Morphological Study of Bacillus thuringiensis Crystals and Spores

In this paper, we report a morphological study of the crystals and spores of different shapes synthesized by seven different strains of Bacillus thuringiensis. Crystals and spores were separated after 48 h of culture on T3 agar medium and imaged under a scanning electron microscope (SEM). Sizes of the crystals and spores were determined using Image J software. The results showed that crystal and spore sizes were normally distributed. In addition, the volumes and aspect ratios of the crystals and spores were calculated. The statistical analysis of the data showed the variability of the size distribution and morphological data of the crystals produced by the analyzed strains. Furthermore, variations in spore size and shape within the same serovar were observed, indicating that, perhaps, there are still some unexplored differences between strains of this serovar, making them less identical than what was believed so far.


Introduction
Bacillus thuringiensis (B. thuringiensis), a Gram-positive, soil-dwelling bacterium, is part of B. cereus sensu lato, a group of nine phylogenetically close, but ecologically diverse bacteria [1]. Its three most studied members are the biopesticide B. thuringiensis, the potentially food-borne toxi-infectious pathogen B. cereus sensu stricto [2] and the causative agent of anthrax, potential bio-weapon B. anthracis [3]. Thanks to its status as an "eco-friendly and safe" biopesticide, B. thuringiensis-based products form ca. 90% of the microbial pesticides currently available [4]. It owes its entomopathogenic capacities to a crystallin inclusion, produced during the sporulation phase, consisting of a constellation of toxins, also dubbed δ-endotoxins, active against a wide array of insect larvae [5]. After sporulation is complete, the crystal is released in the environment alongside the spore and eventually ingested by insect larvae. Once in the insect's gut, the crystal will be solubilized, and toxins will be activated by intestinal proteases. They subsequently bind to receptors on the intestinal midgut, forming pores and leading to the death of the insect larvae [6]. Crystal proteins include those belonging to the pore-forming Cry toxins with or without the presence of cytolytic Cyt toxins. Cry toxins interact specifically with a receptor in the apical membrane of the insect midgut, whereas Cyt toxins act in a detergent-like manner on the membrane phospholipids [7].
Therefore, on the one hand, a variation in a crystal's Cry components underlines a variation in a strain's activity spectrum. So far, 73 Cry and seven Cyt families are registered in the recently updated B. thuringiensis nomenclature database [8]. Each Cry family has a signature shape and activity spectrum; e.g., Cry1 family proteins form a bipyramidal crystal and are active against lepidopteran larvae [9], whereas Cry4 form a spherical crystal and are toxic to dipteran larvae [10].
On the other hand, B. thuringiensis strains are classified according to their serotypes or serovars (var.), which were once based on their respective immunological reactions caused by their flagellar H-antigen proteins, flagellin [11]. Nowadays, this classification is based on pairwise sequence comparison of the latter protein [12]. The two most studied serovars are (i) the reference for anti-dipteran activity, B. thuringiensis var. israelensis, and (ii) the reference for anti-lepidopteran activity, B. thuringiensis var. kurstaki. The former produces spherical crystals containing a set of proteins from the following families: Cry4, Cry10, Cry11, Cyt1 and Cyt2 [13][14][15]. As for the latter, its crystal proteins belong to the Cry1 and Cry2 families, resulting in bipyramidal and cubic crystals, respectively [9,16]. Other than its crystal components, each serovar is characterized by a set of plasmids, toxin-carrying, conjugative or mobilizable, providing serovar-specific peculiarities [14,17].
An in-depth investigation of all aspects involved in B. thuringiensis biopesticidal activity-be they genomic, proteomic, morphological, related to crystal and spore production kinetics and stability or otherwise-is a necessity created by the increased use of B. thuringiensis spore-crystal mixtures in the environment. One of the most recent techniques used to monitor the growth kinetics and production of B. thuringiensis spores and crystals is laser speckle imaging, a reliable and non-invasive technique based on light scattered by an illuminated medium [18]. In order to have accurate results and to understand the interaction between the light and the bacterial density, the size and shape of the scatterers in question should be known, along with whether these scatterers are in suspensions [19] or embedded in agar plates [20]. In the case of B. thuringiensis spores and crystals, their size and morphology vary according to the culture conditions and the presence of additional chaperone proteins. For the latter, an example is that of the B. thuringiensis var. israelensis chaperone; others are accessory proteins p19 or p20, well-known for their association with anti-dipteran toxins. Both proteins were shown to be indispensable for correct folding of Cry and Cyt proteins [21,22]. In fact, the effect of the 20 kDa protein on anti-coleopteran crystal toxins was highlighted in a 2011 study, in which this protein enhanced the volume, yield and solubility of Cry3A [23]. As for variations in culture conditions (culture media composition, pH, available oxygen and initial inoculum volume in the culture), several studies reported significant effects on the production and morphology of crystal inclusions, and the spore count at the start of the sporulation phase [24][25][26][27].
In this study, we established a protocol for mining the morphological data and size distribution of B. thuringiensis spores and crystals of different shapes. Our focus was on the morphological analysis of spores and crystals of three B. thuringiensis var. kurstaki strains with bipyramidal and cubic crystal; three var. israelensis strains with spherical crystals; and one B. thuringiensis strain belonging to an unknown serovar, but that is phylogenetically close to B. thuringiensis var. sichuansis. The null hypothesis tested was that there were no significant differences in the various morphological aspects considered between the analyzed B. thuringiensis strains.

Bacterial Strains and Spore/Crystal Preparation
Seven B. thuringiensis strains were used in this study, whose characteristics and crystal components are summarized in Table 1. Three B. thuringiensis var. kurstaki strains are considered: HD-1, the reference anti-lepidopteran strain used in the commercial biopesticide Thuricide ® [28]; the LIP MKA strain, isolated from Lebanese soil [9]; and the BLB1 strain, isolated from Tunisian soil [29]. All three are active against lepidopteran larvae and produce bipyramidal (B) and cubic (C) δ-endotoxin crystals. In addition, three strains of B. thuringiensis var. israelensis were also used: AM65-52, kindly provided by Pr. Christina Nelsen LeRoux from the GME laboratory of the National Institute for Agronomical Research (INRA-Jouy-en-Josas), as the anti-dipteran reference strain used in the commercial product Vectobac ® [30]; the AR23 strain isolated from Lebanese soil [15]; and the BUPM98 strain isolated from Tunisian soil [31]. All three are active against dipteran larvae-particularly mosquitoes-and produce spherical (S) crystals. Finally, a non-cytolytic novel anti-dipteran strain H3, synthesizing spores and spherical crystals, was also considered in this study [32]. The seven strains were kept as 30% glycerol stock solutions at −80 • C. They were defrosted on LB (Lysogenic broth) agar medium at 30 • C, from which a colony was isolated then plated on a T3 agar medium to simulate the sporulation [33]. Cultures were grown in the following optimized conditions: pH 6.8 and 30 • C for 48 h. At the beginning of our study, the culture medium was inoculated with a number of B. thuringiensis cells corresponding to an optical density of 0.15, which was subsequently spread on the petri dishes containing T3 agar medium.
After 48 h, the crystals and spores of the different strains of B. thuringiensis were separated according to a protocol developed by Monsef et al. [34].

Scanning Electron Microscopy (SEM) Image Acquisition
A Hitachi S3200N scanning electron microscope was used to capture the crystal and spore images. With a voltage of 15 KV, the distance between the sample and the target was set at 4 mm. A simple preparation protocol was put in place in order to always have the same concentration of crystals and spores for the different strains when taking SEM images: 100 µL of the stock solution (solution of pure crystals or spores, or a mixture of crystals and spores) was added to 1900 µL of distilled water, to obtain a total of 2 mL. The tube was sonicated for 5 min at a frequency of 37 kHz to break up the aggregation and then the tube was vortexed. After that, 50 µL was taken from the tube and placed on the sample holder covered with cleaved Mica. The sample holder was then placed in an oven at 37 • C to dry. After drying, the sample was metallized, forming a 2.5 nm gold-palladium deposit to allow the crystals and spores to become conductive. Finally, the sample was placed in the vacuum chamber of the SEM and images were taken.

Spore/Crystal Size Measurement Protocol
Spore and crystal size measurements were performed using the ImageJ (NIH, USA) software. Figure 1 illustrates examples of crystal and spore size measurements. For bipyramidal crystals, the height of the bipyramid (2 h) and the side of the square constituting the bipyramid base (BS) were measured (Figure 1a,b). For cubic crystals, the side of the square (SS) was measured ( Figure 1a). For spherical crystals, the perimeter of the circle observed in the images was measured and the mean radius (r) of the sphere synthesized by the israelensis strains and strain H3 was calculated ( Figure 1c). Finally, as shown in Figure 1d, the lengths (L) and the widths (W) of the spores were measured. The aspect ratio for the bipyramidal crystals was calculated by dividing the base of the square of the bipyramid (BS) by its height (2 h). For the spores, the aspect ratio was determined by dividing the width of the spore (W) by its length (L) [35]. The uncertainty (Unc) was calculated according to Student's law while taking into account the number of measurements N(larger than 100) and the standard deviation of the N measurements σ(X i ) as follows: The histograms were then established by choosing the number and widths of bins according to Rice's rule [36].

Statistical Analysis of the Data
XLSTAT software was used for the statistical analysis of the morphological data. Normality distributions of the variables were assessed with Kolmogorov-Smirnov tests. An ANOVA single factor test was conducted, followed by multiple comparison Tukey-Kramer post-hoc tests (HSD). The level of significance was set at α = 0.05.

Anti-Lepidopteran Strains: Collection of Bipyramidal and Cubic Crystals
The following depicts the distribution of crystal sizes of the B. thuringiensis var. kurstaki strains in histograms plotted according to the normal distribution.
The size distribution of the bipyramidal height was plotted for bipyramidal crystals and the size distribution of square side for the cubic crystals. The x-axis corresponds to the size ranges in micrometers, the left-hand y-axis represents the percentage of crystals and the right-hand y-axis represents the cumulative percentage, as shown in Figures 2-4 for B. thuringiensis var. kurstaki strains HD1, LIP MKA and BLB1, respectively. These histograms show a Gaussian distribution for the sizes of bipyramidal and cubic crystals. The average heights of the bipyramidal crystals were 1.508 ± 0.009 µm for the reference B. thuringiensis var. kurstaki strain HD1, 1.549 ± 0.016 µm for the Lebanese LIP MKA strain and 1.531 ± 0.014 µm for the Tunisian BLB1 strain. For the cubic crystals, the average cube edges were 0.573 ± 0.009 µm for the HD1 strain, 0.478 ± 0.014 µm for the LIP MKA strain and 0.614 ± 0.006 µm for the BLB1 strain.   Morphological data for the three kurstaki crystals, and their volumes and aspect ratios are shown in Table 2. ANOVA univariate analysis revealed significant differences between the three morphological characteristics of bipyramidal crystals with p-values lower than 0.05. The null hypothesis was therefore rejected. Although bipyramidal crystals have roughly the same bipyramidal height (2 h) for the three kurstaki strains, the HSD test revealed a statistically significant difference only between the LIP MKA height value and the HD1 one. As for the volume and the aspect ratio that reflects the shape of bipyramid, the post-hoc test highlighted statistically significant differences between the three kurstaki strains. The bipyramidal crystals of LIP MKA were the most stretched and the least voluminous crystals of the three strains, and those of HD1 were the shortest and the most voluminous (Table 2a). Finally, for cubic crystals, the null hypothesis was also rejected (p = 0), and the HSD test displayed significant differences among the three kurstaki strains: BLB1 synthesized the most voluminous cubic crystals and LIP MKA the least voluminous ones.
These three strains carry Cry protein coding genes from the same families, with small variations in the amino acid compositions of the proteins. This includes four Cry1 coding genes, whose produced proteins are responsible for the bipyramid-shaped crystals, and two Cry2 coding genes, producing a cubic crystal [9,20].

Anti-Dipteran Strains: Spherical Crystals by Israelensis and Kin
The three israelensis strains, AM65-52, AR23 and BUPM98, and H3, a novel nonisraelensis anti-dipteran strain belonging to a different phylogenetic group, produce spherical crystals. Histograms in Figures 5-8 represent the size distributions of the crystal radius, calculated as shown in Figure 1c. ANOVA univariate analysis showed significant differences between the strains (p = 0). Multi-comparison HSD tests indicated significant differences among the four strains. The sphere radius increased from the lowest value of 0.370 ± 0.003 µm for the Tunisian BUPM98 strain, to 0.381 ± 0.003 µm for the Lebanese AR23 strain and 0.394 ± 0.009 µm for the reference B. thuringiensis var. israelensis strain AM65-52; and reached the highest value of 0.411 ± 0.003 µm for the Lebanese non-israelensis strain H3.    The crystal components of the israelensis strains are highly different than that of H3, with H3 having a peculiar set of crystal proteins ( Table 1).
The morphological data of the spherical crystals and their volumes are summarized in Table 3. ANOVA univariate analysis (p = 0) followed by HSD test revealed a small, and yet statistically significant variation among spherical crystal volumes. Crystals of B. thuringiensis var. israelensis strain AM65-52 showed the largest volume of 0.283 ± 0.021 µm 3 . Furthermore, the crystal volume of the non-israelensis strain H3 was significantly larger than those synthesized by the israelensis strains AR23 and BUPM98, as indicated by the HSD tests.

B. thuringiensis Spore Morphology and Size Distribution
After analyzing the morphological results for the crystals of the different strains, we analyzed the morphological data of their spores, in order to see if we could find any differences in size and shape among the different strains of B. thuringiensis. Histograms of the size distribution of the spore length for the seven strains are presented according to the normal distribution. The x-axis corresponds to the spore size ranges, the left-hand y-axis represents the percentage of the number of spores belonging to the ranges and the right-hand y-axis represents the cumulative percentage ( Figure 9). While comparing spore size distribution histograms, a variation in the mean size values of the different strains was noticed. For instance, when comparing kurstaki strains, the highest frequency of measured spores for HD1 has a size of [ AR23 and BUPM98, respectively. This difference was also obtained in the variation of the average spore width observed for each of the seven B. thuringiensis strains, as shown by the morphological data reported in Table 4. Since the length mean values between the seven strains were significantly different (p = 0), the null hypothesis was rejected. The HSD test shows that spores of LIP MKA were statistically different from the other kurstaki strains. This significant difference was also observed between the length mean values of H3 spores and those of the israelensis strains, and between AM65-52 on one hand and AR23 and BUPM98 on the other. No statistical difference was detected between AR23 and BUPM98 strains.
In order to better analyze the morphological results of the spores, we represent in Figure 10 the variation of the aspect ratio and the spores' volume for the seven considered strains. The aspect-ratio of the spores produced by anti-dipteran strains presents a statistically significant difference (p = 0). HSD post-hoc test revealed significant difference between AM65-52 and BUPM98, whereas no difference between BUPM98 and AR23 was detected. Among the israelensis strains, AM65-52 spores are the most spherical. H3 strain synthesizes spores that are more spherical in shape than the israelensis strains, since they presented the highest aspect ratio of 0.754 ± 0.072. This was also revealed by the post-hoc HSD test. As for the anti-lepidopteran strains, a significant difference was also detected (p = 0). HSD test showed no significant difference between the aspect ratios of LIP MKA and BLB1, meaning that they form spores with identical shapes. However, the spores synthesized by the reference kurstaki strain HD1 are more spherical since they are significantly different from those of the two other kurstaki strains.
Concerning the spore volumes, two remarks are noteworthy. First, the reference B. thuringiensis var. kurstaki strain HD1 and BLB1 strain produce spores with much smaller volumes than their fellow kurstaki strain LIP MKA , with average volumes of 0.476 ± 0.104, 0.465 ± 0.032 and 0.603 ± 0.018 µm 3 respectively. Indeed, a significant statistical difference (p = 0) was highlighted and the HSD test distinguished the volume of the LIP MKA strain from those of the two other kurstaki strains. Second, for the strains synthesizing spherical crystals (p = 0), the reference strain AM65-52 produced the less voluminous spores with an average volume of 0.426 ± 0.036 µm 3 , compared to more voluminous spores produced by BUPM98, AR23 and H3, with average volumes of 0.540 ± 0.015, 0.584 ± 0.012 and 0.620 ± 0.049 µm 3 , respectively. In this case, the null hypothesis was rejected (p = 0), and the HSD test showed a difference in the statistical data between the spore volume of AM65-52 and the other israelensis strains, whereas no difference was revealed between AR23 and BUPM98. Furthermore, HSD test exhibited a significant difference between AM65-52 and H3.

Discussion
In this study, the morphologies and size distributions of the crystals and spores of seven different strains of Bacillus thuringiensis were analyzed, with an emphasis on the two most used and studied serovars: the anti-lepidopteran reference serovar, kurstaki, and the anti-dipteran one, israelensis, in addition to a strain with anti-dipteran activity, but that is phylogenetically different from the latter group.
On one hand, we found that all kurstaki strains synthesize bipyramidal crystals with sizes that are statistically different. HD1 produces the most voluminous and shortest bipyramidal crystals, and LIP MKA produces the least voluminous ones with a more stretched form than the other two-a characteristic related to the way that the crystal proteins are associated together. As for the cubic ones, a significant variation was observed in crystal size between strains, with BLB1 showing the largest cubic crystal and LIP the smallest one. Composition of the bipyramidal and cubic crystal protein coding genes of the three B. thuringiensis var. kurstaki strains HD1, LIP MKA and BLB1 showed little alteration in the coding DNA sequences and the promotor regions. In fact, these strains, being from the same serovar, showed high DNA sequence similarity in their toxin-carrying plasmids. However, the level of expression of the toxin coding genes, particularly those encoding proteins of the Cry2 family and producing cubic crystals, needs further investigation to elucidate the reason behind the larger size of the BLB1 cubic crystals.
As for the B. thuringiensis var. israelensis strains, it is remarkable that AM65-52 crystal showed a larger volume than those of AR23 and BUPM98, despite having one less Cry4B protein, thanks to a pseudogeneized version of a cry4Ba gene in AM65-52, vs. a functional one in AR23 and BUPM98 [15]. This pseudogeneized cry4Ba gene in AM65-52 means that the corresponding protein is absent from this strains' crystal, but is present in those of AR23 and BUPM98, as shown in previous studies [15,28]. In this case as well, a difference in the expression rate and copy numbers of the protein variants in the crystal, found in all three strains, cannot be excluded and requires further transcriptomic and proteomic analysis.
Furthermore, the H3 strain synthesizes larger spherical crystals than the israelensis strains. This is most likely due to a different crystal protein composition and the fact that H3 belongs to a different phylogenetic cluster [32].
Nevertheless, we must keep in mind the fact that a change of culture medium does affect the production of B. thuringiensis crystals and their size [37]. Hence, this morphological analysis should be conducted after production in various culture media in order to select the optimal one for the best crystal yield.
On the other hand, in this study, we took a closer look at the intra-and inter-serovar differences of the size and shape (length, width, volume and aspect ratio) of B. thuringiensis spores, whereas previous studies focused on the spore size distribution among various Bacillus species [35,38,39]. Nonetheless, the variations of morphological attributes of spores among B. thuringiensis serovars were yet to be elucidated, and that was the focus of our study. Our results did show variations within all these parameters, proving that the spores of the various strains within the same serovar are not identical. Therefore, we first calculated the aspect ratios of the spores of the various strains, allowing us to show that H3 spores are more circular than all the others. As for the intra-serovar variability, HD1 and AM65-52 produce spores that are less ellipsoidal than their fellow Lebanese and Tunisian kurstaki and israelensis strains, respectively. In addition, the spores of the reference B. thuringiensis var. israelensis strain AM65-52 have a smaller volume to the israelensis Lebanese AR23 and Tunisian BUPM98 strains, much like the kurstaki strains in which the reference strain HD1 produces less voluminous spores than the Lebanese LIP MKA . While their number is affected by the change of culture media, the size of B. thuringiensis spores is not [37]. The causes of variation in spore size and morphology are still unknown. Nevertheless, a key player in this variation might be related to the spore forming units, from the inside-out, endospore, cortex, spore-coat, interspace and exosporium basal layer and nap [40]. A change in the size of one of these units, particularly the thickness of the exosporium layers and/or the interspace between the cortex and the exosporium basal layer, can cause a change in the spore size. In B. anthracis spores, the genes responsible for the formation and volume change of each spore unit are well described. Indeed, there is chromosomal genetic relatedness between B. antharcis and B. thuringiensis, and a closeness in spore morphology, allowing the use of B. thuringiensis spores as B. anthracis surrogates in aerosolization tests [39]. However, a detailed description of B. thuringiensis spore forming units and genes is yet to be presented. In this study, we were not able to extract a direct correlation between this variation and the genomic components of the strains, e.g., plasmid size and nature.
With regard to the prospects of this study, a thorough genomic, transcriptomic and proteomic study should be conducted to elucidate the variations of crystal and spore sizes within the same family. In addition, this study could pave the way to future morphological analysis of B. thuringiensis crystals and spores using another optical method, such as laser speckle imaging.