Exploring a bioinformatics approach to assess the binding affinities of probable COVID-19 therapeutic decoys

In a recent study published in Scientific reportsresearchers developed a computational workflow based on molecular dynamics (MD) simulations and an artificial neural network (ANN) to assess the protein receptor binding domain (RBD) of severe acute respiratory syndrome coronavirus 2 (SARS -CoV-2) – human angiotensin converting enzyme 2 (hACE2) binding affinities of SARS-CoV-2 variants.

Background

Studies have reported that S-hACE2 binding interactions facilitate SARS-CoV-2 entry and subsequent replication in the host. Thus, coronavirus disease 2019 (COVID-19) can be prevented by inhibiting S-ACE2 binding.

Accordingly, human soluble ACE2 (hsACE2) that binds to SARS-CoV-2 virions before SARS-CoV-2 entry can prevent COVID-19; however, the approach requires optimization and adaptation to new SARS-CoV-2 variants.

About the study

In the present study, researchers designed a workflow by combining regular methods with scatterplot-based technology to optimize the development of SARS-CoV-2 variant-specific therapeutic decoys.

MD simulations were performed to identify human angiotensin-converting enzyme 2 amino acid substitutions that enhance S RBD-hACE2 interactions, for which an empirical scoring function (ESF) closely related to the technique LIE (linear interaction energy) was used. In vitro SARS-CoV-2 neutralization assays were performed to assess inhibition of the wild-type strain of SARS-CoV-2 and transmission of the beta variant by hACE2 variants that were bound to the crystallizable fragment region ( Fc) of human immunoglobulin G1 (hACE2-Fc).

A few hACE2-Fc variants have also been expressed in the Nicotiana benthamiana plant to study the feasibility of large-scale production. Molecular dynamics run data was combined with hACE2 halos and S RBD halos for ANN (artificial neural network) training. The model was used to estimate binding affinities of SARS-CoV-2 S with hACE2 variants based on S RBD and hACE2 halos. If a new variant emerged, hACE2 variants could be screened quickly by the artificial neural network and verified by MD simulations so that COVID-19 treatment strategies could be tailored based on which human soluble ACE2 variant has the highest affinity to binding with the new SARS-CoV-2 strain.

The potential of the system to estimate the effects of S RBD variant mutations for the same hACE2 decoys was evaluated using the SARS-CoV-2 Omicron variant subvariants BA.1 and BA.2 as examples. All probable hACE2 mutations were screened and the 300 most promising estimates were validated by MD simulations. In addition to wild-type hACE2, promising variants of hACE2, with a C-terminal human IgG Fc tag, have been expressed in Chinese hamster ovary (CHO) cells.

SARS-CoV-2 RNA was quantified by quantitative reverse transcription-polymerase chain reaction (RT-qPCR) and immunohistochemistry (IHC) analysis. The SARS-CoV-2 neutralizing potential of hACE2 variants expressed in Nicotiana benthamiana plant leaves (hACE2-Fc K31W_NB) was tested using enzyme immunoassays (ELISA). In-silico analyzes were performed to assess the binding affinities of the hACE2 variants with the Omicron BA.3, BA.4/5 and Omicron BA.2.75 RBD proteins.

The crystal structure of wild-type SARS-CoV-2 S RBD linked to hACE2 was downloaded from the Protein Data Bank (PDB) database. The ΔG value estimated by the model was calculated based on electrostatic and van der Waals forces. Sequences used for ANN training included S RBD sequences (n=1,165) and hACE2 sequences (n=95) retrieved from visual examination, literature search, or the database of the Global Initiative on Sharing All Influenza Data (GISAID) by January 4, 2022.

Results

The hACE2-Fc K31W, hACE2 T27Y_L79T_N330Y_K31W and hACE2 T27Y_L79T_K31W hACE2 variants were identified as high binding affinity candidates. Candidates produced in N. benthamiana showed 5.0 times lower and 6.0 times lower CI₅₀ (half-maximal inhibitory concentration) compared to the same variant produced in wild-type CHO and hACE2-Fc cells, respectively. The results indicated that hACE2-Fc variants with correct folding could be produced in N. benthamiana and plant-derived soluble variants of ACE2 represent a promising and cost-effective therapeutic option against SARS-CoV-2.

ESF estimates have been validated in vitro by virus neutralization tests. The experimental data are well correlated with the estimated ΔG_before (Gibbs free energies) in the model. Compared to wild-type hACE2, the majority of hACE2 variants showed enhanced binding affinities with SARS-CoV-2 beta variant, Delta variant and BA.1 subvariant and BA subvariant. 2 from Omicron. hACE2-K31W was the only mutant with very less Gibbs free energy, indicating that the K31W mutation may contribute to S RBD interactions. The presence of the K31W mutation was observed in most high binding affinity mutants.

Variants with 3.0 to 5.0 mutations showed the greatest S RBD binding. Both hACE2 T27Y_L79T_K31W and hACE2 T27Y_L79T_N330Y_K31W showed remarkably high binding affinities for BA.2 S RBD (ΔG_before value -71.0 kJ/mol) compared to wild-type hACE2 (-52.0 kJ/mol). With estimated binding affinities of -62.0 and -67.0 kJ/mol, hACE2 variants T27Y_L79T_K31W and hACE2 T27Y_L79T_N330Y_K31W were the highest high affinity variants for BA.3, and binding affinities for Omicron BA. 4/5 and Omicron BA.2.75 were lower. The highest outliers (MD ΔG values ​​<-70 kJ/mol) were mapped by the model, to the highest observed binding affinity value.

The results indicated that ANN was not only able to better estimate values ​​closer to the core of the binding affinity distribution than extrapolate from closely related variants, but also reliably map the high affinity variants at the highest affinity slice of -68.0 kJ/mol. The artificial neural network could learn significant physical information from Halos with significantly better performance than just learning a regression to the mean or a copy function, and the model could combine information acquired from inputs relatively different (distant SARS-CoV-2 sequences). The model identified single mutants comparable to the best hACE2 mutant found in early MD trials.

Overall, the study results highlighted a bioinformatics approach combining MD simulations, in vitro competitive inhibition assays, live virus and ANN infection assays, for rapid, cost-effective and efficient assessment of hACE2 decoy binding affinity to new strains of SARS-CoV-2 at an early stage, reducing times adaptation of the hACE2 decoy and examples of requirements for in vitro selections.