Bacillus thuringiensis var israelensis crystal δ-endotoxin: effects on insect and mammalian cells in vitro and in vivo

During the process of sporulation of the gram-positive bacterium Bacillus thuringiensis a parasporal, proteinaceous, crystalline inclusion is synthesized within the sporangium (Somerville, 1978; Bulla et al. 1980). The insecticidal properties of this protein crystal δ-endotoxin have stimulated intensive study over the past 30 years leading to the commercial production of the δ-endotoxin and its widespread use as a biological control agent. Based on the antigenic properties of their flagellar H-anti-gens, 19 varieties of B. thuringiensis, belonging to 14 serotypes, have been identified. Thirteen of the serotypes are active against more than 182 species of insect, in particular against pest species of the order Lepidoptera. The fourteenth serotype, B. thuringiensis var israelensis (Goldberg & Margalitt, 1977) is of potential importance since, unlike the other serotypes, it is highly toxic for certain Diptera; notably, mosquitos and blackfly (de Barjac, 1978a). The prospect of employing the crystal δ-endotoxin from this serotype as a biological control agent against these insects, which present a major third-world health problem, has stimulated considerable interest in characterizing the B. thuringiensis var israelensis crystal toxin and discovering the molecular basis of its mode of action.

The biochemistry of B. thuringiensis var israelensis crystal δ-endotoxin has been investigated and compared with that of the crystal from other B. thuringiensis serotypes (Tyrell et al. 1981).

Lepidopteran larvae when fedß. thuringiensis (serotypes l,3a,4a,4b,6,7,8,9) show rapid gut paralysis, sluggish behaviour, cessation of feeding and finally death (Heim-pel & Angus, 1960; Cooksey, 1968). Histological studies (Angus, 1970; Ebersold, Lüthy, Geiser & Ettlinger, 1978; Endo & Nishiitsutsuji-Uwo, 1980) have indicated that the toxin causes gross disruption of the midgut epithelium; cells balloon and lyse with severe disruption of the gut wall. Similarities have been observed between this typical pathology and the symptoms induced by the parasporal crystal δ-endotoxin of B. thuringiensis var israelensis. de Barjac observed that ng amounts of israelensis crystal δ-endotoxin caused cessation of movement and subsequent death of Aedes aegypti larvae (de Barjac, 1978a). de Barjac also presented histopathological evidence that the gut epithelium is the primary target; the gut epithelium showed the typical swelling followed by general tissue lysis (de Barjac, 1978b).

In vitro tissue culture of insect cell lines has previously been used in the investigation of the mode of action of B. thuringiensis serotypes active against Lepidoptera. It was found that cultured insect cells from three types of Lepidoptera responded to enzyme-digested parasporal crystal δ-endotoxin with swelling, vesicle formation and lysis (Murphy, Sohi & Fast, 1976; Geiser, 1979). Similar effects were observed with the Lepidopteran TN-368 cell line (Nishiitsutsuji-Uwo, Endo & Himeno, 1979). As yet there are no reports of the in vitro activity of B. thuringiensis var israelensis in insect cell lines. This paper reports the cytological effects of an alkali-solubilized fraction from the israelensis parasporal crystals on several insect and mammalian cell lines. The results of intravenous and subcutaneous inoculation and oral feeding of the alkali-solubilized crystal δ-endotoxin to Balb.c mice are also described. Results with the israelensis crystal δ-endotoxin are compared with the results of parallel experiments using an alkali soluble fraction from the crystal δ-endotoxin of B. thuringiensis var kurstaki.

B. thuringiensis var israelensis (IPS 78) was obtained from Professor H. de Barjac (Institut Pasteur, Paris). B. thuringiensis var kurstaki (HD1) was obtained from Dr H. D. Burges (Glasshouse Crops Research Institute, England).

Growth and sporulation of both serotypes were as described for Bacillus megaterium KM (Stewart, Johnstone, Hagelberg & Ellar, 1981).

Spores and crystals were harvested by transferring the culture into 500 ml centrifuge bottles on ice and pelleting the spores and crystals at 6000– 7000gfor 5– 10min at 4°C. After centrifugation, sporangia! debris was removed from the pellet by washing with ice-cold deionized water. This was repeated up to six times until spores and crystals were free from debris when viewed by phasecontrast microscopy. The final pellet was resuspended in 50mM-TrisHCl (pH7·5). Spores and crystals were separated using differential ultracentrifugation through a discontinuous sucrose density gradient. A spore/crystal mixture (50 mg) was layered on top of a 30 ml discontinuous sucrose gradient, comprising 10ml each of 67%, 72% and 79% (w/v) sucrose in 50 mM-Tris·HCl (pH 7·5) containing 10 mM-KCl, Centrifugation was carried out in a Beckman L5-50 ultracentrifuge in a SW25·l rotor operating at 80000g 14 h at 4°C. Crystals formed a major band at the interface between the 72% and 79% (w/v) sucrose, while the spores formed a discrete pellet at the bottom of the tube. The crystal band was removed and washed three times in ice-cold 50mM-Tris·HCl (pH 7·5) by centrifugation at 15 000 g for 5 min at 4°C, to remove the sucrose. The final pellet was resuspended in deionized water and stored frozen at −20°C. Purity of the crystal δ-endotoxin preparations was monitored by phase-contrast microscopy and by the production of bacterial colonies on CCY agar plates (Stewart et al. 1981) by serial dilution of crystal preparations.

The method was the same as that for var israelensis except for the following: after harvesting, the final spore/crystal pellet was resuspended in deionized water. Spores and crystals were separated on a discontinuous sucrose gradient of 45%, 67% and 87% (w/v) sucrose in deionized water. The crystals formed a major band at the interface of the 67% and 87% (w/v) sucrose. These were removed from the gradient and washed in ice-cold deionized water to remove sucrose.

Protein concentration was determined by the method of Lowry, Rosebrough, Farr & Randall (1951).

Sodium dodecyl sulphate (SDS)/polyacrylamide gel electrophoresis was conducted as described by Laemmli & Favre (1973), an acrylamide (Fisons)-methylenebisacrylamide (BDH) ratio of 100: 1 being used. Protein samples were completely solubilized by incubation for 5 min at 100°C in SDS/sample buffer (50mM-Tris·HCl (pH7·5), 1% (w/v) SDS, 25 mM-dithiothreitol, 2mM-phenylmethanesulphonyl fluoride, 1 mM-ethylenediaminetetraacetic acid, 10% (w/v) glycerol, 0· 0025% (w/v) Bromophenol Blue). Samples were applied to a 13% gel run at 20mA constant current. The gel was stained overnight at room temperature with 0· 1% (w/v) Coomassie Brilliant Blue R in 50% (v/v) methanol, 10% (v/v) acetic acid and destained with several changes of 10% (v/v) propan-2-ol, 10% (v/v) acetic acid. Molecular weights were estimated by the relation of log M_r to the RF of Sigma SDS/polyacrylamide gel electrophoresis marker proteins.

Crystal δ-endotoxin at 2mg/ml was treated with 50mM-Na₂CO₃·HCl (pH 10·5) by incubation at 37°C for 60 min. Insoluble material remaining was removed by centrifugation at 10000 g for 10 min and the supernatant containing soluble δ-endotoxin material was removed for subsequent bioassay. The pellet was resuspended in phosphate-buffered saline (pH7·0). Protein estimations were carried out on each fraction.

Crystal δ-endotoxin was solubilized by the method of Huber, Lüthy, Ebersold & Cordier (1981).

Aedes albopictus (Dipteran, mosquito adult ovarian tissue) were grown in Mitsuhashi Maramorosch medium (GIBCO) containing 10% foetal calf serum (GIBCO). Choristoneura fumiferana 63 CF1 (Lepidopteran, spruce budworm, trypsinized larval tissue) obtained from Dr S. S. Sohi (Canadian Forest Pest Management Institute, Ontario) were grown in 86% Graces medium (GIBCO) containing 14% foetal calf serum, and 0·25% (w/v) tryptose broth (DIFCO).

Spodoptera frugiperda (Lepidopteran, fall army worm, pupal ovary) and Trichoplusia ni (Lepidopteran, cabbage looper adult ovary) were grown in TC 100 medium (Gardiner & Stockdale, 1975), containing 10% foetal calf serum. Mouse fibroblasts (L929) were obtained from Dr A. A. Newton (Department of Biochemistry, Cambridge). Mouse epithelial carcinoma cell lines (EC2, EC5 and EC6) were obtained from Miss E. Robertson (Department of Genetics, Cambridge). Primary pig lymphocytes were obtained from Miss S. Felber (Department of Biochemistry, Cambridge). All media contained 50μg/ml gentamicin (Nicholas laboratories) and 50μg/ml fungizone (GIBCO).

Cell lines were grown in 25 cm² tissue culture flasks (Nunc), or in 250 ml spinner culture flasks on a Belco magnetic stirrer, at 27 °C. Confluent monolayers were washed once in 5 ml phosphate-buffered saline (PBS), the test sample was applied and then 2 ml medium without foetal calf serum was added. Results were recorded by photography using phase-contrast microscopy and an Olympus OM2 camera attachment.

Prior to the addition of test samples, A. albopictus and C.fumiferana 63 CF1 cells grown in 250 ml spinner culture flasks, were decanted, pelleted by low-speed centrifugation and resuspended at a cell concentration of l– 3(×10⁶)/ml in 5 ml of tissue-culture medium without foetal calf serum. At various time intervals after the addition of the test sample, cell viability was measured by vital staining (0· 1% (w/v) Trypan Blue (Sigma), 0·002% (v/v) glacial acetic acid in PBS) using a haemocytometer counting chamber.

Erythrocytes were washed three times in 50% (v/v) PBS, 2·25% (w/v) glucose by low-speed centrifugation for 5 min in a graduated centrifuge tube. The volume of packed cells was noted and a 01% (v/v) suspension was made in 50% (v/v) PBS, 0·05% (w/v) gelatin, 2·25% (w/v) glucose.

Twofold serial dilutions of test samples were made in round-well microtitre plates (Linbro) in 10 0μl of 5 0% (v/v) PBS, 0· 05% (w/v) gelatin, 2·25% (w/v) glucose. A sample (100μl) of 0· 1% erythrocyte solution was then added to each well. Dilution of the test samples and addition of erythrocytes was carried out either at room temperature or at 4°C. The 4°C plates were left at 4°C for 3 h, after which time they were placed at room temperature. B. thuringiensis var israelensis undiluted test samples titrated in horse, human and sheep erythrocytes were: (1) 20μg of alkali-solubilized crystal δ-endotoxin protein (lμg/μl) in 50mM-Na₂CO₃·HCl (pH 10· 5); (2) 20μg of native crystal δ-endotoxin; (3) 20μg of the insoluble protein remaining after incubation of native crystal δ-endotoxin with 50 mM-Na₂CO₃·HCl (pH10-5); and (4) 20μl of 5OmM-Na₂CO_3HCl (pH 10·5). B. thuringiensis var kurstaki undiluted test samples titrated in horse and human erythrocytes were: (1) 50μg of alkali-solubilized crystal δ-endotoxin protein (Huber et al. 1981), precipitated with saturated ammonium sulphate and resuspended in 50μl of 50mM-Na₂CO₃·HCl (pH 9·5); (2) 50 μg of native crystal δ-endotoxin; (3) 50μg of the insoluble protein remaining after solubilization of the native crystal δ-endotoxin in 50mM-Na2CO3·HCl (pH9·5), 10mM-dithio-threitol; and (4) 50μl of 5OmM-Na₂CO₃·HCl (pH9·5).

All test samples were made up to a final volume of 100 μl with PBS prior to serial dilution in the microtitre plates.

B. megaterium KM protoplasts were prepared as described by Bernheimer & Schwartz (1965).

In vivo toxicity of the B. thuringiensis serotypes for adult Balb.c mice (30g) was tested by administration either intravenously by tail vein inoculation, or by oral feeding, of: (1) 0·lmg, 0·5 mg and 1 ·0 mg of solubilized crystal δ-endotoxin, concentrated by precipitation with saturated ammonium sulphate on ice for 1 h (the precipitates were washed and resuspended in 0·2– 0·4 ml of 50 mM-Na₂CO₃·HCl (pH 10·5); (2) 0·5 mg and l·0mgof native crystal δ-endotoxin in 0·4ml PBS; and (3) an equivalent volume of 50 mM-Na₂CO₃HCl (pH 10·5). Toxicity of lower concentrations of these samples was also tested by subcutaneous inoculation in 1-to 2-week-old suckling Balb.c mice.

The crystalline δ-endotoxin from both serotypes of B. thuringiensis was conveniently separated from spores and purified by differential ultracentrifugation on discontinuous sucrose density gradients as described in Materials and Methods. Residual spore contamination in the purified crystal δ-endotoxin was determined as ≤0·2% by viable colony counts on CCY agar.

Bioassay of the purified crystal δ-endotoxin in A. aegypti 3rd instar larvae (Tyrell, Davidson, Bulla & Ramoska, 1979) gave an LC₅₀; (concentration that will be lethal to 50% of the larvae) of 5– 50 ng protein/ml.

Total solubilization of the crystal δ-endotoxin was achieved by incubation at 100 °C for 5 min in SDS/sample buffer. The polypeptide profile of the crystal δ-endotoxin analysed by SDS/polyacrylamide gel electrophoresis is shown in Fig. 1, track 4. The major israelensis polypeptide had a M_r of 28 000 in agreement with previous reports (Tyrell et al. 1981).

Incubation of the crystal δ-endotoxin with 50mM-Na₂CO₃·HCl (pH 10·5) for 30min at 37°C resulted in the partial solubilization of several crystal components (Fig. 1, track 5) including the 28 000M_r polypeptide. These alkali-soluble polypeptides represented 40% of the total crystal protein. The insoluble protein recovered after centrifugation is represented by Fig. 1, track 6. It is of particular interest that the two high molecular weight polypeptides (M_r 98 000 and 93 500) are recovered exclusively in the pellet.

Bioassay of the purified crystal δ-endotoxin in Pieris brassicae 3rd instar larvae (Cooksey, 1968) gave an LC5o of 10– 100ng protein/ml.

Complete solubilization of the crystal δ-endotoxin was achieved by incubation at 100 °C for 5 min in SDS/sample buffer. Analysis of the total polypeptide profile by SDS/polyacrylamide gel electrophoresis is illustrated in Fig. 1, track 3. The major polypeptide components have M_r values of 126 000 and 63 000. Fig. 1, tracks 1 and 2 represent the soluble and insoluble crystal δ-endotoxin protein obtained after incubation of the native crystal with 50 mM-Na₂CO?·HCl (pH9·5), 10mM-dithio-threitol for 60min at 37°C (Huber et al. 1981). The 126 000 M_r polypeptide was found almost exclusively in the soluble fraction while the majority of the 63 000 M_r polypeptide remained insoluble. The soluble polypeptides represented 90% of the total crystal protein.

In contrast to B. thuringiensis var israelensis, B. thuringiensis var kurstaki was insoluble in the alkaline buffer in the absence of dithiothreitol. It should be noted that the addition of 10mM-dithiothreitol to the alkali-solubilization regime forß. thuringiensis var israelensis resulted in the solubilization of the 98 000 and 93 500 M_r polypeptides (data not shown).

Soluble B. thuringiensis var israelensis crystal δ-endotoxin protein, obtained by extraction with 50mM-Na₂CO?·HCl (pH 10·5) as described above, caused rapid rounding up of A. albopictus cells followed sequentially by swelling (Fig. 2B), bleb-bing of membrane vesicles from the exterior of swollen cells and finally cell lysis. A protein concentration of 5μg/ml caused complete cell lysis in 30-40 min. Higher concentrations of the soluble israelensis crystal protein (25μg/ml) were required to trigger the same cytological sequence (Fig. 2D) in C. fumiferana 63 CF1 cells with essentially complete lysis occurring after 120 min. Using Trypan Blue staining as a measure of cell integrity the effect of these soluble proteins was demonstrable at earlier times. After 3 min and 10 min, 66% and 93%, respectively, of A. albopictus cells (2 × 10⁶/ml) were found to be non-viable after exposure to 5μg soluble israelensis protein/ml (Fig. 3). C. fumiferana 63 CF1 cells (10⁶/ml) exposed to 25 μg soluble protein/ml were 25% and 65% non-viable after 15 min and 47 min, respectively (Fig. 4).

The israelensis crystal ó-endotoxin protein soluble in 50mM-Na₂CO₃·HCl (pH 10’5) triggered a similar sequence of cytological changes in two other insect cell lines, S. frugiperda and T. nt (Fig. 5B,D).

Studies with a variety of mammalian cell types revealed that mouse fibroblasts, primary pig lymphocytes and three mouse epithelial carcinoma cell lines (Fig. 5F) were also susceptible to similar concentrations of the soluble israelensis crystal ô-endotoxin protein. In contrast to these results, protoplasts prepared from the grampositive bacterium B. megaterium KM, when viewed microscopically, appeared to be unaffected by high concentrations (O·4 mg/ml) of the crystal δ-endotoxin protein soluble in 50 mM-Na₂CO₃HCl (pH 10·5).

No in vitro effects were observed when the various insect and mammalian cell types, or B. megaterium KM protoplasts, were exposed to either the same concentration of native crystal δ-endotoxin or the same volume of 50 mM-Na₂CO₃·HCl (pH 10·5) buffer (Fig. 2A,C; Fig. 5A,C,E).

It was found that the soluble B. thuringiensis var kurstaki crystal δ-endotoxin protein prepared by a similar extraction procedure (Huber et al. 1981) was toxic only to C. fumiferana 63 CF1 cells. All other cell types so far described appeared unaffected. The C. fumiferana 63 CF1 cells exhibited the typical sequence of cytological changes; swelling, BLebbing and eventual cell lysis as previously reported by Murphy et al. (1976), Nishiitsutsuji-Uwo et al. (1979) and Geiser (1979). In a typical experiment with C. fumiferana 63 CF1 cells, 55μg/ml of soluble protein caused total cell lysis of 2 × 10⁶ cells/ml in 200 min (Fig. 4).

The soluble crystal δ-endotoxin protein from B. thuringiensis var israelensis prepared by incubation with 5OmM-Na₂CO₃·HCl (pH 10·5) was found to be haemolytic for horse, human, sheep, rat, mouse and rabbit erythrocytes, at room temperature. No haemolytic activity could be detected when the protein was incubated with erythrocytes at 4°C. However, when the microtitre plates, which had been at 4°C, were subsequently incubated at room temperature, haemolysis was observed within 3 h. When the haemolytic activity of the crystal δ-endotoxin protein soluble in 50mM-Na₂CO₃·HCl (pH 10·5) was titrated by twofold serial dilution, using a standardized suspension of horse, human and sheep erythrocytes as previously described, complete haemolysis occurred down to dilutions of 1:16, 1: 64 and 1: 2, respectively. No haemolysis was observed at room temperature when the erythrocytes were challenged with any of the other test samples (see Materials and Methods). None of the test samples of B. thuringiensis var kurstaki caused haemolysis of erythrocytes.

Bioassay of native B. thuringiensis var israelensis crystal δ-endotoxin, and the soluble and insoluble material obtained after incubation of the crystal δ-endotoxin in 50mM-Na₂CO₃·HCl (pH 10·5) in A. aegypti 3rd instar larvae (Tyrell et al. 1979) resulted in an LC50 of 5–50ng protein/ml, 0·6–6μg protein/ml and 50–500ng protein/ml, respectively. P. brassicae 4th instar larvae fed these samples per os (2 μg protein/larvae) showed no inhibition of feeding and no other apparent ill effects (Table 1).

The results of experiments with Balb.c mice are summarized in Table 1. In a series of preliminary toxicity tests with adult and suckling Balb.c mice the nativeB. thurin-giensis var israelensis and var kurstaki crystal δ-endotoxin s produced no pathological effects when introduced by intravenous inoculation, or when fed per os. Although the israelensis crystal δ-endotoxin protein soluble in 50 mM-Na₂CO₃·HCl (pH 10·5) had no effect when fed per os to adult mice (0·5 mg protein/animal), or suckling mice (200 μg protein/animal) it was found to be toxic when introduced by intravenous inoculation. Adult Balb.c mice (average body weight 30g) were killed within 12 h by intravenous inoculation of l·Omg of israelensis soluble crystal δ-endotoxin protein. Within 1 h of injection animals developed paralysis in their hind-quarters and became relatively immobile within 3 h. Their breathing and heart-rate also increased. Animals inoculated intravenously with 0·5 mg of soluble protein developed similar paralysis in their hind-quarters, but those animals that died, did so after 36–48 h. Although none of the animals inoculated intravenously with 04 mg of soluble protein died, three animals developed paralysis in their left hind-leg and a further three showed a stiffening of movement in their hind-quarters. Suckling mice (average body weight 5·7g) were killed within 3 h by subcutaneous inoculations of 80—200μg protein/animal. A severe haematoma was observed in the area around the site of injection. This effect was not observed in animals receiving control injections.

In contrast to these results, nativeB. thuringiensis var kurstaki crystal δ-endotoxin and soluble protein prepared by incubation of B. thuringiensis var kurstaki crystal δ-endotoxin in 50 mM-Na₂CO₃ ·HC1 (pH9·5), 10mM-dithiothreitol produced no pathology in adult Balb.c mice (10mg/animal), or suckling Balb.c mice (200 μg/animal) by intravenous, or subcutaneous inoculation, or by feeding per os.

The molecular basis of the insecticidal activity of the crystal δ-endotoxin produced by B. thuringiensis serotypes is not known. Previous studies of B. thuringiensis serotypes active against Lepidoptera have demonstrated a characteristic disruptive effect on insect gut epithelium in vivo (Angus, 1970; Ebersold et al. 1980; Endo & Nishiitsutsuji-Uwo, 1980). The native crystal δ-endotoxin is non-toxic until ingested by susceptible insects. Pathogenicity appears to involve dissociation of the crystal δ-endotoxin in the alkaline conditions of the insect gut and/or proteolytic modification by gut enzymes to yield active toxin (Faust, Hallam & Travers, 1974; Lecadet, 1970). Activation of the crystal δ-endotoxin has been achieved in vitro by incubation of the crystals with gut proteases (Lecadet & Dedonder, 1966; Lecadet & Martouret, 1967). Crystal δ-endotoxin from B. thuringiensis var kurstaki and var aizawai, activated in this way, produced characteristic cytopathological effects on certain insect cell lines in vitro (Murphy et al. 1976; Nishiitsutsuji-Uwo et al. 1979; Johnson, 1981). These and other data suggest that the plasma membrane of susceptible cells may be the primary target for the δ-endotoxin (Fast, Sohi & Murphy, 1978). Subsequent cell lysis may be the result of specific inhibition of certain membrane proteins responsible for the maintenance of the correct K⁺ balance across the midgut membrane (Griego, Moffet & Spence, 1979; Harvey & Wolfersberger, 1979; Cioffi & Harvey, 1981). Alternatively, the crystal δ-endotoxin may resemble the cytolytic bacterial toxins such as Staphylococcus aureus ő-toxin in its mode of action (Theles-tam & Moellby, 1975a,b).

The recently isolated israelensis strain of B. thuringiensis has excited considerable interest because of its pathogenicity for the larvae of certain Diptera including the genera Aedes and Culex (Goldberg & Margalitt, 1977), which are vectors of malaria and filariasis, respectively. In a study of the in vivo effects of the israelensis crystal δ-endotoxin on mosquito larvae, de Barjac (1978b) demonstrated the typical swelling and lysis of gut cells that characterises the action of crystal δ-endotoxin from other B. thuringiensis serotypes on Lepidopteran gut cells. To our knowledge the present report is however, the first description of an investigation of the in vitro cytopathological and haemolytic effects of the israelensis crystal δ-endotoxin.

It has been reported that the varieties of B. thuringiensis crystal δ-endotoxin can be solubilized by alkaline buffer (pH > 12) (Bulla et al. 1981), or a combination of alkaline buffer (pH 9–10) and reducing conditions (Huber et al. 1981). Crystal δ-endotoxin preparations solubilized by these methods retained in vivo toxicity, but to date in vitro toxicity of these soluble preparations on isolated cell cultures has not been observed unless the alkali-treated toxin is additionally exposed to proteolytic enzymes (Fast, 1981). A similar alkaline regime was developed and used in the present study to solubilize B. thuringiensis var israelensis crystal δ-endotoxin. By incubating israelensis crystal δ-endotoxin with 50mM-Na₂CO₃·HCl (pH 10·5) for 30min at 37°C, 40% of the crystal protein was extracted in a soluble form, which retained toxicity in vivo and in vitro. Analysis of the solubilized polypeptides by SDS/polyacrylamide gel electrophoresis (Fig. 1, track 5) revealed that a substantial amount of the major israelensis crystal δ-endotoxin polypeptide (M_r 28 000) was extracted by this method. By contrast, the two high molecular weight polypeptides (M_r 98 000 and 93 500) remained completely insoluble. It is interesting to note certain similarities between the crystal δ-endotoxin and the granulosis and nuclear polyhedrosis insect viruses. In both cases the primary site of action is the insect gut epithelium. Summers & Egawa (1973) reported the solubilization of a protein (M_t 180 000) from the crystalline proteinaceous structure that surrounds the granulosis virus of T. ni using a 70mM-sodium carbonate buffer (pH 10·7). Also, Eppstein, Thoma, Scott & Young (1975) reported that the crystalline matrix protein (M_r 28 000) of the nuclear polyhedrosis virus of T. ni could be solubilized by incubation with 100mM-sodium carbonate buffer (pH 11). For these occluded viruses, therefore, the alkaline insect gut environment is also thought to be a factor in the solubilization of viral proteins, and the subsequent release of infective particles into the gut during the primary infection (Harrap, 1970). However, these proteins do not appear to have the characteristic effect on gut cells seen with B. thuringiensis.

The israelensis crystal δ-endotoxin soluble in 50mM-Na₂CO?·HCl (pH 10·5) retained toxicity for mosquito larvae and produced rapid cytopathological changes in all the insect cell lines examined. There was clearly a difference between the LC₅₀ for mosquitos of intact israelensis crystal δ-endotoxin and the soluble crystal protein. This may reflect the fact that the larvae are filter feeders and that the particulate crystals are more efficiently ingested than the alkali-solubilized toxin. Alternatively, the in vitro alkali-mediated solubilization/activation regime used in this study may only partially simulate the in vivo activation, which requires a combination of the high pH of the insect gut and proteolytic cleavage by gut proteases.

The observed cytopathology paralleled that observed by several authors using protease-activated crystal δ-endotoxin from otherß. thuringiensis serotypes (Murphy et al. 1976; Geiser, 1979; Nishiitsutsuji-Uwo, Endo & Himeno, 1980). The latter authors reported no cytopathic effect of the activated δ-endotoxin on mammalian cells. However, Prasad & Shetna, (1976) and Seki et al. (1978) have reported cytopathological effects of B. thuringiensis var thuringiensis crystal δ-endotoxin on the Yoshida ascites sarcoma tumour line maintained in Wistar rats and sarcoma 180 ascites cells. It is significant, therefore, that in the present study the alkali-soluble israelensis δ-endotoxin was lethal to the various mammalian cell lines examined. In addition this δ-endotoxin showed haemolytic activity against a wide range of erythrocytes. Interestingly, only one cell type, viz. bacterial protoplasts (derived from an organism unrelated to the producer organism), failed to show the typical cytopathic response to the soluble israelensis crystal δ-endotoxin.

In parallel studies with the crystal δ-endotoxin from B. thuringiensis var kurstaki it was found that a reducing agent was additionally required to achieve solubilization (90%) in 50mM-Na₂CO₃·HCl (pH9·5). In this respect there are again similarities in the conditions required for solubilization of the crystalline bacterial toxin and the inclusion-body protein of the Entomopox insect viruses, which also requires a reducing agent in addition to dilute alkali for dissolution (Vaughn, 1974). The solubilized preparation retained its toxicity to Lepidoptera in vivo and produced the typical cytological effect on C. fumiferana 63 CF1 cells in vitro. However, it was without effect on mosquito-derived cells, mammalian cells, bacterial protoplasts, or the other Lepidopteran cell lines. Moreover, it showed no haemolytic activity against the range of erythrocytes. It would appear that the in vitro activity of the solubilized kurstaki crystal δ-endotoxin is restricted in its host range to cells derived from susceptible insects. In contrast, the data for the israelensis crystal δ-endotoxin indicate that provided it is rendered soluble by incubation with 50 mM-Na₂CO₃·HCl (pH 10·5) it is toxic not only to mosquito cells, but also to a wide variety of cell types outside its in vivo host range, including mammalian cells and erythrocytes.

The failure of the israelensis toxin to affect bacterial protoplasts, or to kill 4th instar P. brassicae larvae, suggests that the bacterial plasma membrane and the Lepidopteran gut epithelium may lack the target/receptor for the soluble δ-endotoxin. Alternatively, the resistance of the Lepidoptera may reflect some inhibitory effect of the larval gut contents on the toxin.

The above differences in the spectrum of activity of israelensis and kurstaki alkali-solubilized crystal δ-endotoxin s were reinforced by the results of mammalian in vivo toxicity tests. The native crystal δ-endotoxin from either israelensis or kurstaki produced no ill effects in Balb.c mice when administered per os, subcutaneously, or intravenously. Similarly the solubilized kurstaki crystal δ-endotoxin was also nontoxic when administered by these routes. Although the soluble israelensis crystal δ-endotoxin was without effect when administered per os, it proved to be extremely toxic when inoculated intravenously into Balb.c mice, or subcutaneously into suckling Balb.c mice, at doses of 15–35 μg soluble protein/g body weight.

Although the 28 000M_r polypeptide is the major component of the israelensis ô-endotoxin protein soluble in 50mM-Na₂CO₃·HCl (pH 10·5), we cannot at present correlate the observed insect and mammalian cell toxicity and haemolytic activity solely with this component. It is possible that the different lytic activities may reside on different polypeptides. Further purification of this soluble preparation is being carried out to resolve this question.

In view of the obvious potential of B. thuringiensis var israelensis as a biological control agent it is important to emphasize that although toxic effects resulted from inoculation of the alkali-soluble israelensis δ-endotoxin into mice, the same preparations produced no observable ill effects when administered orally. This important observation is verified by the absence of in vivo toxicity of this variety in oral tests with a large number of non-target animals, in which only groups very closely related to mosquitos (e.g. chironomids) were affected (Burges, 1982). The native crystal ő-endotoxin and spores have also been injected parenterally into many test animal species without effect. It would be of value to examine the effect of parenteral application of the soluble toxin of more strains on a range of animal species.

The broad spectrum of toxicity shown by the israelensis toxin clearly distinguishes it from the kurstaki toxin and suggests that the molecular mechanisms of the two toxins in Lepidoptera and Diptera may be quite different. It will now be important to determine to what extent the other B. thuringiensis strains, reported to be pathogenic for Diptera, display the same spectrum of toxicity observed here for var israelensis.

This work was supported by a grant from the Agricultural Research Council. We would like to thank Dr H. D. Burges for his helpful discussion. Also, Mr D. Funnell (Shell Research Limited) for the A. aegypti eggs, Mr B. Gardiner for the P. brassicae larvae, Mr L. Jewitt for his assistance with the figures and Mrs L. Futter and Mrs C. Dean for secretarial assistance.