ABSTRACT
Bacillus thuringiensis isolates from the phylloplane of organically cultivated cabbage were characterized using molecular and phenotypic methods. Of the 58 isolates under study, 31 belonged to serovar kurstaki, 16 did not react with any of the currently recognized antisera, 7 reacted with known antisera, and 4 could not be serotyped as they were nonmotile. Round crystals were found in 26 isolates, while bipyramidal crystals were found in the remaining 32 isolates, all of which had activity to lepidopteran larvae. Further, one isolate with unknown serotype and round crystals had lepidopteran activity. Colony hybridization was found to be a useful tool for screening the isolates for specific gene homologies and showed good correlation with the phenotypic observations. Polymerase chain reaction (PCR) was used for confirmation of the colony hybridization data, in most cases with concordant results. However, in one case some of the colony hybridization data could not be confirmed by PCR, due to DNA sequence variations in the binding area of one of the primers. The random amplified polymorphic DNA (RAPD) analysis showed that isolates otherwise indistinguishable could be distinguished by this method. However, the method was not able to distinguish the 31 kurstaki isolates. Further, the kurstaki isolates could not be distinguished from the B. thuringiensis serovar kurstaki HD-1 strain used in commercial products for lepidopteran control. One of the isolates was a serovar israelensis, but no genes encoding dipteran activity could be detected, and the RAPD analysis revealed that the DNA fingerprint of this israelensis isolate deviated from the israelensis ONR60A isolate used in commercial products. In conclusion we find that a molecular method like colony hybridization is suitable for screening large collections of bacteria. When colony hybridization data are combined with RAPD analyses isolates can be grouped based on genetic potential and DNA fingerprint, whereby further characterizations by PCR and the more labourious phenotypic methods can be performed more effectively. Copyright 1998 Academic Press.
