Isolation and identification of Bacillus sphaericus strains pathogenic for mosquito larvae

JOURNAL

OF INVERTEBRATE

PATHOLOGY

isolation and Identification

57,

325-333 (191)

of Bacillus sphaericus for Mosquito Larvae

Strains Pathogenic

MAGALI GUERINEAU, BRIAN ALEXANDER,’ AND FERGUSG. PRIEST Department of Biological Sciences, Heriot Watt University, Edinburgh EH14 4AS, Scotland, United Kingdom ReceivedApril 16, 1990; accepted June 20, 1990 Three selective media for the isolation of Bacillus sphaericus have been compared. BATS medium and a formulation employing adenosine as the principal carbon source were the most effective for the recovery of spores of strain 1593. Anthramlic acid as the principal carbon source was less efftcient. Eighty-four strains were isolated from mud samples using these media and were identified by computer. Identitications were confirmed for representative strains using DNA sequence homology. Most were B. sphaericus sensu strict0 or members of an unnamed group. However, one strain (BSE 18) was identified as the DNA homology group IIB and this organism was found to be highly toxic toward larvae of Culex pipiens. Southern hybridization of BSE 18 DNA to a probe prepared from the cloned toxin gene from strain 1593 revealed that BSE 18 contained a typical gene for the 41.9-kDa toxin. Q 1%~Academic press, IN. KEY WORDS: Bacillus sphaericus; Culex pipiens.

Some strains of Bacillus sphaericus are toxic toward selected mosquito larvae, particularly certain Culex, Anopheles, and Mansonia species. These highly larvicidal strains synthesize a crystal during sporulation which contains a toxic protein of 41.9 kDa (Baumann et al., 1987, 1988). B. sphaericus has the advantage over B. thuringiensis var. israelensis, the most popular microbiological mosquito control agent, in that it appears to persist in the environment longer than Bacillus thuringiensis var. israelensis and thus can establish a longer lasting control of larval populations. Largely for this reason, B. sphaericus is being evaluated as a mosquito control agent by the WHO (World Health Organization, 1985). DNA sequence homology studies (Krych et al., 1980) and numerical analysis of phenotypic features (Alexander and Priest, 1990) have demonstrated that strains of Bacillus which differentiate into spherical spores (the traditional description of B. sphaericus) should be classified into at least six species. Within these six taxa, strains ’ Present address: Inveresk International Research Ltd, Inveresk Gate, Musselburgh, East Lothian, UK.

pathogenic for mosquito larvae have been allocated to a discrete group by DNA sequence homology (group IIA) (Krych et al., 1980) and by numerical taxonomy of phenotypic features (phenon 3a) (Alexander and Priest, 1990). This is highly related to the DNA homology group IIB (phenon 3b; Bacillus fusiformis) which until now was not considered to contain insect pathogenic strains. It would seem that these taxa represent distinct species, although official status has not yet been sought. There is a need for a simple scheme for the accurate identification of environmental isolates of round-spored bacilli from environmental samples which distinguishes the insect pathogens from the nonpathogenic taxa. This would enable potential insect pathogens from selective isolation plates to be screened, and only those strains belonging to the DNA homology group II need be subject to the expensive and time-consuming procedure of toxicity testing. In this study we tested two computerized identification schemes, one based on 27 phenotypic tests and a second which uses only 13 tests. Eighty-four strains were isolated using various selective media designed for the recovery of insect pathogenic strains of B. 325 0022-201 l/91 $1 SO Copyright 0 1991 by Academic Press, Inc. All rights of reproduction in any form resewed.

326

GUERINEAU,

ALEXANDER,

and were identified by computer. One strain was identified as being closely related to the insect pathogenic taxon (phenon 3b; DNA group IIB). Feeding trials revealed it to be highly toxic toward larvae of Culex pipiens. AND METHODS

Bacterial strains and plasmids. Reference strains, their characteristics, and their sources are listed in Table 1; all other strains were isolated during the study as described below. Rifampicin-resistant (Ril’) mutants were spontaneous mutants isolated on nutrient agar containing 25 pg/ml rifampicin (Sigma). Media. BATS medium was prepared as recommended (Yousten et al., 1985). Adenosine and anthranilate media were based TABLE STRAINS

DNA homology Strain

grOUP”

Phenonb

DSM 28

I

1

BS 30 BS 100 1593 SSII-1 2297 WHO 1883 WHO 2013.6

IIA IIA IIA IIA HA

3a 3a 3a 3a 3a

IIA IIA IIA IIA IIA IIA

3a 3a 3a 3a 3a 3a

IIB

3b

IV

3c 3c 6 10

WHO WHO WHO WHO WHO WHO

2115 2173 2314.2 2315 2377 2500

ATCC 7055 BS 126 BS 127 NRS 400 BS 82

USED

Phage type

PRIEST

on the minimal medium of Anagnostopou10s and Spizizen (1962) supplemented with yeast extract (1 g/liter), casein hydrolysate (2 g/liter), streptomycin (100 kg/ml), and adenosine or anthranilate (4 g/liter). Isolation of strains. Soil and mud samples (1 g) were suspended in sterile water (5 ml) and incubated at 80°C for 10 min. Aliquots were either plated directly onto selective media after appropriate dilution in sterile physiological saline or 1 ml was inoculated into 25 ml of nutrient broth and shaken at 30°C for 1 hr (pregermination) before plating onto selective media. Plates were incubated at 30°C for up to 5 days. Identification of isolates. Bacteria were identified using a computerized scheme exactly as described previously (Alexander and Priest, 1990). Both 13- and 27-test iden-

sphaericus

MATERIALS

AND

1

IN THIS

STUDY

Serotype

5a 5b 2 25 2 5a 5b 6 NTd NT NT

27 26a 26b 9 26a 26b

3

5a 5b

Comments

Source’

Type strain of Bacillus sphaericus

1

High toxicity strain Low toxicity strain Contains the 41 .PkDa toxin gene

Contains the 41.9-kDa toxin gene Type strain of B. fusiformis

a DNA homology group according to Krych et al. (1980). * Phenon according to Alexander and Priest (1990). c Sources: 1, Deutsche Sammhmg von Microorganismen; 2, Alexander and Priest (1990); 3, A. A. Yousten, Department of Biology, Virginia Polytechnic Institute and State University, Blacksburg, Virginia. d Nontypeable.

IDENTIFICATION

@cation matrices were used and two coefficients of identification were computed using the MATIDEN program (Sneath, 1979): Wilcox probability (>0.99 is an acceptable identification) and standard error (SE) of taxonomic distance (0) which should be less than 7 for an accurate identification (for details see Priest and Alexander, 1988). DNA hybridization. Total DNA was prepared and sequence homology was determined using random oligonucleotideprimed, ‘*P-labeled probe DNA as described previously (Alexander and Priest, 1989). For Southern hybridization, restriction enzymes were purchased from NBL (Cramlington, Northumberland, UK) or Boehringer and used according to the manufacturers’ instructions. DNA (6 p,g) was electrophoresed in 0.7% agarose gels and transferred to nitrocellulose by standard procedures (Maniatis et al., 1982). Probe DNA was derived from plasmid pJC 1593-1 which contains the 41.9-kDa toxin gene from B. sphaericus 1593 as a 1.5kb Sau3AI fragment ligated into the BamHI site of pUC12 (Hindley and Berry, 1987). Plasmid was prepared by standard procedures (Maniatis et al., 1982), purified on PZ 523 columns (NBL Ltd), digested with &I, and electrophoresed in low-melting temperature agarose (Sigma; 0.8%). Two bands, one of 3.2 kb and the second of 1.O kb, were observed and the latter, containing about 1 pg DNA, was removed from the gel and dissolved in boiling water (5 X volume). This material was used directly for oligonucleotide-primed labeling (Feinberg and Vogelstein, 1983). Typical incorporation was about 5 x lo6 cpm/pg DNA. Hybridization was at 25°C below Tm. RESULTS Evaluation of selective media. BATS medium was developed for the selective isolation of insect pathogenic strains of B. sphaericus. This uses arginine as a carbon and nitrogen source but our previous studies suggested that adenosine or anthranilic acid may be more highly selective (Carboulec and Priest, 1989). We therefore

327

OF Bacillus sphaericus

compared two media formulated from these carbon sources with BATS medium. Initially, we prepared a spore suspension of a rifampicin-resistant mutant of strain 1593 and seeded this into five different soil samples at a rate of approximately lo3 spores/g. The soils were suspended in water, heated to 80°C for 10 min, diluted, and plated onto the three selective media. Colonies were picked at random onto rifampicin plates (between 50-100 per sample) and examined for growth (Table 2). BATS medium consistently recovered high numbers of 1593 Rif colonies, while anthranilate was less effective. Adenosine was almost as efficient as BATS. It seemed likely that one reason for the relative poor recovery of strain 1593 on anthranilate and adenosine media was due to poor germination of spores while it was established that BATS medium allowed germination of strains 1593 and 2362 (Yousten et al., 1985). Comparative counts of spore preparations of 1593, SSII-1, and 2297 with and without pregermination in nutrient broth (see Materials and Methods) showed that BATS medium did indeed support spore germination (counts were about the same with and without pregermination for all three strains) while anthranilate medium showed two times (strain 1593), four times (strain SSII-l), and six times (strain 2297) the number of colonies after pregermination. Adenosine medium showed 1.3 (strain 1593) to 2 times the number of colonies after pregermination. Thus the low

RECOVERY

Soil sample 1 2 3 4 5 Mean

OF

TABLE 2 SPORESOF Bacillus sphaericus 1593 Rif’ ADDED TO SOIL Medium

BATS 86 100 95 72 83 87

Adenosine 80 76 90 63 75 77

Anthranilate 45 38 40 20 20 33

Note. Figures are percentage recovery of Rif’ colonies from isolation media. For details see text.

328

GUERINEAU,

ALEXANDER,

recovery of the Rif strains on anthranilate medium was probably due to poor germination and in future experiments spores were pregerminated before use of this medium. Isolation of strains. Twelve mud samples were gathered from various lakes or rivers in East Scotland and bacteria were isolated using the three media with and without pregermination (Table 3). Random microscopic examination of 50 colonies from each medium showed that all were roundspore formers from the BATS and anthranilate media but that only 70% from adenosine plates formed round spores, with the remainder showing oval spores. Very different colony counts and morphologies were apparent using the three media and depending on pregermination. Eighty-four colonies were picked from the various media and purified by replating on nutrient agar . Zdentijication of strains. The 84 strains were examined for 29 phenotypic characters and identified by computer using two TABLE ISOLATION

AND

AND

PRIEST

related identification matrices. The large matrix included 14 “species” identified by 27 tests and the small matrix comprised 7 species recognized by just 13 tests (Alexander and Priest, 1989). An identification was acceptable if the identification score (Wilcox probability) was greater than 0.99 (which reflects a maximum error rate of 1 in 100 strains) and if the standard error of taxonomic distance (SED) was less than 7 (Priest and Alexander, 1988). Using these stringent criteria we could identify 46 (55%) of the isolates using the large matrix (Table 3). Most (23 of the identified strains) were identified as B. sphaericus sensu strict0 (DNA homology group I) with phenon 10 (an undescribed species), the next most frequent with 17 strains. Bacteria recovered from adenosine medium were also identified to B. fusiformis (phenon 3b; DNA homology group IIB), phenon 3c, and phenon 6 (DNA homology group IV). There was no correlation between medium and species isolated. 3

IDENTIFICATION

OF STRAINS

Number of strains with acceptable identification scoresb Medium Adenosine medium

Adenosine medium with pregermination Anthranilate medium with pregermination BATS medium

Number of strains 30

24 4 26

Identification” Bacillus sphaericus’ B. fusiformis Phenon 3c Phenon 6 Phenon 10 Unidentified B. sphaericus Phenon 10 Unidentified B. sphaericus Phenon 10 Unidentified B. sphaericus Phenon 10 Unidentified

Large matrix 7 (23)d 1 (3) 3 (10)

2 G-9 2 (6) 15 (50) 8 (33)

Small matrix 6 1 3 2 18 7

6 (25) 10 WY 1 (25) 2 (50) 1 (25) 7 (27) 7 (27) 12(46)

17 1 3 5 21

n Phenons as described by Alexander and Priest (1990). b Acceptable identification scores are Wilcox probability >0.99 and standard error of taxonomic distance <7. ’ B. sphaericus is B. sphaericus sense stricto, DNA homology group I. d Numbers in parentheses are percentages of strains isolated for each medium.

IDENTIFICATION

OF

The small matrix performed well. There was an increase in unidentified strains due to phenon 10 not featuring in this matrix and these strains could therefore not be identified, but all other strains identified to the appropriate phenon. DNA sequence homology. We confirmed some of the computer identifications using DNA sequence homology (Table 4). We also included eight reference insect pathogenic strains which we had previously identified by computer as phenon 3a (equivalent to DNA homology group IIA) (Alexander and Priest, 1990). Hybridizations were per-

DNA

329

Bacillus sphaericus

formed under optimum (25°C below Tm) and stringent (15’C below Tm) conditions. We considered an identification to be >60% sequence homology to one taxon and 60% at optimum and stringent temperatures) to the two reference strains: 1593, a highly toxic

TABLE 4 REASS~CIATIONBE~EENREFERENCESTRAINSANDSOMEENVIRONMENTALISOLATES

Strain DNA 1 BSESO DSM 28 Phenon 2 BS 30 BS 100 DNA IIA WHO 1883 WHO 2013.6 WHO 2115 WHO 2173 WHO 2314.2 WHO 2315 WHO 2377 WHO 2500 1593 SSII-1 DNA IIB BSE 18 ATCC 7055 Phenon 3c BS 126 BS 127 BS 6 DNA IV BSE 8 NRS 400 Phenon 10 BSE 81 BS 82

A. hydrophila

DSM 28

1593

1593”

SSII-1

SSII-1”

ATCC 7055

ATCC” 7055

NRS 400

BS127

BS82

96 100

29 22

32 30

31 38

30 30

30 25

32 18

33 40

12 15

14 19

18 17

24 28

37 22

21 21

34 28

38 24

25 21

32 30

12 15

11 15

31 27 35 36 24 20 19 11 34 26

59 64 63 68 73 79 71 62 100 81

62 78 77 92 83 44 75 69 100 77

81 84 63 73 63 66 79 69 69 100

64 70 100 64 76 100 85 88 82 100

65 60 75 55 60 56 66 56 70 68

47 40 43 37 48 64 43 50 47 54

34 22 33 27 29 33 35 29 28 36

12 12 15 13 10 14 13 11 13 13

13 15 17 26 22 17 16 17 16 18

20 17

59 55

57 42

46 42

36 49

74 100

80 100

33 36

17 13

16 19

23 26 23

35 28 24

13 11 31

28 35 35

18 38 28

32 37 27

20 24 22

34 31 23

82 100 61

12 15 14

10 36

30 35

31 24

24 31

42 10

35 8

44 6

70 100

12 15

12 19

17 15 15

10 18 15

24 18 18

21 19 9

12 28 3

35 26 20

3 34 2

31 29 17

15 12 14

32 100 15

Note. All figures are percentage reassociation and represent means of four deviations of the means were generally less than 7. D Reassociations performed at stringent (15’C below Tm) temperature.

separate

determinations.

Standard

330

GUERINEAU,

ALEXANDER.

strain, and SSII-I, a tow toxicity strain and relatively low homology to reference DNA from group IIB. In the few examples where homology to group IIB was higher than might be expected, the values were much lower at stringent temperature, suggesting that poorly matched hybrids were responsible. Similarly, the reference strain of group IIB and the one environmental isolate (BSE 18) which identified to this phenon with a Wilcox probability of 0.999 and a SED score of 5.3 showed low sequence homology to reference strains from group IIA and high relatedness to group IIB. Phenon 3c reference strains and the isolate BSE 6, which identified with a Wilcox probability of 0.999 and a SED of 3.6, formed a tight DNA homology group. Similarly, strain BSE 8, the isolate identified as phenon 6 (DNA homology group IV) with a Wilcox probability of 0.999 and a SED of 3.4, showed high sequence homology to the reference strain of this taxon. Finally, strain BSE 81 was identified as phenon 10 with scores of 1 (Wilcox probability) and 0.8 (SED) and also showed exclusive sequence homology with the reference strain of this taxon. Detection of toxin genes. Strain BSE 18, which identified to B. fusiformis, and the reference strains of DNA group IIA were examined for the presence of toxin genes which would hybridize to a probe derived from the cloned toxin gene from strain 1593. DNA from the reference strains of the insect pathogens, strain SSII-1 as a negative control, 1593 as a positive control, and two group IIB strains, BSE 18 and ATCC 7055, were digested with EcoRI and electrophoresed in agarose gels. Southern blots were probed with a 1-kb CluI fragment representing most of the 41.PkDa toxin gene and part of the upstream open reading frame (see Hindley and Berry, 1987) and autoradiographed. Hybridization was observed with DNA from strains 1593, WHO 2500, WHO 2013.6, and BSE 18 only (Fig. 1) in the form of a band corresponding to a fragment of 4.4 kb. DNA from strains SSII-

AND

PRIEST

00 w z kb 4.4

0 m 1. Southern blots of EcoRI-digested DNA from various strains of the DNA homology group IIA (phenon 3a) hybridized with a probe encoding the 41.9kDa toxin gene of strain 1593. FIG.

1, WHO 2377, WHO 2314.2, WHO 2315, WHO 2173, WHO 2115, WHO 1183, and ATCC 7055 did not show detectable hybridization. Hind111 digests of DNA from the positive strains were probed with the same fragment and showed hybridization of a 3.5-kb band (Fig. 2). Insect toxicity and other characteristics of strain BSE 18. Strain BSE 18 is unusual because it is the only strain isolated from a temperate environment that shows the presence of a toxin gene. Moreover, it is unique in being a toxin-producing member of the DNA homology group IIB. The strain was allocated to phage group 3 by A. A. Yousten and serotype 5a Sb by H. de Batjac. Larval feeding trials showed that it was highly toxic to fourth instar larvae of C. pipiens with a LC& of 1.2 x 10e6 of final whole culture equivalent to about 1 t&ml. DISCUSSION

There is a need for efficient selective media for the recovery of insect pathogenic

kb 4.4 3.5

-

--

--

--

FIG. 2. Southern blots of EcoRI- and HindIIIdigested DNA hybridized with the 41.9~kDa toxin gene probe.

IDENTIFICATION

OF Bacillus

sphaericus

331

other than B. sphaericus sensu strict0 or strains of B. sphaericus from the environment, thus aiding the monitoring of added phenon 10 did so using both matrices. Thus control agents and the isolation of new the small matrix provides an ideal procestrains. BATS medium is effective in this dure for preliminary identification of interrespect but the arginine supplied as carbon esting strains isolated during screening proand nitrogen source permits the growth of grams. To make it more useful in this restrains of several of the B. sphaericus ho- spect, we are currently expanding the small mology groups (Yousten et al., 1985). Ade- matrix to include all taxa described in our nine (here used as adenosine) should be numerical classification (Alexander and more selective and should restrict the Priest, 1990) and would suggest that strains growth of strains of homology groups I, III, are examined initially using this matrix. and V and anthranilic acid does not permit Subsequently, if a more exact identification the growth of homology groups IIB and IV is required, the additional tests for the large (Carboulec and Priest, 1989). These should matrix should be performed. We believe therefore provide some advantages over that these matrices offer a very powerful BATS medium. However, BATS proved to procedure for preliminary characterization support the best growth of B. sphaericus of isolates in screening programs. Those strains in general, and anthranilate was dif- strains that identify to phenon 3 (DNA ficult to use being poorly soluble. Moregroup II) can be readily identified and exover, it permitted only weak growth and amined for larval toxicity. germination of spores was poor necessitatEnvironmental isolates that were identiing pregermination of samples. We con- lied were almost exclusively confined to B. clude that BATS medium is the best general sphaericus sensu strict0 and phenon 10. Inpurpose selective medium but that replacclusion of streptomycin in the selective meing arginine by adenosine may have advan- dia will prevent recovery of DNA homoltages in certain instances. Although B. ogy group III strains which are streptomysphaericus strains can use adenosine as a cin sensitive (Alexander and Priest, 1990), carbon or nitrogen source, it cannot be uti- and the lack of representatives from groups lized as both a carbon and nitrogen source II, IV and V presumably reflects the scarso it is essential to include a nitrogen source city of these bacteria in Southern Scotland. Generally, probabilistic identifications in the medium when using this substrate. Identification of the environmental iso- are accepted as accurate, but in view of the difficulties inherent in identifying roundlates was successful with the majority (55%) allocated to a group using the strinspored bacilli (Gordon et al., 1973), we gent identification scores of Wilcox proba- evaluated the phenotypic identifications usbility >0.99 and SED <7 with the 27-test ing DNA sequence homology. In all inmatrix. This rate compares favorably with stances, DNA sequence homology conother probabilistic identification schemes firmed the computer identifications (Table such as those for mycobacteria (Wayne et 4) and was consistent with the previously al., 1980), Streptomyces (Williams et al., defined DNA homology groups (Krych et al., 1980). Not only does this fully validate 1983), and Bacillus (Priest and Alexander, 1988). Nevertheless, we were prepared to the accuracy of our phenotypic identiticaforego some accuracy in favor of a more tions but it also offers some additional inrapid identification scheme, and the 13-test sight into the systematics of these bacteria. matrix performed remarkably well. The Phenotypic identification of strains to pheonly strains it failed to identify were those non 3c, phenon 6 (DNA group IV), and pheDNA seof phenon 10 for which the data were not non 10 and high, intra-phenon quence homology suggests that these taxa included in the matrix. Of particular importance, all strains which identified to taxa should be raised to species status. The sit-

332

GUERINEAU,

ALEXANDER,

uation, however, is still not clear regarding DNA homology subgroups IIA and IIB. Phenotypic identifications to these taxa performed here and previously (Alexander and Priest, 1990) are accurate and were validated by DNA homology, but there are few consistent characters that separate the groups. Moreover, DNA sequence homology between strains from the two subgroups was sometimes high (>60%), although this was generally reduced at the more stringent reassociation temperature. We suggest at this stage that they should be considered as varieties or subspecies of a single species until more data become available. Of particular importance will be the distribution of toxicity among IIA and IIB strains and the involvement, if any, of plasmids (Singer, 1988). The isolation of strain BSE 18 is the first example of a high toxicity strain isolated from a temperate climate in which the mosquito distribution will be very different from tropical environments. It is therefore interesting that it should be a group IIB strain isolated, and that it synthesizes a typical 41.9-kDa toxin. Toxicity toward fourth instar larvae of C. pipiens was similar to other high toxicity strains (de Bajac et al., 1988) and its allocation to serological group 5a 5b is consistent with its pathogenicity. A strain of group IIB (NRS 1192), which is nontoxic, and strain BSE 18 have both been allocated to phage group 3 (A. A. Yousten, pers. commun.), the group which contains most of the highly toxic B. sphaericus strains (Yousten, 1984). It may be that as we study more strains the correspondence between toxicity, the DNA homology subgroup, the phage group, and serotype will begin to consolidate or alternatively we may need to devise new criteria for classifying these bacteria. ACKNOWLEDGMENTS This work was supported by the Deutsche Sammlung von Microorganismen. M.G. thanks the ERASMUS Program of the European Community for financial assistance. We thank A. A. Yousten for pro-

AND PRIEST

vision of strains, for phage typing strain BSE 18, and toxicity testing of this strain. We also thank H. de Bajac for serotyping and toxicity testing of strain BSE 18 and C. Berry for kindly providing the cloned toxin gene.

REFERENCES C., AND SPIZIZEN, J. l%l. Requirements for transformation in Bacillus subtilis. J. Bacterial., 81, 741-746. ALEXANDER, B., AND PRIEST, F. G. 1989. Bacillus glucanolyticus, a new species that degrades a variety of p&cans. Znt. J. Syst. Bacterial., 39, 112115. ALEXANDER, B., AND PRIEST, F. G. 1990. Numerical classification and identification of Bacillus sphaericus including some strains pathogenic for mosquito larvae. J. Gen. Microbial., 136, 367-376. ANAGNOSTO~~ULOS,

DE BARJAC, H., THIERY, I., COSMAO DUMANOIR, V., FRACHON, E., LAURENT, PH., OFORI, J., CHARLES, J.-F., AND HAMON, S. 1988. Another Bacillus sphaericus serotype harbouring strains very toxic to mosquito larvae: serotype H6. Ann. Microbial.,

139, 363-377. BAUMANN, P., BAUMANN, AND BROADWELL, A. H.

L.,

BOWDITCH,

R. D.,

1987. Cloning of the gene

for the larvicidal toxin of Bacillus sphaericus 2362: Evidence for a family of related sequences. J. Bacteriol., 169, 4061-4067. BAUMANN,

L., BROADWELL,

A. H., AND BAUMANN,

P. 1988. Sequence analysis of the mosquitocidal toxin genes encoding 51.4 and 41.9 kilodalton proteins from Bacillus sphaericus 2362 and 2297. .I. Bacterial., CARBOULEC,

170, 2045-2050.

N., AND PRIEST, F. G. 1989. Phenotypic characterization of some strains of Bacillus sphaericus. Lett. Appl. Microbial., 9, 113-116. FEINBERG, A. P., AND VOGELSTEIN, B. 1983. A technique for radiolabelling DNA restriction endonuclease fragments to high specific activity. Anal. Biothem., 132, 6-13. GORDON, R. E., HAYNES, W. C., AND PANG, C. H.-N. 1973. “The Genus Bacillus.” United States Department of Agriculture, Washington, DC. HERTLEIN, B. C., LEVY, R., AND MILLER, T. W., JR. 1979. Recycling potential and selective retrieval of Bacillus sphaericus from soil in a mosquito habitat. J. Znvertebr. Pathol., 33, 217-221. HINDLEY, J., AND BERRY, C. 1987. Identification, cloning and sequence analysis of the Bacillus sphaericus Microbial., KRYCH,

1593 41.9 kD larvicidal 1, 187-194.

V. K., JOHNSON,

toxin

gene.

J. L., AND YOUSTEN,

Mol. A. A.

1980. Deoxyribonucleic acid homologies among strains of Bacillus sphaericus. Znt. Syst. Bacterial., 30, 476-484.

IDENTIFICATION MANIATIS, T., FRITSCH, E. F., AND SAMBROOK, J. 1982. “Molecular Cloning: A Laboratory Manual.” Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. PRIEST, F. G., AND ALEXANDER, B. 1988. A frequency matrix for the probabilistic identification of some bacilli. J. Gen. Microbial., 134, 301 l-3018. SNEATH, P. H. A. 1979. Basic program for the identification of an unknown with presence-absence data against an identification matrix of percent positive characters. Computers & Geosciences, 5, 195-213. THEIRY, I., AND DE BARJAC, H. 1989. Selection of the most potent Bacillus sphaericus strains based on activity ratios determined on three mosquito species. Appl. Microbial. Biotechnol., 31, 522-581. WAYNE, L. G., KRICHEVSKY, E. J., LOVE, L. L., JOHNSON, R., AND KRICHEVSKY, M. I. 1980. Taxonomic probability matrix for use with slowly-

OF Bacillus

sphaericus

growing mycobacteria. Znt. J. Syst. Bacterial., 528-538. WILLIAMS,

333 30,

S. T., GOODFELLOW, M., WELLINGTON, E. M. H., VICKERS, .I. C., ALDERSON, G., SNEATH, P. H. A., SACKIN, M. J., AND MORTIMER, A. M. 1983. A probability matrix for identification of some streptomycetes. J. Gen. Microbioi., 129, 1815-1830. WORLD HEALTH ORGANIZATION. 1985. “Information Consultation on the Development of Bacillus sphaericus as a Microbial Larvicide.” TDR/ BCVISPHAERICUS185.3 (mimeo.doc.) World Health Organization, Rome. YOUSTEN, A. A. 1984. Bacteriophage typing of mosquito-pathogenic strains of Bacillus sphaericus. J. Znvertebr. Pathol., 43, 124-125. YOUSTEN, A. A., FRETZ, S. B., AND JELLEY, S. A. 1985. Selective medium for mosquito-pathogenic strains of Bacillus sphaericus. Appl. Environ. Microbiol., 49, 1532-1533.