Discovery of Multitarget-Directed Ligands Against Influenza A Virus From Compound Yizhihao Through a Predictive System for Compound-Protein Interactions

Abstract

Influenza A virus (IAV) is a threat to public health due to its high mutation rate and resistance to existing drugs. In this investigation, 15 targets selected from an influenza virus-host interaction network were successfully constructed as a multitarget virtual screening system for new drug discovery against IAV using Naïve Bayesian, recursive partitioning, and CDOCKER methods. The predictive accuracies of the models were evaluated using training sets and test sets. The system was then used to predict active constituents of Compound Yizhihao (CYZH), a Chinese medicinal compound used to treat influenza. Twenty-eight compounds with multitarget activities were selected for subsequent in vitro evaluation. Of the four compounds predicted to be active on neuraminidase (NA), chlorogenic acid, and orientin showed inhibitory activity in vitro. Linarin, sinensetin, cedar acid, isoliquiritigenin, sinigrin, luteolin, chlorogenic acid, orientin, epigoitrin, and rupestonic acid exhibited significant effects on TNF-α expression, which is almost consistent with predicted results. Results from a cytopathic effect (CPE) reduction assay revealed acacetin, indirubin, tryptanthrin, quercetin, luteolin, emodin, and apigenin had protective effects against wild-type strains of IAV. Quercetin, luteolin, and apigenin had good efficacy against resistant IAV strains in CPE reduction assays. Finally, with the aid of Gene Ontology biological process analysis, the potential mechanisms of CYZH action were revealed. In conclusion, a compound-protein interaction-prediction system was an efficient tool for the discovery of novel compounds against influenza, and the findings from CYZH provide important information for its usage and development.

Keywords: biological process analysis; compound yizhihao; in vitro evaluation; influenza A virus; multitarget; virtual screening.

Figure 1

Scheme for model construction, identification of potential anti-influenza ingredients, and elucidation of the mechanisms of CYZH, based on network pharmacology approaches.

Figure 2

The workflow for the construction of multi-target models against IAV.

Figure 3

Diagrams of the non-bonded interactions between the target protein and the co-crystallized ligands. (A,D) show the interaction between HA and o-sialic acid. N- [4-chloranyl-5- [4- [ [3-(2-methoxyphenyl)-5-methyl-1,2-oxazol-4-yl] carbonyl] piperazin-1-yl]−2-nitro-phenyl] pyridine-2-carboxamide shows powerful interactions with NP mainly through π-π T-shaped, π-π stacked, alkyl, and hydrogen bond interactions in (B,E). The non-bonded interactions between M2 and rimantadine (C,F) were strong with the key amino acid residues VAL27 and SER31.

Figure 4

(A) Number of compounds corresponding to each target. (B) Proportion of compounds acting on different numbers of targets. (C) The network of 28 multitarget-directed ligands in CYZH and targets based on the prediction system against IAV. Blue circles represent drug nodes and red circles represent protein nodes.

Figure 5

Effect of CYZH compounds on TNF-α expression in virus-infected A549 cells. Data are expressed as mean±SD (n = 3). *P < 0.05, **P < 0.01, ***P < 0.001 vs. Model group.

Figure 6

The network of compound-target-biological process analysis for CYZH.