold to the target ligand. Improving this sensitivity, particularly for endogenous ligands, has been a persistent challenge. Traditional methods like site-directed mutagenesis struggle to account for complex, long-range mutational effects that influence ligand binding and sensor response. This paper tackles this limitation by exploring the application of computational design to rationally engineer enhanced sensor sensitivity.

Methods

The researchers employed a multi-pronged approach:

  • Computational Modeling: They generated homology models of the dopamine D1 receptor in both inactive and active states, using existing structures of related receptors. This allowed them to identify key residues involved in ligand binding and allosteric communication.
  • Computational Design:
    • Allosteric Designs: In silico deep scanning mutagenesis was performed on predicted allosteric "hub" sites. Mutations were selected based on their ability to maximize differences in allosteric coupling between active and inactive states while maintaining receptor stability.
  • Orthosteric Designs: Rosetta's Ligand design protocol was used to identify mutations within 8Å of the docked ligand that could enhance binding affinity.
  • Protein Engineering and Validation: Designed mutations were introduced into the dLight1.3b sensor backbone using molecular cloning techniques. The engineered sensors were then expressed in HEK cells and characterized in vitro using confocal microscopy and dopamine/norepinephrine titrations to assess their fluorescent responses.

Datasets

Not discussed in the paper. The study focuses on designing and validating sensor variants based on existing knowledge of GPCR structures and computational design tools.

Key Insights and Ideas

  • Allosteric Sites are Key for Sensitivity Enhancement: The most significant improvements in sensor potency were achieved by targeting allosteric regions, rather than the direct orthosteric binding site. This highlights the importance of allosteric communication pathways in modulating GPCR function and sensor performance.
  • Nature Isn't Always Optimal: The study suggests that even naturally evolved GPCRs, like the dopamine D1 receptor, may not be optimally tuned for ligand sensitivity. Computational design can reveal and exploit these suboptimalities to create significantly improved sensors.
  • Sensitivity vs. Selectivity Trade-off: Enhanced sensitivity in the designed sensors came at the expense of selectivity. The engineered sensors showed increased responses to norepinephrine, suggesting a potential trade-off that needs to be considered in future sensor development.
  • Computational Design as a Powerful Tool: The research demonstrates the efficacy of computational approaches in guiding rational protein engineering for biosensor development, offering a more efficient and targeted strategy compared to traditional mutagenesis.

Key Results

  • Two designed variants of the dopamine sensor dLight1.3b showed significantly enhanced potency:
  • Up to a 5-fold increase in potency for dopamine.
  • Up to a 15-fold increase in potency for norepinephrine.
  • Mutations in allosteric positions were successful in improving sensor responses, while mutations in orthosteric positions did not yield significant improvements and sometimes led to reduced sensitivity.
  • The absolute dynamic range (max ΔF/Fmax) for norepinephrine was largely unaltered in the high-affinity mutants, but there was a marked reduction in the absolute dynamic range for the two high-affinity mutants.

Limitations and Future Work

  • Selectivity Trade-off: The most significant limitation noted is the observed decrease in selectivity for the engineered sensors, with increased sensitivity to norepinephrine. Future work should focus on decoupling sensitivity and selectivity, potentially through further computational design or by exploring different allosteric pathways.
  • Orthosteric Design Challenges: The study found it challenging to improve binding through direct receptor-ligand contacts in the orthosteric site, likely due to the small size of the ligands and their existing strong interactions with the native receptor.
  • Reprogramming Allosteric Signaling and Selectivity: The authors suggest that the independent reprogramming of allosteric signaling and selectivity remains a promising area for future investigation.
  • Broader Applicability: The computational allosteric design approach is proposed to be broadly applicable to other GPCR-based sensors, offering a rational blueprint for exploring and understanding allosteric properties in GPCRs and accelerating the development of novel biosensors for in vivo applications.

Summary updated

Abstract

Abstract Genetically-encoded fluorescent sensors for neuromodulators are increasingly used molecular tools in neuroscience. However, these protein-based biosensors are often limited by the sensitivity of the protein scaffold towards endogenous ligands. Here, we explored the possibility of applying computational design approaches for enhancing sensor sensitivity. Using the dopamine sensor dLight1 as proof of concept, we designed two variants that boost the sensor’s potency (EC50) for dopamine and norepinephrine by up to 5- and 15-fold, respectively. Interestingly, the largest effects were obtained through improved designed allosteric transmission in the transmembrane region of the sensor. Our approach should prove generally useful for enhancing sensing capabilities of a large variety of neuromodulator sensors.
BibTeX
@misc{keri2021computationally,
  author = {Dániel Kéri and Reto B. Cola and Zacharoula Kagiampaki and Tommaso Patriarchi and Patrick Barth},
  title = {{Computationally-guided tuning of ligand sensitivity in a GPCR-based sensor}},
  year = {2021},
  doi = {10.1101/2021.09.21.461282},
  howpublished = {bioRxiv},
}
Figure