The strains and plasmids, as well as the primers used in this study, were listed in Table 1 and Table S1, respectively. Escherichia coli were cultured at 37°C in lysogeny broth (LB) medium (g/L: tryptone, 10; yeast extract, 5; NaCl, 10). The medium was adjusted to pH 7.0 before autoclaving at 121°C for 15 min. Unless otherwise specified, B. thuringiensis strains were cultured at 28°C in the GYS medium (g/L: glucose, 1.00; yeast extract, 2.00; K₂HPO₄⋅3H₂O, 0.66; (NH₄)₂SO₄, 2.00; MgSO₄⋅7H₂O, 0.04; MnSO₄⋅H₂O, 0.04; CaCl₂, 0.08). The medium was autoclaved at 115°C for 30 min after adjusting the pH to 7.8.

To construct the plasmids for phaC (BMB171_RS06655, old locus lag BMB171_C1161) and phaZ (BMB171_RS16270, old locus lag BMB171_C2975) deletions, upstream homologous arms of UphaC and UphaD and downstream homologous arms of DphaC and DphaZ (homologous to the 5′ and 3′ uncoding regions of the target genes) of approximately 750 bp were amplified from the BMB171 genomic DNA by PCR, using primer pairs of DphaCU F/DphaCU R, DphaCD F/DphaCD R, DphaZU F/DphaZU R and DphaZD F/DphaZD R, respectively (Table S1). Then, UphaC (or UphaZ) was inserted into plasmid pRP1028 between the Mlu I and BamH I sites to construct the plasmid pRP1028-UphaC (or pRP1028-UphaZ). Next, DphaC (or DphaZ) was inserted into plasmid pRP1028-UphaC (or pRP1028-UphaZ) between the BamH I and Kpn I sites to construct the integrating plasmid pRP1028-UDphaC (pRP1161) [or pRP1028-UDphaZ (pRP2975)] (Figure S1). pRP1028 was a shuttle plasmid with a temperature-sensitive suicide B. thuringiensis replicon. The resulting integrating plasmids were further verified by sequencing.

The markerless gene deletion system was successfully developed for BMB171 based on an I-SceI mediated replacement method as established in B. anthracis by Janes and Stibitz (2006). The detailed procedures have been well described in a previous publication (Zheng et al., 2015), and will only be briefly accounted here taking deletion of phaC as the example. Firstly, BMB171 (recipient strain), E. coli DH5α containing the integrating plasmid pRP1028-UDphaC (donor strain) and E. coli DH5α containing the helper plasmid pSS1827 (helper strain) used for conjugational transfer (triparental mating) were grown in LB medium. The plasmid pRP1028-UDphaC was integrated into the BMB171 chromosome to form the plasmid-integrated strain by homologous single cross-over recombination. Plasmid-integrated strains were verified by PCR using primer pairs IphaC F/UniversalI R (Table S1). Secondly, the expression plasmid pSS4332 (in DH5α-pSS4332), which produced the restriction endonuclease I-SceI, was conjugationally transfered into the plasmid-integrated strain by a triparental mating system with the help of E. coli DH5α-pSS1827 as well. The I-Sce I endonuclease can recognize and cleave the highly specific 18 bp DNA target site within the integrated plasmid, resulting in chromosomal double-stranded break and thus stimulated the host genetic repair by homologous recombination between the flanking repeat sequences. As a result, there is approximately 50% possibility to delete the target gene from chromosome. Positive deletion strains of phaC were verified by PCR and sequenced (Figures S2 and S3). Finally, the plasmid pSS4332 in the phaC deletion strain was eliminated by continuous passage for ten times in the LB medium at 28°C. Those strains in which the pSS4332 plasmid was removed were sensitive against kanamycin. Deletion of phaZ was performed using the same method (Figures S2 and S4).

Twenty mililiter of a cultured sample at 48 h in LB medium was collected, and pelleted cells were grounded in liquid nitrogen. The procedures for RNA extraction and cDNA synthesis were carried out as previously described (Wang et al., 2013b; Zheng et al., 2015). Each experiment was repeated in triple.

Strains of BMB171-cry, ΔphaC-cry and ΔphaZ-cry were obtained by transformation of the cry1Ac10 promoter and cry gene-containing plasmid (pBMB43-304) (Qi et al., 2015) into the BMB171, ΔphaC and ΔphaZ strains, respectively. pBMB43-304 is a derivative of the low copy number shuttle plasmid pHT304 (Arantes and Lereclus, 1991), into which a cry gene was inserted between the two HindIII sites. The plasmid pBMB43-304 was maintained by growing the bacteria in the presence of 25 μg/mL erythromycin. For the extraction of the ICP (Cry1Ac10), strains were grown at 28°C and 200 rpm for 20 h in GYS medium supplemented with 25 μg/mL erythromycin. After optical density measurement, cells were diluted to a final OD₆₀₀ of 1.0, and 20 mL of each culture was separately collected by centrifuge at 6000 × g for 15 min (AG Eppendorf, Hamburg, Germany). Procedure for the extraction of the ICP was carried out according to a previous study (Wang et al., 2013a). Finally, the ICP was visualized by SDS-PAGE and the ICP concentrations were measured by the Bradford method.

The PHB contents of B. thuringiensis cells were determined by UV absorption spectroscopy at 245 nm (Law and Slepecky, 1961) using PHB extracted from BMB171 as standard.

Sample preparation for transmission electron microscopy was carried out according to the method described in the literature (Tian et al., 2005) with slight modification. Strains were collected at 9 and 19 h, centrifuged at 6000 × g for 3 min, and washed three times with PBS buffer (pH 7.2). Pellets were then resuspended in 2.5% glutaraldehyde phosphate buffer, fixed at 4°C for 24 h. After washing three times with PBS buffer (pH 7.2), cells were dehydrated gradually using different concentrations of ethanol (20%, 50%, 70%, 80%, 90%, and 100%). Each process was carried out twice for 15 min each. Samples were stored in vacuum freeze drier overnight. Thin sections were examined on a HITACHI H-7000FA transmission electron microscope (Hitachi, Ibaraki, Japan).

The larvae of Heliothis armigera were reared at 28°C with a light/dark cycle of 12:12 h. Artificial diet comprised 4 g yeast extraction, 7 g bean flour, 0.5 g vitamin C, 1.5 g agar, 36% acetic acid and 1 g penicillin per 100 mL of water. The medium was transferred into 24-hole cell culture plates (Corning, USA), 1 mL per well. 200 mL of cells grown in GYS medium for 20 h were harvested by centrifugation (8,500 × g, 5 min, 4°C) and suspended in 20 mL distilled water. Cells were then diluted 10,000-fold, with 100 μl of diluted cells applied to each well, allowing it to dry automatically. One first-instar larvae was used for feeding assay per well, with mortality recorded at indicated dates. Meanwhile, body length and weight were measured at the 6th day. Three repeats per bioassay were performed using 24 larvae for each strain (Fiuza et al., 2013).

Besides BMB171-cry, ΔphaC-cry and ΔphaZ-cry strains, BMB171 and BMB171-pHT (BMB171 harboring empty plasmid pHT304) strains were used as controls. OD₆₀₀ of each strain grown in the LB medium at 28°C for 72 h were measured, and diluted to a final OD₆₀₀ of 1.0. Cells were then heated to 65°C for 30 min, followed by gradient dilution (ten times), after that 100 μL of a series of diluents were spreaded onto LB plates. CFU (Colony-forming units) per mL were then counted.

In previous reports, phaC and phaZ were identified as the genes encoding PHB synthetase and degradation enzyme in B. thuringiensis respectively (Tseng et al., 2006; Chen et al., 2010). To disrupt the PHB metabolism pathway, phaC and phaZ were deleted from the parent strain BMB171 by the markerless gene deletion method, with the corresponding strains named as ΔphaC and ΔphaZ, respectively. The growth curves of BMB171, ΔphaC andΔphaZ strains in GYS medium at 28°C were determined. No obvious difference was observed for all three strains, and they all entered the stationary phase at approximate 11 h (Figure 1).

The amount of PHB produced was measured during the whole growth period. For the parent strain BMB171, it was found that its intracellular PHB level started to accumulate from 3 h and reached to a maximum level at 11 h, after which it descended rapidly to half amount at about 16 h and approximately to zero at 21 h (Figure 2). No PHB could be detected in the ΔphaC mutant over the whole growth cycle, indicating that PHB accumulation in the ΔphaC mutant was totally abolished. This data indicated that PhaC was absolutely required for the PHB synthesis (Figure 2). As for the ΔphaZ mutant, the accumulation stage was similar, but there was a delay in the degradation stage. It began to degrade at 16 h, with about half amount left at 21 h. Since PHB degradation wasn’t completely abolished in ΔphaZ, we speculated that there were some other PHB degradation enzymes or pathways existing in B. thuringiensis.

BMB171 is an acrystalliferous strain that doesn’t produce ICPs (Li et al., 2000; He et al., 2010). To investigate its morphological changes in the absence of phaC and phaZ, the ICP coding gene cry1Ac10-containing plasmid pBMB43-304 (Qi et al., 2015) needs to be introduced into this strain, and the ΔphaC and ΔphaZ mutants. The corresponding strains were named as BMB171-cry, ΔphaC-cry and ΔphaZ-cry, respectively. Cell morphologies during the exponential phase and sporulation phase were observed using transmission electron microscope. The cell size and shape of the mutants were found to be similar as the parent strain (Figure 3). The spore and parasporal crystal morphologies of ΔphaC-cry and ΔphaZ-cry also exhibited no major difference compared to those of BMB171-cry under the similar growth conditions.

To be more quantitative, we have further measured the ICP amounts to investigate the influence of phaC or phaZ deletion on ICP formation. As shown in Figure 4, all the three strains produce the ICP without conspicuous difference in their amounts (“ns” indicates not significant, P > 0.05). It is well known that ICPs are virulent to the Helicoverpa armigera larvae. To test whether the toxicities differ in these strains, virulence assays were also performed and the survival rates of the H. armigera larvaes fed with these bacteria were measured (H₂O and BMB171 were set as control groups). Table 2 shows that the survival rates of BMB171-cry, ΔphaC-cry and ΔphaZ-cry were almost similar even after six days (ns, P > 0.05). The body length and weight also didn’t differ much among these three strains (ns, P > 0.05). Together, these results indicate that the insecticidal activities of the ICP produced by the three strains were comparable.

As observed by the transmission electron microscopy in the Figure 3, both the ΔphaC-cry and ΔphaZ-cry strains are able to sporulate similar to the BMB171-cry strain, which is different to the previous result that sporulation was abolished when phaC was deleted from BMB171 (Chen et al., 2012). It is thus important to learn what caused the difference. To make the comparison more objective, three B. thuringiensis strains were cultivated in the same condition as Chen’s experiment, and their spore numbers were counted. No significant difference for the BMB171-cry, ΔphaC-cry and ΔphaZ-cry strains (Figure 5) seems to be present (ns, P > 0.05). In addition, several sporulation-related genes were selected to measure their relative expression levels by RT-qPCR experiments using 16S rRNA gene as an internal control. The sporulation-related genes include those expressed in the pre-divisional cells (spoIIAB and spoIIGA), early mother cells (spoIID) and late mother cells (cotH and gerE) (de Hoon et al., 2010). If the sporulation pathway was impaired in those strains, gene expression level will change. Our data showed that the transcription levels of most sporulation-related genes such as spoIIAB, spoIIGA, spoIID, cotH and gerE didn’t change significantly (Figure 6, ns, P > 0.05). Taken together, these results indicate that the spore-forming ability and transcription of sporulation-related genes were not much impaired in the three strains demonstrating that there is no relationship between the PHB accumulation and sporulation in B. thuringiensis.

To further explore the relationship between PHB metabolism and sporulation, we expanded our investigation further through bioinformatics analysis. Since almost all species in the genus Bacillus except B. infantiscan sporulate, we wondered whether those species can also produce PHB. Distribution of phaC and phaZ genes in the 79 strains of the genus Bacillus with complete genomes from NCBI were investigated (Table S2). Screening of metabolic (KEGG) and genomic (NCBI) database for phaC and phaZ genes revealed that they were absent in the B. subtilis group and some other strains in genus Bacillus (Table S3). Since PhaC is required for PHB synthesis, we speculate that there is no PHB production in those strains lacking the phaC gene. Therefore, no relationship between the PHB accumulation and sporulation could be revealed from this bioinformatics study.