Introduction
DNA-dependent RNA polymerase (RNAP) is an enzyme that is essential to life. The core of the bacterial RNAP consists of five subunits (α2ββ′ω). The RNAP associates with the transcription initiation factor, σ, to form the RNAP holoenzyme (Ebright, 2000). The antibiotic rifampicin, which is used to treat multiple types of bacterial infections, exerts its effect by inhibiting RNAP. The crystal structure and genetic and biochemical data suggest that rifampicin binds to RNAP at a site adjacent to its active center, thereby physically blocking the formation of phosphodiester bonds in the RNA backbone. Rifampicin inhibits any RNA extension greater than two or three nucleotides (Campbell and Holt, 2001).
Resistance to rifampicin (Rifr) arises from mutations in the rpoB gene, which encodes the β subunit of RNAP. These mutations decrease the affinity of RNAP for rifampicin (Xu et al., 2005). In Escherichia coli, the majority of resistance mutations in the RpoB protein are located in three clusters. Cluster I comprises amino acids (aa) 507–533, cluster II comprises aa 563–572, and cluster III comprises aa 687 (Jin and Gross, 1988; Goldstein, 2014). Homologous mutations to those in E. coli have been observed in the RpoB protein of many other bacteria, such as Mycobacterium tuberculosis, Bacillus subtilis, Neisseria meningitidis, Staphylococcus aureus, and Streptomyces coelicolor (Alifano et al., 2015). The majority of the identified mutations are in the cluster I region of RpoB (Goldstein, 2014).
Rifr mutations have pleiotropic phenotypes due to their effects on transcription (Jin et al., 1988a,b; Maughan et al., 2004). For instance, the Rifr mutations S531F, Δ532A, L533P, and T563P in RpoB of E. coli lead to a hyper-temperature-sensitive phenotype in dnaA46 and rpoD800 backgrounds (Jin and Gross, 1991; Zhou and Jin, 1997). In addition, the Rifr mutations Q469R, H482R, and S487L in B. subtilis have global effects on growth rate, competence for transformation, sporulation, and germination (Maughan et al., 2004; Perkins and Nicholson, 2008). Moreover, the Rifr mutations S442L and H437R in Streptomyces spp. lead to a stringent response and an increase in antibiotic production (Hu et al., 2002). Therefore, Rifr mutations have been used to better understand the regulatory mechanisms that are underlying bacterial physiology and virulence, and to manipulate gene expression (for strain improvement and drug discovery) in bacterial species that are of industrial interest (Alifano et al., 2015).
Bacillus velezensis is a relatively novel species. It was first described by Wang L.T. et al. (2008) as a heterotypic synonym of Bacillus amyloliquefaciens. Recently, based on comparative genomics and DNA relatedness calculations, B. amyloliquefaciens subsp. plantarum, Bacillus methylotrophicus, and Bacillus oryzicola were reclassified as heterotypic synonyms of B. velezensis (Dunlap et al., 2015). The distinctive characteristics of B. velezensis include methanol utilization, plant growth promotion, and biocontrol capacity (as a result of the production of multiple antibiotics) (Dunlap et al., 2015). Although, extensive research has been conducted on this species to increase its production of bioactive metabolites (Koumoutsi et al., 2007; Jha et al., 2016; Suthar and Nerurkar, 2016) and/or plant colonization capacity (Wang et al., 2014), it remains unclear whether its biocontrol capacity can be enhanced by mutating the global transcription machinery, RNAP, to produce beneficial biocontrol phenotypes.
In this study, we obtained seven spontaneous mutations in rpoB that confer resistance to rifampicin in an endophytic strain of B. velezensis, CC09. This strain produces the non-ribosomal peptide iturin A (a secondary metabolite with antibacterial and antifungal properties), and it exhibits broad antifungal activity against several phytopathogens, including Glomerella glycines, Rhizoctonia solani, and Alternaria alternata (Cai et al., 2013; Yang et al., 2014). We also evaluated the effects of these mutations on cell growth, pellicle formation, swarming motility, sporulation, and iturin A production.
Materials and Methods
Bacterial Strains and Media
The strain used in this study was B. velezensis CC09, which was previously isolated from an evergreen tree, Cinnamomum camphora, and deposited in the China General Microbiological Culture Collection Center (CGMCC no. 4669).
The medium used for B. velezensis cultivation was either Luria-Bertani (LB) medium (5 g/L yeast extract, 10 g/L tryptone, and 10 g/L NaCl; pH 7.0) or Gailiang LB (GLB) medium (3.75 g/L yeast extract, 11.25 g/L tryptone, 5 g/L starch, and 1 g/L NaCl; pH 7.0). The medium used for the fungi was potato-sucrose (PS) medium (200 g/L potato and 20 g/L sucrose). The solid media were prepared by adding 7–20 g/L agar to the liquid media.
Screening of Rif<sup>r</sup> Mutants and Mapping of the Mutations in the <italic>rpoB</italic> Gene
The wild-type (WT) strain was cultured overnight and spread on LB agar plates (200 μL/plate) containing 50 μg/mL rifampicin. The plates were incubated at 37°C for 48 h to generate spontaneous Rifr mutants. The mutants were cultured in the LB medium containing rifampicin and then stored at -80°C.
Three pairs of primers (Supplementary Table S1) were used for PCR amplification of the rpoB gene. The samples were initially denatured at 94°C for 5 min, followed by 30 cycles of denaturation at 94°C for 30 s, annealing at 55°C for 30 s, extension at 72°C for 2 min, and a final extension at 72°C for 10 min. The PCR products were purified using Axygen AxyPrep Mag PCR Clean-Up Kits (Axygen Biosciences, Inc., Union City, CA, USA), sequenced by BGI (Shenzhen, China), and analyzed using BioEdit software (Hall, 1999) to identify the mutations in rpoB.
Allelic Replacement Associated with the Point Mutations in <italic>B. velezensis</italic> CC09
To verify that the rpoB mutations identified in B. velezensis were sufficient to cause the Rifr phenotypic change, the following experiments were performed. First, chromosomal DNA from six strains with single point mutation (H485Y, H485C, H485D, H485R, Q472R, S490L) and one strain with double point mutation (S490L/S617F) was prepared, and then it was amplified using PCR with the primer pair PMRif1/PMRif2 (Supplementary Table S2) flanking the mutated nucleotide, including upstream (∼800 bp) and downstream (∼1000 bp) sequences (Supplementary Figure S1).
For each mutation, the PCR product was ligated to a pMAD vector (a temperature-sensitive integrative plasmid) that had been digested by EcoR I/Bgl II using an In-Fusion HD Cloning Kit (Clontech Laboratories, Inc., Mountain View, CA, USA) (Sleight et al., 2010). The constructed plasmid, pMADrif, was introduced into competent cells of the WT strain by electroporation. Transformants were obtained after 12 h at 30°C on LB plates containing 5 μg/mL erythromycin. A pool of individual clones was cultured at 40°C in the LB medium containing the same concentration of erythromycin that used in the stationary phase. Two more cycles of growth were carried out by diluting the stationary phase culture into antibiotic-free fresh LB medium (Arnaud et al., 2004). Finally, single colonies were obtained by plating the diluted culture onto LB agar plates free of antibiotics. The bacteria were further screened using a replica plating technique with LB agar plates containing antibiotics (erythromycin or rifampicin) to obtain erythromycin-sensitive but rifampicin-resistant colonies. The colonies were confirmed by sequencing. Regarding the S617F mutation (phenylalanine instead of serine) at 1849 bp that was associated with the double mutant, the mutation was prepared using PCR with two primer pairs (PMRif1/PMRif3 and PMRif4/PMRif2) (Supplementary Table S2). The mutation was cloned into a pMAD vector that was subsequently introduced into competent WT cells.
Determination of Minimum Inhibitory Concentration (MIC)
For each mutant, the minimum inhibitory concentration (MIC) of rifampicin was measured using a previously described method (Clinical and Laboratory Standards Institute [CLSI], 2012).
Observation of Multicellular Growth Pattern
The WT strain and Rifr mutants were cultured overnight. Subsequently, 1 mL of each culture was centrifuged at 14,940 × g for 10 min. For each strain, the pellet was resuspended in an appropriate amount of phosphate buffer (pH 7.0) to reach a concentration of approximately 1 × 108 cells/mL. The resuspended cells were placed at the center of GLB plates (5 μL/plate) containing 2% agar. The plates were incubated at 32°C for 24 h. The multicellular growth pattern of each mutant was photographed.
Measurements of Growth Rate, Swarm Motility, and Sporulation
The vegetative cells of the WT strain and Rifr mutants were cultured in liquid GLB at a temperature of 32°C and a rotational speed of 120 rpm for 10 h for use in the following experiments.
Growth Rate
To assess the growth rate, 1% (vol/vol) of the culture was inoculated in 100 mL GLB at 32°C and 120 rpm in a 250-mL flask for 26 h. In order to plot a growth curve, 1 mL of the culture was taken every 30 min, and the optical density at 600 nm (OD600) was measured using a spectrophotometer (GENESYS 10S UV-Vis Spectrophotometer, Thermo Fisher Scientific, Inc., Madison, WI, USA). The specific growth rate (μ) was calculated using the following formula: μ=2.303(lg OD2−lg OD1)(t2−t1)
where μ represents the specific growth rate, and OD2 and OD1 represent the OD of the culture at sampling time t2 and t1, respectively.
Swarming Motility
To assess swarming motility, 5 μL of the culture was placed in the center of GLB plate containing 0.7% agar, followed by incubation at 32°C for 5 h. The swarming diameter was then measured. The mean swarming rate (mm/h) was calculated by dividing the swarming diameter (mm) by 5 (h) (Kearns and Losick, 2003).
Sporulation
To assess sporulation, the cell culture conditions that were used for the growth curve analysis were replicated. After 26 h, the numbers of spores inside and outside the cells (that were related to the vegetative cells) were counted using a hemocytometer (model XB-K-25, Shanghai Anxin Optical Instrument Manufacture, Co., Ltd, Shanghai, China) (Schaeffer et al., 1965). For each strain, three replicates were performed and a sporulation curve was plotted. The number of hours to 50% sporulation (S50) was calculated based on the sporulation curve.
Determination of Inhibitory Activity against Fungal Spore Germination
For each strain, the ability of the culture filtrate to inhibit the germination of Aspergillus niger spores was assessed. Inhibition was calculated as the ratio of the number of colonies on the plates in the presence of the filtrate to the number of colonies in the absence of the filtrate. Culturing was carried out at 28°C for 2 days, based on a previously described method (Yang et al., 2014).
Quantification of Iturin A
Approximately, 1% (vol/vol) of the overnight culture of each strain was inoculated in 50 mL GLB medium in a 250-mL flask at 32°C and 120 rpm for 48 h. Subsequently, 1 mL of the cell suspension was centrifuged at 14,940 × g for 10 min. The resultant supernatant was analyzed using high-performance liquid chromatography (HPLC), following a previously described method (Cai et al., 2013).
Observation of Pellicle Formation
For each strain, the formation of pellicle was evaluated based on the method proposed by López and Kolter (2010). Briefly, overnight culture was centrifuged at 14,940 × g for 10 min. The pellet was resuspended in GLB to reach a concentration of 1 × 108 cells/mL. The resuspended cells were added (1%, vol/vol) to 3 mL sterile GLB medium in the wells of a 12-well tissue culture plate. The cells were incubated at 32°C for 16 h to observe whether pellicle developed, and, if so, its appearance.
The pellicle weight was assessed in terms of its dry weight using a digital weighing balance (Sartorius TE1502S Talent Analytical Balance, Sartorius AG, Göttingen, Germany) according to a previously described method (Graba et al., 2013). Briefly, the pellicle was carefully harvested from each well using sterile pipette tips, rinsed gently with sterile distilled water, dried in an oven at 120°C until the weight stopped decreasing, and finally weighed using the digital balance.
Quantitative Real-Time PCR (RT-qPCR) Analysis
Based on their antifungal activity, three strains were selected for RT-qPCR analysis: CC09, CC09-RIF5, and CC09-RIF3, that is, the WT strain, the S490L mutant (which exhibited decreased iturin A production), and the H485R mutant (which exhibited increased iturin A production), respectively. The strains were grown in GLB medium at 32°C until the pellicle was completely formed (16 h). Cells were harvested by centrifugation at 14,940 × g for 1 min and stored in liquid nitrogen. The total RNA was extracted using Invitrogen TRIzol Reagent (Thermo Fisher Scientific, Inc., USA). Reverse transcription was used to synthesize cDNA using a random primer from the TaKaRa RT-PCR Kit, D6110A (TaKaRa Bio, Inc., Kusatsu, Japan).
Sixteen B. velezensis CC09-specific genes were amplified: 15 that encode proteins associated with antibiotic production and pellicle formation, and the gene that encodes ribosomal 16S RNA (as the internal reference). The primers used for the amplification of the genes were designed according to the genome sequence of B. velezensis CC09 (GenBank no. CP015443) (Cai et al., 2016) (Supplementary Table S3).
For each mutant, the fold-change in the expression of each gene was calculated by dividing the value for the mutant by the value for the WT strain using a previously described method (Schmittgen and Livak, 2008). A fold-change >2- or <0.5 compared to the WT strain was considered to represent a significant degree of differential expression. The set of samples for each strain involved three biological and two technical replicates.
Data Analysis
GraphPad Prism version 3.02 (GraphPad Software, Inc., San Diego, CA, USA, www.graphpad.com) software was used to assess the significance of the differences using one-way analysis of variance (ANOVA) and Tukey’s multiple comparison test (p < 0.05 was considered statistically significant). A dendrogram of hierarchical clustering was used to illustrate the possible correlation between the mutations in rpoB and the phenotypic changes using an R package for weighted correlation network analysis (Langfelder and Horvath, 2008).
Discussion
The aim of this study was to isolate, sequence, and phenotypically characterize several Rifr mutations in the rpoB gene of B. velezensis, an endophytic bacterium with biocontrol properties. We obtained six strains with single and one with a double point mutation in rpoB that conferred resistance to rifampicin (all of which were in the cluster I region of RpoB). These Rifr mutations have been described in other bacteria, such as E. coli (Jin et al., 1988b), S. lividans (Hu et al., 2002), B. subtilis (Ingham and Furneaux, 2000), M. tuberculosis (Mokrousov et al., 2003), and S. aureus (Wichelhaus et al., 1999). Moreover, these Rifr mutations in B. velezensis influence a wide range of processes, including cell growth, pellicle formation, swarming motility, sporulation, and iturin A production, which might be a result of the regulation of related genes.
Most of the Rifr mutations that have been identified thus far map directly to the domain of, or regions adjacent to, the rifampicin-binding pocket of the RpoB (Jin and Gross, 1988, 1991; O’Neill et al., 2000). The aa substitutions at D516, H526, and S531 (using the E. coli RpoB numbering system) frequently confer high levels of resistance to refamipicin in Rifr mutant clinical isolates (Alifano et al., 2015). Among the 14 Rifr mutants sequenced in this study, eight (57.1%) had aa substitutions at H485, and two (14.3%) had aa substitutions at S490, which correspond to homologous substitutions at H526 and S531, respectively, in E. coli RpoB (Table 1). These results suggest that the frequency of Rifr mutations in the rpoB gene of B. velezensis was probably similar to that in other bacteria such as E. coli. In addition, all the aa substitutions at H485 in RpoB of B. velezensis led to the highest MIC of rifampicin among the test strains (Table 2), which is consistent with the findings of a study by Alifano et al. (2015).
The substitution of the glutamine (Q), histidine (H), and serine (S) residues in the cluster I region of RpoB (Q513, H526, and S531 in E. coli; Q469, H482, and S487 in B. subtilis; Q472, H485, and S490 in B. velezensis) are the most effective mutations in terms of the promotion adaptive responses. They cause global changes in the bacteria, such as alternative substrate utilization, secondary metabolism, sporulation/germination, and antibiotic heteroresistance (Zhou and Jin, 1998; Maughan et al., 2004; Alifano et al., 2015). However, the effects of Rifr mutations on phenotypes vary both within and between bacterial species. For instance, compared to the WT strain, the S487L mutation in the rpoB gene of B. subtilis led to a lower specific growth rate, while Q469R and H482R/Y mutations did not affect the growth of the strain when it was cultured in LB medium (Maughan et al., 2004). The Q469R and S487L mutations dramatically inhibited sporulation while H482R/Y exhibited a weak impact or no impact on spore formation in B. subtilis (Moeller et al., 2012). In this study, almost all of these mutations (Q469R, S487L, and H482R/Y) in B. velezensis exhibited inhibitory effects on spore formation.
Interestingly, we isolated a B. velezensis CC09 strain with a double point mutation (CC09-RIF7), which had S490L and S617F substitutions in RpoB. Although the S617F mutation found in the double mutant CC09-RIF7 was not sufficient to confer resistance to rifampicin, it exhibited compensatory effects on the phenotype defects (e.g., defects in pellicle formation, sporulation, and iturin A production) caused by the S490L mutation. This was probably due to a functional or structural interaction between the two sites (Jin and Gross, 1988).
According to the classification system proposed by Alifano et al. (2015), the Rifr mutations in the cluster I region of RpoB of B. velezensis can be clustered into three groups: (i) H485 mutations that lead to a decrease in transcription termination; (ii) Q472 mutations that result in an increase in transcription termination; and (iii) S490 mutations that cause a decrease in antitermination. This classification is relatively consistent with the dendrograms of hierarchical clustering based on the phenotypes (Figure 4) and the relative expression of the 15 functional genes (Supplementary Figure S5) that we observed. Mutants with substitutions in RpoB at aa position H485 (H485Y, H485C, H485R, and H485D) were classified into a group with a high level of iturin A production and increased inhibition of fungal spore germination. In fact, several studies on multiple species revealed that H485 mutations are closely associated with the initiation of, and increase in, antibiotic production (Ingham and Furneaux, 2000; Hu et al., 2002; Wang G. et al., 2008; Tanaka et al., 2013). We intend to explore the correlations between the phenotypes (e.g., pellicle formation, sporulation, and iturin A production) and gene expression in a future study.
The effect of the Rifr mutations on antibiotic production is dependent on the type of antibiotic and bacterial species. For example, the Rifr mutations H437R and S442L significantly enhance the production of streptomycin by S. griseus (Tanaka et al., 2013), while homologous mutations inhibit the production of iturin A in B. velezensis. The H439R/D mutations in RpoB increase the production of erythromycin, while Q426R reduces the production of erythromycin in Saccharopolyspora erythraea (Carata et al., 2009). These mutations (i.e., H485R/D and Q472R) had similar effects in B. velezensis CC09 with respect to iturin A production (Table 2).
In order to understand the possible relationship between iturin A production and gene expression, we performed an RT-qPCR analysis of two genes, ituA and ituB in the iturin A operon, which encode synthetases in CC09-RIF3 (H485R), CC09-RIF5 (S490L), and WT strains of B. velezensis. Surprisingly, the production of iturin A (Table 2) was inconsistent with the expression of ituA and ituB in the mutants (Table 3). However, the relative expression of yczE, which encodes an integral membrane protein, in the CC09-RIF3 (H485R) strain was significantly higher than that in the CC09-RIF5 (S490L) and WT strains. YczE has been reported to be a candidate for anchoring the synthetases at the membrane and enhancing the production of bacillomycin D, which belongs to the iturin A family (Koumoutsi et al., 2007). Thus, the production of iturin A by B. velezensis may be regulated at the post-transcriptional stage (Koumoutsi et al., 2007).
Mature pellicles are embedded in an extracellular polymeric matrix, which is composed of exopolysaccharides, proteins, adhesins, and, occasionally, DNA (Luary et al., 2014). They have a significant impact in medical, industrial, and agricultural settings; for example, they can enhance bacterial colonization of plants (Hallstoodley et al., 2004). Many extracellular metabolites such as surfactants (e.g., iturins and surfactins), polysaccharides, γ-polyglutamic acid (PGA), and proteins (e.g., CheV) are associated with pellicle formation.
The formation of pellicles is mainly regulated via Spo0A/AbrB, SinI/SinR, and DegS/DegU regulatory systems in Bacillus species (Chu et al., 2008; Wang et al., 2015). Our RT-qPCR data (based on the analysis of 15 genes associated with antibiotic production and pellicle formation) also provided evidence that these regulatory systems play important roles in the formation of pellicle in B. velezensis CC09. For instance, the relative expression of abrB and sinI, which negatively regulate the expression of genes related to extracellular polysaccharide synthetases, dramatically increased in both the CC09-RIF3 (H485R) and CC09-RIF5 (S490L) mutants, leading to a defect in pellicle formation. The delayed formation and decreased biomass of pellicle in CC09-RIF5 (S490L) may have been caused by down regulation of the expression of degS and degU. This may in turn have resulted in a higher expression of pgdS (Table 3) and, consequently, a reduction in γ-PGA. It has reported that the production of γ-PGA is positively regulated by DegU/DegS and negatively regulated by PgdS (Ohsawa et al., 2009; Scoffone et al., 2013). Therefore, we speculate that the more severe defect in pellicle formation in CC09-RIF5 (S490L) compared to CC09-RIF3 (H485R) resulted from the down regulation of degS and degU together with the up regulation of abrB and cheV.
Compared to the WT strain, the Rifr mutants exhibited various multicellular growth patterns (Figure 2). It has been reported that the formation of multicellular growth patterns is an exceptionally complex process (e.g., it involves cell–cell interaction, communication via chemotaxis signaling, and swarming motility) that requires the regulation of many genes at all levels of gene regulation (Martínez and Vadyvaloo, 2014). For example, the DegS/DegU regulatory system has been reported to control a variety of cell phenotypes, including extracellular protein production, swarming motility, pellicle formation, and the formation of complex structures based on multicellular growth patterns (Verhamme et al., 2007, 2009). The down regulation of degS and degU in the CC09-RIF5 (S490L) strain might have led, at least in part, to the morphological change in its multicellular growth pattern compared to the WT strain. However, the morphological change in the multicellular growth pattern of CC09-RIF3 (H485R) was not due to a change in degS/degU expression (Table 3), which indicates that bacterial multicellular growth patterns are regulated by more than one regulatory system. However, further studies are required to elucidate how the Rifr mutations lead to changes in the multicellular growth patterns of B. velezensis.