Biochemical characteristics that presented variable response among the bacterial isolates. Molecular characterization is also showed (see below).
Bacillus thuringiensis (Bt) is a spore-forming bacterium well-known for its insecticidal properties associated with its ability to produce crystal inclusions during sporulation. These inclusions are proteins encoded by cry genes and have shown to be toxic to a variety of insects and other organisms like nematodes and protozoa . The primary action of Cry proteins is to lyse midgut epithelial cells through insertion into the target membrane and form pores . Once ingested, crystals are solubilized in the alkaline environment of midgut lumen and activated by host proteases . On the other hand, the involvement of Bt proteases in processing inactive protoxins is also reported . These toxins are also highly specific and completely biodegradable, hence no toxic products are accumulated in the environment. In fact, Calderón et al.  suggest the potential use of some crystal proteins as adjuvants for the administration of heterologous antigens.The activity spectrum of Bt toxins continually increases as result of the ongoing isolation of new strains around the world.
The fall armyworm, Spodoptera frugiperda (S. frugiperda) (Lepidoptera: Noctuidae), and the variegated cutworm, Peridroma saucia (P. saucia) (Lepidoptera: Noctuidae), are two lepidopteran pests that cause severe damage to a variety of crops. While the first one mainly attacks corn, rice, peanuts, cotton, soybeans, alfalfa and forage grasses , the second one targets peanuts, sunflowers, soybeans and grapevines, among others . Currently, control of this pest relies on chemical insecticides. Nevertheless, the rapid increases in resistance to insecticides together with the potential adverse environmental effects produced by these chemicals have encouraged the development of alternative methods for Lepidoptera control [7,8]. Among these methods the use of Bt as a biocontrol agent has shown to be extremely valuable. The diversity of Cry toxins produced by Bt allows the formulation of a variety of bioinsecticides by using the bacteria themselves or by expressing their toxin genes in transgenic plants. To date, many plant species have been genetically modified with cry genes, resulting in transgenic plants with a high level of resistance to insect . However, it has been reported that several pests have developed resistance against Cry proteins [9, 10]. The current approach used to delay evolution of resistance to transgenic crops uses a “high dose” and “refuge” strategy [9, 11]. In addition, it is important to use a combination of cry genes and/or other genes encoding insecticidal proteins within the same transgenic crop [12, 13]. Due to extensive use of transgenic crops in developing countries based on cry-type genes, there is a need for alternative cry gene sequences to meet the challenge of novel insect resistance . Crucial to this development is the identification of novel and more active strains with respect to insect pests of economically important crops .
The cry genes of Bt strains are known to be related to their toxicity [15, 16] and identification of these genes by means of PCR has been used to characterize and predict insecticidal activity of the strains [17, 18]. Nevertheless, a more complete characterization should include alternative methods. Phenotypical analysis such as protein profile determination provides useful information for typing and comparative studies . The literature data report the possibility of using the whole-cell protein profile as a discriminating method with potency similar to RAPD with combined DNA patterns . However, there is not always a good correlation between these factors and insecticidal activity of Bt strains [20, 21]. In addition, there is a need to develop knowledge about the biological properties and diversity of Bt isolates since these data allow a better understanding of the biological factors that determine insecticidal properties. Extracellular factors such as phospholipases, proteases and chitinases have shown to contribute to insecticidal activity of Bt .
During a screening programme of Bt isolates native to Argentina and toxic against Lepidoptera, several strains were characterized according to different biological parameters. In addition, promising isolates regarding their useful in biological control programmes -an environmentally safe technology of pest control- were exhaustively studied [14, 19, 23]. The present work showed most relevant results obtained during a course of those investigations. The discovery of highly pathogenic isolates against devastating insect pests reveals the usefulness of screening studies for novel Bt strains.
2. Biochemical characterization of B. thuringiensis isolates and assessment of toxicity
Crystalliferous spore-forming bacteria were isolated from both S. frugiperda larvae showing disease symptoms and soil samples collected in Argentina . These samples came from maize, sorghum, wheat, grape or sugarcane cultivated fields. Briefly, larvae and soil sample suspensions were made in distilled water, heated at 80 °C for 15 min and then plated onto LB-agar. Plates were incubated at either 30 or 55 °C for 24 h. Colonies that did not grow at 55 °C were then analyzed for the presence of parasporal crystals by microscopic examination . From a total of 254 colonies isolated from 490 different environmental samples, 14 were identified as crystal producer strains, giving a mean Bt index of 0.05. This result suggested that samples analyzed contained a high background level of other spore-forming bacteria. One crystalliferous strain came from sorghum cultivated field, while the others came from maize cultivated field. Concerning the source of isolation, 50% of crystal producers came from soil samples and the other 50% came from ill larvae. Interestingly, the last source provided the most pathogenic strains (Table 1).
Bacteria -characterized by conventional microbiological methods- possessed typical cellular and colonial morphologies, as well as physiological, biochemical and nutritional features that resembled Bacillus spp. They were motile and produced ellipsoidal endospores, located at sub-terminal position in the sporangia, and formed cream-colored colonies with irregular or circular edges on LB agar.
|Phenotypical and molecular characterization||Strains|
|Sub terminal spore||-||-||-||-||-||-||-||+||-||-||-||+||-||-||-|
|Growth at pH 9||+||+||-||+||+||+||+||+||+||+||+||-||+||+||+|
|b e Growth in 0.2 % chitin||-||+||+||+||-||+||+||+||-||+||-||-||+||+||+|
|b c e CMC hydrolysis (0.5 %)||+||+||+||+||+||-||-||+||+||+||-||+||+||-||-|
|b e Chitin hydrolysis (0.2 %)||-||-||+||-||-||+||+||-||-||-||-||-||+||+||+|
|b Gelatin hydrolysis (12 %)||+||-||-||-||-||-||+||-||+||-||-||+||-||-||-|
|b Starch hydrolysis (2 %)||-||-||+||+||+||+||+||-||+||-||+||+||+||+||+|
|Gas production in glucose||-||+||-||+||-||+||-||+||-||-||-||-||+||+||+|
From a biochemical point of view, the 14 strains were catalase-positive, reduced nitrate and produced acetyl methyl carbinol in Voges-Proskauer broth; growth was observed at pH 7 on LB agar supplemented with 2, 3 and 5% NaCl and on LB agar at 30, 37 and 45 ºC. The strains also hydrolyzed casein and were motile on soft LB agar. Negative results for all strains were obtained in several tests: no growth was observed on LB agar at pH 4 or at 50 ºC and none of the strains hydrolyzed carboxymethyl cellulose (CMC) and urea. Antibiotic sensitivity tests revealed a resistance profile to penicillin, oxacillin, trimethoprim and a sensitive profile to erythromycin, vancomycin, levofloxacin, minocycline, chloramphenicol and teicoplanin. Phenotypic features that presented variability among the strains are showed in Table 1. The positive or negative result of each biochemical assay was entered in a 1-0 matrix. These data were subsequently analyzed through correspondence multivariate analysis, using Multivariate Statistical Package (MSVP) software (version 3.13). A cluster diagram based on these variable biochemical properties (that represented 54% of data variability) revealed that the strains formed two main groups (Figure 1). Group A comprised nine crystalliferous isolates which were clustered together with the reference strain B. thuringiensis kurstaki 4D1 (Bt 4D1). A second group (B) included three Bt strains while the remaining two strains presented more divergent features and hence were not included in any group. Isolates from the same sample and/or the same geographic region differed in their phenotypic features and consequently were not grouped together. This indicates that there was no clear association between Bt strains biochemical profile and the environments from which they were obtained .
Although most of the native isolates presented similar biochemical and phenotypical characteristics compared with reference strain Bt 4D1 (Group A) (Figure 1), they differed in their toxicity to S. frugiperda. Our results showed that the mortality on S. frugiperda neonate larvae was variable , ranging among values corresponding to Bt strains of bioinsecticides action low to moderate (Figure 2). However, strains named RT, LSM and LQ were found to be highly pathogenic, two of them, even more than reference strain Bt 4D1 which was selected for this analysis given it is the most widely used microorganism to control lepidopteran pests  (Table 2). This strong biological effect was represented by both a shorter LT50 and a higher mortality, which reached 100% in the case of RT strain on S. frugiperda, after five days of treatment. This result is extremely relevant considering that S. frugiperda is believed a pest with low sensitivity to Bt toxins . In addition, when this strain was assayed against first instar larvae of P. saucia, reached 93% of mortality (Figure 3) suggesting that RT strain native to Argentina could possibly be employed in biological control of lepidopteran pests [19, 23]. It is important to stress that the high levels of mortality in the present work were obtained with a concentration of a spore-crystal suspension that was lower than some commercial Bt formulations; while our crystal spore suspensions presented a dose of 107 c.f.u. ml-1, Bt kurstaki preparations generally present a dose of 109 c.f.u. ml-1 .
|Bt strain||*,a Mortality (%) ± SD||b LT50 (h) (95% fiducial limits )||*Specific biomass bound protease activity (± SD) (U g dry wt -1)|
|RT||100 ± 0 a||9.2 (10.4 –16.0)||1.98 ± 98 b|
|LSM||90.0 ± 7.3 a||37.7 (27.8 – 46.2)||1.80 ± 93 b|
|LQ||73.0 ± 5.7 c||79.6 (68.2 – 90.7)||1.14 ± 25 a|
|Bt 4D1||86.0 ± 15.1 b||58.7 (50.4 – 66.0)||946 ± 14 a|
|control||1.0 ± 3.1 d|
3. Numerical analysis of insecticidal crystal proteins of B. thuringiensis
In order to differentiate native crystalliferous isolates and to evaluate the relationship between the toxicity assays against S. frugiperda, isolated bacteria and parasporal crystals were characterized by the whole-cell protein profiles in SDS-PAGE . Protein bands were individually identified by their specific migration rates in the gels. Once bands were properly and distinctively identified, binary (0/1) matrices were constructed to compare the patterns. Electrophoretic analysis revealed the presence of 53 distinct bands with molecular weights ranging from 266 to 20 kDa (Figure 4). Numerical analysis clearly showed two distinct clusters (Figure 5). Cluster A comprised 11 isolations and the reference strain B. thuringiensis var. thuringiensis 4A4 (Bt 4A4) as well as those crystalliferous isolations that had no or very low toxicity against S. frugiperda first instar larvae. Interestingly, this group of native microorganisms produced proteins from 28 to 31 kDa but not proteins of ~135 and ~65 kDa. These lower molecular mass could correspond to Cyt toxins, entomocidal crystal proteins highly active against Diptera larvae . On the other hand, all isolations with high toxic activity against S. frugiperda (RT, LSM and LQ strains) (Table 2) were located in cluster B, as well as the reference strains Bt 4D1 and B. thuringiensis var. kurstaki 4D3 (Bt 4D3). The isolation RT had a protein profile similar to Bt 4D1 with proteins of ~140 and ~70 kDa. Strains LSM and LQ showed protein bands of ~100 and ~81 kDa. These results demonstrate that the whole cell protein profiling not only allowed the differentiation of Bt at strain level but also revealed a possibility to apply protein profile analysis in classification of toxicity patterns.
4. Molecular characterization of B. thuringiensis strains and crystal morphology
Although the presence of parasporal crystals is a diagnostic characteristic of Bt strains , the taxonomic identity of the toxic crystalliferous isolates was confirmed by amplification and partial sequencing of their 16S rDNA genes  (Table 3). The partial 16S rDNA sequences were tested by BLAST analysis against the GenBank data base. Bt LSM strain showed exact BLAST matches with the sequence from Bacillus thuringiensis var kurstaki (1 hit, 100% of identity, accession number EF638796). Bt LQ strain produced 4 hits (99% of identity, accession number EF638798), all of these corresponding to Bt species. Similarly, Bt RT (best hits, 13; 99% of identity, accession number EF638795) also shared a close relationship with others Bt strains, including Bt LSM.
Generally, B. thuringiensis insecticidal protein toxin genes (cry) reside on large self-transmissible plasmids, and individual B. thuringiensis strains can harbor a diverse range of plasmids that can vary in number from 1 to 17 and in size from 2 to 80 MDa [29,30], although it has also been suggested that they are present in the chromosome . In this context, to study the plasmid profiles of Bt strains is an important parameter to determine their identity, since the number and size of these is associated with a particular Bt strain. Comparison between strains belonging to the same serotype showed a great difference in variability . Some serotypes (e.g., israelensis) showed the same basic pattern among all its strains, while other serotypes (e.g., morrisoni) showed a great diversity of patterns. These results indicate that plasmid patterns are valuable tools to discriminate strains below the serotype level . The profile of extrachromosomal elements in Bt is influenced by a number of stressful growth conditions, which determine its stability and heritability (e. i. high temperatures determine the plasmid loss), therefore it is neccesary to take some care. In this study, cultures were routinely grown at 30 °C to avoid this phenomenon. Detection and isolation of plasmid DNA was conducted following the method of Kado and Liu . DNA plasmid samples were electrophoresed on 0.8 % (wt vol-1) agarose gel. Our results showed that selected Bt strains present a complex plasmid proﬁle (Figure 6).
In this experiment, the plasmid DNA was not linearized and therefore the same plasmid can produce as many as three different bands in the agarose gel. This made it difficult to determine the precise number of plasmids present in each complex plasmid proﬁle. For this reason, we will refer to the number and size of plasmidic bands rather to plasmids themselves. An intense band above the chromosomal band (C) was observed in Bt RT, Bt LSM and Btk 4D1 suggesting that a large plasmid or plasmids is/are found in this strains which might be responsible for production of parasporal bodies. Compared with the other bacteria, Bt LQ presented a very different profile array, suggesting a different cry-genotype.
Identiﬁcation of cry genes by means of PCR has been used to predict insecticidal activity of the strains [17,18] and to determine the distribution of cry genes within a collection of B. thuringiensis strains [20, 33]. In this context, our crystalliferous strains were characterized in terms of presence of cry1 and cry2 genes by amplification with general primers. The most toxic Bt strains RT, LSM and LQ were characterized through additional PCR with specific
primers to identify the presence of cry1Aa, cry1Ab, cry1Ac, cry1Ad, cry2Aa, cry2Ab and cry2Ac genes (Table 3). PCR analysis showed presence of cry1 and/or cry2 genes in most of the isolates (Table 1). Specific PCR showed identical cry gene profile in both Bt LSM and the reference Bt 4D1, while cry gene content of Bt RT was different from them. DNA of Bt LQ was not amplified under the current reaction conditions (Table 1).
In addition, amplified fragments corresponding to cry1 and cry2 genes from Bt RT were sequenced and compared with cry genes sequences available from GenBank. This sequences had 99 and 95% identity with cry1Ab (EU220269) and cry2Ab (EU094885) genes, respectively. As shown in Figure 7, cry1 and cry2 partial sequences from Bt RT and Bt 4D1 were also aligned with ﬁve and six GenBank published cry sequences, respectively. The phylogenetic analysis revealed that cry1 partial sequences from Bt RT and Bt 4D1 possess almost the same level of evolutionary distance (Figure 7A), while cry2 partial sequence from Bt RT lies on a separate diverse branch not only of cry2 from Bt 4D1 but also of the others analyzed cry2 sequences (Figure 7B). Considering the phylogenetic analysis, it could be expected toxicity mediated by Cry1 rather than Cry2 crystal protein. In fact, cry2 partial sequence from Bt RT shared a 95% homology with cry2 sequence from a Colombian native Bt strain active against Tecia solanivora (Lep:Gelechiidae) (EU094885).
As mentioned before, cry genes are a family of genes associated with the toxicity of Bt against insects. While cry1 encodes for proteins forming bipyramidal crystals and are related to toxicity to Lepidoptera  cry2 encodes for cuboidal proteins, toxic to Lepidoptera and Diptera . Our molecular and electron microscopy analyses of Bt RT are in agreement with all this evidence, since this highly pathogenic strain has both genes (Table 1) and both kinds of proteins (Figure 8A). In contrast, and although Bt LSM showed amplification products with cry2 general and specific primers (Table 1) no cuboidal proteins were identified (Figure 8B). This suggests that a modification in the regulation of the gene would be responsible for the lack of protein product of this gene. Although the experimental growth conditions employed could also explain the lack of cuboidal proteins, the production of these proteins by Bt RT under identical experimental conditions argue against this possibility . Cloning and sequencing the putative toxins with surrogate production made help clarify this issue as well as to confirm toxicity. In addition, Bt LQ showed no amplification products of cry1 and cry2 gene in several attempts (Table 1), despite the presence of bipyramidal crystals (Figure 8C). Noguera e Ibarra,  found that cry genes of a Bt strain isolated in Argentina that showed elongated bipyramidal crystals  presented 98% identity with cry5Ba genes. Therefore, Bt LQ may have Cry proteins other than Cry1 that form bipyramidal crystals.
|Primer pairs||Nucleotide sequence||Reference|
|16S: 27F 1492R||5´-AGAGTTTGATCCTGGCTCAG-3´|
From a methodological point of view, washing of crystal suspensions with absolute ethanol/distilled water (Figure 9B) was more appropriate for microscopic observation than washing with distilled water (Figure 9A).
As mentioned above, identiﬁcation of cry genes by means of PCR has been used to predict insecticidal activity of the strains [17,18]. However, but since the primers are designed against known genes, the technique presents limitations in the search of novel cry genes. Moreover, the reliability of the prediction of insecticidal activity based on PCR results is dependent on the expression of the genes. In this context, a more complete characterization of Bt strains should include alternative PCR ﬁngerprinting methods. Among them, assessment of length polymorphism of intergenic transcribed spacers (ITS) between the 16S and 23S rDNA genes has been shown to be an important tool for differentiating bacterial species and even prokaryotic strains . In this context, ITS-PCR was performed as previously described by Daffonchio et al.. Used primer pairs are showed in Table 3. Evaluation of ITS length polymorphism revealed an identical pattern among Bt RT, LSM and LQ strains and also with Bt 4D1 suggesting that ITS exhibited no polymorphism among the strains (Figure 10). In connection with this, Reyes-Ramirez and Ibarra  studied ITS proﬁles of 31 Bt strains and found them to be insufﬁcient to discriminate between isolates.
5. Assessment of enzyme activities in B. thuringiensis
Phenotypic characterization of selected strains allows identification of properties that are relevant at the moment of selecting bacteria for their use in environmental and agricultural microbiology. Synthesis of lytic enzymes by Bacillus species during the early sporulation phase is one of these properties. Secreted microbial enzymes may function as virulence factor that are essential for survival and spread in the host [44,45]. Our results indicate that the native Bt strains responded diversely regarding proteolytic, cellulolytic and/or chitinolytic activity (Table 1). Chitinolytic activity is a contributing factor in Bt pathogenicity  which our most pathogenic strains possessed. The enzymes involved would act on the peritrophic membrane of the host, which facilitates the entry of pathogens into the haemocoel of susceptible insects . In addition, these strains showed no cellulolytic activity in medium supplemented with carboxymethyl cellulose (CMC), one of the products used as a matrix to protect Bt spores against high temperatures and UV exposure prevailing in natural environments . This lack of cellulolytic activity is a desirable property given that gelled CMC will not be degraded at the time of Bt formulation, and therefore it can be employed for this purpose .
In Bt, high levels of protease activity are associated with both crystal and spore formation  and this activity may contribute in processing inactive Cry protoxin to active toxin . While there is a reasonable understanding of soluble midgut proteases in toxin activation, little is known about the role of Bt protease in entomotoxicity. In this connection, during our investigations, biomass-bound protease and extracellular protease activities were determined in the toxic strains, which were process according to . Proteolytic activity was assayed by using azocasein as substrate . Table 2 shows that Bt RT, Bt LSM and Bt LQ displayed high biomass-bound protease activity. To our knowledge, the presence of this naturally immobilized enzyme activity has not been reported in Bt . On the other hand, extracellular protease activity was observed when crude extracts of Bt strains were electrophoresed on SDS-PAGE containing gelatin powder . The gels were then processed according to  for proteolysis to occur. Gel was stained with 0.1% (wt vol-1) Coomasie Blue R-250. Proteinase K (10 mg ml-1) was used for comparative analyze. All strains presented a clear zone of proteolytic activity which were larger in both Bt RT and Bt LSM (Figure 11). Although no correlation between protease activities and mortality values was initially detected, this result could be complementary information to consider in commercial Bt formulations, since the cell structure may act as a natural matrix able to protect the biomass-bound enzymes from the possible negative action of external agents; and therefore it could be that an increased percentage of Bt protease may actually reach the larvae midgut. Finally, it would be useful to explore the role of the extracellular and biomass-bound protease activities in crystal protein modification during Bt fermentation, the synergy of this protease source with insect entomotoxicity and the possible addition of vegetative cells in the final Bt formulation .
Protein profiles are a useful tool to discriminate among strains, as they provide information about the proximity between species, subspecies and biovars [18, 52]. Considering this, characterization of microorganisms by means of their extracellular isoenzymes showing high polymorphism, as is the case of esterase, is particularly appealing. To determine extracellular esterase profiles, Bt strains were processed according to . Briefly, strains were cultured on LB plates during 48 h at 30 ºC and crude extracts were recovered from solid media. Then, extracts were separated by native-PAGE. Esterase activity was assayed using 1.3 mM of α-naphthyl acetate (C2) derivative as substrate. Known electrophoretic esterase profiles of Bacillus pumilus A55 (EF638794.1) (Bp A55) were used for comparative analysis. The electrophoretic profiles of esterase activity showed differences among strains (Figure 12). Bt LQ showed a unique band/enzyme of 40 kDa as well as Bt LSM, but of about 60 kDa while Bt RT presented two bands of 95 and 60 kDa. Our results are in accordance with those by Norris , since it was possible to differentiate Bt strains by comparing the electrophoretic migration profiles of esterase produced during the vegetative growth phase (Figure 12) .
Lepidoptera causes some of the most devastating insect pests in important crops in America. Since economy of these regions depends largely of agriculture, their control is a priority as well as a necessity. In this context, use of environmentally safe technology to reduce crop damage like B. thuringiensis would be extremely valuable. Consequently, we set out to establish and characterize a collection of Bt isolates from samples collected in different Argentinean localities in order to find novel strains toxic against insect pests of economically important crops (like soybean and maize).
Fourteen Bt strains were isolated and phenotypically, genetically and biologically characterized. Analysis of larvicidal activity indicated that three strains exhibited high toxicity against lepidopteran larvae; this toxicity was in most, higher than that of the reference strain Bt 4D1.
The discovery of a highly toxic isolates reveals the usefulness of screening studies for novel Bt strains. The future application of these strains in biological control programmes requires optimization of the production conditions of the microorganisms using low-cost substrates. In this context, characterization of phenotypic and biochemical properties as evaluated in this study is highly relevant.