Host Adaptation of Codon Usage in SARS-CoV-2 From Mammals Indicate Natural Selection

The outbreak of COVID-19, caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infections, spread across hosts from humans to animals, transmitting particularly effectively in mink. How SARS-CoV-2 selects and evolves in the host, and the differences in the evolution of different animals are still unclear. To analysis the mutation and codon usage bias of SARS-CoV-2 in infected humans and animals. The SARS-CoV-2 sequence in mink (Mink-SARS2) and binding energy with receptor were calculated compared with human. The relative synonymous codon usage of viral encoded gene was analyzed to characterize the differences and the evolutionary characteristics. A synonymous codon usage analysis showed that SARS-CoV-2 is optimized to adapt in the animals in which it is currently reported, and all of the animals showed decreased adaptability relative to that of humans, except for mink. The neutrality plot showed that the effect of natural selection on different SARS-CoV-2 sequences is stronger than mutation pressure. A binding affinity analysis indicated that the spike protein of the SARS-CoV-2 variant in mink showed a greater preference for binding with the mink receptor ACE2 than with the human receptor, especially as the mutation Y453F and N501T in Mink-SARS2 lead to improvement of binding affinity for mink receptor. In summary, mutations Y453F and N501T in Mink-SARS2 lead to improvement of binding affinity with mink receptor, indicating possible natural selection and current host adaptation. Monitoring the variation and codon bias of SARS-CoV-2 provides a theoretical basis for tracing the epidemic, evolution and cross-species spread of SARS-CoV-2.

The glycosylation in SARS-CoV-2 and its receptor ACE2

Coronavirus disease 2019 (COVID-19), a highly infectious disease caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), has infected more than 235 million individuals and led to more than 4.8 million deaths worldwide as of October 5 2021. Cryo-electron microscopy and topology show that the SARS-CoV-2 genome encodes lots of highly glycosylated proteins, such as spike (S), envelope (E), membrane (M), and ORF3a proteins, which are responsible for host recognition, penetration, binding, recycling and pathogenesis. Here we reviewed the detections, substrates, biological functions of the glycosylation in SARS-CoV-2 proteins as well as the human receptor ACE2, and also summarized the approved and undergoing SARS-CoV-2 therapeutics associated with glycosylation. This review may not only broad the understanding of viral glycobiology, but also provide key clues for the development of new preventive and therapeutic methodologies against SARS-CoV-2 and its variants.

Involvement of Chemerin and CMKLR1 in the progesterone decrease by PCOS granulosa cells

Polycystic ovarian syndrome (PCOS) is the main cause of infertility in women. It is frequently associated with reduced progesterone production by human luteinised granulosa cells (hlGCs). However, the molecular mechanisms involved in these steroidogenesis alterations in PCOS patients are unclear. In a dihydrotestosterone-induced PCOS mouse model, steroid production is maintained in the setting of chemokine-like receptor 1 (CMKLR1) knockout. Thus, chemerin and chemerin receptors in terms of expression and progesterone regulation could be different in control and PCOS hlGCs. We firstly confirmed that progesterone levels in both plasma (p < 0.0001) and follicular fluid (FF) (p < 0.0001) were significantly reduced in PCOS normal weight women compared to control women. These data were associated with a lower STAR mRNA expression in both in vivo (p < 0.0001) and in vitro (p < 0.0001) hlGCs from PCOS women. Secondly, chemerin FF levels (p < 0.0001) and RARRES2 (p < 0.05) and CMKLR1 (p < 0.0001) mRNA levels in GCs were higher in PCOS normal weight patients. Thirdly, treatment of hlGCs with a specific nanobody (the VHH CA4910) targeting the human receptor for CMKLR1 leading to its inactivation, abolished chemerin-induced progesterone inhibition, suggesting the involvement of CMKLR1 in this process. Furthermore, the inhibition of progesterone secretion induced by chemerin was two-fold higher in PCOS hlGCs (p < 0.05). Moreover, the VHH CA4910 reinstated a normal progesterone secretion with lower concentrations in PCOS hlGCs, suggesting a different chemerin sensitivity between PCOS and control hlGCs. Thus, chemerin, through CMKLR1, could be involved in the steroidogenesis alterations in PCOS hlGCs.

AI-driven prediction of SARS-CoV-2 variant binding trends from atomistic simulations

We present a novel technique to predict binding affinity trends between two molecules from atomistic molecular dynamics simulations. The technique uses a neural network algorithm applied to a series of images encoding the distance between two molecules in time. We demonstrate that our algorithm is capable of separating with high accuracy non-hydrophobic mutations with low binding affinity from those with high binding affinity. Moreover, we show high accuracy in prediction using a small subset of the simulation, therefore requiring a much shorter simulation time. We apply our algorithm to the binding between several variants of the SARS-CoV-2 spike protein and the human receptor ACE2. Graphic abstract

Identifying key determinants and dynamics of SARS-CoV-2/ACE2 tight interaction

SARS-CoV-2 virus, the causative agent of Covid-19, has fired up a global pandemic. The virus interacts with the human receptor angiotensin-converting enzyme 2 (ACE2) for an invasion via receptor binding domain (RBD) on its spike protein. To provide a deeper understanding of this interaction, we performed microsecond simulations of the RBD-ACE2 complex for SARS-CoV-2 and compared it with the closely related SARS-CoV discovered in 2003. We show residues in the RBD of SARS-CoV-2 that were mutated from SARS-CoV, collectively help make the RBD anchor much stronger to the N-terminal part of ACE2 than the corresponding residues on RBD of SARS-CoV. This would result in a reduced dissociation rate of SARS-CoV-2 from human receptor protein compared to SARS-CoV. The phenomenon was consistently observed in simulations beyond 500 ns and was reproducible across different force fields. Altogether, our study adds more insight into the critical dynamics of the key residues at the virus spike and human receptor binding interface and potentially aids the development of diagnostics and therapeutics to combat the pandemic efficiently.

Computational Insights into the Interaction between Cytoadherence Receptor gC1qR and the DBLβ12 Domain of a Plasmodium falciparum PfEMP1 Ligand

Human receptor gC1qR is a 32 kD protein that mediates the cytoadherence of Plasmodium falciparum-infected erythrocytes (IEs) to human brain microvascular endothelial cells (HBMEC) and platelets. The cytoadherence of IEs to gC1qR has been associated with severe malaria symptoms. The cytoadherence to gC1qR is mediated by the Duffy binding-like β12 (DBLβ12) domain of Plasmodium falciparum erythrocyte membrane protein 1 (PfEMP1), PFD0020c. Here, we report the structural insights into the binding of the DBLβ12 domain of PfEMP1 with the human receptor gC1qR using computational methods. A molecular model of the DBLβ12 domain was generated and used for protein–protein docking with the host receptor gC1qR. The protein–protein docking revealed that the DBLβ12 asymmetrically interacts with two subunits of the gC1qR trimer at the solution face of gC1qR. A total of 21 amino acid residues of DBLβ12 interact with 26 amino acid residues in the gC1qR trimer through 99 nonbonding interactions and 4 hydrogen bonds. Comparative analysis of binding sites on the DBL domain fold for the two receptors gC1qR and ICAM1 showed that the two sites are distinct. This is the first study that provides structural insights into DBLβ12 binding with its receptor gC1qR and may help in designing novel antisevere malaria interventions.

Glycan remodeled erythrocytes facilitate antigenic characterization of recent A/H3N2 influenza viruses

During circulation in humans and natural selection to escape antibody recognition for decades, A/H3N2 influenza viruses emerged with altered receptor specificities. These viruses lost the ability to agglutinate erythrocytes critical for antigenic characterization and give low yields and acquire adaptive mutations when cultured in eggs and cells, contributing to recent vaccine challenges. Examination of receptor specificities of A/H3N2 viruses reveals that recent viruses compensated for decreased binding of the prototypic human receptor by recognizing α2,6-sialosides on extended LacNAc moieties. Erythrocyte glycomics shows an absence of extended glycans providing a rationale for lack of agglutination by recent A/H3N2 viruses. A glycan remodeling approach installing functional receptors on erythrocytes, allows antigenic characterization of recent A/H3N2 viruses confirming the cocirculation of antigenically different viruses in humans. Computational analysis of HAs in complex with sialosides having extended LacNAc moieties reveals that mutations distal to the RBD reoriented the Y159 side chain resulting in an extended receptor binding site.

Machine Learning augmented docking studies of aminothioureas at the SARS-CoV-2—ACE2 interface

The current pandemic outbreak clearly indicated the urgent need for tools allowing fast predictions of bioactivity of a large number of compounds, either available or at least synthesizable. In the computational chemistry toolbox, several such tools are available, with the main ones being docking and structure-activity relationship modeling either by classical linear QSAR or Machine Learning techniques. In this contribution, we focus on the comparison of the results obtained using different docking protocols on the example of the search for bioactivity of compounds containing N-N-C(S)-N scaffold at the S-protein of SARS-CoV-2 virus with ACE2 human receptor interface. Based on over 1800 structures in the training set we have predicted binding properties of the complete set of nearly 600000 structures from the same class using the Machine Learning Random Forest Regressor approach.