Network Pharmacology analysis of orally bioavailable SARS-COV2 protease inhibitor shows synergistic targets to improve clinical efficacy

Introduction: Orally bioavailable SARS-CoV2 antiviral drugs will significantly improve the clinical management of the disease. PF07321332 (PF32) one such orally bioavailable SARS-CoV2 protease inhibitor which can be helpful to prevent viral replication in the host. Material: and methods : Hence this study evaluated the network pharmacology of PF32 using established methods to predict its potential safety and efficacy. Results: PF32 was selective against SARS-CoV2 proteases without any affinity against SARS-CoV2 RNA polymerase or its spike protein. While PF32 showed pharmacologically relevant affinity against several targets in human tissues. The target profiling of PF32 indicated a fourfold selectivity towards several proteases in human tissues with an affinity (IC 50 ) ranging from 26 to 41 nM. Conclusion: The predicted inhibitory effects of PF32 against both host and viral proteases may have synergistic effects for superior clinical efficacy.


Introduction
The pandemic caused by coronavirus (SARS-CoV2) which originated in the Wuhan region of China has caused over 150 million human infections (COVID-19) globally with a mortality rate of ~2%. [1][2][3][4] Several measures to treated the COVID-19 continues to be explored amid the spread of various mutants collateral to the mass vaccination efforts. One such approach is the development of drugs to inhibit SARS-CoV2 replication by targeting its protease. [4][5][6][7] Several synthetic and natural compounds have shown variable e cacy against SARS-CoV2 protease. Recently PF07321332 (PF32) is reported as an orally bioavailable protease inhibitor and is currently in clinical trials (Clinical trial ID NCT04756531; http://www.clinicaltrials.gov) to evaluate its safety and e cacy in the treatment of SARS-CoV-2 infections. PF32 speci cally inhibits SARS-CoV2 replicase polyprotein 1ab which is a multifunctional protein essential to transcription and replication of the coronavirus. [3,8,9] The orally bioavailability of PF32 will be major advantage in the clinical management of COVID-19. Like all drugs, PF32 may also have off target effects. Understanding of these off targets is essential to envisage potential synergistic and/or adverse effects. Hence in this study network pharmacology of PF32 was assessed using well established tools to get an insight into the all potential targets of PF32 in human cells.

Materials And Methods
Drug structure and target analysis The structure of PF32 ( gure 1A) was reconstructed in Swiss target prediction database (http://www.swisstargetprediction.ch) using its SMILES identity. [10,11] The database was searched for the all human speci c targets of PF32 and the probability scores of the targets were analysed.
Protein structure and molecular docking analysis: The protein data bank (PDB; https://www.rcsb.org) was searched for the 3D structures of identi ed targets of PF32 and the data was processed as reported before. [7,12] The PDB le of PF32 was generated using the Chimera software and used for the analysis of its molecular interactions (number of hydrogen bonds) with all its identi ed targets as reported before. [12][13][14][15] The reported a nity of homologous structure of PF32 with Fibroblast activation protein alpha (PDB ID IZ68) was used as reference and the a nity of the PF32 with its identi ed targets was predicted based on the differences in the ratio of hydrogen bonds compared to that of the reference standard as reported before. [7,12,16] In similar lines the a nity of PF32 against SARS-COV2 targets was also estimated as reported before. [12,16] Expression of PF32 targets in human tissues Protein expression of PF32 targets in various human tissues was assessed from the human protein atlas database (https://www.proteinatlas.org) on 6 th May 2020 as described before [17,18] .

Results
A majority (49%) of the PF32 targets in humans tissues are proteases, which is followed by electrochemical transporters (12%) and family A G protein coupled receptors (11%) ( gure 1B). The insilco analysis of the PF32 structure ( gure 1A), indicated several drug/lead like features including its oral bioavailability ( gure 1C). In the swiss target prediction analysis, 19 proteins showed probability scores ranging from 0.11-0.12 ( gure 1D), which will be referred to as identi ed targets of PF32. These identi ed targets were further analysed for their molecular interactions (hydrogen bonds) with PF32 in Chimera software. The number of hydrogen bonds ( gure 1E, 2A) between PF32 and its identi ed targets did not correlate with their respective probability scores, suggesting other molecular interaction (probably Van der Waals forces) may also in uence these interactions. The number of hydrogen bonds between PF32 and its identi ed targets ranged from 0 to 178 ( gure 1E). Previous reports have indicated an a nity (IC 50 ) of 73.2±0.5 between PF32 homologue and Fibroblast activation protein alpha (FAP) and this was used as a reference to predict the a nity of PF32 with its identi ed targets ( gure 1f). A nity (IC 50 ) of PF32 against the various receptors ranged from 26 to 4745 nM.
The following proteins showed higher a nity (4UFA, 1XU9, 3DDU, 1H8D, 1DUZ, 2RA3) with their IC 50 values ranging from 26 to 41 nM (table 1). PF32 showed higher a nity with Identi ed targets 4UFA and 1XU9 and least a nity with 4A5S (table 1). Nevertheless the a nities predicted with 11 of the Identi ed targets were in pharmacologically relevant concentrations (table 1).
The a nity of PF32 against various SARS-COV2 speci c targets were also assessed. Representative images of molecular interaction of PF32 against selected identi ed and SARS-COV2 speci c targets are shown in gure 2A. PF32 showed highest a nity (IC 50 45-60 nM) against SARS-COV2 main protease and Replicase polyprotein 1ab ( gure 2B). As several 3D structures of Replicase polyprotein 1ab are reported in the PDB databased, a representative of each of the variable 3D structure reported was screened in this study. PF32 did not show any a nity against SARS-COV2 spike protein, RNA polymerase and some variable structures of Replicase polyprotein 1ab (table 2).
The expression pattern of the identi ed targets of PF32 was evaluated in the human tissues. Expression pattern of representative identi ed targets (ACE, HSD11B1, PREP and FAP) are shown in gure 3. While some identi ed targets were expressed only in selected tissues (ACE and HSD11B1) others were ubiquitously expressed (PREP and FAP) ( gure 3).

Discussion
PF32 is recently developed orally bioavailable SARS-COV2 protease inhibitor which has entered clinical safety and e cacy evaluation phase (Clinical trial ID NCT04756531) (https://go.drugbank.com/drugs/DB16691). This study reports the network pharmacology analysis to identify human tissue and SARS-COV2 speci c targets the PF32 molecule can interact with. Knowledge of these interactions will be essential to understand the safety and e cacy of PF32 as a supplement to that identi ed in clinical trials. [19][20][21] The target pro ling of PF32 indicated a fourfold selectivity towards proteases with an a nity (IC 50 : 26 to 41 nM) which was pharmacologically relevant. In the human tissue the a nity of PF32 was maximum towards Angiotensin-converting enzyme (ACE). Considering the reports of SARS-COV2 using the ACE2 as receptor for entering into host cell, [3,22] the a nity of PF32 towards ACE may evince synergistic effects by both inhibiting the virus multiplication as well as preventing virus entry into host cells. Besides ACE, several other proteases were also overserved to have pharmacologically relevant a nity with PF32. Although the clinical relevance of these interactions are unclear at present, considering the systemic in ammation evinced by SARS-COV2, [23][24][25] the broader protease inhibitory potential of PF32 observed in this study may facilitate synergistic clinical bene ts.
Similar to the broader a nity of PF32 against several human tissue speci c proteases, PF32 was observed to have pharmacologically relevant a nity against SARS-COV2 main protease as well as its Replicase polyprotein 1ab. This selective inhibition of SARS-COV2 proteases but not its RNA polymerase or spike protein with higher a nity (IC 50 : 45 to 60 nM) together with its bioavailability/cell permeability may potentiate clinical safety and e cacy of PF32. However unlike the SARS-CoV2 targets, the identi ed targets of PF32 in human tissues had a wider tissue expression pro le, which paraphs re ects the wider pharmacological pro le of PF32. Depending on the tissue speci c virus presence and considering associated systemic in ammation the diffused pharmacological pro le of PF32 may prove to be clinically bene cial.
In summary the network pharmacology analysis of PF32 in this study identi es its relevant targets in human tissues and SARS-CoV2, which may have synergistic effects for superior clinical e cacy.    between PF32 and its identi ed targets at 10Å distance. F) A nity (IC50) of PF32 against its identi ed targets (direction of arrow is from low to high a nity, targets in grey have no a nity).

Figure 2
Molecular interactions and a nity of PF07321332 (PF32) with its targets. A) Selected images showing interactions of PF32 with its identi ed targets in humans and SARS-COV2 (yellow lines indicate the hydrogen bonds). B) A nity (IC50) of PF32 against various SARS-COV2 targets (direction of arrow is from low to high a nity, targets in grey have no a nity).

Figure 3
Anatomogram of selected targets of PF32 in human tissues. The graphs show the expression levels of the respective protein in various humans tissues.