Deep-Learning-Enhanced Diffusion Imaging Assay for Resolving Local-Density Effects on Membrane Receptors

G-protein-coupled receptor (GPCR) density at the cell
surface is
thought to regulate receptor function. Spatially resolved measurements
of local-density effects on GPCRs are needed but technically limited
by density heterogeneity and mobility of membrane receptors. We now
develop a deep-learning (DL)-enhanced diffusion imaging assay that
can measure local-density effects on ligand–receptor interactions
in the plasma membrane of live cells. In this method, the DL algorithm
allows the transformation of 100 ms exposure images to density maps
that report receptor numbers over any specified region with ∼95%
accuracy by 1 s exposure images as ground truth. With the density
maps, a diffusion assay is further established for spatially resolved
measurements of receptor diffusion coefficient as well as to express
relationships between receptor diffusivity and local density. By this
assay, we scrutinize local-density effects on chemokine receptor CXCR4
interactions with various ligands, which reveals that an agonist prefers
to act with CXCR4 at low density while an inverse agonist dominates
at high density. This work suggests a new insight into density-dependent
receptor regulation as well as provides an unprecedented assay that
can be applicable to a wide variety of receptors in live cells.