Activity-based protein profiling guided new target identification of quinazoline derivatives for expediting bactericide discovery

Graphical abstract

Highlights

• •New quinazoline derivatives decorating with acrylamide moiety were prepared for probing the bactericide discovery.
• •The β-ketoacyl-ACP-synthase II (FabF) is first confirmed as new target via activity-based protein profiling.
• •The optimal antibacterial quinazoline 7a can selectively and covalently bind with FabF.
• •The FabF-targeted antibacterial behavior of quinazoline 7a is investigated in-depth.

Abstract

Introduction

The looming antibiotic-resistance problem has imposed an enormous crisis on global public health and agricultural development. Even worse, the evolution and widespread distribution of antibiotic-resistance elements in bacterial pathogens have made the resurgence of diseases that were once easily treatable deadly again. The development of antibiotics with novel mechanisms of action is urgently required.

Objectives

Inspired by charming activity-based protein profiling (ABPP) technology and increasing attention to quinazolines in the development of antibacterial agents, this study engineered a series of new quinazoline derivatives, assessed their antibacterial profiles, and first identified the possible target.

Methods

The target identification and their possible binding sites were verified by ABPP technology, molecular docking, and molecular dynamic simulations. The fatty acid synthesis process was analyzed by gas chromatography, propidium iodide staining, and scanning electron microscopy. The physicochemical properties and fungicide-likeness were evaluated using the Fungicide Physicochemical-properties Analysis Database.

Results

Compound 7a, an acrylamide-functionalized quinazoline derivative, exhibited excellent antibacterial potency against Xanthomonas oryzae pv. oryzae with an EC₅₀ value of 13.20 µM. More importantly, ABPP technology showed that β-ketoacyl-ACP-synthase Ⅱ (FabF) was the first identified quinazolines’ potential target. Compound 7a could selectively bind to the Cys151 residue of FabF through covalent interaction, suppress fatty acid biosynthesis, and damage the cell membrane integrity, thereby killing the bacteria. The pot experiment results showed that compound 7a demonstrated protective and curative values of 49.55 % and 47.46 %, surpassing controls bismerthiazol and thiodiazole copper. Finally, compound 7a exhibited low toxicity towards non-target organisms. These unprecedented performances contributed to excavating new quinazoline-based bactericidal agents.

Conclusion

Our research highlights the superiority of ABPP technology, for the first time, identifies the target of engineered quinazolines in pathogenic bacteria, and their potential target fished by ABPP tools holds great promise for the development of quinazoline-based and/or FabF-targeted bactericides.

Introduction

Bacterial pathogens are ubiquitous and can colonize a broad range of hosts, including humans and plants, leading to a perpetual increase in morbidity and mortality of humans as well as a tremendous economic burden for crops, which has been the widespread attention to research issues [1], [2], [3]. Over the past decades, the use of antibiotics has been the major accepted approach to control bacterial infections in both fields of agriculture and medicine [4]. However, the long-term and imprudent use of antibiotics has contributed to the escalating tide of bacterial resistance at an alarming rate [5], [6]. Consequently, it is a critical priority to discover new antimicrobial agents with novel chemical structures and distinct modes of action, which not only plays a vital role in tackling antibiotic-resistance problems but also can stimulate target-based antibiotic discovery [7], [8].

Quinazolines, as the most significant nitrogen-containing fused heterocycles [9], comprise over 200 natural alkaloids. Their multifaceted biological activities, including anticancer [10], antiviral [11], antihypertensive [12], and antibacterial properties [13], [14], endow themselves with discovering newer, more highly active drugs with enormous progress. Over the past few decades, quinazoline and its derivatives were found in approved drugs/pesticides that demonstrated their great potential in the field of medicine and crop protection [15], [16], [17]. For instance, several antitumor drugs featuring the pharmacophore of quinazoline have received FDA approval for clinical use, such as gefitinib (1) (Fig. 1A) and erlotinib [15]. Additionally, fenazaquin (2) (Fig. 1A) is categorized as quinazoline group of pesticide, and known for its acaricidal activity. It works by inhibiting electron transport at site Ⅰ of the mitochondrial respiratory chain, making it an effective acaricide primarily utilized for controlling mites in vegetables and fruits [16]. Notably, our previous study has reported that compound 3 (Fig. 1A) demonstrated outstanding in vitro inhibitory activity with the half maximal effective concentration (EC₅₀) value of 7.13 μg mL⁻¹ against Xanthomonas oryzae pv. oryzae (Xoo) that exerted the potential application in agriculture [17]. Therefore, quinazolines are the important pharmacodynamic framework for the development of antibacterial agents [18]. Although quinazolines have garnered popularity among researchers globally [19], [20], their underlying action mechanism, especially antibacterial behavior, is still indistinct [21].

Fig. 1.

Bioactive molecules with the quinazoline or acrylamide scaffold, the designed quinazoline-acrylamide hybrids, and the workflow of target engagement of the target compounds. (A) The chemical structures of bioactive molecules containing the quinazoline scaffold. (B) The chemical structures of acrylamide-based bioactive compounds. (C) The design strategy for acrylamide-functionalized quinazoline-4-amines. (D) The workflow for enriching and identifying cellular targets using activity-based protein profiling (ABPP) tactic in the context of pathogens. SDS-PAGE, sodium dodecyl sulfate polyacrylamide gel electrophoresis; LC-MS/MS, liquid chromatography tandem mass spectrometry.

Acryloyl group, a private warhead that forms covalent bonds with amino acids, is extensively found in drug molecules including EGFR inhibitors (afatinib, osimertinib) and Bruton’s tyrosine kinase (BTK) inhibitors (ibrutinib, zanubrutinib) [22]. In detail, the EGFR inhibitor, afatinib (4), can covalently and irreversibly bind with the HS-group of Cys797 in the position of adenosine triphosphate (ATP) binding cleft of EGFR and the covalent, irreversible, and highly potent small-molecule BTK inhibitor, ibrutinib (5), binds to Cys481 in the ATP-binding domain of BTK (Fig. 1B) [23], [24]. Besides, their antimicrobial activities are also prominent [25], [26], [27]. For example, compound 6 exhibits good antibacterial activity with a minimum inhibitory concentration (MIC) value of 0.5μg mL⁻¹ against Staphylococcus aureus (S. aureus) (Fig. 1B) [27]. As a result, we hypothesize the hybridization of quinazoline with acrylamide moiety may provide a promising strategy to identify new antibacterial agents capable of addressing bacterial resistance issues. Meanwhile, identifying the targets and probing the possible action mechanism of quinazolines toward the bacteria is also critically prominent for the discovery of antibacterial agents.

Inspiringly, the activity-based protein profiling (ABPP) strategy represents a classical and resultful “panacea” for probing the underlying action mechanism of acryloyl-based covalent molecules [28], [29], [30]. Notably, our previous studies have identified that dihydrolipoamide S-succinyltransferase (DLST) was the candidate target of the antibacterial molecule jiahuangxianjunzuo (JHXJZ) in Xanthomonas by utilizing the click chemistry-ABPP (CC-ABPP) technique. JHXJZ was identified as the first covalent DLST inhibitor that interrupted the energy production in Xanthomonas [31], [32]. Thus, the ABPP technique demonstrated the unprecedented potential for target validation and the study of biological mechanisms.

To date, various quinazolines were discovered to have excellent bactericidal activity. However, their underlying antibacterial behavior remained unknown. Therefore, considering the superiority of target identification of ABPP technology, it was employed to excavate the possible target protein of quinazolines. Herein, new quinazolines possessing acrylamide moiety were synthesized, their molecular targets were identified using the CC-ABPP tactic, and their antibacterial mode of action was investigated for the first time (Fig. 1C). Xoo, known for inducing a devastating rice disease that impedes rice quality and yield, was chosen as a model system to evaluate the antibacterial activity of this hybridization. To elucidate the antibacterial mechanism of quinazolines, clickable chemical probes were synthesized to globally profile the targets in cell lysates and living cells using the CC-ABPP strategy (Fig. 1D). Additionally, compound 7a could selectively and potently bind to the Cys151 residue of β-ketoacyl-ACP-synthase Ⅱ (FabF) through covalent interaction, suppress the biosynthesis process of fatty acid, and damage the cell membrane integrity, thereby killing the bacteria. Compound 7a showed no toxicity to rice, with favorable druggability and significant antibacterial activity in vivo. Overall, our research for the first time identifies the quinazolines-interacting target in pathogenic bacteria that will provide an unprecedented basis for understanding the antibacterial mechanism of quinazolines and stimulate new bactericide discovery for combating bacterial infection.

Materials and methods

Materials

The reagents used in this study were obtained from commercial sources. The Xanthomonas oryzae pv. oryzae (Xoo) strain was graciously provided by Mingguo Zhou (Nanjing Agricultural University, Nanjing, China).

Synthesis, characterization, and target engagement of compound 7a in Xoo

The target compounds 7a-7 g, 8a-8 g, probes P1-P8, and negative probe (NP) were synthesized through various reactions and characterized using nuclear magnetic resonance (NMR) spectra and mass spectra (supplementary information). The in vitro antibacterial evaluation, target analysis, target identification, and target confirmation in Xoo were conducted following our established protocols [17], [31], [32], [33], [34]. The detailed experiment procedures are available in the supplementary information. The primers used in this study are shown in Table S1.

Molecular docking

Homology modeling was implemented to predict the 3D structure of β-ketoacyl-ACP-synthase (FabF) in Xoo (XooFabF) with Vibrio cholerae O1 as the template using SWISS-MODEL online server [35]. Then, a covalent molecular docking on probe P1 and compound 7a with the predicted XooFabF model, respectively, was performed using CovDock in the Schrödinger package (version 2021–3). Concerning protein preparation, XooFabF underwent the addition of hydrogens, water removal, charge calculation, and restrained minimization using the Protein Preparation Wizard module. The LigPrep protocol with default parameters was employed for ligand preparation. Covalent docking utilized standard default settings, with Michael addition selected as the reaction type and Cys151 as the reactive residue. Finally, the obtained docked poses were analyzed and displayed using PyMOL [36].

Molecular dynamic (MD) simulations

The MD simulations of the docked complex were used by GROMACS (2020) software. Based on the protocol recommended CHARMM36 force field and TIP3P water model, complex topology was built in the presence of probe P1 or compound 7a covalently interacted with XooFabF residue Cys151. Then, the complex system was neutralized by adding suitable ions (Na⁺ or Cl^–). The steepest descent algorithm was conducted to minimize system energy, equilibrated with NVT/NPT simulations (100 ps) in a velocity rescaling thermostat and a Berendsen barostat. The MD simulations for every complex were 2 fs and 100 ns. The complex free energy landscape (FEL) was built by the Gromacs gmx_sham tool. Principal component analysis was employed to identify complex conformational projection by using Gromacs gmx_covar and gmx_anaeig tools. All complex simulation procedure was carried out with a GROMACS tutorial (https://www.mdtutorials.com/gmx/). The root mean square deviation (RMSD), radius of gyration (Rg), root mean square fluctuation (RMSF), hydrogen bond number, and solvent accessible surface area (SASA), were presented by Xmgrace software [37], [38].

Functional analysis

The analysis of the growth curve, fatty acid components, scanning electron microscopy (SEM), and propidium iodide (PI) staining are presented in the supplementary information [39], [40].

Evaluation of the safety, physicochemical properties, fungicide-likeness, and in vivo antibacterial activity of compound 7a

The protocols of phytotoxicity on rice, cytotoxicity evaluation, acute toxicity test on zebrafish, physicochemical properties, fungicide-likeness assessment, and in vivo testing evaluation against Xoo for compound 7a were shown in the supplementary information [41], [42], [43], [44], [45].

Statistical analysis

All experiments were conducted in biological triplicate and values were expressed as mean ± standard deviation (SD) unless otherwise indicated. Data analysis was calculated through a one-way analysis of variance (ANOVA) using Origin 2021 software (OriginLab, Corporation, USA) followed by Tukey’s post hoc test and considered significant differences when p < 0.05. Data visualization was performed using GraphPad Prism version 8.0.2 (GraphPad Software Inc., La Jolla, CA, USA).

Results

Synthesis and in vitro antibacterial evaluation of target molecules and designed probes

We initially synthesized a series of quinazoline derivatives 7a-7 g by introducing variously substituted 4-amine quinazolines (Fig. 2A). Subsequently, we conducted in vitro antibacterial evaluations against Xanthomonas oryzae pv. oryzae (Xoo), with bismerthiazol (BT) and thiodiazole copper (TC) as reference compounds. Compounds 7a, 7b, and 7c exhibited significant inhibition activity whereas the other four molecules, 7d-7 g, displayed lower antibacterial capacity against Xoo in comparison to BT and TC at concentrations of 50 and 100μg mL⁻¹ (Fig. S1A). The remarkable potency of compound 7a prompted the exploration of quinazoline derivatives with various substituents on the quinazoline’s benzenoid ring (Fig. 2B). Compounds 8a-8 g exhibited noteworthy primary antibacterial activity against Xoo when compared to BT and TC at concentrations of 50 and 100μg mL⁻¹ (Fig. S1A). Furthermore, the half maximal effective concentration (EC₅₀) values for compounds 7a, 7b, and 7c were 13.20 µM, 30.72 µM, and 39.73 µM, respectively, and the EC₅₀ values for compounds 8a-8 g ranged from 8.59 µM to 39.03 µM (Fig. 2C).

Fig. 2.

The chemical structures of target compounds and designed probes, along with their antibacterial activity against Xanthomonas oryzae pv. oryzae (Xoo). (A and B) The synthetic scheme for the target compounds 7a-7 g and 8a-8 g·THF, tetrahydrofuran; IPA, isopropanol; DMF, N, N-dimethylformamide; DIPEA, N, N-diisopropylethylamine. (C) The half maximal effective concentration (EC₅₀) values of compounds 7a-7c and 8a-8 g towards Xoo. (D) The chemical structures of designed probes. (E) The EC₅₀ values of designed probes towards Xoo. (F) The chemical structures and the EC₅₀ values of probe P1 and negative probe (NP) towards Xoo. Different letters above the bars indicate significant differences (p < 0.05). Conversely, bars marked with the same letter denote no significant differences.

Encouraged by the potent in vitro antibacterial activity of quinazoline derivatives 7a, 8a-8 g against Xoo, we derivatized them by introducing the terminal alkyne groups onto the NH-group at the 4-position of the quinazoline ring as a “clickable” tag to monitor the target engagement (Scheme S1). Thus, in our initial research, probes P1-P8 and negative probe (NP) were synthesized and evaluated for their antibacterial capacity. The results showed that probes P1-P8 exhibited significant antibacterial activity at concentrations of 50 and 100μg mL⁻¹ against Xoo (Fig. S1B). Notably, probes P1-P7 showed low EC₅₀ values, maintaining their activities in comparison with their parent compounds (Fig. 2D). Probe P1 exhibited the highest antimicrobial potential, while NP displayed little inhibitory effect, consistent with our speculation (Fig. 2E).

Selective labeling of p42 in Xoo by probe P1

In an effort to monitor the target engagement of the quinazolines, we profiled the reactivity profiles of the probes featuring diverse substituents at the 6-position of the quinazoline ring (referred to as P1, P4, P7, P2, and P8). The five probes at a concentration of 50 μM showed consistent target preferences (Fig. S2A), suggesting that the acrylamide-decorated quinazoline moiety, rather than modifications on the benzenoid ring moiety, is the primary factor responsible for protein labeling. Besides, the labeling intensity decreased both in situ and in vitro as the EC₅₀ values of probes P1, P4, P7, P2, and P8 increased, with values of 20.89, 22.94, 23.28, 39.98, and 146.60 μM, respectively. Notably, probe P1 showed superior labeling compared to the other probes (Fig. S2A). We also evaluated the proteome reactivity of the probes with various −Cl substituents at positions 5, 6, 7, and 8 of the quinazoline ring, namely P4, P5, P3, and P6. The findings unveiled comparable protein targets that were consistent with their close EC₅₀ values of 22.94, 29.95, 30.87, and 32.68 μM, respectively (Fig. S2B). Therefore, probe P1 was used as a representative probe for further studies in target engagement. Interestingly, probe P1 exhibited prominent labeling at ∼ 42 kDa in situ or in vitro and the labeling signal revealed concentration-dependent trends with concentrations ranging from 5-100 μM (Fig. 3A). As expected, NP failed to label any proteins because of the poor antibacterial efficacy. A time-course experiment demonstrated that the target protein (p42) was rapidly visualized after 5 min and completely labeled in 2 h by P1 (Fig. 3B). The labeling was dose-dependently competed upon compound 7a preincubation of cell lysates and cells (Fig. 3C). Taken together, P1 can potently, selectively, and specifically engage the p42 protein of unknown identity both in cell lysates and living cells.

Fig. 3.

Proteome profiling and target validation with probe P1 in Xoo cell lysates and live Xoo cells. (A) Concentration-dependent labeling with representative probe P1 in Xoo cell lysates and live Xoo cells unveiled potent engagement of probe P1 towards an unknown target of ∼ 42 kDa (p42). The unknown target cannot be labeled by NP. (B) Time-course labeling revealed rapid engagement of probe P1 to p42 in live Xoo cells. (C) Dose-dependent blockade of labeling in Xoo cell lysates and live Xoo cells with probe P1 by pretreatment of compound 7a. Coomassie brilliant blue (CBB) staining was employed as a loading control to verify the equal amounts of protein across the gel. (D) The streptavidin pull-down gel of live Xoo cells labeled by probe P1 was shown. (E) The schematic diagram illustrating the workflow of enriching and identifying cellular targets by probe P1. DMSO, dimethyl sulfoxide; CuAAC, copper (I)-catalyzed azide-alkyne cycloaddition. (F) Recombinant β-ketoacyl-ACP-synthase Ⅱ (FabF) in Xoo can be labeled by probe P1 in a dose-dependent manner and cannot be labeled by NP. (G) In vitro competitive labeling of probe P1 to XooFabF pre-treated with various concentrations of compound 7a. Silver staining indicates an equal loading. (H) Confirmation of the major labeling band as FabF by Western blot analysis against a specific anti-FabF antibody, analysis input and after pull down on streptavidin beads with probe P1 at a concentration of 200 μM. (I) Antagonistic effect of recombinant XooFabF on the antibacterial activity of probe P1 at concentrations of 0, 2.5 × EC₅₀, and 5 × EC₅₀ and compound 7a at concentrations of 0, 2.5 × EC₅₀, and 5 × EC₅₀ (The concentration of recombinant FabF was 1 mg mL⁻¹).

Identification of p42 as FabF by qualitative chemical proteomics

Our attention next turned to unraveling the identity of ∼ 42 kDa using qualitative chemical proteomics. The Xoo cells were treated with P1 or dimethyl sulfoxide (DMSO) for 2 h, followed by lysis and copper (I)-catalyzed azide-alkyne cycloaddition (CuAAC) mediated conjugation to a biotin-azide tag. Biotin-linked target proteins were then captured by streptavidin-conjugated beads, washed stringently, and resolved by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) (Fig. 3D). The corresponding protein band was excised, trypsinized, and subjected to liquid chromatography tandem mass spectrometry (LC-MS/MS) analysis [46]. By employing spectral counting to quantify the fold-enrichment for proteins derived from P1-treated in contrast with DMSO-treated cells, 3-oxoacyl-ACP-synthase II (also known as β-ketoacyl-ACP-synthase Ⅱ, with FabF as a common abbreviation) was pinpointed as the probable target of P1 (Table S2). FabF possesses a predicted molecular weight of 41.6 kDa, in agreement with the results of our above proteomic labeling experiments.

Validation of compound 7a-FabF interaction

To confirm the identified targets, we initially conducted a series of validation experiments involving recombinant FabF and probe P1 (Fig. 3E). As shown in Fig. 3F, recombinant FabF could be labeled by P1 in a dose-dependent manner. When FabF was preincubated with excess compound 7a prior to the addition of P1, the labeling intensity was dose-dependently blocked (Fig. 3G). Then, western blotting analysis, employing a FabF antibody to detect the enriched proteins post-pulldown, revealed a prominent band in the P1-treated sample but not in the DMSO-treated control, further substantiating that FabF was the target of P1 (Fig. 3H). Finally, we explored the in vitro antibacterial effects of compound 7a and P1 preincubated with FabF against Xoo. When both compound 7a and P1 were preincubated with FabF, they showed decreased inhibition rates compared to the compounds that were tested alone. These results indicated antagonistic effects on the antibacterial activity of FabF and confirmed their interactions between the compounds and FabF (Fig. 3I). Notably, FabF alone almost didn’t exert this effect on Xoo. Taken together, multiple lines of evidence coalesced to corroborate FabF as the target of P1 and compound 7a in Xoo cells.

Identification of the binding sites of compound 7a

Studies have shown that the Cys163-His303-His340 catalytic triad serves as the active site of FabF in Escherichia coli, being highly bound by FabF inhibitors [47], [48]. Notably, these residues are conserved in XooFabF with Cys151-His291-His328, as revealed by sequence homology analysis between XooFabF and EcFabF (Fig. S3, highlighted in black box). Meanwhile, based on structure–activity relationship analysis [22], [49], we speculated that both probe P1 and compound 7a could covalently bind to the sulfhydryl group of the active site Cys151 in FabF via Michael addition through α, β-unsaturated amide (Fig. 4A). Therefore, first, a competitive experiment was carried out using excessive compounds iodoacetamide (IAA) or N-ethylmaleimide (NEM) (active alkylating reagents of cysteine). Interestingly, IAA and NEM competed away the labeling of P1 to FabF (Fig. 4B-C), indicating P1 could interact with the sulfhydryl group of Cys151. Then, covalent docking was performed to further gain insight into the binding models of compound 7a and P1 with FabF. The structure of FabF from Vibrio cholerae O1 (PDB code: 4JRH), sharing a sequence identity of 59.35 % with XooFabF (Fig. S4), was chosen as the template for generating the 3D structure of XooFabF through homology modeling using the SWISS-MODEL online server [35], [50], [51]. The quality of the predicted FabF model was assessed using diverse techniques. PROCHECK analysis showed that 91.1 % of residues were distributed in the most favorable regions of the Ramachandran plot (Fig. S5), indicating that the model had good quality [52]. VERIFY3D results revealed that 80.95 % of the residues have scored ≥ 0.2 in the 3D/1D profile (Fig. S6A), exceeding the limitation (≥80 %) [53]. Evaluation of FabF with ProSA-web displayed a compatible Z-score value of −8.82 within the distribution range of native proteins (Fig. S6B) and the energy map illustrated predominantly negative residue energies in the FabF model (Fig. S6C) [54]. Overall, these findings imply that the predicted model is highly accurate for subsequent investigations. As shown in Fig. S7, the acrylamide motif of P1 formed a covalent bond with Cys151, aligning with the covalent mechanism [55]. Moreover, the pyrimidine motif interacted with His291 through a Pi-Pi stacking interaction and the bromine atom formed a halogen bond with Gly388. Similarly, compound 7a established a covalent bond with Cys151 and a Pi-Pi stacking interaction with His291 (Fig. 4D-E). Moreover, a nitrogen atom of the pyrimidine motif formed a hydrogen bond with Thr293 and a hydrogen of the acrylamide motif formed two hydrogen bonds with His291 and Phe385, respectively. The stronger binding ability of compound 7a compared with P1 aligns with its superior antibacterial activity. Taken together, these findings imply that P1 and compound 7a bind to FabF and covalently target Cys151.

Fig. 4.

Validation of the binding sites between compound 7a and XooFabF. (A) Reaction schemes of the irreversible covalent interaction of the nucleophilic cysteines with compound 7a or probe P1 containing α, β-unsaturated carbonyl electrophiles. (B and C) In vitro competitive labeling of recombinant XooFabF pre-treated with iodoacetamide (IAA) and N-ethylmaleimide (NEM) followed by labeling with probe P1 at a concentration of 50 μM. Silver staining indicates equal loading. The 3D binding model (D) and 2D binding model (E) for compound 7a at the active site of XooFabF. (F) The root mean square deviation (RMSD) comparation of XooFabF-7a and XooFabF-P1 complexes. (G) The radius of gyration of gyration (Rg) in XooFabF-7a and XooFabF-P1, respectively. (H) The number of XooFabF-7a or XooFabF-P1 hydrogen bonds. The solvent accessible surface area (SASA) values of XooFabF-7a and XooFabF-P1 (I), and its probability distribution (J). (K) The root mean square fluctuation (RMSF) values of XooFabF residues.

Fig. 6.

Evaluation of disrupting the membrane integrity of compound 7a. (A) Bacterial growth curve of Xoo cells in the absence or presence of 2.5 × EC₅₀ and 5 × EC₅₀ compound 7a from 0 to 24 h. OD, Optical density. (B) The fatty acid composition of Xoo cells incubated with compound 7a at concentrations of 0 and 5 × EC₅₀ for 12 h was analyzed using gas chromatography (GC). (C) Scanning electron microscopy (SEM) images (a-c) and fluorescent dye propidium iodide (PI) assay (d-f) of Xoo cells after incubation with compound 7a at various concentrations for 12 h (a-c, bacterial morphology affected by compound 7a at concentrations of 0, 2.5 × EC₅₀, and 5 × EC₅₀, Scale bar = 1 μm; d-f, fluorescence images of Xoo cells treated with compound 7a at concentrations of 0, 2.5 × EC₅₀, and 5 × EC₅₀, Scale bar = 10 μm).

In drug design, molecular dynamics simulation is a crucial tool to predict and analyze the interactions between small molecules and target proteins. By simulating the movement of the ligands located in the active pocket, researchers can observe how they bind to the protein, which is crucial for understanding the binding affinity and stability of the ligand to the protein. Here, GROMACS is performed to simulate the stabilized state of the covalent complex. Protein backbone conformation shifts from the original structure can be detected with the root mean square deviation (RMSD) value; this is a reference to judge complex system equilibration and stability during limited simulation. Compared with the original structure, the lower RMSD score reflects less fluctuation. As shown in Fig. 4F, XooFabF-7a and XooFabF-P1 systems demonstrated a relatively stable state in a 100 ns simulation. XooFabF-7a almost reached equilibrium after a fluctuation of 50 ns. Whereas XooFabF-P1 suffered slight ups and downs in the whole dynamics. The average protein backbone RMSD scores of XooFabF-7a and XooFabF-P1 were 0.26 ± 0.05 nm and 0.37 ± 0.15 nm, respectively. To gain a deeper understanding of the stability of the protein–ligand complexes, the compactness of the complex structure was analyzed by calculating protein gyration radius. The radius of gyration (Rg) movement of XooFabF-7a and XooFabF-P1 is stabilized during the simulation (Fig. 4G). They present an average Rg value of 2.10 ± 0.01 and 2.08 ± 0.01 nm, respectively, demonstrating a dynamic complete structure. Moreover, hydrogen bonds are important intermolecular interactions in protein–ligand interactions. The formation and stability of hydrogen bonds can greatly affect the affinity and specificity of a protein–ligand complex. From the hydrogen bond plot, the ligands have more hydrogen bonds with the key target areas in XooFabF-7a than in XooFabF-P1 (Fig. 4H). Furthermore, solvent accessible surface area (SASA) refers to the total surface area of a protein molecule that is accessible to solvent molecules, namely, solvent molecules can interact with specific regions of the protein, forming hydrogen bonds and hydrophobic interactions, which can stabilize the protein's structure and prevent it from unfolding or denaturing. It could be seen from Fig. 4I-J that the binding of compound 7a or probe P1 to the target protein did not significantly change the protein SASA value. Their respective mean SASA values were 181.9 ± 3.7 nm and 180.0 ± 3.8 nm, indicating that the introduction of propargyl (probe P1) in the ligand amino group (compound 7a) exhibits a weak effect on SASA at the active site. Overtly, the number and density distribution of hydrogen bonds formed by the XooFabF-7a complex are more intensive in the simulation, indicating that compound 7a may act on protein active pockets stronger, which is in line with its better antibacterial activity than probe P1. In addition, the root mean square fluctuation (RMSF) is a measure of the average deviate on of atomic positions within a molecular system during a simulation. Higher values of the RMSF typically indicate regions of loops or terminal residues. While lower RMSF values present a more rigid protein conformation. The RMSF values reflected that XooFabF-7a and XooFabF-P1 displayed equilibrium fluctuations, and their RMSF plots exhibited a mean atomic fluctuation < 0.14 Å for amino acid residues, suggesting a stable secondary structure (Fig. 4K). Comparatively, XooFabF-7a showed smaller amino acid residue fluctuations over the simulation period than XooFabF-P1.

To date, the free energy landscape (FEL) plots have been became the crucial tool, which reflects the most stable conformational protein structure in dynamic simulation. Based on the joint determination of residue and radius, XooFabF-7a showed a constrictive cluster basin (Fig. 5A) while XooFabF-P1 showed three broad ensembles (Fig. 5D). We extracted the trajectory of the lowest energy states of XooFabF-7a and XooFabF-P1, and superimposed them with the original docked complex conformation, with RMSD values of 1.089 Å (Fig. 5C) and 1.073 Å (Fig. 5F), respectively. The dynamic conformational landscape of the complex (Fig. 5B and E) showed that the collective motion of the XooFabF-7a and XooFabF-P1 was distributed in a relatively narrow landscape space, illustrating that the different substituents on the ligand did not interfere with protein movement. Moreover, the Gibbs free energy computed for XooFabF-7a and XooFabF-P1 was 15.8 and 13.1 kJ/mol, respectively. The number of cluster basins and energy calculations illustrated that XooFabF-7a appeared a superior interaction with protein.

Fig. 5.

Free energy analysis of XooFabF-7a and XooFabF-P1. (A) Free energy landscape (FEL) scope of XooFabF-7a. (B) The gyrate-RMSD conformational landscape of XooFabF-7a and XooFabF-P1. (C) The RMSD comparison of the originally docked XooFabF-7a (green) with the lowest energy state XooFabF-7a (cyan). (D) FEL scope of XooFabF-P1. (E) The PC1-PC2 conformational landscape of XooFabF-7a and XooFabF-P1. PC, principal component. (F) The RMSD comparison of the originally docked XooFabF-P1 (wheat) with the lowest energy state XooFabF-P1 (cyan).

Overall, molecular dynamic simulation demonstrated that compound 7a or probe P1 could be stable bound to protein and hardly cause prominent conformational transformation.

Fatty acid analysis and membrane damage

FabF, a β-ketoacyl-ACP synthase as type II fatty acid synthesis (FASII) in bacterial fatty acid synthesis, enables the elongation of long-chain fatty acids to form vital cellular components including cell envelopes, phospholipids, and lipoproteins [56]. To further confirm the impact of compound 7a on the fatty acid composition and content of Xoo membranes, gas chromatography (GC) was employed to quantitatively analyze. Of note, compared with the control group, the growth of Xoo showed a decreased tendency with the concentrations of compound 7a increased (Fig. 6A). To accomplish the following assay, 12 h was chosen as the optimal incubation time due to Xoo cells in the control group allowed to grow in the log phase (an OD value of 0.8). As reported by literature, the Xoo cell membranes consisted of carbamic acid, palmitoleic acid, palmitic acid, oleic acid, and stearic acid [44], [57]. The outcomes of GC showed that the ingredient of Xoo cell membranes of palmitic acid, palmitoleic acid, 17 acid, and stearic acid were confirmed in the control group and compound 7a; whereas a reduction in palmitic acid, palmitoleic acid, 17 acid, and stearic acid contents was observed upon the addition of compound 7a (Fig. 6B), indicating that compound 7a could affect the fatty acid synthesis process by targeting XooFabF.

Based on the above-mentioned outcomes of fatty acid production analysis, it could be inferred that compound 7a might inhibit the bacterial fatty acid biosynthesis and further impede the formation of the bacterial cell membrane. Therefore, scanning electron microscopy (SEM) and fluorescence microscopy were conducted to test the impact of compound 7a on the membrane. Fig. 6C showed that the Xoo cells in the control group revealed a smooth, rod-shaped surface with an undamaged appearance. However, the compound 7a-treated bacteria displayed rougher, wrinkled, and even formation of pores on the bacterial surfaces. To further ascertain the membrane damage ability triggered by compound 7a, propidium iodide (PI) fluorescence staining displayed that compound 7a-incubated Xoo cells exhibited the luminous red color in comparison to the control group, illustrating that compound 7a could disrupt the membrane integrity by suppressing the XooFabF-mediated bacterial fatty acid biosynthesis. The underlying antibacterial behavior of compound 7a was revealed in Fig. 7.

Fig. 7.

The proposed mechanisms of compound 7a exerting antibacterial effects. Compound 7a could target FabF in Xoo, further suppress the biosynthesis process of fatty acid, and damage the cell membrane integrity, thereby resulting in killing the bacteria.

Safety assessment on plant, aquatic organisms, human cell lines, and in vivo antibacterial activity of compound 7a

Based on the significant antibacterial activity and antibacterial mechanism of compound 7a, it is imperative to investigate the agricultural applications. Therefore, the safety assessments are conducted. First, we evaluated the phytotoxicity of compound 7a on rice. Notably, compared to the control group, the leaves treated with 7a exhibited no signs of damage, confirming the non-toxic nature of the compound toward rice (Fig. 8A). Then, we evaluated the environmental risk of compound 7a, considering pesticides can infiltrate water sources through rainfall, drift, and runoff, thereby posing risks to environmental organisms in agricultural applications [58]. Brachydanio rerio (B. rerio) was selected as a model organism to study the safety of compound 7a in aquatic organisms. Notably, compound 7a exhibited the concentrations that killed 50 % of the fish (LC₅₀) values of 0.53, 0.48, and 0.48 mg/L against zebrafish following 48, 72, and 96 h of exposure, respectively, indicating the time-dependent nature of these LC₅₀ values (Table 1). The results indicated that compound 7a has a weak toxicity to zebrafish. Besides, we evaluated the potential cytotoxicity of compound 7a toward non-target organisms, namely human cell line and normal rat kidney cell line. Compound 7a exhibited low cytotoxicity with the half maximal inhibitory concentration (IC₅₀) value of 23.05 μM towards A549 cells (Table 2), surpassing the reference agent 5-fluorouracil (5-FU) with an IC₅₀ value of 1.42 μM. Consistency was observed in the results for normal rat kidney cells (NRK-52E) with an IC₅₀ value of 17.14 μM compared to those for human lung cancer cells (A549), indicating good profiles in both cell lines of compound 7a.

Fig. 8.

A comprehensive assessment of the phytotoxicity, physicochemical properties, and in vivo antibacterial activity of compound 7a. (A) The phytotoxicity of compound 7a towards rice was evaluated under greenhouse conditions at concentrations of 0, 200, and 500 μg mL⁻¹. (B) The radar plots of physicochemical properties for compound 7a: molecular weight (MW), log of the octanol–water partition coefficient (ALogP), number of hydrogen bond acceptors (HBA), number of hydrogen bond donors (HBD), number of rotatable bonds (ROB), and number of aromatic bonds (ARB). The pink region signifies the upper boundary for fungicidal bioavailability, the white region delineates the lower bound range, whereas the blue hexagon illustrates the six calculated properties associated with compound 7a. (C) The pie chart for quantitative analysis of compound 7a: the relative drug likelihood (RDL), quantitative estimate of fungicide-likeness (QEF), and Gaussian score function (GAU). The disease symptoms, lesion lengths (D), disease index (E), and control efficiency (F) of compound 7a at concentrations of 0 and 200 μg mL⁻¹ were evaluated for curative and protective activity under greenhouse conditions at 14 days post-inoculation (dpi) with Xoo when infected by the scissor-clipping method in conjunction with bismerthiazol (BT) and thiodiazole copper (TC) as reference compounds. Different letters above the bars indicate significant differences (p < 0.05). Conversely, bars marked with the same letter denote no significant differences.

Table 1.

LC₅₀ values of compound 7a to B. rerio at different periods.

Time (h)	Toxic regression equation	R	LC50 (mg/L)	95 % confidence interval (mg/L)
48	y = 17.024x + 4.735	1.00	0.53	0.51 – 0.54
72	y = 3.536x + 11.232	0.99	0.48	0.44 – 0.54
96	y = 3.536x + 11.232	0.99	0.48	0.44 – 0.54

Note: LC₅₀, the concentrations that killed 50% of the fish; B. rerio, Brachydanio rerio.

Table 2.

The cytotoxicity of compound 7a against A549 and NRK-52E in vitro.

Compd.	A549					NRK-52E
	Toxic regression equation	r	IC50 (µM)	95 % confidence interval (μM)		Toxic regression equation	r	IC50 (µM)	95 % confidence interval (μM)
7a	y = 5.551x − 7.564	0.98	23.05	21.94 – 24.71		y = 10.322x − 12.739	0.98	17.14	17.02 – 18.29
5-FU	y = 0.271x − 0.041	1.00	1.42	0.82 – 3.47		y = 0.849x − 0.913	0.99	11.92	9.12 – 19.80

Note: A549, human non-small cell lung cancer cells; NRK-52E, normal rat kidney cells; Compd., compound; 5-FU, 5-fluorouracil, representing the positive control.

Subsequently, to overcome the challenges encountered in the later stages of new pesticide-lead compound development, we also conducted evaluations of pesticide-like properties of compound 7a. The Fungicide Physicochemical-properties Analysis Database (FungiPAD) is freely accessible to all researchers and is a comprehensive database that not only calculates physicochemical descriptors of small molecules but also provides qualitative and quantitative assessments of their potential as fungicides [42]. Utilizing the FungiPAD database, we evaluated the fungicide-likeness of compound 7a, with the computation panel segmented into three sections. The first section exerted the values of 37 physicochemical descriptors (Table S3). In the second section, histograms and radar plots were illustrated for the qualitative appraisal of the fungicide-likeness of compound 7a (Fig. 8B-C). The six properties that are related to size, lipophilicity, polarity, flexibility, and photostability, including molecular weight (MW), log of the octanol–water partition coefficient (ALogP), number of hydrogen bond acceptors (HBA), number of hydrogen bond donors (HBD), number of rotatable bonds (ROB), and number of aromatic bonds (ARB), were chosen for the pesticide-likeness evaluation to describe if a particular molecule has satisfying bioavailability [59]. Notably, the distribution of ARB represents the photostability, which is one of the most important properties for pesticides to be considered because of their use mainly at outdoors [60]. The histograms depict the frequency distribution of the six calculated properties (Fig. S8), revealing that the calculated properties of compound 7a have a high-frequency distribution. For the radar chart, the pink area represents the upper limit range for fungicidal bioavailability, and the white area stands for the low limit range. The blue hexagon corresponds to the values of the six calculated properties of compound 7a, all falling within the optimization range (Fig. 8B), including MW = 276.985 Da, ALogP = 0.743, HBA = 3, HBD = 1, ROB = 3, ARB = 13 (Table S3). Overall, they conform to the pesticide-likeness rules established by Hao et al. In the third section, the pie chart visually demonstrates three fungicide-likeness scores–relative drug likelihood (RDL), quantitative estimate of fungicide-likeness (QEF), and Gaussian score function (GAU)–for quantitative analysis. MW, ALogP, HBA, HBD, ROB, number of nitrogen atoms (nN), number of oxygen atoms (nO), number of rings (Rings), and topological polar surface area (TPSA), were used to display fungicide-likeness functions for GAU. MW, ALogP, HBA, HBD, ROB, and number of aromatic rings (arR), were used to illustrate fungicide-likeness functions for RDL and QEF. The RDL infers the probability of a molecule being a fungicide and the score range of QEF is 0 to 1, meaning the potential of a molecule to be a fungicide. The GAU ranges from 0 to 9, with higher scores indicating a greater likelihood of a compound being a fungicide candidate. The scores of compound 7a for RDL, QEF, and GAU are 1.266, 0.652, and 6.718, respectively, suggesting compound 7a’s strong fungicidal potential (Fig. 8C).

Finally, the effectiveness in combating Xoo against rice plants of compound 7a was assessed to explore its viability for agricultural application. As illustrated in Fig. 8D-F, compound 7a with a concentration of 200 μg mL⁻¹ exhibited superior antibacterial activity in vivo compared to BT and TC. Compound 7a demonstrated protective and curative values of 49.55 % and 47.46 %, surpassing BT with values of 41.04 % and 37.29 %, as well as TC with values of 36.75 % and 36.43 %, respectively.

Discussion

In the relentless pursuit of innovative agrochemicals, the discovery of new agrochemicals with novel structures, excellent biological activity, and unique targets is crucial [61], [62], [63]. Notably, the amalgamation of at least two molecular structures from distinct biologically active compounds through molecular hybridization is recognized as a key element in pesticide innovation [64], [65]. Therefore, we synthesized a series of bioactive molecules via the incorporation of two important scaffolds–quinazoline and acrylamide. Among the compounds 7a-7 g, compound 7a exerted the most potential antibacterial activity, whereas compounds 7b-7 g showed gradually reduced activity in vitro against Xanthomonas oryzae pv. oryzae (Xoo). We hypothesized that the reduced activity of compounds 7b-7 g might be due to the increasing steric clash of substitution group at the NH of acrylamide-tailored quinazolin-4-amines. Considering the above-mentioned findings, compound 7a was made for an interesting point to synthesize a series of quinazoline derivatives with various substituents on the quinazoline’s benzenoid ring, namely compounds 8a-8 g. The antibacterial effects of compounds 8a-8 g are similar to that of compound 7a, implying that the varying substitutions on the quinazoline ring do not significantly influence the antibacterial activity.

Due to the potent antibacterial activity of compounds 7a and 8a-8 g, the click chemistry-activity-based protein profiling (CC-ABPP) approach was employed to profile the targets and investigate the antibacterial mechanism [31], [32], [66]. Notably, the essence of CC-ABPP lies in designing activity-based probes with smaller alkyne or azide groups. The groups serve as benign tags for conjugation with different reporter groups through copper (I)-catalyzed azide-alkyne cycloaddition (CuAAC), facilitating the visualization, enrichment, and identification of protein targets [67], [68]. Therefore, a subset of cell-permeable probes (P1-P8) and negative probe (NP) were synthesized based on the parent compounds 7a and 8a-8 g. Interestingly, probe P1 was chosen for target engagement as it exhibited the highest labeling efficiency among the probes both in Xoo cell lysates and living cells, correlating with superior antibacterial activity. Probe P1 could bind the protein targets in a concentration- and time-dependent manner and selectively engaged a prominent unidentified target of ∼ 42 kDa in both cell lysates and living cells, implying that the protein of ∼ 42 kDa may be the potential target of probe P1. Notably, the competitive labeling revealed that compound 7a selectively inactivated the ∼ 42 kDa band in a dose-dependent manner, suggesting that P1 competes for the same target with compound 7a, thus distinguishing specific bands from non-specific targets. Furthermore, the pull-down experiment and qualitative chemical proteomics analysis identified β-ketoacyl-ACP-synthase Ⅱ (FabF) as the target of compound 7a. The dose-dependent, competitive labeling with recombinant FabF and the pull-down/western blotting co-confirmed the covalent binding of both P1 and compound 7a to FabF [31], [32]. Besides, the antibacterial efficacy of both P1 and compound 7a exhibited a decline with the addition of FabF, elucidating the interaction between these compounds and FabF. Notably, compound 7a contains an α,β-unsaturated carbonyl moiety, which is capable of cysteine modification [69], [70]. Competitive labeling with iodoacetamide (IAA) and N-ethylmaleimide (NEM), molecular docking, and molecular dynamic (MD) stimulation all suggested that Cys151, in the active pocket of FabF, was one of the binding sites of both P1 and compound 7a. Overall, these experimental results revealed that compound 7a covalently could bind to the sulfhydryl group of the active site Cys151 in FabF via Michael addition through α, β-unsaturated amide. Despite these experimental results in our study that first identify the quinazolines' covalent targets, the identification of non-covalent or reversible binding proteins of quinazolines is also worth studying.

Previous studies have highlighted bacterial FabF as a broad-spectrum target capable of catalyzing the formation of β-ketobutyryl-ACP from acyl-ACP and malonyl-ACP through the claisen condensation reaction, triggering the initiation of the elongation cycle in type II fatty acid synthesis [71], [72], [73]. Then the key question pertains to the antibacterial impact of compound 7a mediated by FabF. The gas chromatography (GC) technique was conducted to evaluate the effect of compound 7a on the fatty acid composition and content of Xoo cell membranes and the results showed that a reduction in contents of palmitic acid, palmitoleic acid, 17 acid, and stearic acid—the ingredient of Xoo cell membranes, were observed in the treatment of compound 7a in comparison with the control group. These findings revealed that compound 7a could interact with XooFabF and further affect the fatty acid synthesis process. Considering the function of FabF that enables the elongation of long-chain fatty acids to form vital cellular components including cell envelopes, phospholipids, and lipoproteins [56], and the abovementioned results, the effect of compound 7a on the formation of cell membranes was then investigated. The subsequent analysis of scanning electron microscopy (SEM), and propidium iodide (PI) staining experiments co-showed that compound 7a disrupted the membrane integrity of Xoo. These interesting outcomes implied that compound 7a could interact with FabF, impair enzyme functionality, disrupt fatty acid biosynthesis, and lead to cell membrane rupture, thus resulting in bacteria death.

Furthermore, because of the extensive existence of FabF in bacteria [71], compound 7a holds promise as a potential antibiotic for combatting a broader spectrum of bacterial infections by binding with FabF. Moreover, compounds 8a-8 g, exhibiting similar structures and potent antibacterial properties akin to compound 7a, could serve as covalent and irreversible FabF modulators. At present, the existing FabF inhibitors like cerulenin, thiolactomycin, platensimycin, and platencin [74], lack a shared substructure with compound 7a, thus mitigating the risk of cross-resistance. Additionally, the results of our study suggest that compound 7a has promising potential for use in paddy fields as a potential fungicidal agent with low toxicity towards rice, aquatic organisms, and human cell lines. Further research is needed to fully explore the efficacy and safety of compound 7a in agricultural settings, but these initial findings are encouraging. Additional studies could focus on optimizing the application method and dosage of compound 7a to maximize its effectiveness while minimizing any potential negative impacts on the environment or human health.

Conclusions

Our findings unveiled that compound 7a, a newly discovered quinazoline, could selectively bind covalently to FabF, inhibit the fatty acid biosynthesis process, and induce perturbations in the bacterial membrane integrity, thereby culminating in the demise of the bacterial cell and reducing infectivity in vivo, suggesting that FabF may be a viable therapeutic target for the treatment of bacterial infections. In a broader sense, the discovery of FabF-based bactericides may provide breakthroughs in the prevention and treatment of more antibiotic-resistant bacteria. To conclude, our research, for the first time, identifies the target of engineered quinazolines in pathogenic bacteria, and their potential target fished by ABPP tools holds great promise for the development of new antibiotics.

Funding

This research was financially supported by National Natural Science Foundation of China (32372610, U23A20201, 32160661, 32202359), Guizhou Provincial Foundation for Excellent Scholars Program No.GCC[2023]072, National Key Research and Development Program of China (2022YFD1700300), the Central Government Guides Local Science and Technology Development Fund Projects [Qiankehezhongyindi [2024]007], and GZU (Guizhou University) Found for Newly Enrolled Talent (No. 202229).

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Biographies

Song Yang: State Key Laboratory of Green Pesticides, Key Laboratory of Green Pesticide and Agricultural Bioengineering, Ministry of Education, Center for R&D of Fine Chemicals of Guizhou University, Guiyang, 550025, China (Corresponding Author)

Zhenchao Wang: State Key Laboratory of Green Pesticides, Key Laboratory of Green Pesticide and Agricultural Bioengineering, Ministry of Education, Center for R&D of Fine Chemicals of Guizhou University, Guiyang, 550025, China. Or School of Pharmaceutical Sciences, Guizhou University, Huaxi District, Guiyang, 550025, China. ORCID: 0000-0003-1859-0128 (Corresponding Author)

Xiang Zhou: State Key Laboratory of Green Pesticides, Key Laboratory of Green Pesticide and Agricultural Bioengineering, Ministry of Education, Center for R&D of Fine Chemicals of Guizhou University, Guiyang, 550025, China. ORCID: 0000-0002-3589-9074 (Corresponding Author)

Gui-Ping Ouyang: School of Pharmaceutical Sciences, Guizhou University, Huaxi District, Guiyang, 550025, China (Corresponding Author)

Jiao Meng: State Key Laboratory of Green Pesticides, Key Laboratory of Green Pesticide and Agricultural Bioengineering, Ministry of Education, Center for R&D of Fine Chemicals of Guizhou University, Guiyang, 550025, China. ORCID: 0009-0006-0785-5820; Email

Ling Zhang : State Key Laboratory of Green Pesticides, Key Laboratory of Green Pesticide and Agricultural Bioengineering, Ministry of Education, Center for R&D of Fine Chemicals of Guizhou University, Guiyang, 550025, China

Xin-Xin Tuo: State Key Laboratory of Green Pesticides, Key Laboratory of Green Pesticide and Agricultural Bioengineering, Ministry of Education, Center for R&D of Fine Chemicals of Guizhou University, Guiyang, 550025, China.

Yue Ding: State Key Laboratory of Green Pesticides, Key Laboratory of Green Pesticide and Agricultural Bioengineering, Ministry of Education, Center for R&D of Fine Chemicals of Guizhou University, Guiyang, 550025, China.

Kun-Lun Chen: State Key Laboratory of Green Pesticides, Key Laboratory of Green Pesticide and Agricultural Bioengineering, Ministry of Education, Center for R&D of Fine Chemicals of Guizhou University, Guiyang, 550025, China.

Mei Li: State Key Laboratory of Green Pesticides, Key Laboratory of Green Pesticide and Agricultural Bioengineering, Ministry of Education, Center for R&D of Fine Chemicals of Guizhou University, Guiyang, 550025, China.

Biao Chen: School of Pharmacy, Guizhou University of Traditional Chinese Medicine, Guiyang 550025, China.

Qing-Su Long: State Key Laboratory of Green Pesticides, Key Laboratory of Green Pesticide and Agricultural Bioengineering, Ministry of Education, Center for R&D of Fine Chemicals of Guizhou University, Guiyang, 550025, China.

Appendix A. Supplementary data

The following are the Supplementary data to this article: