Novel anti-SARS-CoV-2 mechanisms of fusion broad range anti- infective protein ricin A chain mutant-pokeweed antiviral protein 1 (RTAM-PAP1) in silico

A deadly pandemic named COVID-19 caused by a new coronavirus SARS- CoV-2 has emerged in 2019 and is still spreading globally at a dangerous pace. As of today, there are no proven vaccines, therapies or even strategies to ght off this virus. Here, we describe the in silico results of a novel broad range anti-infective fusion protein RTAM-PAP1 against the various key pro- teins of SARS-CoV-2 using the latest protein-ligand docking software. RTAM-PAP1 was compared against the SARS-CoV-2 B38 antibody, ricin A chain, pokeweed antiviral protein from leaves and the lectin grithsin using CoDockPP special COVID-19 version. These experiments revealed novel binding mechanisms of RTAM-PAP1 with high anity to numerous targets with anti-SARS-CoV-2 effects. RTAM-PAP1 was further characterized in a preliminary toxicity study in mice and was found to likely be a potent anti- SARS-CoV-2 agent. These ndings might lead to the discovery of novel SARS-CoV-2 targets and therapeutic protein structures.


Brief Report
A new pandemic has emerged and is still spreading at alarming rates at the time of this article. The socalled new coronavirus COVID-19 is currently causing damaging in ammatory response, fever, severe respiratory illness, and death globally. COVID-19 closely relates the severe acute respiratory syndrome coronavirus (SARS-CoV) based on the latest phylogenetic analysis. There are nonetheless some major essential differences in their genetic make-up that led to their very different behaviors [8][9][10] . Indeed, COVID-19 (or SARS-CoV-2) appears to have a high transmissibility from person to person and antibodies that could inhibit SARS-CoV are not functional on COVID-19 9,[11][12] . Despite global efforts, we still lack an effective antiviral strategy, drug or vaccine to ght off this virus with the growing fear that SARS-COV-2 may become another endemic virus in our communities.
In order to lower the costs of and speed up the drug discovery phase, numerous researchers have utilized in silico tools such as protein-ligand docking software to screen for traditional compounds that could bind to and inhibit the key proteins present in SARS-CoV-2, highlighting their potential antiviral activity 13 . The major targets for these compounds include SARS-CoV-2 key proteins 3-chymotrypsin-like protease (Mpro), papain like protease (PLpro), RNA-dependent RNA polymerase (RdRp), small envelope protein (E), membrane protein (M) and spike (S) proteins. The S proteins interact directly with human angiotensinconverting enzyme (ACE2), allowing the virus to enter the cells. The S protein is a class I fusion protein consisting of S1 and S2 domains with the receptor binding domain (RBD) located on the S1 domain 11 .
The RBD is the main target of antibodies and fusion inhibitors in development such as the human convalescent COVID-19 patient-origin B38 antibody (B38) and plant lectin gri thsin (GRFT).
Here, we report the potential potent antiviral activity against SARS-CoV-2 of a previously discussed novel broad spectrum anti-infective fusion protein between a mutant of the ricin A chain and pokeweed antiviral protein isolated from leaves (RTAM-PAP1) from Ricinus communis and Phytolacca americana respectively 1 . RTAM-PAP1 activity was compared with that of the B38, ricin A chain (RTA), pokeweed antiviral protein isolated from leaves (PAP1) and GRFT. Their binding and inhibiting capacity was evaluated against the major key proteins of SARS-CoV-2 in silico using the latest peptide-ligand docking software 2-7 .
The 3D structure of RTAM-PAP1 was obtained as previously described 1 and those of RTA, PAP1, B38, and GRFT were retrieved in protein data bank (PDB) format from the Research Collaboratory for Structural Bioinformatics (RCSB) website (https://www.rcsb.org/). A knowledge-based scoring docking prediction was done for all the compounds against S, S1 RBD and M using CoDockPP global docking. An additional run was done for ACE2 and human SARS antibody CR3022 against S1 RBD as reference 12 . The 3D structures of all the key proteins and ACE2 were already available at the software site in this "covid-19 targets docking only" version. The peptide/antibody-ligand version was used as small molecules docking software is not suited for these types of compounds. The generated 3D model of B38, ACE2 and CR3022 bound to S1 RBD were comparable to available crystallography of the same complexes in RCSB (access:7BZ5, 6M0J and 6W41 respectively) with some deviations (results not shown). However, the greater binding a nity and fusion inhibiting activity of B38 compared to CR3022 for S1 RBD was observed in accordance with published in vitro results 8,10,12 . Indeed, B38 was found to have a dissociation constant of 70.1 nM with a complete inhibition of ACE2 binding to S1 RBD compared to CR3022's dissociation constant of 115 nM with no inhibition of ACE2 binding. The difference in inhibition of ACE2 binding to S1 RBD being due to their binding conformation to S1 RBD. However, ACE2 binding to S1 RBD was found to have the smallest dissociation constant in literature with a value ranging from 4 to 15 nM. The results for the rst and last model (out of the top ten generated) of each compounds in complex with S, S1 RBD and M are presented in Table 1.a. B38 has the highest overall binding a nity of the lot with a binding energy ranging from -449 to -300 kcal/mol, as expected. ACE2's binding energy was between -314 to -246 kcal/mol for S1 RBD. RTAM-PAP1 is a close second, very comparable to B38, with an overall higher binding a nity (lower binding energy) than all of the other compounds tested against the S, S1 RBD and M key proteins, sometimes higher than that of B38 with -469 kcal/mol for M, for example. The high binding a nity of RTAM-PAP1 and B38 to S, S1 and M may be explained by the fact that the M epitope is very similar in structure to S1 RBD ( gure 1.a) 8,10 . RTA binding a nity is similar to RTAM-PAP1 to a certain extent and GRFT and PAP1 are very comparable. The same higher binding a nity behavior for RTAM-PAP1 was observed with Mpro, PLpro, E and RdRp when compared to PAP1, GRFT and RTA (Table 1.b). All of the tested compounds showed potentially inhibiting biding conformations to the various key proteins based on the 3D structures of the complexes formed (results not shown). These results seems to indicate that the fusion between RTAM and PAP1 allowed RTAM-PAP1 to be more stable across the different possible binding conformations with a higher binding a nity than either of its moieties alone when in complex with SARS-CoV-2 key proteins.
B38 was found to have a concentration for inhibition of the cytopathic effect of 50% (EC50) against SARS-CoV-2 simultaneous infection in Vero Cells in vitro of 0.177 ug/ml. It was further demonstrated that B38 was effective in mice post-infection 8 . GRFT was found to have low pre-infection EC50 on different strains of SARS in cytoprotection (CPE) assays in vitro (0.6-1.2 ug/ml) and effective in mice pre-infection 14 . RTA was shown in literature to have high binding a nity to many viral proteins [15][16] .
PAP1 has a broad range antiviral activity against numerous infections both in vitro and in clinical trials [17][18] . An earlier different version of RTAM-PAP1 was shown to have potent broad range antiviral activity at low post-infection EC50 (0.002-12.3 ug/ml) against human immunode ciency virus-I (HIV), hepatitis B virus (HBV), hepatitis C virus (HCV), Zika virus (Zika) and human coronavirus 229E (HCoV229E) in CPE assays in vitro 1,19 . RTA and PAP1 have been observed to have a drastic increase in viral inhibition activity if administered pre-infection both in vitro and in vivo at sub-toxic dosages [20][21][22][23] with potent antiviral mechanisms, from viral DNA/RNA depurination, viral proteins synthesis inhibition, viral cell entry inhibition to apoptosis induction of infected cells via a preferential virus infected cell entry mechanism 1 .
Nonetheless, such high a nity of RTAM-PAP1 to that many key proteins of SARS-CoV-2 is uncommon.
Yet, the most surprising part of the generated models was the discovery of unique binding mechanisms of RTAM-PAP1 with potential inhibiting activity by hindering viral entry and cellular machinery. This discovery might explain the previously observed gain of function of RTAM-PAP1 1 by means of the acquired ability to simultaneously bind the target with both moieties with high a nity, i.e. increasing the docking sites from 86 to 102 for single moiety binding and simultaneous binding to S1 RBD respectively, for example.
In order to con rm these ndings, RTAM-PAP1 was run against SARS-CoV-2 S1 and M using different docking programs (Zdock and HADDOCK) with the known active residues in RCSB. The synergetic binding of RTAM-PAP1 was con rmed and the generated models for M are shown in gure 1.b-d. Although the model generated by HADDOCK returned a more important role for PAP1 than RTAM, the simultaneous binding of both moieties can clearly be seen when in complex with M with an increase in docking sites from 62 for single moiety binding to 96 for simultaneous binding of both moieties (Zdock model). This might signi cantly increase disruption of viral cellular machinery by RTAM-PAP1. It can be concluded from these results and those previously acquired in vitro that the fusion of RTA and PAP1 via the exible linker conferred greater structure stability, enhanced activities, new binding sites and mechanisms and also, potentially, novel functions to RTAM-PAP1.
For those reasons, a short toxicity study of RTAM-PAP1 was conducted in BALB/c mice to determine the potential maximum tolerated dosage. Adverse clinical signs were observed at a single bolus intravenous administration of 3 mg/kg of RTAM-PAP1 with up-regulation of IP-10, KC and MCP-10 chemokines from 14 cytokines/chemokines assessed (results not shown). These results are in line with previously described homopolymers of ribosome inactivating proteins and con rm an in vivo behavior intermediate between that of native ribosome inactivating proteins and of immunotoxins [24][25] .
In conclusion, given the very high a nity for SARS-CoV-2 key proteins, the previous antiviral results in vitro, the newly discovered mechanisms, the preliminary in vivo pro le, potent bioactivities across the assays and preferential entry into virus infected cells as opposed to non-infected cells, it is the opinion of the authors that this novel N-glycosidase ribosome inactivating fusion protein be tested against SARS-CoV-2 in vitro and in vivo. It would be the rst tested therapeutic utilizing this particular strategy against COVID-19 and might make a difference at subtoxic dosages and open the doors for the discovery of novel SARS-CoV-2 targets and therapeutic protein structures. This table represents the binding energies in kcal/mol for the models generated by CoDockPP for each compounds in complex with the outer virus envelope proteins. The top1 is the model with the lowest binding energy (highest binding affinity) and the top10 is the 10 th model with the lowest binding energy. The lowest energy for the top1 and top10 models for each complex is in bold. This table represents the binding energies in kcal/mol for the models generated by CoDockPP for each compound less the antibody with the viral proteins important for cellular machinery. The top1 is the model with the lowest binding energy (highest binding affinity) and the top10 is the 10 th model with the lowest binding energy. The lowest energy for the top1 and top10 models for each complex is in bold.

E. coli in vivo expression system and Rabbit Reticulate
Lysate protein synthesis inhibition 2.1.1. Protein expression and puri cation RTAM-PAP1 was produced and puri ed as previously described 1 . In brief, the vector pET30a-6H-RPAP1 was generated and validated by DNA sequencing before being transformed into E. coli BL21(DE3) cells (NEB). Expression of the proteins were examined from individual clones and analyzed by either Western blot using a monoclonal antibody speci c to ricin A chain (ThermoFisher, RA999) or SDS gel stained with Comassie blue (ThermoFisher). Optimal conditions were determined and protein production induced in the presence of 1mM IPTG from 1L culture. The bacteria were then harvested by centrifugation, followed by lysing the cell pellets with 50ml of lysis buffer (50mM Tris-Cl, 150mM NaCl, 0.2% Triton X100 and 0.5mM EDTA). After sonication (3x2min), the soluble lysates were recovered by centrifugation at 35K rpm for 40min. The soluble proteins were then recovered by centrifugation (please contact the authors for more details). The puri cation of the native RTAM-PAP1 from soluble lysate was achieved by a nity versus His-tag on Ni-sepharose column (GE Healthcare). RTAM-PAP1 proteins were eluted with the elution buffer (20mM Tris-Cl, pH7.9, 100mM NaCl, 1mM EDTA and 300mM Imidazole). A second puri cation step using Hydroxylapatite column (GE Healthcare) was used to further separate RTAM-PAP1 from co-puri ed host proteins. A third puri cation step, gel ltration on a fast protein liquid chromatography (FPLC) column of Superose 12 (GE Healthcare), was necessary to completely get rid of degraded or/and premature protein products. The resulting mixture went through endotoxins removal by using a propriatery technology developed by AscentGene until endotoxin level was less than 10EU/ml. The nal product was formulated in the buffer containing 20mM HEPES-Na, pH7.9, 200mM NaCl, 0.2mM CaCl2 and 0.5mM EDTA. Dosing was staggered to allow for an initial assessment of tolerability at a particular dose level prior to escalating to a higher dose level. Mice were weighted and evaluated daily for clinical signs for 8 days following administration of RTAM-PAP1. Data were analyzed using GraphPad Prism (GraphPad Software, Inc., San Diego, CA, USA). Statistical signi cance of the difference between groups was calculated by 1factor or 2-factor ANOVA followed by post-hoc analysis. Differences were considered to be not signi cant with p>0.05. indicate where the van der Waals radii of atoms overlaps and the colors how close the contact is: yellow = close, orange = touching, and red = overlapping. For b-d, the rainbow colors (blue to red) depict the N terminal to C terminal of the colored protein (models viewed using Jmol).