Dissimilar Crystal Proteins Cry5Ca1 and Cry5Da1 Synergistically Act against Meloidogyne incognita and Delay Cry5Ba-Based Nematode Resistance

ABSTRACT

Cry proteins of Bacillus thuringiensis (Bt) have been successfully used as biopesticides and in transgenic crops throughout the world. However, resources against the most serious agricultural pathogens, plant root-knot nematodes, are limited. The genomes of several highly nematicidal virulent Bt strains from our laboratory have been sequenced, facilitating the identification of novel Cry proteins and other virulence factors. We identified two novel Cry proteins, Cry5Ca1 and Cry5Da1, that exhibit high toxicity against Meloidogyne incognita. Using the Caenorhabditis elegans model, the two Cry5 toxins were shown to negatively affect nematode life span, fertility, and survival. The 50% lethal concentrations (LC₅₀s) of Cry5Ca1 and Cry5Da1 were 57.22 μg/ml and 36.69 μg/ml, respectively. Moreover, a synergistic effect (synergism factor, 1.61 to 2.04) was observed for nematicidal toxicity of Cry5Ca1 and Cry5Da1, which is accordant with the phylogenetic results suggesting that domain II of the two novel Cry5 toxins evolved into two independent clades. Through comparison of the depressed degree of toxicity in the β-methylgalactoside detoxification test, we found that the novel toxin Cry5D possesses a different galactose-binding epitope; meanwhile, the finding that Cry5D does not share a motif (GXXXE) in the corresponding loop of domain II with Cry5B could explain the different galactose binding performance. Additionally, low-level cross-resistance of C. elegans bre mutant strains was evident between Cry5B and Cry5D. These results suggest that Cry5D can be used as an alternative to delay the potential resistance of nematodes to Cry5B.

IMPORTANCE Although proper gene resources for Bt crops against the most serious agricultural pathogens, plant root-knot nematodes, are limited, we have identified two novel nematicidal toxins, Cry5Ca1 and Cry5Da1, against M. incognita, which have supplied more gene candidates for Bt crops designed against nematodes. Moreover, the association of the dissimilarity between Cry5Da1 and Cry5Ba1 and their low cross-resistance can be attributed not only to a low sequence similarity of domain II but also to the structural difference of the key motif and receptor-binding epitope in the loops. This association facilitates the selection of a proper candidate for the prospective design of pyramided Bt crops that can delay potential resistance.

INTRODUCTION

The agricultural economic damage caused by more than 4,100 species of plant-parasitic nematodes (PPNs) has been estimated to cost US$80 billion per year (1, 2), and a large proportion of this cost is due to the serious damage caused by the plant-parasitic nematode Meloidogyne incognita (3). Although many nematophagous fungi have been identified, available bacterial resources other than Pasteuria commercial products are limited. The most successful commercially available biopesticide for agricultural pest biocontrol, Bacillus thuringiensis (Bt), is a ubiquitous Gram-positive bacterium that produces insecticidal and nematicidal parasporal crystals during the sporulation phase (4, 5). Combined with other assistant virulence factors, crystal proteins contribute to its pathogenesis during infection of the host insect or worm (6, 7). Although the number of identified nematicidal Bt strains is increasing (8) (relative to the amount of insecticidal Bt), additional nematode control resources are needed.

In addition to the wide use of bioinsecticides, the use of transgenic crops with insecticidal crystal (cry) genes has contributed to the excellent control of agricultural pests, along with improvements in crop yield (5) and tremendous benefits for the environment and farmers. However, the recently observed resistance of some Lepidoptera and beetle pests has reduced these benefits (9) and is becoming the most serious threat to the continued efficacy of Bt crops. The “pyramided” strategy has been widely used to significantly delay resistance (10, 11). For the sustainable use of nematode-resistant Bt crops, the dissimilar two-toxin gene pyramided transgenic strategy should be proactively applied for nematode control. To optimize the rules that select proper Bt toxin genes in the pyramided strategy, it is proposed that the similarity of domain II sequences between toxins is associated with cross-resistance and antagonism between the toxins used in pyramids (12). However, that theory is based on the field application for insecticidal toxins, yet little is known about selective criteria of proper nematicidal toxins applied in prospective design pyramided Bt crops against plant-parasitic nematodes.

Over the past few decades, a large number of cry genes have been identified and characterized, and more than 700 Cry toxin sequences have been classified into 74 distinct groups (http://www.btnomenclature.info/) according to a previously proposed nomenclature system (13). Although many Cry toxins are known to target various insects, only a few nematicidal crystal protein genes, particularly those targeting plant-parasitic nematodes (PPNs), have been found and utilized, of which only cry5Ba and cry6Aa have been applied into transgenic tomatoes (14–16). These nematicidal Cry toxins cluster together in phylogenetic subfamilies (17); the Cry5, Cry13A, Cry14A, and Cry21 proteins share the same evolutionary ancestry and are thus clustered into the Cry5 subfamily, which includes the nontoxic proteins Cry5A and Cry21G (6). Out of these nematicidal toxins, only Cry6Aa and Cry55Aa differ from the three-domain (3D) toxin Cry5 subfamily based on phylogenetic analysis and bioassays (18). Given these results, in nematode toxic gene pools, the different Bt toxin genes applied to crops are rare. To overcome the scarcity of nematicidal proteins used as biopesticides and transgenic candidates, the identification of more novel nematicidal crystal proteins is needed. In addition, various dissimilar receptor-binding toxins need to be identified to delay evolving resistance among nematodes in the future.

Due to the rapid development of sequencing technology and decreasing costs, mining the novel crystal protein genes via genome sequencing is becoming more and more efficient. Our previous studies have sequenced the genome of strain YBt-1518, from which three nematicidal Cry proteins (19), as well as Bmp1, chitinase (20), and other assistant virulence factors (21), were identified and appraised. Additionally, within the genome of nematicidal B. thuringiensis DB27, several Cry protein genes against nematodes were also identified (6, 22).

In this study, we identified two novel nematicidal crystal proteins, Cry5Ca1 and Cry5Da1, by analyzing the genome of nematicidal Bt strain Sbt003 and the toxicity against M. incognita, both of which possess synergistic effects on toxicity against Caenorhabditis elegans. Additionally, low-level cross-resistance of nematode bre mutant strains was proven to exist between Cry5Ba and Cry5Da. Cry5Da has different galactose-binding epitope-related β-methylgalactoside (β-MeGal) detoxification results and Cry5Da does not share a motif (GXXXE) in the corresponding loop of domain II with Cry5Ba, which may explain the presentation of low-level cross-resistance in Cry5Ba-related bre mutant nematodes.

RESULTS

Genomic insights into the pathogenesis of B. thuringiensis Sbt003 against nematodes.

Our previous work aimed to isolate numerous Bt strains targeting nematodes from the soil, including B. thuringiensis Sbt003, that exhibit high toxicity toward C. elegans (see Fig. S1 in the supplemental material). To investigate the virulence factors that account for this high toxicity against nematodes and to exploit additional toxin genes that may be used as biopesticides and/or in transgenic crops, we sequenced the B. thuringiensis Sbt003 genome (23). Eight potential crystal protein genes were identified via BtToxin_scanner (24), of which only four full-length toxin genes were obtained by domain structure analysis (23). Meanwhile, the genome of Sbt003 revealed a series of potential virulence factor-encoding genes based on the MvirDB database via MP3 bioinformatic prediction (see Table S1 in the supplemental material). Similar to other insecticidal Bt strains (25), both the crystal proteins and the series of potential virulence factors may contribute to and are indispensable for the observed high toxicity in Sbt003.

Cry5Ca1 and Cry5Da1 are novel crystal proteins.

Previous studies have demonstrated that in the Cry5 class (name beginning with Cry5), Cry5Ba toxins exhibit high toxicity against nematodes whereas Cry5A toxins fail to do so. Given that the Sbt003 genome contains four full-length Cry toxin genes, whether or not their coding sequences are separately expressed and possess nematicidal toxicity needs to be determined. The Cry5Da1 protein sequence shares 53% identity with that of nontoxic Cry5Ac, which is the highest identity among the existing toxins, whereas the Cry5Ca1 protein sequence shares 48% identity with that of Cry5Ba1. To clarify the relationships among the members of the Cry5 class, we constructed their three-domain phylogenetic trees along with all of the other members in the Cry5 subfamily. The trees showed two obvious monophyla of domain II (Fig. 1A, a and b) and domain III (Fig. 1A, c and d) in the Cry5 subfamily. Nematicidal toxins, Cry21Fa1, Cry21Ha1, Cry14Aa1, etc., were grouped into one clade (Fig. 1A, a and c) with Cry5Ba, and the others were grouped into another clade (Fig. 1A, b and d) with Cry5A toxins. More strikingly, Cry5Ca1 was clustered into one clade (Fig. 1A, a and c), whereas Cry5Da1 was clustered into another (Fig. 1A, b and d). The separately evolved clades in the crucial domains of nematicidal toxins implied that Cry5Da is dissimilar to Cry5Ca and Cry5Ba.

FIG 1.

Molecular characterization of crystal proteins Cry5Ca1 and Cry5Da1 from Bacillus thuringiensis Sbt003. (A) Phylogenetic relationships of the separate domain sequences within 3D (three domains) of Cry5B-related Cry toxins. The reliability of the tree was evaluated by the bootstrap method with 500 replications using the maximum-likelihood method. Two clades (Aa and Ab) are indicated in the tree of domain II of Cry5B-related Cry toxins; two clades (Ac and Ad) are also indicated in the tree of domain III. The solid triangle refers to proteins with demonstrated nematicidal toxicity, whereas those indicated by a hollow triangle show no toxicity to nematodes. (B) Sequence alignment and conserved domain and block analysis. The Cry5 subclass of toxins and the structurally characterized Cry5Ba1 were analyzed using Expresso in the T-COFFEE online services (49). The secondary structure of Cry5B is indicated on the top of the four sequences. Horizontal red boxes indicate the GXXXE and GG sequences in domain II. The red squiggles indicate the five conserved blocks of the Cry5 subclass of toxins. Residues in red have high identity; the residues indicated in yellow are similar. The η symbol refers to a 3₁₀-helix. α-Helices and β-strands are rendered as arrows, strict β-turns are shown as TT letters, and strict α-turns are indicated with TTT. acc represents the accessibility of each residue within the Cry5B structure (dark blue means the residue is accessible, cyan means it is intermediate, and white means it is buried).

Following a previously described Bt-pHT304 expression system, Cry5Ca and Cry5Da proteins were expressed in recombinant BMB1361 and BMB1362, presenting with a molecular mass of 130 kDa through SDS-PAGE (Fig. 2A). Both Cry5Ca and Cry5Da form typical bipyramidal crystals through microscopic observation (Fig. 2B and C), which differs from the oval and rice shaped crystal of strain Sbt003.

FIG 2.

Analysis and bioassay of crystal proteins Cry5Ca1 and Cry5Da1 from recombinant strains of B. thuringiensis. SDS-PAGE analysis of crystal proteins (A) and ordinary optic micrographs of crystals from recombinant strains of B. thuringiensis (B and C). Lane M, molecular mass standard; Lane 1, crystal protein Cry5Ca1 expressed in recombinant BMB1361; Lane 2, crystal protein Cry5Da1 expressed in recombinant BMB1362. (B and C) Ordinary optic micrographs (recoded under optical microscopy with an oil immersion objective lens) of parasporal crystals from BMB1361 and BMB1362 with a bipyramid shape. S, spore; C, crystal. (D) The toxicity of the original strain Sbt003 against the nematode C. elegans mainly contributes to the separately purified Cry5 toxins. The killing of nematodes by toxins and the wild-type strain Sbt003 is time dependent. (E) Morphological observation of nematodes exposed to the crystal-spore mixtures of different Cry5 toxins from BMB1361 and BMB1362. Intestinal shrinkage and the accumulation of a few spores occurred. Acrystalliferous strain BMB171 with an empty vector pHT304 was used as the control. The bar indicates 30 μm. (F) Bioassay of purified Cry5 toxins against plant parasite nematode M. incognita (second-stage juvenile [J2] animals; 40 to 60 in each well). The bioassays were repeated a minimum of three times, and the number of nematodes at each time was more than 30. The error bars shown in panels D and F indicate the standard errors of the mean (SEMs).

Cry5Ca1 and Cry5Da1 exhibit toxicity against the nematodes C. elegans and M. incognita.

Because the results indicate that the two novel Cry5 proteins contain five conserved blocks in the three-domain region, which is the same as that of Cry5B (Fig. 1B), here the toxicities of the purified Cry5Ca and Cry5Da proteins were tested against the nematodes C. elegans (Fig. 2D) and M. incognita (Fig. 2F). Lethal dose assays (Table 1) performed in C. elegans revealed that Cry5Da (50% lethal concentration [LC₅₀] = 36.69 μg/ml) and Cry5Ca (LC₅₀ = 57.22 μg/ml) exhibit the same level of toxicity with Cry5Ba (26). Moreover, the mortality of plant root-knot nematode M. incognita was also directly influenced by the dose of crystal proteins. At concentrations of up to 350 μg/ml of Cry5Ca or Cry5Da toxins, approximately 55% of nematodes were dead (Fig. 2F).

TABLE 1.

Lethality of Cry5Ca and Cry5Da toxins against the nematode C. elegans

Crystal protein	Regression equation	Regression coefficient	LC50 (95% confidence interval)a (μg/ml)
Cry5Ca	y = 1.68 + 1.12x	0.938	57.22 (5.19–83.85)
Cry5Da	y = 1.59 + 1.15x	0.974	36.69 (0.01–74.65)
Cry5Ba	y = 1.49 + 1.25x	0.948	23.21 (0.23–61.08)

^aThe LC₅₀s were determined by PROBIT analysis, and the 95% confidence intervals were added.

We also examined the susceptibility of C. elegans growth inhibition to Cry5 proteins. Synchronized larval stage 1 (L1) nematodes were fed different concentrations of the toxins. After 36 h, the nematodes fed no toxin (Escherichia coli OP50 only) grew into adults. In contrast, the growth of the nematodes intoxicated with Cry5Ca and Cry5Da was delayed at the first or second stage (Fig. 3A). As shown in the plot (Fig. 3B), the extent of repression of the worms' development was aggravated with increasing concentrations of the Cry5 toxins. Cry5Da also exhibited increased inhibition compared to that of Cry5Ca but decreased inhibition compared to that of Cry5Ba.

FIG 3.

Effect of Cry5Ca1 and Cry5Da1 toxins on the nematode C. elegans N2. (A) Microscopic observation of synchronized nematodes (N2) exposed to different Cry toxins at a concentration of 100 μg/ml compared with normal growing N2 fed E. coli OP50. The scale bar indicates 30 μm. (B) The worm size percentage relative to the control (HEPES) was affected by a serial concentration gradient of the Cry5 toxin relative to that observed in the absence of toxin. The data represent the averages and standard deviations from 50 measurements for each concentration of toxin. (C) The brood size of the L4 C. elegans nematode was affected by different concentrations of Cry5 toxins. The brood size was normalized to the average brood size of the control; n = 3; error bars denote the SEMs.

It has been previously mentioned that nematode fertility is also deemed a judgment criterion (27). Therefore, a brood size inhibition assay was performed, and it was found that Cry5 toxins affected offspring birth rates. Cry5Ba affected nematode fertility, Cry5Da also had an intermediate impact on nematode fertility, and Cry5Ca exerted the weakest effect (Fig. 3C). These results are consistent with the toxicity observed against nematodes in the quantitative bioassay.

Cry5Ca1 and Cry5Da1 toxins lead to intestinal damage in C. elegans.

It is evident that Cry5Ba toxins bind to the glycolipid receptor in intestinal cells, which leads to intestinal breakdown and, ultimately, death (28). To confirm whether Cry5Ca1 and Cry5Da1 can also destroy the midgut tissue, morphology observation of nematodes feeding on Cry5Ca and Cry5Da crystal spore mixtures was carried out (Fig. 2E). The intestinal lumen shrank when nematodes were fed the Cry5 toxins, and the formation of a bubble and separation of the intestine from the body wall were observed. Moreover, the pharynx was severely affected, reflecting feeding behaviors. In comparison, the nematodes fed BMB171 (crystal-negative mutant B. thuringiensis) lived normally with an intact intestine, and the lumen was glutted with spores. Conversely, only several spores were found in the shrunken lumen of pathogenic nematodes fed Cry5 toxins, which was believed to be due to feeding cessation (28). We therefore propose that Cry5Ca and Cry5Da toxins can disrupt the intestine and cause feeding inhibition of nematodes.

Cry5Ca1 and Cry5Da1 toxins exhibit synergistic nematicidal activity.

To verify whether Cry5Ca and Cry5Da toxin genes can be used as potential pyramided Bt crop candidates, we examined their synergistic activity against C. elegans using the mixture of Cry5 purified toxins. As reflected in Table 2, all of the observed LC₅₀ values of the Cry5Ca and Cry5Da mixture toxins were significantly lower than the expected LC₅₀ toxicity values. In addition, the combination of Cry5Ca and Cry5Da at a ratio of 2:1 resulted in the most significant synergistic effect in the bioassay (approximately 2-fold higher toxicity against C. elegans than expected). Thus, it is likely that Cry5Ca and Cry5Da have separate receptor-binding manners, which is concordant with the results of the domain II phylogenetic analysis.

TABLE 2.

Cry5Ca and Cry5Da toxins show synergistic nematicidal activity against C. elegans N2

Proportion of Cry5Ca/Cry5Da	LC50 (μg/ml)		Synergism factorc
	Observeda	Expectedb
1:0	57.22 (5.19–83.85)
0:1	36.69 (0.01–74.65)
1:1	10.15 (0.05–19.32)	14.90	1.47
1:2	6.82 (0.02–11.63)	13.89	2.04
2:1	9.97 (0.07–16.31)	16.08	1.61

^aLC₅₀s were analyzed by a PROBIT model, and the 95% confidence intervals are given in parentheses.

^bThe expected LC₅₀s were calculated following Tabashnik's equation based on the additive effect of the two components.

^cThe synergism factor was calculated by dividing the expected value by the observed LC₅₀.

Low-level cross-resistance of bre mutant strains of C. elegans to Cry5Da1 and Cry5Ba1.

It is known that Cry5Ba binds to the glycolipid receptor involved in repressive nematicidal activity to bre mutant strain worms (29, 30). The above analysis of the phylogenetic relationships between domain II of all Cry5 subfamily members showed that Cry5Ca is closely related to Cry5Ba in clade A and that Cry5Da in clade B is distant from Cry5Ba. Here, we assessed whether Cry5Ca and Cry5Da toxins have nematicidal activity to bre mutant strain nematodes (deficient in glycolipid receptors). C. elegans bre-5(ye17) mutant strains exhibited similar levels of resistance to Cry5Ba and Cry5Ca over a range of concentrations, but both toxins were highly toxic to wild-type nematode strain N2 (Fig. 4A). Meanwhile, Cry5Da was lethal to bre-5(ye17) mutant strain nematodes in a clear dose-dependent manner, with 50% lethal concentration (LC₅₀) values of 36.7 μg/ml and 93.5 μg/ml for strain N2 and bre-5(ye17) mutant strain nematodes, respectively. The LC₅₀s of the bre-5(ye17) mutant strains to Cry5Da showed no significant difference from those found for N2, but a 50-fold significant difference was observed compared with the LC₅₀s of those treated with Cry5Ba (Fig. 4B). Furthermore, growth inhibition of all five bre mutant strain worms was observed under treatment with toxins, and all bre mutant strains still exhibited susceptibility to Cry5Da, similar to wild-type nematode N2 (Fig. 4C), but did have resistance to Cry5Ca and Cry5Ba. These results indicate a high level of cross-resistance between Cry5Ca and Cry5Ba for bre mutant strains but very low cross-resistance between Cry5Ba and Cry5Da. This difference implies that Cry5Da may be used as a supplement to Cry5Ba to enhance nematicidal activity or as an alternative to Cry5B to overcome the potential resistance of nematodes.

FIG 4.

Resistance of C. elegans bre mutant strains toward Cry5Da1 toxicity. (A) L4-stage C. elegans bre-5(ye17) mutant strains were resistant to Cry5Ca1 and Cry5Ba1 in contrast to the sensitivity of C. elegans N2 to these toxins. (B) C. elegans bre-5(ye17) mutant strains were susceptible to Cry5Da1, exhibiting nearly the same toxicity as N2, which was significantly different (P < 0.001) from that of the Cry5Ba toxin. (C) Synchronized L1 nematodes were examined to determine their growth on recombinant BMB171 plates with cells expressing the Cry5 toxin. Cry5Da1 showed a high level of inhibition of bre mutant strain growth; the bre-2(ye31) mutant strain showed little resistance to Cry5Da1 in contrast to the other mutants. At least 30 nematodes were used for each condition, and photographs were taken with a compound microscope. The bar indicates 100 μm. All error bars represent the standard errors of the mean over three independent experiments.

β-Methylgalactoside protects C. elegans from Cry5Da intoxication, but Cry5Ca has low detoxification.

The galactose analog β-methylgalactoside (β-MeGal) can inhibit Cry5Ba binding to the entire bre-dependent glycolipid, indicating that it may serve as an antidote to Cry5Ba, yielding diminished Cry5Ba toxicity against the growth of C. elegans hermaphrodites (30). We subsequently examined the function of β-MeGal as an antidote to Cry5Ca and Cry5Da toxins to further determine whether their receptors bind in different manners. All Cry5 toxins exhibited different levels of toxicity against C. elegans at the same concentration of β-MeGal (21 μg/ml) or without exogenous carbohydrate (Fig. 5). The extent of β-MeGal detoxification to Cry5Ca is equal to that of Cry5Ba, which was based on the context of the same level of growth inhibition (50%) with different concentrations of toxins. Moreover, the same dosage of the antidote, β-MeGal (an excessive concentration), protected C. elegans from growth inhibition, but this effect was observed only when the level of Cry5Da (80 μg/ml) was 8-fold higher than that of Cry5Ba (9 μg/ml) (Fig. 5). Thus, detoxification using the same carbohydrate proved that the receptors binding the three Cry5 toxins were via the same galactose terminus-mediated epitope, while the same dosage of the antidote β-MeGal may be able to bind more Cry5Da toxins, which suggests that Cry5Da has fewer galactose terminus-binding epitopes. Moreover, it implies that the β-galactose-rich terminus of the receptor is the Cry5Ba- and Cry5Ca-binding epitope, whereas the Cry5Da-binding epitope probably possesses a low-galactose terminus, which may, to some extent, explain why Cry5Da possesses little sensitivity to the bre-2(ye31) mutant strain worms (Fig. 4C).

FIG 5.

The galactose analog β-methylgalactoside (β-MeGal) protects C. elegans from Cry5 intoxication. All toxins against L1 C. elegans N2 synchronized worms were grown in the presence of the same concentration of β-MeGal (21 μg/ml). To inhibit N2 worms at the same level (50%), different concentrations of the toxins were used: 9 μg/ml Cry5Ba1, 15 μg/ml Cry5Ca1, and 80 μg/ml Cry5Da1. β-MeGal detoxification to Cry5Da1-mediated N2 toxicity was more susceptible than that observed with Cry5Ba1, although all Cry5 toxins were detoxified by the galactose analog β-MeGal. The error bars denote standard deviations from the mean based on data from three independent experiments. Statistical comparisons between two groups were performed using a paired t test, and significant differences were determined according to a threshold (*, P < 0.05; **, P < 0.01).

Cry5Da1 possesses a different glycan-binding motif compared to that of Cry5Ba, contributing to its low cross-resistance.

The similarity of domain II (Table 3) between Cry5Da and Cry5Ba (18%) is slightly higher than that between Cry1Ac and Cry2Ab (15%; high cross-resistance), which were proposed as the two insecticidal toxins with the lowest similarity with cross-resistance from the positive association model (12). Thus, whether cross-resistance exists between Cry5Ba and Cry5Da may not be easily deduced according to that model. To determine whether Cry5Ca and Cry5Da differ from Cry5Ba with respect to glycolipid-receptor binding, we performed a pairwise alignment with other Cry5 nematicidal toxins and predicted theoretical three-dimensional structures of Cry5Ca and Cry5Da using Phyre2 (31). Domain II of Cry5Da is less similar to those of other nematicidal proteins, which is consistent with its classification in a distinct clade in the phylogenetic tree (Fig. 1A). Cry5Ca and Cry5Da do not share the GXXXE motif with Cry5Ba in loop 1 of the β1 and β2 strands (Fig. 1B). Additionally, another binding motif, GG, was not found in the loop of β11 and β12 (32). Instead, Cry5Ca forms a shorter motif (GXXE), and Cry5Da does not contain a motif in the relevant loop 1 but possesses a GXXXD motif (a standard motif in BanLec) in the loop of the β7 and β8 strands that is close to loop 1, which may alternatively function as receptor binding (see Fig. S2 in the supplemental material).

TABLE 3.

Amino acid sequence similarity between pairs of Cry5Ca/Cry5Da and other crystal proteins

Domain	Crystal protein	Similaritya (%) to:
		Cry5Aab	Cry5Bac	Cry5Cad	Cry5Dad	Cry5Ead	Cry21Aac	Cry21Bac	Cry21Cac	Cry21Dac	Cry21Fac	Cry21Gab	Cry21Hac	Cry12Aab	Cry13Aac	Cry14Aac
II	Cry5Ca	7	35		28	32	40	39	44	46	43	30	43	35	39	34
	Cry5Da	28	18	28		27	17	19	13	18	7	32	27	23	16	16
III	Cry5Ca	47	46		33	31	42	44	39	51	48	41	40	36	41	43
	Cry5Da	46	50	33		50	51	49	55	35	38	44	38	37	49	47
Overall	Cry5Ca	38	44		34	35	43	44	44	53	47	38	43	37	40	40
	Cry5Da	44	38	34		43	39	38	42	37	38	43	39	42	37	39

^aThe Needleman-Wunsch algorithm hosted by EMBL-EBI Pairwise Sequence Alignment was used to calculate the similarity score; overall comparisons were based on the sequence with the C-terminal halves deleted, and the Blosum62 matrix and other gap penalties were determined using default parameters.

^bToxin did not show toxicity.

^cToxin presented toxicity against C. elegans.

^dToxicity was undetermined.

Based on the characteristics associated with galactose inhibition, we predicted that the same potential sites exist in all nematicidal Cry toxins by positive selection analysis, which has been widely used to predict the active sites of functional proteins under the same natural selection pressure. Thirteen residues out of 19 sites under positive selection are located in domain II of Cry5Ca and Cry5Da (see Table S2 in the supplemental material). Thr356 in Cry5Ba is next to but not within the predicted GXXXE motif. Gly432, Asp433, and Pro436 are located at the β5 and β6 loop, and Ser484 and Gly485 are located at the loop of β9 and β10, whereas Leu530 is located next to the predicted GG motif. It is hypothesized that all of the potential positive selection positions may form a superposition patch on the surface of the toxin, which may be involved in receptor binding, but more site-directed mutations of Cry5Da toxins in these sites are needed to demonstrate this hypothesis. Our data prove, to some extent, that the Cry5Da toxin does not share the same motif with Cry5Ba toxin, apart from the galactose and other receptor epitope-binding positions mentioned above. Based on comparative analyses of these Cry5 toxins, the secondary structure motifs, the superposition based on the tertiary structure, and the similarities in domain II (Table 3) are largely different between Cry5Da and Cry5Ba, which may explain their association with a low level of cross-resistance.

DISCUSSION

In this study, we identified and expressed two novel Cry5 toxins from Bacillus thuringiensis Sbt003 (23), Cry5Ca1 and Cry5Da1. We also found that the novel Cry5 toxins exhibit significant activity against C. elegans and M. incognita. Furthermore, it is revealed that these toxins not only affect nematode life span, fertility, and survival but also cause irreversible damage to the intestine. Their effects on the pharynx may be recovered after cessation of toxin feeding. Therefore, the novel Cry5 toxin genes are complementary candidates for the application of transgenic crops due to their resistance against plant root-knot nematodes.

In addition to these two novel nematicidal toxins, we found a series of virulence factors in strain Sbt003. Many assistant virulence factors, common virulence factors, and specific virulence factors are predicted to be necessary for the pathogenesis process of a nematicidal bacterium, which may be the same as various assistant virulence factors, Bel, Bmp1, InhA (33), and ColA, from other insecticidal Bt strains, and have also been proven to contribute to toxicity against the insect host with Cry toxins (25). More dissimilar Cry toxins seem to be important gene resources that can be used in pyramided Bt crops against pests. Simultaneously, the exploitation of more virulence factors may constitute another tool that can be applied in binary Bt toxin crops against resistance in insects or nematodes, similarly to Vip3 toxins for transgenic cotton (34).

We revealed a high level of cross-resistance between Cry5Ca and Cry5Ba against bre mutant strains of C. elegans but no cross-resistance between Cry5Ca and Cry5Da, which is in accordance with the results of our domain II and III phylogenetic analyses. Moreover, domain II is involved in 3D-toxin specificity, particularly via the binding of its exposed loops to distinct receptors, while domain III also contributes to receptor binding (4, 12, 32). Cry5Ca and Cry5Da may bind to different receptors because their clades evolved separately in the phylogenetic trees (Fig. 1). Based on the different toxicities and decreasing performance in the β-MeGal detoxification assay and the decreased susceptibility to the Cry5Da toxin in the C. elegans bre-2(ye31) mutant strain, where bre-2(ye31) refers to deficits in galactosyltransferase activity, we inferred that Cry5Da may also bind to the galactose terminus-mediated receptor, which is not the same glycolipid as that of Cry5Ba. Though no direct binding assay can conclude that Cry5Da binds with one receptor rather than glycolipids, Cry5D may exhibit a different nematicidal manner via the binding of a unique galactose-containing receptor through the bioassay results (Fig. 4C and 5). The sequence similarities of domain II that were positively associated with cross-resistance were quantitatively evaluated to determine the efficacy of insecticidal toxins in pyramided Bt crops, which indicated that insecticidal Cry1Ac and Cry2Ab (domain II similarity, 15%) had a high cross-resistance in cotton bollworm (12). On the other hand, our novel Cry5 toxins shared greater domain II similarity (18%) than the aforementioned Cry1Ac and Cry2Ab insecticidal toxins, yet the Cry5 toxins did not share cross-resistance. Thus, a threshold of 15% similarity in domain II is not sufficient to predict an association between sequence and resistance patterns in nematicidal toxins (Table 3). Therefore, the dissimilar nematicidal toxin genes selected to be applied into pyramided Bt crops need additional rules. Furthermore, the predicted motif (GXXXE) in Cry5Ba that functions in glycolipid binding does not exist in both Cry5Ca and Cry5Da, but a similar motif (GXXE) was instead found in Cry5Ca, which was consistent with the low-level cross-resistance between Cry5Da and Cry5Ba, while Cry5Ca showed high-level cross-resistance with Cry5Ba. Thus, more than the similarity of domain II is associated with cross-resistance; the similarities among the domain II topological structures of the toxins, such as the motif identified, as well as the sequence similarities of the loops that link domains II and III, may also be associated with the different resistance levels among these Cry5 toxins.

Although toxicity bioassays were performed on laboratory-selected bre mutant strain nematodes rather than field-evolved resistant plant parasite nematodes, the separate evolutionary relationship of domain II sequences and structures, as well as the significant synergistic nematicidal toxicity displayed by Cry5D and Cry5C, may provide additional data regarding the design of pyramided Bt crops that show resistance to nematodes.

We speculate that Cry5B is a case study because its motif is the same as BanLec (a lectin that binds glucose and mannose), although our toxins and other Cry proteins harbor the lectin-like structure in domain II and the carbohydrate-binding module 35-like structure in domain III (35, 36) in a parasporin ricin-like motif instead. A previous study of CGL2, a nematotoxic fungal galectin (37) ligand with galactose in the intestine, can provide a better receptor-binding structural reference in contrast to the mannose-binding lectin BanLec (38). Because the available information regarding nematicidal crystal protein toxin active sites is scarce, we used the maximum likelihood (ML) model to identify the position under adaptive selection (see Table S2 in the supplemental material), which may help elucidate the mechanisms through which Cry toxins target nematodes and delay the occurrence of resistance. Accordingly, it is important to reconstruct the Cry toxin site-directed evolution mutants in these residues to validate their toxicity improvement.

As shown herein, B. thuringiensis Sbt003 produces two crystal proteins with different characteristics against nematodes. Although its genome harbors four potential nematicidal cry genes, only two of the four have the ability to target nematodes, with the exception of the cancer cell-killing parasporin Cry65Aa1 (39). Based on an analysis of other strains, such as B. thuringiensis L366 with naturally truncated Cry5Ad1 (40); B. thuringiensis DB27 products Cry21Fa1, Cry21Ha1, and naturally interceptive Cry21Ga1 (6); and YBT1518 products Cry6Aa2, Cry55Aa1, and silent Cry5Ba2 (26), we believe that the different characteristic crystal proteins that exist in one Bt may be the result of coevolution between the host and pathogen (41) and that the outcome arises from the natural selection pressure of antagonist nematodes and Bt itself. These natural genomic variations adapt to environmental nutrition pressure (42). The interaction and arms race between free-living nematodes and nematicidal pathogens from the same ecosystem soil are extremely and frequently intense; therefore, highly similar, highly toxic proteins can instantaneously kill the nematode but are not suitable for long-term persistence with the nematode host, and the Bt population would be sharply decreased. Taking into account the fitness costs and trade-off theory, we speculate that dissimilar and natural truncated Cry toxins in some nematicidal Bt strains contribute to variant environmental adaptation.

Although it is easy to get more cry genes from Bt strain genome sequencing, few of them were suitable to apply into pyramided Bt crops against plant parasite nematodes. The reliable toxin gene candidates used in pyramided Bt crops should reflect the same activities or low cross-resistance. To delay and reduce the resistance of Bt toxins leading to agricultural loss, the identification of factors that contribute to the low cross-resistance of toxins may aid the selection of appropriate novel cry genes that can be used in pyramided Bt crops (43).

MATERIALS AND METHODS

Bacterial strains, plasmids, and culture conditions.

B. thuringiensis Sbt003 was isolated by and stored in our laboratory. The acrystalliferous B. thuringiensis BMB171 was used as the host for the expression of the parasporal crystal genes (44). The shutter vector pHT304 (45) was used for cloning and expressing genes. B. thuringiensis and E. coli cells were grown at 28°C and 37°C, respectively, in Luria-Bertani (LB) medium (1% tryptone, 0.5% yeast extract, and 1% NaCl, pH 7.0) with shaking at 220 rpm. B. thuringiensis was sporulated and produced Cry toxins in individual culture sporulation medium (ICPM) (0.6% tryptone, 0.5% glucose, 0.1% CaCO₃, 0.1% MgSO₄·7H₂O, and 0.05% KH₂PO₄, pH 7.0). Ampicillin (100 μg/ml) or erythromycin (25 μg/ml) was added to the medium to select for antibiotic-resistant strains of E. coli or B. thuringiensis, respectively.

Microscopy.

Sporulation cultures were monitored with a Zeiss Photomicroscope III instrument (Zeiss, Oberkochen, Germany). For observation of parasporal crystals, crystals were separated from spores by washing three times with Tris-EDTA (TE) buffer (1 M NaCl, 10 mM EDTA, pH 8.0) and three times with double-distilled water as described previously (18) and then stained with basic fuchsin.

Cloning and expression of cry5Ca and cry5Da genes.

To obtain the gene fragments of cry5Ca and cry5Da genes, two pairs of primers were designed targeting the coding genes and their flanking promoters and terminators for the PCR product of the cry5Ca1 gene with forward primer 5′-TACTGGTTCAAAAGAACTAT-3′ and reverse primer 5′-GGGTTCGTAATGGATAAAAAG-3′ and for the PCR product of the cry5Da1 gene with forward primer 5′-GCTGCTGAGACATTAGCCGA-3′ and reverse primer 5′-CCTCCTGGTACAGCTGTCGGA-3′. Total DNA prepared from strain Sbt003 was used as the PCR template. Amplification products were cloned into the pMD18-T vector and sequenced using an automated DNA sequencer (ABI, USA). The target fragments were purified and ligated into the same endonuclease-digested B. thuringiensis vector pHT304 to generate the recombinant plasmids pBMB1361 and pBMB1362. The two recombinant plasmids were introduced into acrystalliferous B. thuringiensis BMB171. An individual transformant was selected from LB plates containing erythromycin (25 μg/ml) and incubated in ICPM at 28°C until sporulation (90%). The spore-crystal mixtures were pelleted by centrifugation at 12,000 rpm for 5 min and washed three times with TE buffer and three times with sterile distilled water. Finally, the pellets were resuspended in sterile distilled water, and the suspension was prepared for sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) analysis.

Nematode toxicity bioassay.

C. elegans was maintained on standard nematode growth medium (NG) plates at 25°C and fed E. coli OP50 (46). The L1 growth assay was conducted by feeding first larval stage C. elegans different doses of toxins following the worm assay method (47). Each well of the 96-well plate contained 10 μl of freshly hatched L1 worms in M9 buffer (approximately 30 worms total), 10 μl 5-fold toxin or control buffer, 10 μl E. coli OP50 in S medium (optical density at 600 nm [OD₆₀₀], approximately 0.6), and 70 μl S medium at a final volume of 100 μl. Each dose was administered in triplicate. The plate was wrapped in a damp paper towel and placed inside an enclosed box at 20°C for 72 h. The worms (at least 20 worms for each condition) were then photographed at 100-fold magnification on a phase-contrast microscope. The worm was outlined and the worm area analyzed using ImageJ 1.4 software. The average area of the worms under each condition was normalized to that of the no-toxin control.

Brood size assay (fertility) was performed as described above except that the worms were in L4 (the fourth larval stage). The plate was incubated at 20°C for 72 h. The number of progeny in each well was counted. The brood size was normalized to the average brood size of the control.

For morphological observations, synchronized L4 nematodes were picked from the 96-well plates that included proteins from the spore-crystal lysate of Bt recombinant BMB1361 and BMB1362. The pharynx and anterior intestine were visualized under 400-fold differential interference contrast optics after 40 h of feeding toxins.

Phylogenetic analysis.

To investigate the phylogenetic relationship between Cry5Ca1 and Cry5Da1, we retrieved all of the following sequences of nematicidal crystal proteins and related proteins from the Neil Crickmore Bt toxin nomenclature website (http://www.btnomenclature.info/): Cry5Aa1, Cry5Ba1, Cry12Aa1, Cry13Aa1, Cry14Aa1, Cry21Aa1, Cry21Ba1, Cry21Ca1, Cry21Da1, Cry21Ea1, Cry21Fa1, Cry21Ha1, and Cry21Ga1. All of the above selected protein sequences were then aligned with our novel proteins using the MUSCLE algorithm implemented in MEGA 6.0 software. Aligned regions with gaps or poorly aligned sites were trimmed using the TrimAl tool of Phylemon 2.0 with an automated method.

Positive selection position analysis.

Positive selection position analysis was performed using the CodeML program from PAMLX software, with all gaps manually deleted. The coding sequences were the same as those used in the phylogenetic analysis, except for toxins with no nematicidal toxicity. Site Model in CodeML was used to obtain the ratio of nonsynonymous to synonymous evolutionary changes (dN/dS), along with M3 (discrete model) analysis, M1a (nearly neutral model), M2a (positive selection model), M7 (β distribution model), and M8 (β and ω model). M1a allows for an ω of <1 or an ω of 1. M2a and M7 do not allow for positive selection. M8 also allows for an ω of >1 (positive selection). Likelihood ratio tests (LRTs) were used to check the results, which were calculated using the 2ΔlnL method and then compared against the chi-square critical value, with df = 2, to determine the P value (a P value of <0.001 indicates significance). Empirical Bayes analysis was used for more reliable results (48).

Statistical analysis.

The 50% lethal concentration (LC₅₀) values and 95% confidence limits were calculated from three independent experiments using PROBIT from SPSS Statistics (version 22.0; IBM, USA). One-way analysis of variance (ANOVA) followed by Tukey's honestly significant difference (HSD) test was used to determine the significance of the differences among the different C. elegans strains. Statistical differences were considered significant at a P value of <0.05; otherwise, the null hypothesis was retained. All data are presented as the means ± standard deviation (SD) using Origin 9.0 software.