Detection of Bacillus thuringiensis kurstaki HD1 on cabbage for human consumption


  • Niels Bohse Hendriksen,

    1. Department of Environmental Chemistry and Microbiology, National Environmental Research Institute, Frederiksborgvej, Roskilde, Denmark
    Search for more papers by this author
  • Bjarne Munk Hansen

    1. Department of Environmental Chemistry and Microbiology, National Environmental Research Institute, Frederiksborgvej, Roskilde, Denmark
    Search for more papers by this author

  • Editor: Andre Klier

Correspondence: Niels Bohse Hendriksen, Department of Environmental Chemistry and Microbiology, National Environmental Research Institute, Frederiksborgvej 399, PO Box 358, DK-4000 Roskilde, Denmark. Tel.: +45 46301372; fax: + 45 46301216; e-mail:


The objectives of the study were to develop a specific procedure for quantification and identification of Bacillus thuringiensis kurstaki HD1, which is used as a biopesticide, and to quantify its presence in different kinds of cabbage for human consumption. We found that B. thuringiensis kurstaki HD1 can be distinguished from other B. thuringiensis strains by its unique random amplification of polymophic DNA-PCR pattern with the OPA9 primer and the presence of the flagellin genes, as detected by the primers FLAB1 and FLAB2. We detected from one to 100 Bacillus cereus-like bacteria in 10 batches of five different cabbage products for consumption. As many as 73 out of 134 isolates (53.7%) were identical with B. thuringiensis kurstaki HD1. The results show that B. thuringiensis kurstaki HD1 from biopesticides can be found in vegetables for human consumption.


Bacillus thuringiensis is a gram-positive, facultative anaerobic, endospore-forming bacterium. It is characterized by its ability to produce parasporal crystalline inclusions toxic to larvae of different insect orders and other invertebrates. Biocontrol of insect pests by these insecticidal crystal proteins represents one of the most successful uses of biological control agents. The use of B. thuringiensis bioinsecticides has increased during the last years and is estimated to exceed more than 30 000 tonnes annually worldwide. Major targets for B. thuringiensis-based bioinsecticides are herbivorous lepidopteran larvae like cabbageworm, cabbage looper, hornworms, European corn borer, cutworms, some armyworms, diamondback moth, tent caterpillars and Indianmeal moth larvae in stored grain. For this purpose, strains of serotype kurstaki are generally used, especially the strain B. thuringiensis serotype kurstaki HD1 (Copping, 1998).

Bacillus thuringiensis is a member of the Bacillus cereus group, a separate unit within the group I Bacillus species. The Bacillus cereus group contains in addition to B. thuringiensis, B. cereus, Bacillus anthracis, Bacillus mycoides, Bacillus weihenstephanensis and Bacillus pseudomycoides.

The taxonomy of the group is controversial, and some consider the species to be a subspecies of B. cereus sensu lato (Helgason et al., 2000). Owing to the genetic similarities between B. thuringiensis and B. cereus, which might cause gastrointestinal diseases and somatic infections (Hansen & Hendriksen, 2001), the safety of B. thuringiensis as a bioinsecticide has been questioned (Damgaard, 1995), although only very few clinical case reports related to the use of B. thuringiensis have been published (Glare & O'Callaghan, 2000). However, neither medical practice nor the methods used for the detection of food pathogens discriminate between B. thuringiensis and B. cereus as causative agents in connection with food contamination or human diseases (Granum, 1997; Shinagawa, 1990). Therefore, the true proportion of B. thuringiensis involvement in these diseases, thought to be caused by B. cereus, is not known. However, one study exists in which B. thuringiensis has been shown to be involved in an outbreak of gastroenteritis in four persons (Jackson et al., 1995). In this study, the strain was initially identified as B. cereus but later found to be B. thuringiensis.

Only few attempts have been made to isolate B. thuringiensis from different kinds of foods, and none of these attempts have included quantification. Rusul & Yaacob (1995) and Damgaard et al. (1996) recovered B. thuringiensis from a range of foods, including pasta, rice, spices, grains, bread, legumes and milk. The isolates recovered by Damgaard et al. (1996) belonged either to serotype kurstaki or neoleonensis. Bidochka et al. (1987) isolated B. thuringiensis kurstaki from grapes for human consumption and hypothesized that the bacterium could have originated from biopesticide residues. Further, Rosenquist et al. (2005) analysed a large number of ready-to-eat products for B. cereus-like bacteria; they found 31 of 40 randomly selected isolates to belong to B. thuringiensis.

The objectives of this study were to develop a specific procedure for quantification and identification of B. thuringiensis kurstaki HD1 and to quantify the presence in different types of cabbage. Outdoor uses of B. thuringiensis kurstaki HD1 as a biopesticide in northern temperate regions is restricted to few pests, of which control of cabbageworm is the most important.

The procedure takes advantage of the fact that polymorphism in sequences between genes coding for flagellin genes is a common feature between B. thuringiensis strains (Hansen et al., 1998; Yu et al., 2002), and random amplification of polymophic DNA (RAPD)-PCR seems to be discriminative between B. thuringiensis strains, even between strains within the same serotype (Hansen et al., 1998).

Materials and methods

Ten different batches of cabbages, representing five different kinds of cabbage, were bought in triplicate in four different shops in Roskilde, Denmark, on 22 October 2002. Subsamples of the products (outer leaves of the head or parts of the flower, in total between 24.3 and 75.7 g depending on size and kind of product) were placed in sterile Stomacher bags with 50 mL water. Bacteria were extracted from the cabbage by two 30-s periods blending in a Stomacher 80 lab- lender (Seward Ltd, Worthing, UK).

Ten milliliters of an extract was heat treated in a water bath (35 min at 65°C). To detect low numbers of Bacillus cereus-like spores in the samples, 1.0 mL extracts were pipetted onto 135 mm diameter Petri dishes with T3 sporulation agar (Travers et al., 1987) in triplicate. The Petri dishes were incubated for 2 days at 30°C. Colonies having a rugose, ice-crystal-like appearance and a diameter >1 mm were counted as B. cereus-like colonies. In total, 134 B. cereus-like colonies were isolated and subcultured on T3-agar.

The isolates were examined by phase-contrast microscopy for their ability to produce parasporal inclusion bodies (crystals) in the sporangium after growth to sporulation on T3-agar for 3 days.

For DNA preparation, bacteria were plated on Luria–Bertani (LB) agar and incubated overnight at 30°C. An amount of bacteria corresponding to a colony 1–2 mm in diameter was transferred to 200 μL of Tris-EDTA buffer. Bacteria were lysed by incubation at 102°C for 10 min in a temperature-regulated block, and debris was removed by centrifugation at 15 000 g for 3 min. The DNA-containing supernatant was transferred to a new microfuge tube and stored at 4°C. The primer sets used in this study are shown in Table 1. PCR detection of the flagellin genes and the δ-endotoxin gene (cry1a) with the LEP2A and LEP2B primers was performed essentially as described elsewhere (Hansen et al., 1998). One microliter of DNA extract was amplified with 0.5 U of Taq polymerase (Roche, Mannheim, Germany) in a 25-μL reaction mixture using 30 cycles of denaturation at 94°C for 15 s, annealing at 55°C for 45 s and extension at 72°C for 2 min. The RAPD-PCR with the primer OPA9 was performed as described by Hansen et al. (1998). PCR analysis of the 16S-23S rRNA gene spacer region with the ITS-16S-1392-S-15 and ITS-23S-206-A-21 primers (Willumsen et al., 2005) was used as a control of DNA quality and verification of the strains belonging to the B. cereus group (Hansen et al., 2001). PCR products were analysed by 1.5% agarose gel electrophoresis, using MW VI (Roche) as a molecular weight marker.

Table 1.   Primers used in the study
PrimerSequenceSequence accession no.Reference
  1. N=A/G/C/T; V=A/C/G; Y=C/T; ITS, internal transcribed spacer.

FLAB1GCAGCTGACGATGCGGCTX67138,AY029474,X67139Hansen et al. (1998)
FLAB2AGTGTCCAGAGCCGTGATX67138,AY029474,X67139Hansen et al. (1998)
FLAB3CCGAAAGTTGCTCTAGCGX67138,AY029474,X67139This study
ITS-16S-1392-S-15GNACACACCGCCCGTWillumsen et al. (2005)
ITS-23S-206-A-21NCTTAGATGTTTCAGTTCVCYWillumsen et al. (2005)
OPA 9GGGTAACGCC Operon Technologies Inc. (Alameda, Ca)

Further, 41 Bacillus thuringiensis strains (Table 2) and 22 B. cereus strains (Hansen & Hendriksen, 2001) were analysed as reference strains.

Table 2.   Results from PCR analysis with the FLAB1, FLAB2 and FLAB3 primers and for the RAPD Bacillus thuringiensis kurstaki pattern with the OPA9 primers
StrainFLAB1 and FLAB2FLAB1 and FLAB3OPA9 HD1 pattern
  1. B.t., Bacillus thuringiensis; RAPD, random amplification of polymophic DNA.

B.t. finitimus HD 3
B.t. alesti HD 4++
B.t. kurstaki HD 1+++
B.t. kurstaki HD 73++
B.t. dendrolimus HD 7
B.t. dendrolimus HD 106
B.t. sotto HD 770
B.t. galleriae HD 29
B.t. galleriae HD 234
B.t. canadensis HD 224
B.t. entomocidus HD 9+
B.t. entomocidus HD 110+
B.t. entomocidus HD 198+
B.t. morrisoni HD 12
B.t. ostriniae HD 501
B.t. tolworthi HD 537
B.t. tolworthi HD 125
B.t. darmstadiensis HD 146
B.t. darmstadiensis HD 601
B.t. toumanoffi HD 201
B.t. kyushuensis HD 541
B.t. thompsoni HD 542
B.t. pakistani HD 395
B.t. indiana HD 521
B.t. dakota HD 932
B.t. tohokuensis HD 866+
B.t. kuamotoensis HD 867
B.t. tochigiensis HD 868
B.t. colmeri HD 847
B.t. shandogiensis HD 1012
B.t. thuringiensis HD 22
B.t. aizawai HD 131+
B.t. aizawai HD 137+
B.t. aizawai HD 11+
B.t. aizawai HD 112+
B.t. aizawai HD 283+
B.t. kenyae HD 136
B.t. israelensis HD 567
B.t. israelensis 4Q2-72
B.t. israelensis Bta2
B.t. tenebrionis NB-125
B.t. roskildiensis DMU-39


Two different combinations of three primers were used for the detection of two flagellin genes in 41 different Bacillus thuringiensis strains representing 29 different serotypes (Table 2). PCR with the primers FLAB1 and FLAB3 resulted in products from seven strains of four serotypes. PCR with the primers FLAB1 and FLAB2 resulted in products from only three of these strains, having the alesti and kurstaki serotypes. Further, 22 Bacillus cereus strains were analysed for the presence of flagellin genes by two primer combinations. Two of the B. cereus strains were positive with the primer combination FLAB1 and FLAB3, while none were positive with the combination FLAB1 and FLAB2.

Random amplification of polymophic DNA-PCR with the OPA9 primer resulted in 28 different patterns with the 41 B. thuringiensis strains. The pattern for Bacillus thuringiensis kurstaki HD1 is shown in Fig. 1 together with the pattern for the five Bacillus thuringiensis aizawai strains that form the same pattern. This pattern is unique for B. thuringiensis kurstaki HD1 and the B. thuringienisis aizawai strains and distinguishes them even from B. thuringiensis kurstaki HD 73 (Fig. 1). This unique pattern was not found among the patterns produced by 22 B. cereus strains.

Figure 1.

 Random amplification of polymorphic DNA with primer OPA9 and DNA from Bacillus thuringiensis alesti (lane 1), B. thuringiensis kurstaki HD 73 (lane 2), B. thuringiensis kurstaki HD1 (lane 3), B. cereus (type-strain) (lane 4) and B. thuringiensis aizawai (lanes 5–9). Lane M is a molecular size marker.

Between 1 and approximately 100 CFU g−1 were detected on the different cabbage products investigated (Table 3), the highest number was found on Broccoli batch A and Chinese cabbage batch B and the lowest on Broccoli batch B and Cauliflower. All B. cereus-like bacteria isolated during the study were confirmed to belong to the B. cereus-group by a group-specific PCR assay (Table 3). As many as 74 of the 134 isolates produced crystals, and these isolates were especially found among the isolates from Broccoli batch A, Cabbage batch A and Chinese Cabbage batch A. All isolates producing crystals, except one, possess CryIa crystal genes, as evidenced by PCR with the primers LEP2A and LEP2B. The presence of this gene suggests that the isolates are active against lepidopteran larvae. The remaining 73 isolates gave a product of the expected size with the FLAB1 and FLAB2 primers, and 72 of them produced an OPA9 RAPD-PCR pattern identical with the B. thuringiensis kurstaki HD1 pattern. The 73rd isolate produced a pattern identical with the pattern produced by B. thuringiensis HD73. Thus, 54% of the B. cereus-like bacteria isolated from cabbage products could not be distinguished from B. thuringiensis kurstaki HD1. These isolates originated from Broccoli batch A, Cabbage batch A and Chinese cabbage batch A; 96% of the isolates found on Broccoli A were indistinguishable from B. thuringiensis kurstaki HD1. The farm producing this broccoli uses Dipel, containing B. thuringiensis kurstaki HD1 as the active ingredient (personal communication).

Table 3. Bacillus cereus-like bacteria found on different kinds of cabbage for human consumption
(CFU g−1)
Number of
Isolates from B.c.-
group (specific 16S –
23S rRNA gene
spacer PCR pattern)
Isolates with
positive for
Isolates giving
PCR product
with FLAB1
and FLAB2
Isolates with
OPA9 pattern
identical to
Percentage B.t.
isolates (%)
  1. B.t., Bacillusthuringiensis; B.c., Bacillus cerus; RAPD, random amplification of polymophic DNA.

Broccoli A88±6250504848484896
Broccoli B1±1110000
Cabbage A22±1620201413131365
Cabbage B4±1111110
Cabbage C3±233222267
Chinese cabbage A54±899777778
Chinese cabbage B96±43333311113
Chinese cabbage C30±3113130000
Total 1341347473737254


Few attempts have been done until now to examine the occurrence of Bacillus thuringiensis strains from microbial pest control agents in food. This is due to (i) a lack of methods for distinguishing such strains from other B. thuringiensis strains and (ii) the presumed safety for humans of the use of B. thuringiensis as a pest control agent.

We evaluated whether the genetic patterns generated by RAPD-PCR with the OPA9 primer could be used to distinguish between Bacillus thuringiensis kurstaki HD1 from other strains of B. thuringiensis, as well as from Bacillus cereus. This evaluation involved 41 B. thuringiensis strains representing 29 different serotypes and 22 B. cereus strains. The data showed that the OPA9 pattern generated by B. thuringiensis kurstaki HD1 differed from all the other B. thuringiensis and B. cereus strains, except from five B. thuringiensis strains with the aizawai serotype. Multilocus sequence typing has also revealed a very close relationship between B. thuringiensis kurstaki and Bacillus thuringiensis aizawai strains (Priest et al., 2004). Hansen et al. (1998) found that the OPA9 RAPD-PCR could not differentiate B. thuringiensis kurstaki HD1 from strains isolated from five different commercial microbial pest control products used for the control of lepidopteran larvae. This confirms the applicability of RAPD with the OPA9 primer for the characterization of B. thuringiensis kurstaki HD1 strains. de Amorim et al. (2001) were also able to differentiate B. thuringiensis kurstaki HD1 and seven other B. thuringiensis strains and a single B. cereus strain by RAPD analysis. To further distinguish between B. thuringiensis strains, PCR analysis with the primers FLAB1 and FLAB2, detecting flagellin genes, have been very useful, as this primer set exclusively anneals to the genes in the serotypes alesti and kurstaki. It is most likely that this difference is due to polymorphism within these genes between serovars (Hansen et al., 1998; Yu et al., 2002). Sequence analysis by basic local alignment search tool has revealed that FLAB2 will anneal to two of the three homologous fla genes present in B. thuringiensis alesti, but not to the four fla genes present in the B. cereus type-strain. The FLAB1 primer was, on the basis of the published B. thuringiensis alesti sequences of Lövgren et al. (1993) and Ankarloo et al. (1996), chosen to match a conserved region of the genes named flaA and flaB in this serotype, while FLAB2 and FLAB3 were chosen to match specific regions in flaA and flaB, respectively. Thus, B. thuringiensis kurstaki HD1 can be distinguished from other B. thuringiensis strains by its unique RAPD-PCR patterns with OPA9 primers and the presence of fla-genes as detected by the FLAB1 and FLAB2 primers.

We detected from one to approximately 100 B. cereus-like CFU per gram in 10 batches of five different cabbage products for consumption from Danish shops. These densities are below the densities of 103–105 CFU g−1 reported to be able to cause illness by B. cereus (Granum, 2001). These findings are also below the acceptable threshold of 103 CFU g−1 given by the food regulatory agencies in some countries. Just above 50% of the isolates from these cabbage products could be affiliated to B. thuringiensis based on the presence of crystals and the cryIa genes encoding crystal proteins active against lepidopteran larvae as evidenced by PCR analysis (Carozzi et al., 1991).

Of the 74 B. thuringiensis isolates, 72 were identical with B. thuringiensis kurstaki HD1, as they gave rise to a product with the FLAB1 and FLAB2 primers and have an OPA9 RAPD pattern identical to the B. thuringiensis kurstaki HD1 pattern. The B. thuringiensis-based product for the control of lepidopteran larvae available on the Danish market contains a strain identical to B. thuringiensis kurstaki HD1 (Hansen et al., 1998; Winding, 2005). Thus, it seems reasonable to conclude that a major part of these bacteria originate from these products. A field experiment conducted with a B. thuringiensis kurstaki HD1-like isolate on cabbage during summertime in Denmark (Pedersen et al., 1995) revealed an initial half-life of 16 h, and a level of approximately 100 CFU g−1 on the cabbage was reached in about 15 days. Thus, it is likely that B. thuringiensis kurstaki HD1 from products are present at the levels reported here on cabbage products for consumption at harvest. The highest level of B. thuringiensis kurstaki HD1-like bacteria was found on Broccoli from a field that has been treated with a product containing this bacterium. It is unlikely that B. thuringiensis kurstaki HD1 occurring on cabbage products, at the densities demonstrated in this study, is of any concern in relation to public health.


The project was supported by a grant from the Danish Ministry of Food, Agriculture and Fishery (FØS100-DMU-5). We thank Bente R. Hansen and Lillian F. Larsen for excellent technical assistance.