Machine Learning for Prioritization of Thermostabilizing Mutations for G-Protein Coupled Receptors.

Although the three-dimensional structures of G-protein coupled receptors (GPCRs), the largest superfamily of drug targets, have enabled structure-based drug design, there are no structures available for 87% of GPCRs. This is due to the stiff challenge in purifying the inherently flexible GPCRs. Identifying thermostabilized mutant GPCRs via systematic alanine scanning mutations has been a successful strategy in stabilizing GPCRs, but it remains a daunting task for each GPCR. We developed a computational method that combines sequence-, structure-, and dynamics-based molecular properties of GPCRs that recapitulate GPCR stability, with four different machine learning methods to predict thermostable mutations ahead of experiments. This method has been trained on thermostability data for 1231 mutants, the largest publicly available data set. A blind prediction for thermostable mutations of the complement factor C5a receptor 1 retrieved 36% of the thermostable mutants in the top 50 prioritized mutants compared to 3% in the first 50 attempts using systematic alanine scanning.

Prediction of Conformation Specific Thermostabilizing Mutations for Class A G Protein-Coupled Receptors.

G protein-coupled receptors (GPCRs) are highly flexible and prone to denaturation during protein extraction in detergents and purification. This poses a huge challenge to purify a conformationally homogeneous solution of GPCRs. Thermostabilizing mutations have been used widely to purify and obtain crystal structures of several GPCRs. However, identifying thermostabilizing mutations for GPCRs remains a tedious and expensive task as they are not transferable even among closely related GPCRs. Additionally, the mutations stabilizing one conformational state of a GPCR do not always stabilize other conformational state(s) of the same GPCR. Previously we developed a computational method, LiticonDesign, for rapid prediction of thermostabilizing mutations for a specific GPCR conformation. In this study, we have used LiticonDesign to predict thermostabilizing mutations for the agonist bound active-intermediate state of the human adenosine receptor (A2AR) using the structure of the inactive state of the same GPCR and vice versa. Our study shows that the thermostable mutation predictions using LiticonDesign, for an active-intermediate state of a GPCR (A2AR in our case), requires a homology model that is derived from an active/active-intermediate state GPCR structure as a template. Similarly, the homology models derived from inactive state GPCR conformations are better in predicting the thermostable mutations for the inactive state of A2AR. Overall, LiticonDesign method is not only efficient in predicting thermostabilizing mutations for a given GPCR sequence but also can recover conformation specific mutations for a state of interest, if a suitable starting structure of desired conformation is chosen.

Unraveling allostery within the angiotensin II type 1 receptor for Gαq and ß-arrestin coupling.

G protein-coupled receptors engage both G proteins and ß-arrestins, and their coupling can be biased by ligands and mutations. Here, to resolve structural elements and mechanisms underlying effector coupling to the angiotensin II (AngII) type 1 receptor (AT1R), we combined alanine scanning mutagenesis of the entire sequence of the receptor with pharmacological profiling of Gαq and ß-arrestin engagement to mutant receptors and molecular dynamics simulations. We showed that Gαq coupling to AT1R involved a large number of residues spread across the receptor, whereas fewer structural regions of the receptor contributed to ß-arrestin coupling regulation. Residue stretches in transmembrane domain 4 conferred ß-arrestin bias and represented an important structural element in AT1R for functional selectivity. Furthermore, we identified allosteric small-molecule binding sites that were enclosed by communities of residues that produced biased signaling when mutated. Last, we showed that allosteric communication within AT1R emanating from the Gαq coupling site spread beyond the orthosteric AngII-binding site and across different regions of the receptor, including currently unresolved structural regions. Our findings reveal structural elements and mechanisms within AT1R that bias Gαq and ß-arrestin coupling and that could be harnessed to design biased receptors for research purposes and to develop allosteric modulators.

C-NHEJ without indels is robust and requires synergistic function of distinct XLF domains.

To investigate the fidelity of canonical non-homologous end joining (C-NHEJ), we developed an assay to detect EJ between distal ends of two Cas9-induced chromosomal breaks that are joined without causing insertion/deletion mutations (indels). Here we find that such EJ requires several core C-NHEJ factors, including XLF. Using variants of this assay, we find that C-NHEJ is required for EJ events that use 1-2, but not ≥3, nucleotides of terminal microhomology. We also investigated XLF residues required for EJ without indels, finding that one of two binding domains is essential (L115 or C-terminal lysines that bind XRCC4 and KU/DNA, respectively), and that disruption of one of these domains sensitizes XLF to mutations that affect its dimer interface, which we examined with molecular dynamic simulations. Thus, C-NHEJ, including synergistic function of distinct XLF domains, is required for EJ of chromosomal breaks without indels.