Insecticidal Toxins from the Bacterium Photorhabdus luminescens

See allHide authors and affiliations

Science  26 Jun 1998:
Vol. 280, Issue 5372, pp. 2129-2132
DOI: 10.1126/science.280.5372.2129


Transgenic plants expressing Bacillus thuringiensis (Bt) toxins are currently being deployed for insect control. In response to concerns about Bt resistance, we investigated a toxin secreted by a different bacterium Photorhabdus luminescens, which lives in the gut of entomophagous nematodes. In insects infected by the nematode, the bacteria are released into the insect hemocoel; the insect dies and the nematodes and bacteria replicate in the cadaver. The toxin consists of a series of four native complexes encoded by toxin complex loci tca, tcb, tcc, and tcd. Both tca and tcd encode complexes with high oral toxicity to Manduca sexta and therefore they represent potential alternatives to Bt for transgenic deployment.

Photorhabdus luminescens is a bioluminescent Gram-negative bacterium of the Enterobacteriaceae (1). This bacterium lives in a mutualistic association with entomophagous nematodes of the family Heterorhabditae and is released from the gut of the nematode upon invasion of the insect hemocoel by the nematode. The bacteria multiply and kill the host within 24 to 48 hours, and the nematodes feed on both the bacteria and the insect cadaver itself (2). At this stage, the insect host emits light produced by the bacteria. The invading nematodes reproduce within the insect and third-generation larvae then leave the cadaver in search of new hosts.

Photorhabdus luminescens can be cultured away from the host, and 50% insect mortality has been reported with fewer than five bacteria per larva (3). Interestingly, despite the fact that this bacterium also produces crystalline inclusion proteins (4, 5), a variety of antifungal and bacterial compounds (6, 7), and secreted proteases, lipases, and lipopolysaccharides (2, 8–11), our work has shown that insecticidal toxicity is associated with high molecular weight protein complexes secreted directly into the growth medium. Partially purified mixtures of these complexes are active against a wide range of insects (12) from several different orders (Lepidoptera, Coleoptera, and Dictyoptera), unlike differentBacillus thuringiensis (Bt) δ-endotoxins, which often exhibit specificity for a given insect group. In view of recent concerns about the evolution of insect resistance to transgenic crops expressing Bt (13–15), we are therefore investigating the use of the Photorhabdus toxin (Pht) as an alternative.

We purified a toxic high molecular weight protein fraction fromP. luminescens strain W14 broth by sequential ultrafiltration, DEAE anion-exchange chromatography, and gel filtration (16). Subsequent high-pressure liquid chromatography (HPLC) anion-exchange chromatography revealed that this final fraction contained several peaks (Fig. 1A). Peaks A and B run as single or double complexes (A, B1/B2) on a native agarose gel (Fig. 1B) but resolve into a series of unique polypeptides (see Table 1 for NH2-terminal sequences) with SDS–polyacrylamide gel electrophoresis (PAGE) (Fig. 1C). Complex A was responsible for most of the oral activity with smaller peaks of toxicity being associated with complexes C and D (Fig. 1D). Complex A has a median lethal dose of 875 ng per square centimeter of diet after 7 days (17); at doses as low as 40 ng/cm2, larvae gained only 14% of the weight of untreated controls. CryI Bt proteins are also active against Manduca sexta in the nanograms per square centimeter range (18).

Figure 1

Analysis of the high molecular weight toxin fraction by HPLC, native agarose gel, SDS-PAGE, and oral bioassay against M. sexta. (A) HPLC purification into four peaks: A, B, C and D. (B) Native gel shows HPLC peak A to be a single complex (complex A) and peak B resolves into two (B1/B2). (C) SDS-PAGE gel stained with Coomassie brilliant blue shows polypeptides present in each complex. Numbers indicate polypeptides with derived NH2-terminal sequences (Table 1). (D) Bioassay of HPLC-derived fractions showing toxicity associated with peaks A, C, and D. (E) Bioassay of purified complex A showing both percentage mortality data and growth inhibition of surviving larvae (see text for discussion).

Table 1

Characteristics of TcA, TcB, and TcC peptides (34).

View this table:

To clone the genes encoding these toxin complexes, we used monoclonal (C5F2) and polyclonal antisera against purified toxin preparations (19) to screen a plasmid P. luminescens genomic library (20). The antitoxin polyclonal antiserum precipitates the native toxin complexes and neutralizes oral activity against M. sexta (5). Three toxin complex–encoding loci—tca, tcb, andtcc—were cloned. Both tca and tcccontain three long open reading frames (ORFs) transcribed in the same direction and then a short terminal ORF in the opposite orientation (Fig. 2). The tcb locus consists of a single long ORF. Each ORF corresponds to one or more of the polypeptides sequenced (Table 1) from the HPLC-derived peaks (Fig. 1) except for TcaZ, TccC, and TccZ, which were not detected.

Figure 2

(A) Map of the four toxin complex (tc)–encoding loci. tca andtcc show a similar organization with three ORFs (tcaA, tcaB, tcaC or tccA, tccB, tccC) in the same direction and a fourth (tcaZ ortccZ) transcribed in the opposite direction. NH2-terminal sequencing of Tc proteins (numbered arrowheads) purified from growth medium (Table 1) shows that some of the encoded proteins undergo posttranslational processing. Thus, TcaA and TcaB are cleaved and TcaC appears intact. Green shading shows the regions of TcaC and TccA with similarity to theSalmonella virulence proteins spvB and spvA, respectively (see text). Yellow indicates regions of amino acid similarity among TcaB, TcbA, and TcdA surrounding the presumptive protease cleavage site [see (B)]. Asterisk indicates the recognition site of monoclonal antibody C5F2. Restriction maps of all Hind III sites (H) are shown below each locus, alongside other selected restriction sites used in cloning and gene disruption (B, Bgl II; EI, Eco RI; E5, Eco RV; EIII, Eco 47III; N, Ns I; Sp, Sph I; Sa, Sau 3a). Open boxes correspond to sequences deleted in knockout strains and shaded boxes correspond to restriction fragments used as probes in Southern blot analysis of the resulting mutants. (B) Alignment of predicted amino acid sequences from TcaB, TcbA, and TcdA showing similarity around the presumptive protease cleavage sites (arrowheads) in each.

Comparison of the predicted versus the observed NH2-terminal sequences of the different polypeptides shows that several of the polypeptides encoded by the tc loci are found cleaved in the bacterial supernatant. TcaA is cleaved into three polypeptides (TcaAi, TcaAii, and TcaAiii), TcaB is cleaved into two (TcaBi and TcaBii) and TcaC is apparently uncleaved. Similarly, TcbA is also cleaved into three polypeptides (TcbAi, TcbAii, and TcbAiii) whose NH2-termini were again confirmed by sequencing (Table 1).

The tca, tcb, and tcc loci (GenBank accession numbers AF046867, AF047457, and AF047028) do not show overall similarity to any sequences currently deposited in GenBank. However, the predicted NH2-terminal region of TcaC has 47% amino acid identity to the putative spvB protein, and TccA shows similarity to spvA (GenBank accession number S22664). Salmonellaplasmid virulence (spv) genes are required for growth ofSalmonella dublin in bovine macrophages (21).spvB in particular is required for Salmonella typhimurium to induce cytopathology in human monocyte-derived macrophages (22), suggesting that TcaC may attack insect hemocytes.

There is, however, considerable predicted amino acid similarity among components of Tca and Tcb. Thus, a region of 689 amino acids in TcaB (amino acids 501 to 1189) shares 40% overall identity with a 704–amino acid region of TcbA (amino acids 1801 to 2504). This identity is raised to 53% in a 151–amino acid region surrounding the proposed proteolytic cleavage sites of the two proteins (Fig. 2B). The predicted amino acid sequence of all components of Tcb shows 51.6% identity to products of a fourth locus, tcd—another single long ORF (Fig. 2A) (23), which again shares the same putative protease cleavage site (Fig. 2B). Tcd thus may be a homolog of Tcb and undergo similar proteolytic processing. tcb andtcd both encode very large proteins of a size similar to those of Clostridium difficile toxins A and B. Therefore, although TcbA has only a very limited overall predicted amino acid identity to these toxins (17% identity to both toxins A and B) (24, 25), the similarity in the large size and processing of the toxins from both bacteria, and their effect on the gut of the host organism (26), may suggest similar modes of action.

To investigate the cause of oral toxicity to M. sexta, we used heterologous expression of the tc genes inEscherichia coli and disruption of the loci in P. luminescens. Although Tc proteins expressed in E. coliwere recognized by our antibodies, they were not processed or secreted and were not orally toxic. We then deleted or disrupted each of thetc loci from the single P. luminescens strain W14, generating mutant strains tca, tcb , tcc, andtcd (Fig. 2A). Correct disruption of each of the loci was shown by Southern blot analysis and absence of the corresponding toxin complex was confirmed by Western blotting (data not shown). Deletion of either tca or tcddramatically reduced the percentage of mortality and correspondingly increased the relative weight gain of surviving larvae (Fig. 3). Deletion of both loci in a single strain, tca/tcd , completely abolished oral toxicity. These data show that both tca andtcd loci encode orally active toxins and that, together, they comprise the majority of activity against M. sexta. Interestingly, deletion of either tcb or tccalone also reduces mortality (note the differing broth concentrations between the tc mutants and the wild-type W14 in Fig. 3).

Figure 3

Bioassay of Manducaneonates with the tc knockout strains. (A) Percentage mortality is shown at 1×, 5×, and 10× broth concentrations (33). Note that deletion of eithertca (tca ) or tcd(tcd ) reduces toxicity and toxicity is abolished in the tca/tcd double knockout. The median effective concentrations (EC50) oftca, tcd , andtca /tcd were not determinable because of low levels of toxicity (>10×). The EC50 and 95% confidence intervals (expressed at the broth concentration that killed 50% of larvae after 7 days) of W14 and the remaining mutants are as follows: W14, 0.11 × (0.07 to 0.17);tcb , 8.17 × (6.5 to 12.5);tcc , 4.30 × (3.2 to 5.5);tca/tcc , 8.21 × (6.0 to 14.8). (B) Relative weight gain of surviving larvae after 7 days of exposure to diet treated with broths from the knockout strains. Weight is expressed as a proportion of the control weight (165+ 10.5 mg; mean + SE). Note how increasing the concentration of both tca andtcd broths causes a dose-dependent reduction in growth, whereas the combination oftca−/tcd in a single bacterial strain results in no weight reduction.

The complex relationship between the toxicity of the products of all four loci in the gene knockout experiments suggests that there are interactions among the different gene products. In relation to the interaction between Tca and Tcd, several alternative hypotheses can be raised. Complex D may require complex A for enhancement of toxicity, in a fashion analogous to toxin B of C. difficile. Alternatively, complex D may modulate the toxicity of complex A (and possibly other complexes), as suggested by the interactions apparent inFig. 3B. Finally, it should be noted that the above discussion is confined to oral activity against a single lepidopteran. Therefore, the apparent absence of a significant oral effect of deletingtcb (a clear homolog of tcd) and the minor effect of removing tcc may reflect that they are toxins active either against different groups of insects (such as Coleoptera) or by alternative routes of delivery (for example, direct introduction into the hemocoel is required for full toxicity).

To compare Pht with other toxins that are active in the gut [including the δ-endotoxins and vegetative insecticidal proteins of B. thuringiensis (27–30) and cholesterol oxidase (31)], we examined the histopathological effects of purified Tca (Peak A) on the gut of M. sexta(32). After M. sexta feed on Tca-treated diet, large cavities appear in the midgut epithelium and cellular debris appears in the gut lumen. Damage begins in the anterior of the gut and after 48 hours the disruption has spread posteriorly along the midgut, which has become totally disorganized (Fig. 4). Subsequently, the cavities in the midgut epithelium enlarge, there are no recognizable columnar cells remaining, and the lumen of the gut is packed with cell debris (Fig. 4B). Larvae then cease feeding and either die or gain little or no weight. This histopathology resembles that described for other toxins that are active in the gut (27–30).

Figure 4

Cross sections of M. sexta midgut epithelium either untreated (A) or treated (B) with Tca toxin. (A) Anterior midgut epithelium of neonate larvae 48 hours after ingesting untreated diet. Note regular arrangement of goblet cells (gc) and columnar cells (cc) in the midgut epithelium and the lumen (lu) packed with diet. (B) A similar section 48 hours after ingesting Tca from diet dosed at 1350 ng/cm2. Note the complete disorganization of, and large cavities within, the epithelium and the presence of cellular debris (cd) in the lumen. Bar = 50 μm.

In conclusion, these data show that Pht toxins are as potent as the δ-endotoxins of B. thuringiensis and therefore may provide useful alternatives to the deployment of Bt toxins in transgenic plants. Alternation or co-deployment of Pht and Bt toxins would prolong the effective life of both biological insecticides by delaying the evolution of resistance to either component alone. However, the interaction between the products of tca and tcdand the observed processing of the Tc polypeptides may complicate their expression in transgenic plants.

  • * To whom correspondence should be addressed. E-mail: ffrench{at}


View Abstract

Navigate This Article