Wilson, E. A. | bioRxiv Science

bioRxiv2021

The Structure of ChAdOx1/AZD-1222 Reveals Interactions with CAR and PF4 with Implications for Vaccine-induced Immune Thrombotic Thrombocytopenia

Adenovirus derived vectors, based on chimpanzee adenovirus Y25 (ChAdOx1) and human adenovirus type 26 are proving critical in combatting the 2019 SARS-CoV-2 pandemic. Following emergency use authorisation, scale up in vaccine administration has inevitably revealed vaccine related adverse effects; too rare to observe even in large Phase-III clinical trials. These include vaccine-induced thrombotic thrombocytopenia (VITT), an ultra-rare adverse event in which patients develop life-threatening blood clots 5-24 days following vaccination. To investigate vector-host interactions of ChAdOx1 underpinning VITT we solved the structure of the ChAdOx1 capsid by CryoEM, and the structure of the primary receptor tropism determining fiber-knob protein by crystallography. These structural insights have enabled us to unravel key protein interactions involved in ChAdOx1 cell entry and a possible means by which it may generate misplaced immunity to platelet factor 4 (PF4), a protein involved in coagulation. We use in vitro cell binding assays to show that the fiber-knob protein uses coxsackie and adenovirus receptor (CAR) as a high affinity binding partner, while it does not form a stable interface with CD46. Computational simulations identified a putative mechanism by which the ChAdOx1 capsid interacts with PF4 by binding in the spaces between hexon proteins, with downstream implications for the causes of VITT. SummaryWe present the structure of the ChAdOx1 viral vector, derived from chimpanzee adenovirus Y25 at 4.2[A] resolution1. ChAdOx1 is in global use in the AstraZeneca vaccine, ChAdOx1 nCoV-19/AZD-1222, to combat the SARS-CoV-2 coronavirus pandemic. Recently observed, rare, adverse events make detailed mechanistic understanding of this vector key to informing proper treatment of affected patients and the development of safer viral vectors. Here, we determine a primary mechanism ChAdOx1 uses to attach to cells is coxsackie and adenovirus receptor (CAR), a protein which is identical in humans and chimpanzees. We demonstrate the vector does not form a stable CD46 interaction, a common species B adenovirus receptor, via its primary attachment protein. Further, we reveal the surface of the ChAdOx1 viral capsid has a strong electronegative potential. Molecular simulations suggest this charge, together with shape complementarity, are a mechanism by which an oppositely charged protein, platelet factor 4 (PF4) may bind the vector surface. PF4 is a key protein involved in the formation of blood clots2, and the target of auto-antibodies in heparin-induced immune thrombotic thrombocytopenia (HITT)3, an adverse reaction to heparin therapy which presents similarly to vaccine-induced immune thrombotic thrombocytopenia (VITT), a rare complication of ChAdOx1 nCoV-19 vaccination4-6. We propose a mechanism in which the ChAdOx1-PF4 complex may stimulate the production of antibodies against PF4, leading to delayed blood clot formation, as observed in VITT.

molecular biology↗

bioRxiv2020

Mind reading of the proteins: Deep-learning to forecast molecular dynamics

Molecular dynamics (MD) simulations have emerged to become the back-bone of todays computational biophysics. Simulation tools such as, NAMD, AMBER and GROMACS have accumulated more than 100,000 users. Despite this remarkable success, now also bolstered by compatibility with graphics processor units (GPUs) and exascale computers, even the most scalable simulations cannot access biologically relevant timescales - the number of numerical integration steps necessary for solving differential equations in a million-to-billion-dimensional space is computationally in-tractable. Recent advancements in Deep Learning has made it such that patterns can be found in high dimensional data. In addition, Deep Learning have also been used for simulating physical dynamics. Here, we utilize LSTMs in order to predict future molecular dynamics from current and previous timesteps, and examine how this physics-guided learning can benefit researchers in computational biophysics. In particular, we test fully connected Feed-forward Neural Networks, Recurrent Neural Networks with LSTM / GRU memory cells with TensorFlow and PyTorch frame-works trained on data from NAMD simulations to predict conformational transitions on two different biological systems. We find that non-equilibrium MD is easier to train and performance improves under the assumption that each atom is independent of all other atoms in the system. Our study represents a case study for high-dimensional data that switches stochastically between fast and slow regimes. Applications of resolving these sets will allow real-world applications in the interpretation of data from Atomic Force Microscopy experiments.

biophysics↗