INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Photorhabdus luminescens is an entomopathogenic, bioluminescent, gram-negative bacterium related, as determined by its biochemical properties (36) and by 16S RNA analyses, to the Enterobacteriaceae in the gamma subdivision of the purple bacteria (15, 35). This bacterium was first isolated from a light-emitting insect cadaver infected by entomogenous nematodes (22, 30). Infective juvenile (IJ)-stage nematodes of the family Heterorhabditae carry this bacterium in their intestinal tracts (16, 23, 32). The IJ nematodes, enclosed in an outer cuticle layer, do not feed but actively seek out insect prey in the soil. When the IJ nematode encounters a potential victim, it penetrates the outer cuticle of the insect and, upon entering the hemolymph, releases the P. luminescens cells from its intestine. The bacteria multiply rapidly, killing the insect within 1 to 2 days. During this time, as the bacterial population reaches a high level, the insect cadaver becomes red in color and visibly bioluminescent. Antibiotics with antifungal and antibacterial activities are produced by P. luminescens, and it is thought that these substances prevent invasion of the cadaver by other microorganisms, resulting in ideal conditions for growth and reproduction of the nematode (2, 29). The nematodes feed on the bacteria and dead insect tissue, developing into mature reproductive adults. The nematodes complete several reproductive cycles over a 10- to 20-day period. Then, presumably sensing impending nutrient depletion, the nematodes differentiate into the IJ developmental stage and migrate from the cadaver in search of new insect victims. Detailed descriptions of aspects of P. luminescens taxonomy and physiology and the events involved in insect pathogenesis of the bacteria and nematodes are contained in two recent reviews (17, 18).

During growth in the insect cadaver or in various complex growth media, the P. luminescens cells produce intracellular phase-bright protein inclusion bodies. The inclusions were purified and found to be composed of two distinctive low-molecular-weight proteins (7). These proteins do not possess insecticidal activity, and their functions are unknown. The genes coding for the two proteins were cloned and analyzed (4). The genes do not show a significant degree of homology to any previously described gene.

The 50% lethal dose of P. luminescens injected into Galleria mellonella larvae was reported as being less than five cells (27). Injection of cells from a 24-h culture of the bacterium used in this investigation, P. luminescens W-14, into Manduca sexta larvae showed a 50% lethal dose of one cell (unpublished observation). This suggests that P. luminescens is an amazingly effective insect pathogen, able to resist completely the antibacterial defense mechanisms of the insect. Several species of the closely related bacterial genus Xenorhabdus (6) are also lethal to insect larvae when injected at these low levels (17). Xenorhabdus spp. are nematode-borne entomopathogens specifically associated with a different family of nematodes, the Steinernematidae (36).

Little is known about the mechanisms by which Xenorhabdus and Photorhabdus spp. are virulent for insects. Surface components of P. luminescens (12) and Xenorhabdus nematophilus (13) were shown to be implicated in the evasion of hemocytes in insect hemolymph by these bacteria. Lipid A isolated from X. nematophilus cell walls was lethal upon injection into G. mellonella larvae (13). Jarosz et al. (21) reported that extracts of homogenates of G. mellonella larvae that had been infected with the nematode Steinernema feltiae containing X. nematophilus cells and also with the nematode Heterorhabditis bacteriophora containing P. luminescens cells were lethal when injected into larvae of the insect. This publication also reported that injection of samples of cell-free broth cultures of P. luminescens, but not X. nematophilus, was lethal to the insects. The infected-insect extracts and broth culture samples had no adverse effects when orally supplied to the G. mellonella larvae. The toxic activity of the samples was heat labile, and they contained protease activity.

Schmidt et al. (33) purified an alkaline metalloprotease from culture broths of P. luminescens and suggested that the protease might be involved in the virulence of the organism. No evidence supporting this idea was presented. Yamanaka et al. (40) reported injectable insecticidal activity for Spodoptera litura larvae in culture broths of two of four species of the genus Xenorhabdus and a low level of toxicity in one of three Photorhabdus strains. All of the extracts of both genera contained high levels of protease activity, leading the authors to conclude that proteases are not involved in virulence.

A lipase activity purified from broth cultures of Photorhabdus sp. strain K122 showed toxicity for larvae of G. mellonella (8). Culture broths of Escherichia coli that had been transformed with the plasmid-borne lipase gene acquired injectable toxicity for the insect larvae. The lipase was not reported to be toxic when administered orally. Involvement of lipase activity in the virulence of X. nematophilus is also suggested by a report that the autoinducer molecule of bioluminescence in Vibrio harveyi, n--hydroxybutanoyl homoserine lactone, increased lipase activity in avirulent mutants of X. nematophilus coincident with restoration of virulence for G. mellonella larvae (11).

In avirulent mutants of X. nematophilus obtained by Tn5 mutagenesis, no relationship between virulence and lecithinase, lipase, or hemolysin activity was evident (39). A later study of Tn5-derived avirulent mutants of the same bacterium revealed no relationship between extracellular enzyme activities and insect virulence (14). There was also no difference in removal of the virulent and avirulent mutants by G. mellonella hemocytes.

This report describes the purification and characterization of a high-molecular-weight protein complex produced by P. luminescens during growth in peptone broth. The purified protein complex, which has no protease activity and negligible lipase activity, is lethal when injected into or fed to a variety of insect larvae.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Bacterial cultures and growth conditions. The P. luminescens strains used in this investigation were W-14 (ATCC 55397; isolated from South Florida garden soil during this study), Hm (obtained from G. Poinar, University of California, Berkeley), NC-19 (obtained from W. Brooks, University of North Carolina), and Hb (isolated from Hb nematodes purchased from Gardens Alive, Lawrenceburg, Ind.). Stocks were maintained on petri plates containing 2% Proteose Peptone no. 3 (PP3) and 1.5% agar (Difco Laboratories, Detroit, Mich.). The cultures were incubated at 28°C for 4 days and then stored at room temperature for up to 1 month. Primary-form colonies, selected on the basis of colony morphology, bioluminescence, pigmentation, and inclusion protein production, were inoculated into 1-liter flasks containing 200 ml of 2% PP3 broth supplemented with 0.5% polyoxyethylenesorbitan monostearate (Tween 60; Sigma Chemical Co., St. Louis, Mo.). The cultures were incubated for 48 h at 30°C on a rotary shaker at 250 rpm.

Insect rearing. M. sexta larvae were reared from eggs (Carolina Biological Supply Co., Burlington, N.C.) or supplied by N. Beckage or W. Goodman, Department of Entomology, University of Wisconsin, Madison. Larvae were reared using a 16-h light-8-h darkness photoperiod at 25°C and fed a gypsy moth wheat germ diet (ICN Pharmaceuticals Inc., Costa Mesa, Calif.). G. mellonella larvae (obtained from H. Coppel, Department of Entomology, University of Wisconsin, Madison) were reared by the procedure of Mohamed and Coppel (28). German and American cockroaches, obtained from S. C. Johnson Wax Co., Racine, Wis., were reared on dry rat chow (Ralston Purina, St. Louis, Mo.) and tap water. Pharaoh ants, also obtained from S. C. Johnson Wax, were reared on 10% sucrose, tap water, and dead frozen houseflies. A colony of meal worms (Tenebrio sp.) was maintained in our laboratory on a diet of dry oatmeal, apples, carrots, potato slices, granola, and tap water.

Toxin assays. Test fractions were filter sterilized with a 25-mm-diameter, 0.2-µm-pore-size cellulose acetate syringe filter (Nalgene-Nalge Co., Rochester, N.Y.). For determination of the activity of toxins administered orally to M. sexta larvae, various samples, consisting of 20 to 100 µl of solution, were applied to the surfaces of 1-cm³ blocks of gypsy moth wheat germ diet. Bioassays for other insects that were administered toxins orally were performed with a high-performance liquid chromatography (HPLC)-purified toxin fraction in 100 mM KPO₄ buffer, pH 6.9, at a concentration of 2 mg of total protein/ml. Cockroaches were each fed a 1-cm³ block of gypsy moth wheat germ diet to which 100 µl of toxin solution had been applied. The wax moth larvae were fed a diet of dry food to which 100 µl of toxin solution had been added/g. Mealworms were fed several rolled oats which had been moistened with the toxin solution (approximately 10 µl/oat). Pharaoh ants were fed the toxin solution in a 10% sucrose solution. The intrahemocoelic injections of 5-µl toxin samples were performed through the first proleg of fourth- or fifth-instar M. sexta, wax moth, or mealworm larvae. The intrahemocoelic injection of 5-µl toxin samples into the cockroaches was done through the abdomen. A 25-µl gas-tight syringe (Hamilton Co., Reno, Nev.) equipped with a 30-gauge, no. 4-point needle was used for all injections. Different syringes were used for injections of live bacteria and filtered samples. Between injections, each syringe was rinsed three times with sterile deionized-distilled H₂O (ddH₂O), three times with 70% ethanol, and finally three times again with sterile ddH₂O. Each larva was examined for weight, cessation of growth, and death at 24-h intervals for up to 7 days.

Preparation of concentrated culture supernatants. Cultures of the P. luminescens strains grown for 48 h in 2% PP3 plus 0.5% Tween 60 at 30°C were centrifuged at 10,000 × g for 20 min. A sample of the supernatant fluid of each culture was filter sterilized and centrifuged for 4 h at 500 × g and 4°C in a Centriprep 100 ultrafiltration device with a molecular mass cutoff of 100 kDa or a Centricon 30 apparatus with a molecular mass cutoff of 30 kDa (Amicon Inc., Beverly, Mass.). An equal volume of sterile 100 mM Tris-HCl, pH 7.0, was added to the retentate chamber of the Centriprep tube; this was followed by a 1-h centrifugation as described above. The high-molecular-weight material retained by the filters was removed with a pipette and tested for toxicity to M. sexta larvae by oral administration or injection of samples.

Column chromatography. All buffers and stock solutions for column chromatography were filtered through 0.2-µm-pore-size filters and autoclaved before use. Two liters of a 48-h culture grown in 2% PP3 plus 0.5% Tween 60 was centrifuged at 10,000 × g for 20 min. The supernatant fraction was adjusted to 50 mM K₂HPO₄ with 1.0 M K₂HPO₄, and the pH was adjusted to 8.6 by the addition of a small amount of 5.0 M KOH. The supernatant was then mixed with a 250-ml slurry of DEAE-Sephacel (Pharmacia Biotech, Uppsala, Sweden) which had been autoclaved as suggested by the manufacturer and equilibrated with 50 mM K₂HPO₄, with no pH adjustment. The toxin activity was adsorbed to the DEAE resin under these conditions. This mixture was poured into a 2.6- by 40-cm column and then washed with 500 ml of 50 mM K₂HPO₄. The column was next washed with 150 mM KCl prepared in ddH₂O, pH 6.3, and applied at a flow rate of 30 ml/h at room temperature until the effluent had a steady baseline absorbance at 280 nm. The column was then eluted with 300 mM KCl prepared in ddH₂O, pH 6.3. Fractions were collected and tested for insecticidal activity and for the presence of a high-molecular-weight protein complex by HPLC molecular sieving (see below). Fractions containing toxic activity were pooled, filtered through a 0.2-µm-pore-size filter, and concentrated in a Centriprep 100 device.

HPLC. HPLC was performed with a 0.75- by 60-cm TSK-GEL G-4000 SW molecular sieve column (Toso Hass, Montgomeryville, Pa.). The column was eluted with 100 mM KPO₄ (pH 6.9) at a flow rate of 0.4 ml/min at room temperature. The sample loading volume was 250 µl. The eluent absorbance at 218 and 280 nm was routinely monitored. The column was calibrated with proteins of known molecular mass which were dissolved at 2 mg/ml in 100 mM KPO₄, pH 6.9. The protein standards (Pharmacia Biotech) were thyroglobin (669 kDa), aldolase (158 kDa), and ovalbumin (42.7 kDa).

Determination of protein concentrations. Protein concentrations were determined with bicinchoninic acid protein assay reagent (Pierce, Rockford, Ill.), using bovine serum albumin as a concentration standard. The protein concentration of HPLC-purified toxin was estimated both by the bicinchoninic acid method and by comparison of the integrated peak areas (absorbance at 280 nm) of the toxin samples with integrated peak areas of standard proteins at a known concentration (thyroglobin, aldolase, ovalbumin, and bovine serum albumin, each at 2 mg/ml).

Electrophoresis. Nondenaturing native agarose gel electrophoresis with high-resolution Metaphor agarose (FMC BioProducts, Rockland, Maine) was performed by the method of Dean et al. (9). The 15-cm by 12-cm by 4-mm-thick 1.9% agarose resolving gel, buffered with 200 mM Tris-borate (pH 8.3), was allowed to solidify at 4°C. A 5-cm section was cut from one end of the resolving gel and replaced with the 1.5% agarose stacking gel, buffered with 100 mM Tris-HCl (pH 7.0) and containing 0.05% phenol red as a tracking dye. The stacking gel was allowed to solidify at 4°C for 30 min. Protein samples, buffered in 50 mM Tris-HCl (pH 7.0), were then loaded into wells at the cathode end of the gel. The cathode buffer was 0.025 M Tris-0.192 M glycine, and the anode buffer was 1.0 M Tris-HCl, pH 8.3. The gel was electrophoresed horizontally at 6.5 mA with a constant current on a Peltier-cooled platform at 15°C until the tracking dye reached the anode end of the gel.

Analysis of native-gel fractions for insecticidal activity. Proteins separated in native gels were either used directly or eluted into buffer by incubation of macerated sections for 16 h at 30°C in an equal volume of sterile 50 mM Tris-HCl, pH 7.0, in a microcentrifuge tube. Following removal of gel pieces by centrifugation, the buffer (containing the eluted proteins) was removed with a pipette.

Enzymatic activity assays. HPLC-purified toxin was assayed for the presence of proteolytic activity (38), type C phospholipase activity (25), and nuclease activity (24) in accordance with published procedures. Lipase was assayed by cutting wells in Tween agar (34) and filling the wells with toxin samples. A positive result was a zone of precipitation in the agar around the well. Lipase activity was also determined by a colorimetric assay (8). Hemolysin was assayed by cutting wells in sheep erythrocyte blood agar and filling the wells with toxins samples. A positive reaction was a zone of hemolysis (clearing of the erythrocytes) in the agar around the well.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Discovery of insecticidal activity. Approximately 100 cells of P. luminescens NC-19, Hm, or W-14 were injected into each of 10 fourth-instar M. sexta larvae. This number of bacteria was lethal to the larvae within 24 h. At 48 h, the larvae were brick red in color and bioluminescent, and microscopic examination of a sample of hemolymph from each insect revealed large numbers of P. luminescens cells. These cells were easily recognized by the presence of phase-bright intracellular protein inclusion proteins. The moribund larvae were then ground in a tissue grinder in 2.5 ml of sterile saline. The homogenates were then centrifuged, the supernatant fractions were passed through 0.2-µm-pore-size filters, and a 5-µl sample of filtrate was injected into each of 10 insect larvae. All of the insect larvae were dead 24 h later. A similarly prepared homogenate of an insect larva that had not been injected with the bacteria did not kill or inhibit growth following injection into control larvae. The three strains of P. luminescens had produced a toxic material during growth in the larvae.

Purification of insecticidal toxin. Advantage was taken of the apparent large size of the insect-killing activity in broth cultures. Growth liquor from a 48-h PP3-Tween 60 broth culture of strain W-14 was concentrated 15-fold by using a Centriprep 100 device. The concentrate was then applied to a large-pore molecular sieving HPLC column. The column resolved the concentrated liquor into numerous-size fractions (Fig. 1A). Samples of each fraction were tested for insecticidal activity, and activity was found only in the small peak eluted from the column at approximately 31 to 34 min. This result provided two pieces of information: the insecticidal activity is large, since it elutes early from the column, and the low UV absorbance of these fractions suggests that the toxin is present in very small amounts. To obtain larger quantities of toxin, a DEAE-Sephacel chromatography purification step was developed. Culture broth was mixed batchwise with the DEAE-Sephacel (so that the toxin bound to the resin); the mixture was poured into a column, and the toxic activity was eluted with KCl. This step greatly increased the insecticidal activity present in the sample while reducing other proteins (Fig. 1B). Insect toxicity was found exclusively in the peak eluting at 33.8 min, which had increased in size (absorbance) relative to the other peaks present. The material in this KCl fraction from the DEAE column was then applied to a Sephacryl S-400 column (Fig. 1C). Just one major peak, which contained the insecticidal activity, was eluted. The material in this fraction was next passed through a 4000 SW size exclusion column (Fig. 1D). One sharp peak of protein, which eluted from the column at 33.5 min, contained the toxic activity. Injection or feeding of 5-µl samples of this fraction killed all of 10 M. sexta larvae within 48 h (data not shown). Based on the elution times of the standard proteins (Fig. 1E), we estimated the size of the toxic protein molecule to be approximately 1,000 kDa. Calculations based on the amount of protein in the 33.5-min peak fraction (Fig. 1D) indicate that the organisms produce and secrete approximately 2.5 mg of toxin per liter of 72-h culture broth. The toxin activity, purified by the procedure described above, was considered sufficiently pure for further studies of the activity.

View larger version (19K): [in this window] [in a new window]   FIG. 1.   HPLC chromatograms, from the HPLC 4000 SW molecular sieve column, of the toxic fraction following each step of the purification. (A) Crude culture broth; (B) DEAE-Sephacel; (C) Sephacryl S-400HR; (D) HPLC 4000 SW molecular sieve column; (E) standard proteins (37.07 min, thyroglobin, 669 kDa; 47.74 min, aldolase, 158 kDa; 51.40 min, ovalbumin, 42.7 kDa). The arrows below the chromatograms indicate fractions containing the insecticidal activity. The numbers on the chromatograms indicate the elution times in minutes. Absorbance was monitored at 280 nm.

View larger version (52K): [in this window] [in a new window]   FIG. 2.   Analysis of protein components of the insecticidal toxin complex of P. luminescens. (A) SDS-10% PAGE analysis of HPLC-purified P. luminescens insecticidal toxin preparations. The gel was stained with Coomassie brilliant blue. Sta, size standard; tox, purified toxin sample. (B) HPLC-purified toxin separated by native agarose gel electrophoresis; 5 µg of total protein was loaded. The gel was stained with Coomassie brilliant blue. (C) SDS-18% PAGE analysis of protein bands separated by native agarose gel electrophoresis. Lanes: 1, band 1; 2, band 2; 3, band 3. The gel was silver stained. (D) HPLC-purified toxin separated by agarose gel electrophoresis in the presence of 0.1% Triton X-100; 4 µg of total protein was loaded. The gel was stained with Coomassie brilliant blue. (E) SDS-10% PAGE analysis of protein bands separated in native agarose gels in the presence of 0.1% Triton X-100. Lanes: 1, band 1; 2, band 2; 3, band 3. The positions of molecular mass markers are indicated.

Properties of the insecticidal toxin. The insecticidal activity appears to be proteinaceous since it is heat labile (Table 1) and large in size. This supposition was confirmed by exposing the purified toxin to proteolytic digestion. Both the injectable and oral toxicities to larvae were completely eliminated by protease K digestion (data not shown).

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Two families of nematodes, each carrying a specific genus of bacteria, are pathogenic for a wide range of agriculturally important insect pests (for a review, see reference 31). We have discovered that one of these bacteria, P. luminescens, secretes into the growth medium a high-molecular-weight protein complex that is lethal when injected into or fed to members of at least five insect genera representing four orders of the class Insecta. The significance of our discovery is that it makes feasible the protection of plants by direct application of the toxin or by transformation of the toxin gene or genes into plants as a strategy for the control of insect pests. Attempts to exploit the Heterorhabditis and Steinernema nematodes in pest control have been made (for reviews, see references 3 and 31). Problems inherent in mass cultivation of nematodes, their stability during transport and storage, and application of the live nematodes make the use of entomopathogenic nematodes for pest control, especially in large-scale agriculture applications, impractical. Research directed to cloning of the P. luminescens genes and their transformation into plants, with the goal of obtaining transgenic insect-resistant plants, is now in progress.

It is likely that the virulence of P. luminescens for insects is a complex and multifaceted process. It is important to point out that the protein toxin that we purified and characterized shows none of the enzymatic activities, such as lipase, protease, and phospholipase, that others have suggested to be associated with virulence. These enzymes may be virulence factors also, but only the protein complex that we have purified has been shown to be toxic to insects upon oral administration. Work now in progress indicates that the P. luminescens toxin causes disruption of the midgut epithelium of M. sexta larvae (5).

The heat lability and protease susceptibility of the purified toxin complex indicate that its lethality to insects is associated with a protein(s) in the complex. This activity is clearly different from the lipopolysaccharides and lipid A, which were shown to be associated with virulence of Xenorhabdus and Photorhabdus spp. (12, 13).

Many bacterial toxins, including the insecticidal toxin of Bacillus thuringiensis, are known to be composed of multiple protein subunits (20, 26), and many of these toxins require enzymatic processing or activation by the producing cells or cells of the target organism to be fully active. The Photorhabdus insecticidal toxin is composed of a complex of proteins of various sizes that separate on native agarose gel electrophoresis into three subcomplexes. The proteins in these complexes may represent various stages of processing of a single large protein, or they may be derived from three or more individual gene products. Our data indicate clearly that the toxic activity is due to one protein or a combination of proteins that were purified as a multimeric complex from the culture supernatant.

Cells of P. luminescens are often observed to go through a phase variation process in which variants, termed secondary forms, appear at a high frequency (1, 17). These secondary variants have lost many characteristics of the original primary culture, including bioluminescence, pigmentation, dye reduction activity, protease activity, antibiotic production, and intracellular crystalline proteins. We found the secondary variants to be as virulent as primary cells when injected into M. sexta larvae, and the insect toxin was detected in the growth medium (unpublished observations).