Synthesis and quantitative structure–activity relationship (QSAR) studies of novel rosin-based diamide insecticides

Abstract

In a continuing effort to develop novel natural product-based insecticidal agents endowed with low mammalian toxicity, two series of rosin-based diamides have been synthesized, and their insecticidal activities against Plutella xylostella and Mythimna separata were evaluated. Most of the synthesized compounds exhibited moderate to significant insecticidal activity. Among them, thiadiazole-containing diamides 6a–n displayed better activity than others, especially compounds 6f and 6n which exhibited excellent insecticidal activity against P. xylostella, with LC₅₀ values of 0.223 and 0.214 mg L⁻¹, respectively, which approximate to, or are lower than that of, the control flubendiamide (0.222 mg L⁻¹). The preliminary structure–activity relationship analysis indicated that rosin-based diamides with electron-withdrawing groups on the benzene ring showed better insecticidal activity than those with electron-donating groups. Via a best multilinear regression analysis, the generated quantitative structure–activity relationship (QSAR) model (R² = 0.9566) revealed a strong correlation of insecticidal activity against P. xylostella with molecular structures of these compounds. These consequences can be expected to instruct the design and development of new rosin-based insecticides.

---

1. Introduction

Synthetic insecticides that protect crops from plant insects and diseases are still playing an important role in current agricultural practice.¹ However, the continual application and excessive use of all these conventional insecticides over the years has led to environmental problems and the undesired development of resistance to pests.^2,3 Therefore, the search of potential insecticides with new targets and low mammalian toxicity remains desirable in the field of crop protection.^4,5

Diamides constitute a new class of insecticides for the control of the lepidopteron pest.^6,7 Studies have shown that diamides are activators of the ryanodine receptors (RyRs) which regulate Ca²⁺ release from intracellular stores located in the sarcoplasmic reticulum.⁸ Owing to their exceptional activity, unique mode of action, and low mammalian toxicity, diamides have received considerable attention.⁹ So far, three diamide insecticides, i.e., flubendiamide, chlorantraniliprole (Rynaxypyr) and cyantraniliprole (Cyazypyr) have been discovered and commercialized (Fig. 1).^10–12 However, the chemical synthesis of these diamide insecticides is time-consuming and expensive because of their complicated molecular structure and chiral centers. As a feasible solution, a class of secondary metabolites from natural sources can be chemically modified with similar structures and chiral centers.

Fig. 1 Chemical structures of diamide insecticides.

Plant secondary metabolites are obtained in the process of coevolution between plants and the environment.¹³ Currently, a range of secondary metabolites from natural sources such as flavonoids, alkaloids and terpenoids have been developed as the lead compounds for the preparation of potent insecticides.¹⁴ With the advantage of a good environmental profile and the rare development of resistance, these botanical insecticides have been considered as attractive alternatives to synthetic agrochemicals for pest management.

Rosin, a major component of secretion from pine trees, is considered as one of the most abundant and popular natural terpenoid resources in China.¹⁵ The broad-spectrum biological activities of rosin and its derivatives, such as antitumor, antibacterial, antivirus, and hormone regulation have been well documented.¹⁶ Even so, investigations reporting on the systematic study of the quantitative structure–activity relationship (QSAR) of rosin-based insecticidal agents are few. Due to the rising cost, synthesizing all possible compounds and screening a few candidates from thousands of compounds has become economically and practically impossible. One approach to improve this time-consuming and expensive process is to apply QSAR. QSAR methodology is an essential tool, which enables the calculation of numerous quantitative descriptors solely on the basis of molecular structural information. Based on the constructed QSAR models, some vital features responsible for insecticidal activity can be identified, and the potential target sites can even be speculated. Meanwhile, the use of the computational approaches and methodologies of the QSAR study may provide further guidance for the design of novel potent insecticides.^17–22

In order to obtain novel natural product-based insecticides, a series of rosin-based diamides were synthesized on the basis of molecular similarity. In addition, the thiadiazole heterocyclic group, an important pharmacophore, was introduced into the backbone structure of the diamides for designing insecticides with higher activity. The insecticidal activities of the title compounds against diamondback moth (P. xylostella) and oriental armyworm (M. separata) were evaluated. Moreover, the QSAR was also performed on all of the title compounds using the Gaussian and CODESSA software packages, which can account for their structural features responsible for the insecticidal activity. This exploration is expected to improve the application of rosin as an insecticide material.

2. Experimental sections

2.1 Synthetic procedures and characterisations

Gum rosin (grade one) was obtained from a commercial source (Wu Zhou Pine Chemicals Ltd., Guangxi, China) and used without further purification. All other chemicals used were of reagent grade. The IR spectra were taken on a Nicolet IS10 FT-IR (Nicolet, Madison, USA) spectrophotometer. The ¹H NMR spectra were recorded on a Bruker AV-300 (Bruker, Karlsruhe, Germany) nuclear magnetic resonance spectrometer with either CDCl₃ or DMSO-d₆ as the solvent and TMS as an internal standard. The MS spectra were taken on an Agilent-5973 (Agilent, Santa Clara, USA) spectrophotometer. The melting points were determined using XT-5 (Saiao, Beijing, China) melting point apparatus. The elemental analysis (C, H, and N) was carried out on a Vario EL-III (Elementar, Hanau, Germany) elemental analyzer and the results were in good agreement with the calculated values. All reactions were traced by thin layer chromatography (TLC).

2.1.1 Syntheses of 1b. Following the procedures described in previous literature,²³ 1000 g of gum rosin and 5 g of hydroquinone were added to a flask equipped with a stirrer, dropping funnel, N₂ inlet, thermometer, and water trap topped with a water-cooled condenser. The rosin was first heated under a slow stream of nitrogen and then stirred after melting had occurred. The temperature was adjusted to 220 °C, and the acrylic acid was added dropwise within 0.3 h. The mixture was then heated to 230 °C for a residence time of 4 h and the products were collected after cooling to 170 °C. Two isomers of the rosin-acrylic acid adduct (RAAA) (1a and 1b in Scheme 1, with contents of 15% and 55%, respectively, in the products) were found by gas chromatography. The target chemical 1b was obtained by recrystallization with ethanol to give colorless crystals (0.30 mm × 0.20 mm × 0.20 mm) suitable for X-ray single crystal diffraction as shown in Fig. 2 and 3 with the following crystallographic parameters: a = 12.682 (3) Å, b = 12.476 (3) Å, c = 14.629 (3) Å, β = 90.12 (3)°, λ = 0.71073 Å, θ = 9–12°, V = 2314.6 (12) ų, Z = 4, μ = 0.08 mm⁻¹, F₀₀₀ = 868, R = 0.0498, T = 293 K, wR(F²) = 0.194, S = 1.02, final R factor = 7.70%.²⁴

Scheme 1 Preparation of compounds 3 and 4a–n.

Fig. 2 Molecular structure of the compound 1b.

Fig. 3 Packing diagram of the compound 1b.

2.1.2 Syntheses of 16-isopropyl-5,9-dimethyltetracyclo[10.2.2.0^1,10.0^4,9]hexadec-15-ene-5,14-dicarbonyl chloride (2). A solution of 1b (7.50 g; 20 mmol) and 80 mL dichloromethane (DCM) were added to a 250 mL flask equipped with a water-cooled condenser, thermometer, drying tube, and dropping funnel. The solution was stirred until the solid dissolved. After that, thionyl chloride (5.95 g; 50 mmol) was added dropwise through a dropping funnel within 1 h. After refluxing for 4 h at 65 °C, acrylpimaryl chloride (2) was obtained as a yellow oil after removing the DCM and excess thionyl chloride under reduced pressure. Yield: 7.40 g (90%). IR (cm⁻¹): 2926, 2854 (–CH₃, –CH₂); 1741 (C=O). ¹H NMR (CDCl₃ δ/ppm 300 MHz), 5.50 (s, H, C=CH–); 2.52–1.41 (m, 5H, –CH–); 1.86–1.24 (m, 14H, –CH₂–); 1.27–1.10 (m, 12H, CH₃). ESI-MS m/z = 412 [M + H] ⁺. Anal. calcd for C₂₃H₃₂Cl₂O₂: C, 67.15; H, 7.84. Found: C, 67.21; H, 7.89.

2.1.3 Syntheses of 16-isopropyl-5,9-dimethyltetracyclo[10.2.2.0^1,10.0^4,9]hexadec-15-ene-5,14-dicarbonylamide (3). A solution of compound 2 (7.40 g; 18 mmol) in 15 mL of tetrahydrofuran (THF) was added dropwise to a solution of 200 mL of ammonium water (80%) in 40 mL THF within 30 minutes at 0 °C. After reacting for 12 h, the THF solvent and excess ammonium water were removed under vacuum. Then, the mixture was washed with deionized water three times and recrystallized with ethyl acetate to give a yellow solid 3. Yield: 4.71 g (70.4%); yellow powder; m.p. 152.8–153.6 °C. IR (cm⁻¹): 3359 (N–H); 2933, 2849 (–CH₃, –CH₂); 1663 (N–C=O). ¹H NMR (CDCl₃ δ/ppm 300 MHz), 5.78 (m, 4H, CONH₂); 5.49 (s, H, C=CH–); 2.61 (m, H, –CH–(Me)₂); 2.42 (s, H, –CH–C=O–); 2.12–1.43 (m, 3H, –CH–); 1.82–1.26 (m, 14H, –CH₂–); 1.34–1.06 (m, 12H, CH₃). ESI-MS m/z = 373 [M + H]⁺. Anal. calcd for C₂₃H₃₆N₂O₂: C, 74.15; H, 9.74; N, 7.52. Found: C, 74.11; H, 9.69; N, 7.56.

2.1.4 Syntheses of 16-isopropyl-5,9-dimethyltetracyclo[10.2.2.0^1,10.0^4,9]hexadec-15-ene-5,14-dicarboxamides (4a–n). A solution of compound 2 (7.40 g; 18 mmol) in 15 mL of DCM was added dropwise to a solution of 60 mmol of amine and 60 mmol of triethylamine (Et₃N) in 40 mL of DCM within 30 minutes at room temperature. After reacting for 12 h, the mixture was washed using 50 mL of 0.1 M hydrochloric acid and deionized water three times. Purification of the residue by silica gel chromatography [v (ethyl acetate)/v (petroleum ether) = 1 : 10] gave the fourteen resulting derivatives, 4a to 4n. The example data of compounds 4a and 4b are shown as follows, whereas data of compounds 4c–n can be found in the ESI.†
Data for compound 4a. Yield: 7.77 g (70.4%); white powder; m.p. 140.8–141.6 °C. IR (cm⁻¹): 3363, 3259 (N–H); 2926, 2854 (–CH₃, –CH₂); 1659 (N–C=O); 738, 697 (Ar–H). ¹H NMR (DMSO-d₆ δ/ppm 300 MHz): 9.61, 9.06 (m, 2H, CONH–); 7.61–6.96 (m, 10H, Ar–H); 5.29 (s, H, C=CH–); 4.71–4.30 (m, 4H, Ar–CH₂–); 2.57 (s, H, –CH–C=O–); 2.01–1.83 (m, 3H, –CH–); 1.82–1.24 (m, 14H, –CH₂–); 1.53 (m, H, –CH–(Me)₂); 1.14–0.61 (m, 12H, CH₃). ESI-MS m/z = 553 [M + H]⁺. Anal. calcd for C₃₇H₄₈N₂O₂: C, 80.39; H, 8.75; N, 5.07. Found: C, 80.27; H, 9.00; N, 4.86.
Data for compound 4b. Yield: 7.55 g (72%); white powder; m.p. 135.6–137.4 °C. IR (cm⁻¹): 3354 (N–H); 2961, 2861 (–CH₃, –CH₂); 1663 (N–C=O); 754, 691 (Ar–H). ¹H NMR (DMSO-d₆ δ/ppm 300 MHz): 8.08, 7.95 (m, 2H, CONH–); 7.29–7.20 (m, 10H, Ar–H); 5.25 (s, H, C=CH–); 2.68 (s, H, –CH–C=O–); 2.50–1.28 (m, 14H, –CH₂–); 2.11–1.90 (m, 3H, –CH–); 1.57 (m, H, –CH–(Me)₂); 1.17–0.55 (m, 12H, CH₃). ESI-MS m/z = 525 [M + H]⁺. Anal. calcd for C₃₅H₄₄N₂O₂: C, 80.11; H, 8.45; N, 5.34. Found: C, 79.90; H, 8.54; N, 5.08.

2.1.5 Syntheses of 16-isopropyl-5,9-dimethyltetracyclo[10.2.2.0^1,10.0^4,9]hexadec-15-ene-5,14-dicarbonylamide (5). A mixture of 1b (3.75 g; 10 mmol) and thiosemicarbazide (1.90 g; 20 mmol) in phosphorus oxychloride (25 mL) was refluxed. After reacting for 2 h, the mixture was cooled and adjusted to pH = 10 with aqueous sodium hydroxide. The solid was collected by filtration and then recrystallized with ethanol to give compound 5. Yield: 4.60 g (95%); white powder; m.p. 135.9–136.7 °C. IR (cm⁻¹): 3412, 3298, 1614 (N–H); 1635 (C=N); 1099 (C–S–C). ¹H NMR (DMSO-d₆ δ/ppm 300 MHz), 5.44 (s, H, C=CH–); 4.00 (m, 4H, –NH₂); 2.74–1.78 (m, 5H, –CH–); 1.80–1.25 (m, 14H, –CH₂–); 1.33–1.05 (m, 12H, CH₃). ESI-MS m/z = 485 [M + H]⁺. Anal. calcd for C₂₅H₃₆N₆S₂: C, 61.95; H, 7.49; N, 17.34. Found: C, 61.93; H, 7.51; N, 17.05.

2.1.6 Syntheses of 16-isopropyl-5,9-dimethyltetracyclo[10.2.2.0^1,10.0^4,9]hexadec-15-ene-5,14-[1,3,4]thiadiazol-2-yl]-diamide (6a–n). A solution of the chloride (20 mmol) in 15 mL of DCM was added dropwise to a solution of 5 (4.85 g; 10 mmol) and 60 mmol of Et₃N in 40 mL of DCM within 30 minutes at 0 °C. After reacting for 12 h, the mixture was washed using 50 mL of 0.1 M hydrochloric acid and deionized water three times. Purification of the residue by silica gel chromatography [v (ethyl acetate)/v (petroleum ether) = 1 : 5] gave the fourteen resulting derivatives, 6a to 6n. The example data of compounds 6a and 6b are shown as follows, whereas data of compounds 6c–n can be found in the ESI.†
Data for compound 6a. Yield: 3.98 g (70%); white powder; m.p. 235.9–236.7 °C. IR (cm⁻¹): 3370, 3159 (N–H); 2926, 2854 (–CH₃, –CH₂); 1660 (N–C=O); 1633 (C=N); 1080 (C–S–C). ¹H NMR (DMSO-d₆ δ/ppm 300 MHz), 8.01, 7.95 (s, 2H, –CONH–); 5.44 (s, H, C=CH–); 2.80–1.80 (m, 5H, –CH–); 2.02, 1.95 (m, 6H, –COCH₃); 1.86–1.24 (m, 14H, –CH₂–); 1.34–0.98 (m, 12H, –CH₃). ESI-MS m/z = 569 [M + H]⁺. Anal. calcd for C₂₉H₄₀N₆O₂S₂: C, 61.24; H, 7.09; N, 14.78. Found: C, 61.33; H, 7.01; N, 14.75.
Data for compound 6b. Yield: 4.50 g (65%); white powder; m.p. 222.3–223.2 °C. IR (cm⁻¹): 3329, 3100 (N–H); 2961, 2880 (–CH₃, –CH₂); 1659 (N–C=O); 1632 (C=N); 1079 (C–S–C); 751, 690 (Ar–H). ¹H NMR (DMSO-d₆ δ/ppm 300 MHz): 8.18, 8.02 (m, 2H, –CONH–); 7.59–7.44 (m, 10H, Ar–H); 5.25 (s, H, C=CH–); 2.81–1.79 (m, 5H, –CH–); 1.95–1.28 (m, 14H, –CH₂–); 1.37–0.55 (m, 12H, CH₃). ESI-MS m/z = 693 [M + H]⁺. Anal. calcd for C₃₉H₄₄N₆O₂S₂: C, 67.60; H, 6.40; N, 12.13. Found: C, 67.63; H, 6.41; N, 12.05.

2.2 Biological assay

The P. xylostella and M. separata used in the bioassay were provided by the research & development center of biorational pesticide, Northwest A&F University. The bioassays were performed in an artificial greenhouse with a constant temperature (25 ± 0.5 °C), relative humidity (75 ± 5%), and photoperiod (L/D = 14/10). All test compounds were dissolved in DMSO and diluted with sterile water to obtain required concentrations. Assessments were made on a dead/alive basis, and mortality rates were corrected using Abbott’s formula. Evaluations were based on a percentage scale of 0–100, where 0 = no activity and 100 = complete eradication. The standard deviations of the tested biological values were ±5%. LC₅₀ values were calculated by probit analysis. For comparative purposes, the commercial product flubendiamide was tested under the same conditions.^25,26

2.2.1 Larvicidal activity against P. xylostella. The larvicidal activity of the title compounds and contrast compound flubendiamide against P. xylostella were estimated according to the leaf-dip method using the reported procedure.^27–30 Briefly, leaf disks (6 cm × 2 cm) of fresh cabbage leaves were dipped into the test solution for 3–5 s and dried. The treated leaf disks were placed individually into glass tubes. Each dried treated leaf disk was infested with 30 second-instar P. xylostella. Leaves treated with DMSO and sterile water were provided as a control. All experiments were conducted in triplicate to ensure reproducibility at a given concentration.

2.2.2 Larvicidal activity against M. separata. The larvicidal activity of the title compounds and contrast compound flubendiamide against M. separata were also estimated according to the leaf-dip method using the reported procedure.^26,31,32 Briefly, leaf disks (about 5 cm diameter) of fresh corn leaves were dipped into the test solution for 10–15 s. After air drying, the treated leaf disks were placed on plates. Each dried treated leaf disk was infested with 10 third-instar M. separata. Percentage mortalities were evaluated 72 h after treatment. Leaves treated with DMSO and sterile water were provided as a control. All experiments were conducted in triplicate to ensure reproducibility at a given concentration.

2.3 Building and validation of the QSAR model

Firstly, the optimal conformers with the lowest energy of the title compounds were performed at the DFT/6-31G (d) level using a Gaussian 03W package of programs.³⁰ Secondly, the calculated results were changed into a form compatible with CODESSA 2.7.15³³ using Ampac 9.1.3.³⁴ Finally, all the molecular descriptors involved in these compounds were calculated by CODESSA 2.7.15. In order to find out which structural features play an important role in insecticidal activity against P. xylostella, the best multiple regression analysis was selected to generate the QSAR model equation. In this equation, the statistical criteria were indicated by the squared correction coefficient (R²), the squared standard error of the estimates (S²), and the Fisher significance ratio (F). Tested LC₅₀ values were converted into the corresponding log LC₅₀ and used as dependent variables in the QSAR studies. The quality of the final model was determined using both an internal validation and the “leave-one-out” cross-validation methods.

3. Results and discussion

3.1 Synthesis

The syntheses of two series of rosin-based diamide derivatives developed in the present work are illustrated in Scheme 1 and 2 respectively. As a dicarboxylic acid, the RAAA was obtained by the Diels–Alder addition reaction between rosin and acrylic acid; the content of the RAAA in the products was 70%. Additional purifications were required to remove other reactants and separate the two isomers 1a and 1b from the RAAA. According to the literature, the target isomer 1b was separated by recrystallization with an overall yield of 95%.¹⁵ Acrylopimaryl chloride (2) was prepared from the reaction of 1b with thionyl chloride and used without purification. The target compounds 3 and 4a–n were prepared by a simple and convenient three-step procedure starting from rosin. At room temperature, compounds 4a–n were synthesized in high yields (60–90%) by the reaction of acrylopimaryl chloride (2) with various aromatic amines in DCM using Et₃N as the acid acceptor. However, compound 3 was prepared by the reaction of NH₃·H₂O with acrylopimaryl chloride (2) at 0 °C in THF with a satisfactory yield (70%). As shown in Scheme 2, compound 1b and thiosemicarbazide refluxed in POCl₃ to give the corresponding thiadiazole substituted amine 5 via a cyclization reaction. The reaction of 5 with various chlorides can give thiadiazole-containing amides, 6a to 6n. The structures of the title compounds were well-characterized by IR, ¹H NMR, MS, and elemental analysis.

Scheme 2 Preparation of compounds 6a–n.

3.2 Biological activity and structure–activity relationships

3.2.1 Larvicidal activity against P. xylostella. All of the compounds were initially tested at a concentration of 10 mg L⁻¹, and consequently the compounds with high insecticidal potency were investigated further at low concentrations. The result of the larvicidal activity of the title compounds against P. xylostella is summarized in Table 1, from which we can see that compounds 3, 4a–n and 6a–n exhibited moderate to significant insecticidal activity. The introduction of a thiadiazole heterocyclic group increased the insecticidal activity of the title compounds. From Table 1, we can see that at 5 mg L⁻¹, compounds 6c–f, 6h and 6k–n exhibited 90–100% larvicidal activities, and eight of those compounds still possessed 50–80% activities at 1 mg L⁻¹, respectively. It is worth noting that 6f and 6n showed a death rate of 77% at 0.25 mg L⁻¹, which is more effective than flubendiamide (70%) against P. xylostella. The LC₅₀ values correspondingly were 0.223 and 0.214 mg L⁻¹, which were similar to, or lower than, that of the control flubendiamide (0.222 mg L⁻¹). On the one hand, most compounds with electron-withdrawing substituents F, Cl, Br, and –CF₃ displayed a higher larvicidal activity against P. xylostella (the compound 6j, possessing the electron-withdrawing substituent NO₂, which did not display a higher larvicidal activity, needs further study), while compounds with electron-donating substituents –OCH₃ and –OCF₃ led to a significant decrease in activity (4k–n > 4h–j, 6d–f, 6h, 6k–n > 6g, 6i and 6j). It can be concluded that the electronic effect of the substituent on the benzene ring is important in the insecticidal activity of amide groups.²⁵ On the other hand, the steric effect should be considered, when substituents of –OCH₃ and –OCF₃ were introduced, their activities were comparably low, which indicated that the larger substituents in the position have a negative effect on the activity. These observations revealed that substitution patterns on the benzene ring have an important influence on the larvicidal activity.^11,35

Table 1 Insecticidal activity of compounds against P. xylostella

No.	Compound	Larvicidal activity (%) at a concentration of (mg L^−1)								LC50	y = a + bx	R^2	log LC50
		10	5	2.5	1	0.5	0.25	0.1	0.05
1	3	100	67	40	20	0	—	—	—	2.759	y = −1.210 + 2.746x	0.985	0.441
2	4a	100	73	40	7	0	—	—	—	2.949	y = −1.672 + 3.560x	1.000	0.470
3	4b	100	73	40	7	0	—	—	—	2.949	y = −1.672 + 3.560x	1.000	0.470
4	4c	100	77	40	7	0	—	—	—	2.879	y = −1.681 + 3.659x	1.000	0.459
5	4d	100	73	37	10	0	—	—	—	2.923	y = −1.563 + 3.355 x	0.992	0.466
6	4e	100	77	37	10	0	—	—	—	2.853	y = −1.567 + 3.443x	0.986	0.455
7	4f	100	80	57	7	0	—	—	—	2.506	y = −1.463 + 3.667x	0.974	0.399
8	4g	100	73	40	7	0	—	—	—	2.949	y = −1.672 + 3.560x	1.000	0.470
9	4h	70	40	17	3	0	—	—	—	6.193	y = −1.948 + 2.460x	0.998	0.792
10	4i	70	37	17	3	0	—	—	—	6.379	y = −1.968 + 2.446x	0.995	0.805
11	4j	67	40	20	3	0	—	—	—	6.339	y = −1.837 + 2.290x	0.998	0.802
12	4k	100	83	50	23	3	0	—	—	2.131	y = −0.924 + 2.813x	0.983	0.329
13	4l	100	80	50	20	3	0	—	—	2.246	y = −0.988 + 2.812x	0.993	0.351
14	4m	100	77	50	23	3	0	—	—	2.244	y = −0.931 + 2.652x	0.980	0.351
15	4n	100	77	50	27	3	0	—	—	2.185	y = −0.875 + 2.578x	0.964	0.339
16	6a	100	87	50	30	3	0	—	—	2.076	y = −0.922 + 2.906x	0.982	0.317
17	6b	100	87	57	30	3	0	—	—	1.982	y = −0.874 + 2.942x	0.992	0.297
18	6c	100	90	57	27	3	0	—	—	1.879	y = −0.816 + 2.977x	0.983	0.274
19	6d	100	90	70	57	23	7	0	—	1.096	y = −0.084 + 2.115x	0.972	0.040
20	6e	100	90	70	53	23	7	0	—	1.123	y = −0.107 + 2.125x	0.981	0.050
21	6f	100	100	90	80	77	60	33	7	0.223	y = 1.004 + 1.543x	0.953	−0.652
22	6g	100	70	40	10	0	—	—	—	2.927	y = −1.516 + 3.250x	1.000	0.466
23	6h	100	90	70	50	20	7	0	—	1.182	y = −0.155 + 2.136x	0.974	0.073
24	6i	100	67	37	10	0	—	—	—	3.068	y = −1.559 + 3.201x	0.999	0.487
25	6j	100	73	40	7	0	—	—	—	2.949	y = −1.672 + 3.560x	1.000	0.470
26	6k	100	90	70	53	23	7	0	—	1.123	y = −0.107 + 2.125x	0.981	0.050
27	6l	100	90	67	50	20	7	0	—	1.211	y = −0.178 + 2.147x	0.983	0.083
28	6m	100	90	70	57	23	7	0	—	1.096	y = −0.084 + 2.115x	0.972	0.040
29	6n	100	100	90	80	77	60	33	13	0.214	y = 1.001 + 1. 494x	0.966	−0.670
	Flubendiamide	100	100	97	80	73	60	30	10	0.222	y = 1.125 + 1.722x	0.977	−0.654

---

3.2.2 Larvicidal activity against M. separata. The results of larvicidal activity of the title compounds against M. separata are listed in Table 2, from which we can see that compounds 3, 4a–n, and 6a–n exhibited moderate larvicidal activity against M. separata. The death rate of most of these compounds at 50 mg L⁻¹ was about 30–40%, which was less effective than larvicidal activity against P. xylostella.

Table 2 Insecticidal activity of compounds against M. separata

Compd	Larvicidal activity (%) at a concentration of (mg L^−1)
	50	20	10
3	40	0	—
4a	30	0	—
4b	40	0	—
4c	40	0	—
4d	37	0	—
4e	37	0	—
4f	27	0	—
4g	40	0	—
4h	17	0	—
4i	17	0	—
4j	20	0	—
4k	40	0	—
4l	30	0	—
4m	40	0	—
4n	40	0	—
6a	37	0	—
6b	37	0	—
6c	20	0	—
6d	40	0	—
6e	17	0	—
6f	17	0	—
6g	20	0	—
6h	40	0	—
6i	30	0	—
6j	40	0	—
6k	40	0	—
6l	37	0	—
6m	37	0	—
6n	30	0	—
Flubendiamide	100	100	100

---

3.3 QSAR

There are many regression approaches available for CODESSA 2.7.15 software, such as best multi-linear regression, multi-linear regression, principal component analysis, partial least-square regression, and heuristic regression.³⁶ Taking into account the number of samples and descriptors used in this study, the best multi-linear regression was selected for developing the QSAR model. The “breaking point” rule for the determination of the number of the descriptors was employed as described in Fig. 4. The best multi-linear regression showed a significant increase in R² when the number of the descriptors was no more than 5. However, there was negligible change in R² when the number of descriptors increased from 5 to 6. Descriptors with high t values were accepted and those with low t values were rejected. A “breaking point” indicates that the improvement of the regression model has become insignificant (ΔR² < 0.02–0.04).³⁷ In addition, the number of the descriptors complies to the linear regressions given by eqn (1).

N ≥ 3(K + 1) (1)

where N is the number of sample compounds and K is the number of descriptors. Therefore, the final model with five descriptors was selected as the best model. The values of all five descriptors can be found in the ESI (Table S1†).

Fig. 4 The “breaking point” rule results.

The statistically optimized QSAR equation for the log LC₅₀ data has the following statistical characteristics: R² = 0.9566, F = 101.46, S² = 0.0040 (Table 3). This model includes five descriptors in descending order according to their statistical significance (t values), where X and ΔX are the regression coefficients and their standard errors. The comparison between the experimental and predicted log LC₅₀ is listed in Table 4. Fig. 5 shows the plot of predicted versus experimental activity of the 29 compounds. The five-descriptor QSAR model equation and corresponding statistical criteria are described in the following eqn (2) (Fig. 6).

log LC₅₀ = 2.8005 + 3.4038 × HOMO − 9.2784 × DM + 2.7178 × q^O_max − 2.3561 × q^N_min + 2.4803 × μ_h (2)

N = 29, R² = 0.9566, F = 101.46, S² = 0.0040

Table 3 The best five-descriptor model

Descriptor No.	X	±ΔX	t-Test	Descriptor
a Energy of the highest occupied molecular orbit in atomic units.b Dipole moment.c Max. net atomic charge for a O atom.d Min. net atomic charge for a N atom.e Tot hybridization composite of the molecular dipole.
0	2.8005	4.9272 × 10^−1	5.6838	Intercept
1	3.4038	3.0849 × 10^−1	11.0338	HOMO^a
2	−9.2784	9.1786 × 10^−1	−10.1086	DM^b
3	2.7178	4.0462 × 10^−1	6.7169	q^Omax^c
4	−2.3561	2.6446 × 10^−1	−8.9090	q^Nmin^d
5	2.4803	7.6091 × 10^−1	3.2596	μh^e

---

Table 4 The difference between the experimental log LC₅₀ and predicted log LC₅₀

No.	Compd	Calc. log LC50	Exp. log LC50	Difference
1	3	0.508	0.441	0.067
2	4a	0.463	0.470	−0.007
3	4b	0.431	0.470	−0.039
4	4c	0.469	0.459	0.010
5	4d	0.469	0.466	0.003
6	4e	0.474	0.455	0.019
7	4f	0.443	0.399	0.044
8	4g	0.505	0.470	0.035
9	4h	0.779	0.792	−0.013
10	4i	0.749	0.805	−0.056
11	4j	0.803	0.802	0.001
12	4k	0.269	0.329	−0.060
13	4l	0.257	0.351	−0.094
14	4m	0.248	0.351	−0.103
15	4n	0.355	0.339	0.016
16	6a	0.268	0.317	−0.049
17	6b	0.228	0.297	−0.069
18	6c	0.270	0.274	−0.004
19	6d	0.139	0.040	0.099
20	6e	0.118	0.050	0.068
21	6f	−0.619	−0.652	0.033
22	6g	0.563	0.466	0.097
23	6h	0.051	0.073	−0.022
24	6i	0.418	0.487	−0.069
25	6j	0.500	0.470	0.030
26	6k	0.083	0.050	0.033
27	6l	0.154	0.083	0.071
28	6m	0.094	0.040	0.053
29	6n	−0.766	−0.670	−0.096

---

Fig. 5 Experimental log LC₅₀ versus predicted log LC₅₀.

Fig. 6 HOMO energy maps for compounds 6f and 6n from DFT calculation of Gaussian 03W. The green parts represent positive molecular orbitals, and the red parts represent negative molecular orbitals.

The internal validation and “leave-one-out” cross-validation methods were used to validate the developed QSAR model.^37,38 The internal validation was carried out by dividing the compound data into three subsets A, B and C, with 10, 10 and 9 compounds respectively. The compounds 1, 4, 7, 10, etc., went into subset (A); 2, 5, 8, 11, etc., went into subset (B); and 3, 6, 9, 12, etc., went into the third subset (C). Two of the three subsets, (A and B), (A and C), and (B and C), made up the training set while the remaining subset was treated as a test set. The correlation equations were derived from each of the training sets using the same descriptors and then applied to predict values for the corresponding test set. Internal validation results are presented in Table 5. The R_Training² and R_Test² are within 5% for all three sets, and the average values of R_Training² = 0.9575 and R_Test² = 0.9530 were close to the overall R² value. Therefore, the QSAR model obtained demonstrated the predictive power of 3-fold cross-validation. The “leave-one-out” method was completed in a similar manner to the internal validation. In the “leave-one-out” method, a set of seven compounds (4, 8, 12, etc.) was used as the external test set and the remaining compounds were left in the training subset. The QSAR model containing the same five descriptors was obtained with R² = 0.9868 from the training set. When the same QSAR model was applied on the test set, R² = 0.9793 was observed. Therefore, the “leave-one-out” cross-validation results were also satisfactory.

Table 5 Internal validation of the QSAR model^a

Training set	N	R^2	F	S^2	Test set	N	R^2	F	S^2
a Compds A: 1, 4, 7, 10, 13, 16, 19, 22, 25, 28. Compds B: 2, 5, 8, 11, 14, 17, 20, 23, 26, 29. Compds C: 3, 6, 9, 12, 15, 18, 21, 24, 27.
A + B	20	0.9600	100.26	0.0053	C	9	0.9644	112.34	0.0037
B + C	19	0.9599	98.44	0.0031	A	10	0.9433	100.52	0.0039
A + C	19	0.9526	118.20	0.0048	B	10	0.9515	108.75	0.0048
Average		0.9575	105.63	0.0044	Average		0.9530	107.20	0.0041

---

By interpreting the descriptors involved in the model, we gained some insight into the structural features influencing insecticidal activity. The foremost important descriptor was the HOMO energy (the energy of the highest occupied molecular orbital). This descriptor has a significant effect on the activity as the energy of the HOMO is directly related to the ionization potential of the compounds and characterizes the susceptibility of the molecule to electrophilic attack.^39,40 The negative contribution of HOMO energy also suggested that the electron withdrawing substitution groups of rosin-based diamides are favorable for the insecticidal activity against P. xylostella.

The second important descriptor was the dipole moment. This descriptor was important in modulating insecticidal activity against P. xylostella because of the presence of C=O, N–H, and thiadiazole groups, which exhibited permanent polarization due to an electronegativity difference between the atoms.^37,39 The O (C=O), N (N–H/thiadiazole group), and S (thiadiazole group) atoms may be involved in binding interactions with insect cells present at the target site. The dipole moment thus played a critical role in modulating the insecticidal activity of the test compounds.^41,42

The 3^rd and 4^th descriptors obtained in the model were max. net atomic charge for a O atom and min. net atomic charge for a N atom. These two descriptors belonged to electrostatic descriptors, and reflected characteristics of the charge distribution of the molecules.^39,41–43 The 5^th descriptor obtained in the model was tot hybridization composite of the molecular dipole, which belonged to quantum-chemical descriptors. This descriptor reflected the quantitative measure of the lipophilic and hydrophobic properties of the compounds, and was essential for the penetration and distribution of the compounds as well as the interaction of compounds with receptors.^44,45 In eqn (2), appearance with a negative sign in the model indicates that a molecule with a lower descriptor value has a higher log LC₅₀, and contrary to that a positive sign in the model indicates that a molecule with a higher descriptor value has a higher log LC₅₀.

4. Conclusions

In summary, 29 rosin-based diamides 3, 4a–n and 6a–n were synthesized and their insecticidal activities against P. xylostella and M. separata were evaluated. Thiadiazole-containing diamides 6a–n displayed better activity than others, especially compounds 6f and 6n which exhibited excellent insecticidal activities against P. xylostella. A QSAR study indicated that the involved descriptors for rosin-based diamides may account for their structural features responsible for the insecticidal activity. These promising results are of significant importance to the development of insecticides from common, inexpensive, and non-toxic natural products.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (Project no. 31400509 and Project no. 31401783), Forestry Industry Research Special Funds for Public Welfare Projects (Project no. 201404701) and Research Fund for the Doctoral Program of Higher Education of Northwest A&F University (Project no. 2013BSJJ048).

Notes and references

1. G. P. Lahm, T. M. Stevenson, T. P. Selby, J. H. Freudenberger, D. Cordova, L. Flexner, C. A. Bellin, C. M. Dubas, B. K. Smith, K. A. Hughes, J. G. Hollingshaus, C. E. Clark and E. A. Benner, Bioorg. Med. Chem. Lett., 2007, 17, 6274–6279.
2. G. P. Lahm, D. Cordova and J. D. Barry, Bioorg. Med. Chem., 2009, 17, 4127–4133.
3. G. P. Lahm, D. Cordova, J. D. Barry, T. F. Pahutski, B. K. Smith, J. K. Long, E. A. Benner, C. W. Holyoke, K. Joraski, M. Xu, M. E. Schroeder, T. Wagerle, M. J. Mahaffey, R. M. Smith and M. Tong, Bioorg. Med. Chem. Lett., 2013, 23, 3001–3006.
4. L. A. Teixeira and J. T. Andaloro, Pestic. Biochem. Physiol., 2013, 106, 76–78.
5. T. P. Selby, G. P. Lahm, T. M. Stevenson, K. A. Hughes, D. Cordova, I. B. Annan, J. D. Barry, E. A. Benner, M. J. Currie and T. F. Pahutski, Bioorg. Med. Chem. Lett., 2013, 23, 6341–6345.
6. A. Jeanguenat, Pest Manage. Sci., 2013, 69, 7–14.
7. M. Liu, Y. Wang, W. Wangyang, F. Liu, Y. Cui, Y. Duan, M. Wang, S. Liu and C. Rui, J. Agric. Food Chem., 2010, 58, 6858–6863.
8. B. Troczka, C. T. Zimmer, J. Elias, C. Schorn, C. Bass, T. G. E. Davies, L. M. Field, M. S. Williamson, R. Slater and R. Nauen, Insect Biochem. Mol. Biol., 2012, 42, 873–880.
9. C. Gnamm, A. Jeanguenat, A. C. Dutton, C. Grimm, D. P. Kloer and A. J. Crossthwaite, Bioorg. Med. Chem. Lett., 2012, 22, 3800–3806.
10. B. Wang, H. Zhu, Y. Ma, L. Xiong, Y. Li, Y. Zhao, J. Zhang, Y. Chen, S. Zhou and Z. Li, J. Agric. Food Chem., 2013, 61, 5483–5493.
11. M. Feng, Y. Li, H. Zhu, L. Zhao, B. Xi and J. Ni, J. Agric. Food Chem., 2010, 58, 10999–11006.
12. A. K. Isaacs, S. Qi, R. Sarpong and J. E. Casida, Chem. Res. Toxicol., 2012, 25, 1571–1573.
13. Y. Wang, Y. Shao, Y. Wang, L. Fan, X. Yu, X. Zhi, C. Yang, H. Qu, X. Yao and H. Xu, J. Agric. Food Chem., 2012, 60, 8435–8443.
14. C. L. Cantrell, F. E. Dayan and S. O. Duke, J. Nat. Prod., 2012, 75, 1231–1242.
15. J. Li, J. Song, S. Shang, X. Rao and Y. Gao, Nat. Prod. Res., 2012, 27, 702–710.
16. E. J. Alvarez-Manzaneda, R. Chahboun, J. J. Guardia, M. Lachkar, A. Dahdouh, A. Lara and I. Messouri, Tetrahedron Lett., 2006, 47, 2577–2580.
17. G. Liu, X. Ju, J. Cheng and Z. Liu, Chemosphere, 2010, 78, 300–306.
18. A. Speck-Planche, M. Natalia Dias Soeiro Cordeiro, L. Guilarte-Montero and R. Yera-Bueno, Curr. Comput.–Aided Drug Des., 2011, 7, 304–314.
19. A. Speck-Planche, V. V. Kleandrova and M. T. Scotti, Chemom. Intell. Lab. Syst., 2012, 111, 39–45.
20. C. Hansch and R. P. Verma, Eur. J. Med. Chem., 2009, 44, 260–273.
21. P. K. Naik, S. T. Singh and H. Singh, SAR QSAR Environ. Res., 2009, 20, 551–566.
22. H. Xu, J. Wang, H. Sun, M. Lv, X. Tian, X. Yao and X. Zhang, J. Agric. Food Chem., 2009, 57, 7919–7923.
23. J. Li, X. Rao, S. Shang, Y. Gao and J. Song, BioResources, 2012, 17, 1961–1971.
24. H. Wang, S. Shang, Y. Yin, X. Rao and X. Xu, Acta Crystallogr., Sect. E: Struct. Rep. Online, 2009, 65, o1521.
25. Y. Zhou, B. Wang, F. Di, L. Xiong, N. Yang, Y. Li, Y. Li and Z. Li, Bioorg. Med. Chem. Lett., 2014, 24, 2295–2299.
26. Q. Feng, Z. Liu, L. Xiong, M. Wang, Y. Li and Z. Li, J. Agric. Food Chem., 2010, 58, 12327–12336.
27. M. Feng, Y. Li, H. Zhu, J. Ni, B. Xi and L. Zhao, Pest Manage. Sci., 2012, 68, 986–994.
28. J. Shang, W. Wang, Y. Li, H. Song, Z. Li and J. Wang, J. Agric. Food Chem., 2012, 60, 8286–8293.
29. F. Huang, M. Zhao, X. Zhang, C. Wang, K. Qian, R. Kuo, S. Morris-Natschke, K. Lee and S. Peng, Bioorg. Med. Chem., 2009, 17, 6085–6095.
30. M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, J. A. Montgomery, T. Vreven, K. N. Kudin, J. C. Burant, J. M. Millam, S. S. Iyengar, J. Tomasi, V. Barone, B. Mennucci, M. Cossi, G. Scalmani, N. Rega, G. A. Petersson, H. Nakatsuji, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, M. Klene, X. Li, J. E. Knox, H. P. Hratchian, J. B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann, O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, P. Y. Ayala, K. Morokuma, G. A. Voth, P. Salvador, J. J. Dannenberg, V. G. Zakrzewski, S. Dapprich, A. D. Daniels, M. C. Strain, O. Farkas, D. K. Malick, A. D. Rabuck, K. Raghavachari, J. B. Foresman, J. V. Ortiz, Q. Cui, A. G. Baboul, S. Clifford, J. Cioslowski, B. B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I. Komaromi, R. L. Martin, D. J. Fox, T. Keith, M. A. Al-Laham, C. Y. Peng, A. Nanayakkara, M. Challacombe, P. M. W. Gill, B. Johnson, W. Chen, M. W. Wong, C. Gonzalez and J. A. Pople, Gaussian 03, Gaussian, Inc., Wallingford CT, 2004.
31. J. Zhang, J. Xu, B. Wang, Y. Li, L. Xiong, Y. Li, Y. Ma and Z. Li, J. Agric. Food Chem., 2012, 60, 7565–7572.
32. M. Mao, Y. Li, Y. Zhou, X. Zhang, Q. Liu, F. Di, H. Song, L. Xiong, Y. Li and Z. Li, J. Agric. Food Chem., 2014, 62, 1536–1542.
33. A. Katritsky, M. Karelson, V. S. Lobanov, R. Dennington, T. A. Keith and R. D. A. T. Keith, Codessa 2.7.15, Semichem, Inc., Shawnee, KS, 2004.
34. M. J. S. Dewar, A. J. Holder, I. Roy, D. Dennington, D. A. Liotard, D. G. Truhlar, T. A. Keith, J. M. Millam and C. D. Harris, AMPAC 9.3.1, Semichem, Inc., Shawnee, KS, 2004.
35. S. Kang, B. Song, J. Wu, M. He, D. Hu, L. Jin, S. Zeng, W. Xue and S. Yang, Eur. J. Med. Chem., 2013, 67, 14–18.
36. J. Song, Z. Wang, A. Findlater, Z. Han, Z. Jiang, J. Chen, W. Zheng and S. Hyde, Bioorg. Med. Chem. Lett., 2013, 23, 1245–1248.
37. Z. Wang, J. Song, J. Chen, Z. Song, S. Shang, Z. Jiang and Z. Han, Bioorg. Med. Chem. Lett., 2008, 18, 2854–2859.
38. S. Liao, J. Song, Z. Wang, J. Chen, G. Fan, Z. Song, S. Shang, S. Chen and P. Wang, Bioorg. Med. Chem. Lett., 2014, 24, 773–779.
39. P. Sharma, A. Kumar, S. Upadhyay, V. Sahu and J. Singh, Eur. J. Med. Chem., 2009, 44, 251–259.
40. D. Sharma, B. Narasimhan, P. Kumar and A. Jalbout, Eur. J. Med. Chem., 2009, 44, 1119–1127.
41. A. D. Pillai, S. Rani, P. D. Rathod, F. P. Xavier, K. K. Vasu, H. Padh and V. Sudarsanam, Bioorg. Med. Chem., 2005, 13, 1275–1283.
42. K. Kalani, D. Yadav, F. Khan, S. Srivastava and N. Suri, J. Mol. Model., 2012, 18, 3389–3413.
43. P. P. Roy, S. Kovarich and P. Gramatica, J. Comput. Chem., 2011, 32, 2386–2396.
44. R. Kumar, A. Kumar, S. Jain and D. Kaushik, Eur. J. Med. Chem., 2011, 46, 3543–3550.
45. M. Witschel, Bioorg. Med. Chem., 2009, 17, 4221–4229.

---

Footnote

† Electronic supplementary information (ESI) available: IR, ¹H NMR, MS, and elemental analysis data for the target compounds. The values of all descriptors of rosin-based diamides with insecticidal activity against P. xylostella are also available. See DOI: 10.1039/c4ra10125a

---