Gene Replacement for the Generation of Designed Novel Avermectin Derivatives with Enhanced Acaricidal and Nematicidal Activities

Abstract

Avermectin (AVM) and ivermectin (IVM) are potent pesticides and acaricides which have been widely used during the past 30 years. As insect resistance to AVM and IVM is greatly increasing, alternatives are urgently needed. Here, we report two novel AVM derivatives, tenvermectin A (TVM A) and TVM B, which are considered a potential new generation of agricultural and veterinary drugs. The molecules of the TVMs were designed based on structure and pharmacological property comparisons among AVM, IVM, and milbemycin (MBM). To produce TVMs, a genetically engineered strain, MHJ1011, was constructed from Streptomyces avermitilis G8-17, an AVM industrial strain. In MHJ1011, the native aveA1 gene was seamlessly replaced with milA1 from Streptomyces hygroscopicus. The total titer of the two TVMs produced by MHJ1011 reached 3,400 mg/liter. Insecticidal tests proved that TVM had enhanced activities against Tetranychus cinnabarinus and Bursaphelenchus xylophilus, as desired. This study provides a typical example of exploration for novel active compounds through a new method of polyketide synthase (PKS) reassembly for gene replacement. The results of the insecticidal tests may be of use in elucidating the structure-activity relationship of AVMs and MBMs.

INTRODUCTION

Avermectins (AVMs) are a series of 16-membered macrocyclic lactone derivatives with potent anthelmintic and insecticidal properties (1, 2). Ivermectin (IVM), a hydrogenated product of AVM B1, has been one of the best-selling antiparasitics since 1981 (3). However, as insect resistance to AVM (4) and IVM (5–7) is on the rise, alternatives or new products with enhanced potency and expanded spectra of activity are urgently needed.

Milbemycins (MBMs) are another group of 16-membered macrolides that share similar structures with AVMs. The insecticidal activity of milbemectin (a mixture of MBMs A3 and A4) against some parasites is higher than those of AVM and IVM, while the toxicity is significantly lower (8).

The structural differences among AVM, IVM, and MBM are in C-25, C-22-23, and C-13 (Fig. 1) (9). IVM differs from AVM in C-22-23 (the former has a saturated bond, whereas the latter has a double bond in that position), but its toxicity is lower than that of AVM (10), implying that the single bond in C-22-23 should be the better structure in terms of safety. The major structural difference between IVM and MBM is a bisoleandrosyloxy substituent attached at C-13 of IVM, whereas that position is unsubstituted in MBM. Also, there are different alkyl substituents at C-25: in IVM, the substituent can be isopropyl or secondary butyl, while in MBM, it can be methyl or ethyl. However, MBM is more potent against some parasites and is safer than IVM, suggesting that one or both of the moieties in C-25 and C-13 of MBM is (are) the better structure(s) in these ways than that (those) of IVM. Another conclusion is that the two oleandroses in C-13 are important to the antiparasitic activity of IVM, since their removal will reduce the activity significantly, about 30-fold (10). Based on the current understanding, it was postulated that combining all these more desirable structures, i.e., the C-25 substituent of MBM, the C-22-23 substituent of MBM and IVM, and the C-13 substituent of AVM and IVM, could generate novel antibiotics (in this case, 25-methyl-22,23-dihydroavermectin and 25-ethyl-22,23-dihydroavermectin) with better pharmacological properties than those of AVM, IVM, and MBM.

FIG 1.

Structures of avermectins (AVMs), ivermectins (IVMs), milbemycins (MBMs), nemadectin, and meilingmycin.

AVM is produced by Streptomyces avermitilis (11), and MBM can be produced by Streptomyces hygroscopicus (12). Both AVM and MBM are biosynthesized by polyketide synthases (PKSs). The C-22-25 structures of AVM and MBM are determined by aveA1 and milA1, respectively (Fig. 2), each housing a loading domain (LD), module 1 (M1), and module 2 (M2). The differences between the C-22-25 structures of AVM and MBM depend on the composition of domains in the modules: (i) the LDs determine the C-25 substituents of AVM and MBM, and (ii) M2 of the MBM PKS contains an additional enoyl reductase (ER) domain to form a fully saturated chain at C-22-23 (9). This suggests that if the LD and M2 of AVM PKS in S. avermitilis are replaced by those of MBM PKS or if the entire aveA1 gene is replaced by milA1, the desired 25-methyl-22,23-dihydroavermectin and 25-ethyl-22,23-dihydroavermectin will be produced. In this study, we replaced aveA1 with milA1 to produce the designed modified AVMs.

FIG 2.

Comparison of the hypothetical reaction mechanisms of LD, M1, and M2 on AVM PKS and MBM PKS. PKS, polyketide synthase; LD, loading domain; M1, module 1; M2, module 2; AT, acyl transferase; ACP, acyl carrier protein; KS, ketosynthase; KR, ketoreductase; DH, dehydratase.

MATERIALS AND METHODS

Plasmids, strains, and culture conditions.

Conjugative vector pUAmT14 was constructed previously (13). Plasmid pCOS5-87 was a cosmid containing the entire milA1 gene and flanking sequences. Escherichia coli DH5α was used for cloning purpose, and E. coli ET12567 (pUZ8002) (14) was used for conjugation. Both of the E. coli strains were cultured in LB medium (15). S. avermitilis G8-17 (13), an AVM industrial strain derived from MA-4680 (ATCC 31267), and S. hygroscopicus HS023 (CGMCC 7677), an MBM industrial strain, were cultured in YMS agar (per liter, 4 g of yeast extract, 4 g of soluble starch, 10 g of malt extract, 18 g of agar) or in tryptic soy broth (TSB; Becton, Dickinson and Company, USA). Antibiotics were added at the following concentrations: kanamycin, 50 μg/ml; carbenicillin, 100 μg/ml; chloramphenicol, 25 μg/ml; nalidixic acid, 20 μg/ml; and apramycin, 50 μg/ml.

Molecular biology techniques.

Streptomyces genome DNA was isolated using standard protocols (16). PCR amplification with PrimeSTAR HS DNA polymerase (TaKaRa, Japan) was carried out according to the manufacturer's manual. Plasmids from E. coli cells were extracted by alkaline lysis (15). Purification of DNA fragments from agarose gels and solutions was performed using the TaKaRa MiniBEST agarose gel DNA extraction kit and TaKaRa MiniBEST DNA fragment purification kit, respectively. DNA fragments were phosphorylated and blunted by use of the blunting kination ligation kit (TaKaRa) when needed. E. coli cells were transformed by Inoue's method (15), and Streptomyces strains were transformed by conjugation (14).

Construction of genetically engineered strain S. avermitilis MHJ1011.

The gene replacement plasmid pUAmT-AMA was constructed in three steps, as follows (Fig. 3).

FIG 3.

Schematic overview of construction of gene replacement plasmid pUAmT-AMA. (A) Fragments used and plasmids on which they were carried. E, EcoRI; S, SphI; N, NsiI; H, HindIII. (B and C) Seamless splicing strategy of AAU-MA5 and MA3-AAD, respectively. Restriction sites are underlined, and lowercase base pairs show introduced sequences that will be removed after digestion and blunting. Black bars and capital letters indicate sequences from S. avermitilis; gray bars and capital letters indicate sequences from S. hygroscopicus. AAU, 3,098-bp fragment upstream from aveA1 ORF; MA5, 2,899-bp milA1 5′ terminal fragment; MA3, 3,212-bp milA1 3′-terminal fragment; AAD, 3,266-bp fragment downstream from the aveA1 open reading frame.

(i) Fragment AAU, the 3,245 bp upstream from the aveA1 open reading frame (ORF), was amplified from S. avermitilis G8-17 with primers AA1UF31 (5′-aacgaattcTGCGAGTCGCGACACTGGC-3′; EcoRI) (lowercase letters show introduced sequences, and restriction sites are underlined; similar annotation is used hereinafter) and AA1UR32 (5′-aacgcatgCTGAGCTGTGTCCTCACCGCTAGG-3′; SphI), and the 2,899-bp milA1 5′-terminal fragment (fragment MA5) was generated by PCR from S. hygroscopicus with primers MA15F29 (5′-TTGCCCAAAGCCCAGAACGAGTTCG-3′) and MA15R30 (5′-CCGACGGCTTGTCCACGTGC-3′). These two fragments were subcloned into pUAmT14 in turn: fragment AAU was inserted as an EcoRI-SphI fragment, and MA5 was then cloned into the 5′ end of fragment AAU (using the blunted SphI site introduced into the 3′ end of fragment AAU). The plasmid generated, pMHJ05, was verified by restriction digestion to confirm the insertion direction of MA5.

(ii) The 3,207-bp milA1 3′-terminal fragment (fragment MA3) was amplified from S. hygroscopicus HS023 with primers MA13F1 (5′-CAGACCATGTGGCTCGTGGAGC-3′) and MA13R2 (5′-aacatgcaTCAGGAGAGGCCGAGGTCGTTC-3′; NsiI), and the 3,266-bp fragment downstream from the aveA1 ORF (fragment AAD) was generated from S. avermitilis G8-17 by PCR with oligonucleotides AA1DF9 (5′-ACCGGACGCCTGCCACTCCGCCCGTATC-3′) and AA1DR10 (5′-GCCTGTGTCCGCTCCGACGATCGCC-3′). These two fragments were subcloned into pBlueScript SK(+) in the following order: MA3 was phosphorylated and ligated with EcoRV-digested pBlueScript SK(+), and AAD was then inserted into the blunt-ended NsiI site introduced into the 3′ end of MA3. The structure yielded was verified by restriction digestion to confirm the insertion direction and named pMHJ06.

(iii) The seamlessly spliced MA3-AAD was cut from pMHJ06 as an SphI-HindIII fragment and inserted in the same sites of pMHJ05 to give rise to plasmid pMHJ07, and the intact milA3 ORF was recovered by insertion of the 9,410-bp SphI fragment obtained from cosmid pCOS5-87 into the SphI site of pMHJ07, yielding the gene replacement plasmid pUAmT-AMA.

pUAmT-AMA was transformed into S. avermitilis G8-17 by conjugation. From the culture of a randomly selected transformant, an apramycin-sensitive mutant, MHJ1011, was screened, from which a 1,459-bp target fragment (an internal fragment of milA1) could be amplified with primers 027M1EF (5′-TGCATCTGACCGCCTACGCCCAACCG-3′) and 028M1ER (5′-GCGTCGGCAAACCGGTCGTAGACCCC-3′) but no such fragment (an internal fragment of aveA1) could be amplified with primers 025A1EF (5′-GGGAGGAGTTGCTGGAGCTGCTGGGG-3′) and 026A1ER (5′-GTGGCCAACTCGGGTGACATGGGTCG-3′), meaning that the aveA1 had been replaced by milA1.

Fermentation.

The shake flask fermentation conditions were described previously (13). Large-scale production for target metabolite isolation was carried out in a 50-liter fermenter (Shanghai Guoqiang Bioengineering Equipment Co. Ltd., China), and the volumes of seed and fermentation media were increased to 1.5 liters and 30 liters, respectively.

Isolation and purification of the target metabolites.

The final 30 liters of fermentation broth was filtered, and the resulting cake was extracted twice for about 24 h with 10 liters of ethanol. The ethanol extract was then evaporated to 1 liter under reduced pressure at 45°C and subsequently extracted three times using an equal volume of ethyl acetate. The combined ethyl acetate phase was concentrated under reduced pressure to yield 260 g of oily substances. Five grams of the residual oily substance was passed through a silica gel column (200 to 300 mesh; Qingdao Marine Chemical Factory, Qingdao, China) and successively eluted with a stepwise gradient of petroleum ether-acetone (90:10 to 60:40, vol/vol) to afford four fractions (I to IV) based on the thin-layer chromatography (TLC) profiles. Fraction II, eluted with petroleum ether-acetone at 70:30 (vol/vol), was further separated by semipreparative high-performance liquid chromatography (Agilent 1100 with a Zorbax SB C₁₈ column, 5-μm particle size, 250 mm, 9.4-mm inner diameter, 1.5 ml/min, 244-nm wavelength; Agilent, Palo Alto, CA, USA), eluting with CH₃OH-CH₃CN-H₂O (46:46:8, vol/vol), to give compound 1 (retention time [t_R], 17.1 min; 0.79 g) and compound 2 (t_R, 21.5 min; 0.25 g).

Analysis and identification of the target metabolites.

High-performance liquid chromatography (HPLC) analysis was performed as described previously (13). The electrospray ionization mass spectrometry (ESI-MS) and high-resolution electrospray ionization mass spectrometry (HRESI-MS) spectra were taken on a quadrupole time of flight (Q-TOF) micro-LC–tandem mass spectrometry (MS-MS) instrument (Waters Co., Milford, MA, USA). ESI was applied and operated in positive ion electrospray mode. The capillary and cone voltages were set at 3 kV and 30 V. The desolvation temperature was set to 250°C, and the source temperature to 120°C. The cone gas was set to a flow rate of 10 liters/h, and the desolvation gas flow was maintained at 700 liters/h. TLC was performed on silica gel plates (HSGF₂₅₄;Yantai Chemical Industry Research Institute, Yantai, China), with solvent systems of petroleum ether-acetone (3:2). The developed TLC plates were observed under a UV lamp at 254 nm and then revealed with sulfuric acid-ethanol (5:95, vol/vol). ¹H and ¹³C nuclear magnetic resonance (NMR) spectra were measured with a Bruker DRX-400 (400 MHz for ¹H and 100 MHz for ¹³C) spectrometer (Bruker, Rheinstetten, Germany).

Acaricidal activity test.

All bioassays were performed on representative test organisms reared in the laboratory. The larvicidal activities of test compounds against Tetranychus cinnabarinus were tested according to a previously reported procedure (17).

Each test sample was prepared in acetone at a concentration of 1,000 mg/liter and diluted to the required concentrations of 0.01, 0.005, 0.0025, 0.001, and 0.0005 mg/liter with distilled water containing alkylphenol ethoxylates (∼1/1,000, vol/vol). The primary leaves of Vicia faba L. species were infected with T. cinnabarinus. At 2 h after infection, 10 fourth-instar mite larvae were dipped in the diluted solutions of related chemicals for 5 s before the superfluous liquid was removed, and the larvae were kept in a conditioned room. Three replicates were made for each concentration and a blank control. The mortality was evaluated 24 h after treatment by examining the adult mites under a binocular microscope to determine the living and dead individuals. The 50% lethal concentrations (LC₅₀s) of tested compounds were calculated using the probit method.

Nematicidal activity test.

About 10 μl of each sample was added to an aqueous suspension (90 μl) containing approximately 2,500 living nematodes (third-instar and fourth-instar larvae of Bursaphelenchus xylophilus) per milliliter, and the suspension kept at 25°C for 24 h. The blank control group was prepared in the same way but lacked the tested compound. Three replicates were performed in each trial. Finally, the activities of the tested compounds were monitored under a microscope by recording the death rates of nematodes. Nematodes that did not move when prodded with a needle were considered to be dead. The LC₅₀s of tested compounds were calculated using the probit method.

RESULTS

Genetically engineered strain S. avermitilis MHJ1011 produced two novel compounds.

A fermentation sample of MHJ1011 was analyzed by HPLC. As shown by the data in Fig. 4A and B, MHJ1011 produced two novel compounds (compounds 1 and 2, the retention times of which were 8.773 min and 11.066 min, respectively) with a titer of 3,400 mg/liter in total. The sample was then analyzed by ESI-MS, and the data (Fig. 4C and D) showed that the molecular weights of compounds 1 and 2 were 832 and 846, respectively, matching those of the desired 25-methyl-22,23-dihydroavermectin and 25-ethyl-22,23-dihydroavermectin.

FIG 4.

HPLC and MS analytical data from fermentation products. (A) HPLC analytical data from fermentation sample of genetically engineered strain MHJ1011. (B) HPLC analytical data from fermentation sample of parent strain G8-17. (C) ESI-MS spectrum of compound 1. (D) ESI-MS spectrum of compound 2. Boldface 1 and 2, compounds 1 and 2; B1a, avermectin B1a; B2a, avermectin B2a.

The two novel compounds were structurally identified.

The molecular formula of compound 1 was established to be C₄₅H₆₈O₁₄ as deduced from the HRESI-MS and ¹³C NMR data (Table 1). The ¹H NMR (400 MHz, CDCl₃) spectrum of compound 1 displayed five doublet aliphatic methyls at δ 0.84, 1.16, 1.17, 1.26, and 1.27, two olefinic methyl signals at δ 1.50 and 1.87, and two methoxy groups at δ 3.42 and 3.46. The ¹³C NMR and HMQC spectra revealed 45 carbon resonances, including an ester carbonyl carbon at δ 173.9 (s), a ketal carbon at δ 97.6 (s), two acetal carbons at δ 94.5 (d) and 98.5 (d), 5 sp², methines, 3 sp² quaternary carbons, 5 secondary methyls, 2 vinylic methyls, 8 methylenes (one oxygenated), 15 aliphatic methines (12 oxygenated), one oxygenated quaternary carbon, and 2 methoxy groups. Comparison of the ¹H NMR data of compound 1 with those of IVM (10) suggested that compound 1 was structurally related to IVM. The difference between compound 1 and IVM was in C-25, where a methyl group was substituted at C-25 in compound 1. The ¹H–¹H correlation spectroscopy (COSY) (Fig. 5) correlation of H-25 (δ_H 3.32)/H-31 (δ_H 1.16) and the heteronuclear multiple bond correlations (HMBCs) (Fig. 5) from δ_H 1.16 (H₃-31) and 0.84 (H₃-30) to δ_C 71.4 (C-25) confirmed the structural assignment of compound 1. The relative configuration of compound 1 was assigned by analogy with AVM and IVM (18). Furthermore, compound 1 (80 mg) was dissolved in 1% sulfuric acid–methanol (2 ml) and stirred overnight in a nitrogen atmosphere. The solution was diluted with dichloromethane and rinsed with water, then 10% aqueous sodium bicarbonate solution, and then again with water. The organic layer was separated and dried over anhydrous sodium sulfate. The residue obtained after evaporating the solvent was purified by silica column chromatography, eluting with 1% to 5% methanol in dichloromethane, and compounds 1a (7 mg) and 1b (18 mg) were obtained.

TABLE 1.

¹H and ¹³C NMR data of compounds 1 and 2

Position	Value in indicated compound
	δH (J in Hz)		δC (ppm)
	1	2	1	2
1			173.9 sa	173.8 s
2	3.27 br s	3.29 br s	45.7 d	45.7 d
3	5.39 br s	5.41 br s	118.0 d	118.0 d
4			137.9 s	137.9 s
5	4.29 br t (6.4)	4.30 br t (6.5)	67.7 d	67.7 d
6	3.96 d (6.4)	3.97 d (6.5)	79.1 d	79.1 d
7			80.3 s	80.3 s
8			139.7 s	139.7 s
9	5.84 br d (10.8)	5.86 br d (10.2)	120.3 d	120.4 d
10	5.73 m	5.74 m	124.7 d	124.7 d
11	5.73 m	5.74 m	138.0 d	138.0 d
12	2.52 m	2.53 m	39.8 d	39.8 d
13	3.95 br s	3.95 br s	81.3 d	81.6 d
14			134.9 s	135.0 s
15	5.02 br d (9.8)	5.00 br d (10.2)	118.2 d	118.3 d
16	2.27 m	2.27 m	34.2 t	34.2 t
	2.33 m	2.34 m
17	3.64 m	3.66 m	67.3 d	67.3 d
18	0.84 m	0.87 m	37.0 t	37.0 t
	1.79 m	1.78 m
19	5.43 m	5.43 m	68.4 d	68.4 d
20	1.37 t (11.8)	1.39 t (12.0)	41.0 t	41.1 t
	1.98 dd (11.8, 4.3)	2.00 dd (12.0, 4.3)
21			97.6 s	97.4 s
22	1.53 m	1.54 m	35.7 t	35.6 t
	1.68 m	1.68 m
23	1.53 m	1.54 m	27.7 t	27.8 t
24	1.26 m	1.34 m	36.5 d	34.2 d
25	3.32 m	3.14 m	71.4 d	75.9 d
26	1.87 br s	1.88 br s	19.9 q	19.9 q
27	4.65 br d (14.8)	4.66 br d (14.4)	68.5 t	68.5 t
	4.70 br d (14.8)	4.71 br d (14.4)
28	1.17 d (6.0)	1.17 d (6.9)	20.3 q	20.2 q
29	1.50 br s	1.51 br s	15.2 q	15.2 q
30	0.84 d (6.6)	0.85 d (6.7)	17.9 q	17.7 q
31	1.16 d (8.0)	1.36 m	19.4 q	25.6 t
		1.68 m
32		1.00 t (7.3)		10.0 q
1′	4.81 d (3.3)	4.80 d (3.1)	94.5 d	94.7 d
2′	2.29 m	2.27 m	34.7 t	34.6 t
	1.60 m	1.60 m
3′	3.61 m	3.63 m	79.4 d	79.4 d
4′	3.24 t (9.2)	3.25 t (8.9)	80.5 d	80.4 d
5′	3.81 m	3.82 m	67.2 d	67.2 d
6′	1.26 d (6.0)	1.26 d (6.0)	18.4 q	18.4 q
1″	5.39 br s	5.41 br s	98.5 d	98.5 d
2″	1.53 m	1.54 m	34.2 t	34.2 t
	2.33 m	2.34 m
3″	3.48 m	3.49 m	78.2 d	78.2 d
4″	3.16 t (9.1)	3.17 t (9.1)	76.1 d	76.1 d
5″	3.77 m	3.77 m	68.1 d	68.1 d
6″	1.27 d (7.2)	1.28 d (6.4)	17.7 q	17.7 q
3′-OCH3	3.46 s	3.44 s	56.7 q	56.6 q
3″-OCH3	3.42 s	3.43 s	56.4 q	56.4 q

^aAccording to distortionless enhancement by polarization transfer (DEPT) sequence.

FIG 5.

Structures of compounds 1, 2, 1a, and 1b, and key HMBCs of compounds 1 and 2.

The ¹H NMR data of compound 1b (see Fig. S13 in the supplemental material) were consistent with those of 13-α-hydroxy-MBM A4 (19; D. B. Frei, A. C. O'Sullivan, and D. P. Maienfisch, 12 September 1985, European patent application EP0180539), while the ¹H NMR data of compound 1a (see Fig. S19) were identical to those of methyl α-l-oleandroside obtained from AVMs (20). Thus, the stereostructure of compound 1 was elucidated.

The molecular formula of compound 2 was determined to be C₄₆H₇₀O₁₄ (HRESI-MS m/z 869.4683 [M+Na]⁺). The extra 14 mass units compared to the data for compound 1 suggested the presence of a methylene group. Detailed analysis of its NMR data (Table 1) revealed structural similarity to compound 1, except that the C-25 methyl in compound 1 was replaced by an ethyl group in compound 2, which was supported by ¹H–¹H COSY (Fig. 5) correlations of H-25 (δ_H 3.14)/H-31 (δ_H 1.36, 1.68)/H-32 (δ_H 1.00) and HMBCs (Fig. 5) from δ_H 1.00 to δ_C 75.9 (C-25) and 25.6 (C-31). Therefore, the gross structure of compound 2 was established, with its configuration similarly deduced by analogy to compound 1.

The results showed that the structures of compounds 1 and 2 matched those of the expected 25-methyl-22,23-dihydroavermectin and 25-ethyl-22,23-dihydroavermectin (Fig. 5), which were subsequently named tenvermectin A (TVM A) and tenvermectin B (TVM B), respectively.

TVM A (compound 1), C₄₅H₆₈O₁₄, white amorphous powder; melting point 153 to 155°C; [α]²⁰_D −26.7 (c 0.5, ethyl alcohol [EtOH]). UV (EtOH) λ_max nm (log ε): 244 (4.55). Infrared (IR) (KBr), ν_max cm⁻¹: 3,464, 2,931, 1,718, 1,451, 1,381, 1,341, 1,305, 1,198, 1,120, 1,051, 987. ¹H NMR (400 MHz, CDCl₃) and ¹³C NMR (100 MHz, CDCl₃) data are shown in Table 1, and NMR spectra are shown in Fig. S1 to S6 in the supplemental material. ESI-MS m/z 855 [M+Na]⁺. HRESI-MS m/z 855.4551 [M+Na]⁺. Calculated for C₄₅H₆₈O₁₄Na, 855.4501.

TVM B (compound 2), C₄₆H₇₀O₁₄, white amorphous powder; melting point 153 to 155°C; [α]²⁰_D −18.0 (c 0.5, EtOH). UV (EtOH) λ_max nm (log ε): 244 (4.60). IR (KBr), ν_max cm⁻¹: 3,463, 2,931, 1,717, 1,453, 1,380, 1,341, 1,303, 1,197, 1,106, 1,051, 986. ¹H NMR (400 MHz, CDCl₃) and ¹³C NMR (100 MHz, CDCl₃) data are shown in Table 1, and NMR spectra are shown in Fig. S7 to S12 in the supplemental material. ESI-MS m/z 869 [M+Na]⁺. HRESI-MS m/z 869.4683 [M+Na]⁺. Calculated for C₄₆H₇₀O₁₄Na, 869.4658.

Compound 1a, ¹H NMR (400 MHz, CDCl₃), δ 4.76 (1H, d, J = 3.2 Hz, H-1), δ 3.64 (1H, m, H-5), δ 3.48 (1H, m, H-3), δ 3.38 (3H, s, 3-OCH₃), δ 3.32 (3H, s, 1-OCH₃), δ 3.14 (1H, br t, J = 9.2 Hz, H-4), δ 2.26 (1H, dd, J = 12.8, 4.2 Hz, H-2), δ 1.29 (3H, d, J = 6.2 Hz, H-6). NMR spectra are provided as supporting information (see Fig. S19 to S20 in the supplemental material).

Compound 1b, ¹H NMR (400 MHz, CDCl₃), δ 5.82 (1H, m, H-9), δ 5.76 (1H, m, H-10), δ 5.71(1H, m, H-11), δ 5.40 (1H, m, H-19), δ 5.40 (1H, br s, H-3), δ 5.38 (1H, m, H-15), δ 4.72 (1H, d, J = 14.4 Hz, H-27a), δ 4.67 (1H, d, J = 14.4 Hz, H-27b), δ 4.30 (1H, br s, H-5), δ 4.02 (1H, br s H-13), δ 3.98 (1H, d, J = 6.2 Hz, H-6), δ 3.65 (1H, m, H-17), δ 3.30 (1H, m, H-25), δ 3.28 (1H, br s, H-2), δ 2.54 (1H, m, H-12), δ 2.32 (2H, m, H-16), δ 2.01 (1H, dd, J = 12.1, 3.8 Hz, H-20a), δ 1.89 (3H, br s, H-26), δ 1.79 (1H, m, H-18a), δ 1.69 (1H, m, H-22a), δ 1.56 (1H, m, H-22b), δ 1.54 (2H, m, H-23), δ 1.54 (3H, br s, H-29), δ 1.37 (1H, t, J = 11.4 Hz, H-20b), δ 1.26 (1H, m, H-24), δ 1.19 (3H, d, J = 7.3 Hz, H-28), δ 1.17 (3H, d, J = 6.5 Hz, H-31), δ 0.85 (3H, d, J = 6.5 Hz, H-30), δ 0.85 (1H, m, H-18b). ¹³C NMR (100 MHz, CDCl₃) δ 173.6 (s, C-1), δ 139.9 (s C-8), δ 138.6 (s, C-14), δ 137.8 (s, C-4), δ 137.0 (d, C-11), δ 124.8 (d, C-10), δ 120.3 (d, C-9), δ 118.1 (d, C-3), δ 117.3 (d, C-15), δ 97.6 (s, C-21), δ 80.2 (s, C-7), δ 79.2 (d, C-6), δ 77.6 (d, C-13), δ 71.4 (d, C-25), δ 68.5 (d, C-19), δ 68.5 (t, C-27), δ 67.7 (d, C-5), δ 67.4 (d, C-17), δ 45.7 (d, C-2), δ 41.1 (t, C-20), δ 40.2 (d, C-12), δ 36.7 (t, C-18), δ 36.5 (d, C-24), δ 35.6 (t, C-22), δ 34.2 (t, C-16), δ 27.7 (t, C-23), δ 19.9 (q, C-26), δ 19.3 (q, C-31), δ 19.1 (q, C-28), δ 17.9 (q, C-30), δ 14.6 (q, C-29). NMR spectra are provided as supporting information (see Fig. S13 to S18 in the supplemental material).

TVM showed better pharmacological properties than AVM, IVM, and MBM against Tetranychus cinnabarinus.

The acaricidal activities of TVM (mixture of TVMs A and B at an A/B ratio of 3:1), AVM, IVM, and MBM (mixture of MBMs A3 and A4 at an A3/A4 ratio of 3:7) against T. cinnabarinus were tested. The mite mortality of the TVM treatment group was significantly different (P < 0.05) from those of groups treated by AVM, IVM, and MBM (Fig. 6A; also see Table S1 in the supplemental material). The results indicated that TVM possessed the highest activity against T. cinnabarinus among the compounds tested. According to the results, AVM products had a slow-acting effect in mites at the LC₅₀ dosage. However, we noticed that mites became moribund soon after contact with TVM and died over only a 1-day period, while with the other compounds, the time spans were 3 to 4 days. Such a property might make TVM more promising for new agricultural drug development.

FIG 6.

Median lethal concentrations (LC₅₀s) (mg/liter) of compounds against Tetranychus cinnabarinus (A) and Bursaphelenchus xylophilus (B). See also Tables S2 and S3 in the supplemental material. TVM, tenvermectin; AVM, avermectin; IVM, ivermectin; MBM, milbemycin. Bars represent means ± standard deviations, and those marked by the same letter are not significantly different at a P value of 0.05 (which is adjusted by Duncan's new multiple range method). n = 3 replicates.

TVM is more potent against Bursaphelenchus xylophilus than AVM, IVM and MBM.

Acetone solutions of TVM were prepared at concentrations of 2, 5, 10, 20, and 50 mg/liter. For the purpose of comparison, AVM, IVM, and MBM were diluted to the same concentration series. The results showed that the activity of TVM against B. xylophilus was at least 2-fold higher than those of AVM, IVM, and MBM (Fig. 6B; see also Table S2 in the supplemental material). At the lower drug concentrations, the mortality rate increased slowly and leveled off after 24 h. Specifically, TVM displayed the highest nematicidal activity and gave 100% mortality at concentrations greater than 50 mg/liter.

It should be noted that compounds like IVM are effective against Caenorhabditis elegans at concentrations 1,000-fold lower than the concentrations used here (21). However, as we did not have access to more susceptible species, we set B. xylophilus, one of the injurious pests of greatest concern in forestry, as the target for the nematicidal activity test.

B. xylophilus, commonly known as pine wilt nematode (PWN) and likened to “pine cancer,” is a species of nematode that infects pine trees and causes the diseased pines to wilt. Once a pine tree is infected with PWN, there is no effective control measure (22). TVM is more active against B. xylophilus than AVM, IVM, and MBM and might be a potential candidate for PWN control.

DISCUSSION

Relationship between structure and bioactivity.

From the comparisons of acaricidal activities and structures between compound pairs in which the two members differ from each other only in one position, two conclusions can be drawn, as follows: (i) the C-25 substituents of TVM (or MBM) are better structures than those of IVM (or AVM) with regard to activity against T. cinnabarinus, given the higher activity of TVM than of IVM, and (ii) the bisoleandrosyl moiety in C-13 is especially important for the activity against T. cinnabarinus, in contrast to the unsubstituted structure, since the activity of MBM is dramatically lower than that of TVM. The results of the nematicidal activity test against B. xylophilus lead to the same conclusions.

Previous research revealed that removal of the bisoleandrosyl moiety from C-13 of IVM, generating an IVM aglycone containing a 13-α-hydroxy group, reduces the antiparasitic activities in animal health noticeably but that further removal of the 13-hydroxy structural entity from the IVM aglycone could restore the potent anthelmintic activity (23). However, as for the activities against T. cinnabarinus and B. xylophilus, our study gives a divergent conclusion. In addition, no direct evidence has ever shown a relationship between activity and C-25 substituents. Our study provides new insights into the relationship between the structures and antiparasitic activities of AVMs and MBMs.

PKS reassembly gene replacement strategy for novel active compound exploration.

One of the main trends in producing novel rationally designed natural products is to manipulate the known biosynthetic machinery of PKS, namely, combinatorial biosynthesis (24). PKS reassembly can be manipulated on the levels of domain, module, and PKS gene. To date, there are a number of successful precedents for structural modification through manipulation of domains (25–28) and modules (29, 30), but structural modifications through PKS gene replacement are little reported. In fact, domain/module swaps are often only marginally successful or entirely unsuccessful, leading to only very small amounts of the desired compound or no product at all (26–28, 31, 32). A major limitation is the unpredictable loss of enzyme catalytic activity. The polyketide biosynthesis is brought about by complex machinery consisting of a tightly coupled network of core catalytic and structural domains (33). Changes of domains or/and modules will probably cause the PKS complex to fold incorrectly and lose a necessary activity or lead to a modified polyketide chain that is not recognized as a substrate by a subsequent PKS activity (34). As a PKS gene usually contains several modules, the replacement of such a gene would probably magnify the limitations of domain/module swapping. This might be the major obstacle for structural modification through PKS gene replacement. Furthermore, the risk of poor cooperation of the docking domains among the foreign multienzyme and successive native domains might be another negative effect.

To obtain 25-methyl-22,23-dihydroavermectin (TVM A) and 25-ethyl-22,23-dihydroavermectin (TVM B), the following strategy is also feasible and even more convenient for manipulating the shorter target DNA segments: (i) substitute the LD of MBM PKS for that of AVM PKS to change the C-25 moiety of AVM into that of MBM, and (ii) substitute the ER, dehydratase (DH), and ketoreductase (KR) domains of MBM M2 for the DH-KR domains of AVM M2 or perform a domain swap as described previously (27) to gain a single C-C bond in C-22-23 of AVM. However, besides the risk of failure to obtain the desired products in step i, the catalytic activity of AVM M1 would probably be reduced due to the variation of the substrate. In contrast, the MLB M1 would be more efficient, as the alternative substrate is more like the native substrate. Pari ratione, MBM M2 should be a better candidate than the hybrid M2s reassembled in step ii. In fact, the genetically engineered strain constructed by Gaisser et al. (27) produced not only IVM-like compounds in a low ratio but also all components of AVM predominately, implying that the foreign domains did not work so harmoniously with the natives. Therefore, in spite of the technical limitations, replacing LD, M1, and M2 simultaneously, namely, replacing aveA1 with milA1, is no doubt a better strategy than replacing LD and the domains of M2 separately.

In our previous studies, aveA1, together with its promoter, was knocked out in S. avermitilis G8-17, and milA1, together with its promoter, was introduced by shuttle or integrative vector, but no AVM was produced (data not shown). One possible reason is that the promoter of milA1 did not work well in S. avermitilis. Another reason might be that the downstream gene aveA2 is cotranscribed with aveA1 and was influenced by knockout of aveA1, as promoter analysis using the Neural Network Promoter Prediction (NNRP) software (version 2.2; M. G. Reese [http://www.fruitfly.org/seq_tools/promoter.html]) and PROSCAN (version 1.7; Center for Information Technology, National Institutes of Health [http://www-bimas.cit.nih.gov/molbio/proscan/]) showed that there was no promoter between aveA1 and aveA2. Nevertheless, seamless replacement of the aveA1 ORF with the milA1 ORF is the most reliable scheme to make sure that the alien and the native genes (milA1 and aveA2) can be expressed well. Under the aveA1 promoter, milA1 worked efficiently in S. avermitilis, and the desired novel AVMs, TVMs A and B, were produced at a total titer suitable for industrial production.

TVM is different from IVM in the substituent of C-25. Previous research had explored a platform to prepare AVM analogues with a wide range of novel C-25 substituents by feeding carboxylic acids or their biosynthetic precursors to a Streptomyces avermitilis mutant strain which lacked the ability to form the natural precursors of AVM PKS (35). However, the AVM LD cannot recruit acetyl coenzyme A (CoA) or propionyl-CoA as a start unit and, therefore, cannot form the C-25 structure of TVM. Although chemical methods can also modify the C-25 substituent (36, 37) of IVM to give rise to TVM, they are too laborious to apply in industrial production. In contrast, here we report an efficient method to produce TVM.

The results demonstrate that, besides domain and module swaps, gene replacement could also be efficient for PKS reassembly under specific conditions. In this study, the efficiency of gene replacement was mainly in the high similarity between AVM and MBM PKSs. In particular, the docking domains between MilA1 and the following multienzyme, AveA2, still matched well. Efforts were carried out to minimize the deleterious perturbations to the protein-protein interactions introduced by the manipulations of domain/module swapping (38). PKS gene replacement is a reliable strategy to guarantee that the catalytic domains remain in their native environments. Given the proper conditions, this is highly significant for novel active compound exploration with such a strategy. For example, both nemadectin and meilingmycin (Fig. 1) are produced by PKSs that are organized similarly to those of AVM and MBM. The exchange of corresponding PKS genes could produce different kinds of novel hybrids of AVM, MBM, nemadectin, and meilingmycin with better pharmacological properties.