Role of serine/threonine protein phosphatase PrpN in the life cycle of Bacillus anthracis

Reversible protein phosphorylation at serine/threonine residues is one of the most common protein modifications, widely observed in all kingdoms of life. The catalysts controlling this modification are specific serine/threonine kinases and phosphatases that modulate various cellular pathways ranging from growth to cellular death. Genome sequencing and various omics studies have led to the identification of numerous serine/threonine kinases and cognate phosphatases, yet the physiological relevance of many of these proteins remain enigmatic. In Bacillus anthracis, only one ser/thr phosphatase, PrpC, has been functionally characterized; it was reported to be non-essential for bacterial growth and survival. In the present study, we characterized another ser/thr phosphatase (PrpN) of B. anthracis by various structural and functional approaches. To examine its physiological relevance in B. anthracis, a null mutant strain of prpN was generated and shown to have defects in sporulation and reduced synthesis of toxins (PA and LF) and the toxin activator protein AtxA. We also identified CodY, a global transcriptional regulator, as a target of PrpN and ser/thr kinase PrkC. CodY phosphorylation strongly controlled its binding to the promoter region of atxA, as shown using phosphomimetic and phosphoablative mutants. In nutshell, the present study reports phosphorylation-mediated regulation of CodY activity in the context of anthrax toxin synthesis in B. anthracis by a previously uncharacterized ser/thr protein phosphatase–PrpN.

Introduction strain of prpN. Our results identify CodY, a pleiotropic global transcription regulator, as a substrate of serine/threonine phosphatase PrpN and serine/threonine kinase PrkC in B. anthracis. Most importantly, this study demonstrates CodY phosphorylation to be a regulatory switch in the synthesis of anthrax toxins.

Identification and characterization of PrpN protein
BAS0539 was identified as a putative B. anthracis ser/thr phosphatase in the NCBI database's annotation. Conserved domain analysis of the retrieved protein sequence using CDD tool (NCBI) confirmed the presence of a metal-dependent protein phosphatase (MPP) domain characteristic of a large family of mostly eukaryotic-like ser/thr phosphatases (cd00144: MPP_PPP_family), with all the conserved residues of the active and metal binding sites ( Fig  1A). We chose to designate this putative phosphatase as PrpN. The PrpN-encoding gene is flanked in the B. anthracis chromosome by a glycerol-3-phosphate ABC transporter/glycerol-3-phosphate-binding protein and a DNA-binding response regulator ( Fig 1B). The DOOR and MicrobesOnline databases suggested that these genes are not a part of an operon. This prediction was validated using intergenic primers as depicted in the schematic to assess co-transcription of the BAS0538, BAS0539 and BAS0540 genes by RT-PCR ( Fig 1B). A reaction with no reverse transcriptase (NRT) was used as a negative control to check the presence of contaminating genomic DNA in the mRNA sample, while genomic DNA template was used as a positive control. Amplified products of around 1200 bp were observed in lanes 2 and 4 respectively (genomic DNA as template), while no amplification was observed in lane 3 and 5 (NRT) confirming independent transcription of these genes (Fig 1C-left panel). Additionally, rpoB and prpN intergenic primers amplifying a product of around 100-bp were used to ensure successful cDNA synthesis (Fig 1C-right panel). Protein BLAST analysis showed 99.1% sequence identity of PrpN with putative phosphatases in many strains of Bacillus cereus and Bacillus thuringiensis and 95.3% sequence identity with Bacillus subtilis strains (S1A Fig). Evolutionary relationship of PrpN among firmicutes was also examined by generation of a phylogenetic tree (S2 Fig). Interestingly, the observed hits were also annotated as STPs. Also, Streptococcus pneumoniae, a non-sporulating pathogen was clustered in the clade comprising sporulating Bacillus strains. Structural prediction using I-TASSER suggested structural similarity to previously characterized ser/thr phosphatases namely, those of Gallus gallus (PDB ID-1S70) and Escherichia phage lambda (PDB ID-1G5B) and the presence of a manganese ion (Mn +2 ) binding pocket (S1B and S1C Fig).
To biochemically characterize PrpN we cloned, expressed, and purified the protein (as described in Materials and Methods section). We then used the non-specific phosphatase substrate p-nitrophenyl phosphate (pNPP) to confirm enzyme activity and to optimize assay conditions and analyze co-factor requirements. The lone characterized B. anthracis ser/thr phosphatase, PrpC, was used as a positive control and EDTA/EGTA control was used to examine metal ion requirements. The phosphatase assay data indicated Mn +2 as the most preferred metal ion while other ions (zinc, magnesium and calcium) were not able to substitute for Mn +2 in PrpN activation [ Fig 1D-(i)]. Following this, we then used a more specific ser/thr phosphatase assay kit containing ser/thr phosphopeptides and checked for PrpN phosphatase activity in the presence and absence of ser/thr phosphatase inhibitors. The readings were plotted using GraphPad PRISM software and enzyme kinetics parameters for PrpN were calculated using the Michaelis Menten equation [Fig 1D-(ii)]. The above experimental data confirmed PrpN as the second functional ser/thr phosphatase in B. anthracis.

PrpN knockout and complemented strain generation
To study the function of the PrpN ser/thr protein phosphatase in the physiology of B. anthracis, a null mutant of prpN (BAS ΔprpN) was generated in B. anthracis Sterne strain background by a Cre-loxP strategy [47]. The complemented strain (BAS ΔprpN::prpN) was generated by electroporating the BAS ΔprpN strain with a pYS5 shuttle vector carrying the prpN gene along with its predicted promoter sequence (400-bp upstream region) (S3A Fig). Genomic DNAs of BAS WT, BAS ΔprpN and BAS ΔprpN::prpN were used as templates along with prpN gene specific primers to confirm these strains at the genomic level (S1 Table). An amplified product corresponding to prpN gene size (705 bp) was detected in BAS WT and BAS ΔprpN::prpN, while no amplification product was detected in BAS ΔprpN (S3B Fig). Next, the strains were confirmed at the protein level using PrpN specific antibody and purified recombinant PrpN protein as a positive control. A band consistent with the molecular weight of monomeric PrpN protein (~26 kDa) was detected in BAS WT and BAS ΔprpN::prpN lysates and no band was detected in the null mutant strain, confirming the deletion of the prpN gene (S3C Fig).

Temporal expression of PrpN and its role in growth cycle
The role of protein ser/thr phosphorylation in bacterial growth and cell division is a widely investigated topic [48][49][50]. A recent report by our group in B. anthracis showed growth defects in a ser/thr kinase (PrkC) mutant strain, thus establishing a functional correlation between bacterial growth and ser/thr phosphorylation [41]. To examine the role of PrpN in B. anthracis growth machinery, we monitored the growth patterns of BAS WT, BAS ΔprpN and BAS ΔprpN::prpN strains. The bacterial cells were periodically examined microscopically to visualize the vegetative cells morphology. As seen in Fig 2A, BAS ΔprpN strain entered the stationary phase sooner and showed reduced growth yield as compared to BAS WT strain, while the growth pattern of BAS ΔprpN::prpN strain was comparable to BAS WT strain. We also monitored the growth profiles in the presence of salt [1 M sodium chloride (NaCl)] and oxidative [2.5 mM hydrogen peroxide (H 2 O 2 )] stresses and observed an extended lag phase in the absence of prpN (Fig 2B and 2C). Intrigued by the above growth pattern, we next checked for the temporal expression of PrpN in BAS WT by immunoblot analysis at four bacterial growth phases-(i) lag, (ii) exponential, (iii) early stationary (time points indicated in Fig 2A) and (iv) late stationary/sporulation initiation~60 hours [41]. The bacterial cells were visualized microscopically to confirm the growth phase ( Fig 2D-upper panel) prior to protein lysates preparation. Bacterial cells of BAS ΔprpN strain were also examined at the above-mentioned time points to compare the bacterial morphology with BAS WT. The deletion of prpN resulted in twisted and coiled-coil chains and this phenotype was most prominent at stationary time points (Fig 2D-lower panel). GroEL was used as a loading control for densitometer analysis and the results demonstrated maximal expression of PrpN in the stationary growth phase ( Fig  2E and 2F). Interestingly, PrpN showed two bands at stationary growth phase in the was fetched from NCBI GenBank database. (C) Operon prediction by RT-PCR analysis of BAS0538-40 region. BAS0538, 39 and 40 internal gene specific primers as depicted in Fig 1B were

PLOS PATHOGENS
PrpN regulates toxin synthesis in B. anthracis

Absence of PrpN causes structural abnormalities in vegetative bacterial cells
Morphological variations and multi-cellular arrangements such as bacterial chaining, biofilm formation, and host colonization are often linked to bacterial survival and virulence in several pathogenic bacteria including B. anthracis [41,[51][52][53][54]. Bacillus are rod-shape chain-forming bacteria that grows along their longitudinal axes facilitated primarily by re-arrangements of cell wall and cytoskeleton proteins, thereby maintaining a homogenous population of bacterial cells in normal growth conditions [55,56]. Since our previous results implicated PrpN in B. anthracis growth control, we examined BAS WT, BAS ΔprpN and BAS ΔprpN::prpN cells at exponential and stationary phase by phase contrast microscopy ( Fig 3A). A heterogeneous population of bacterial cells and structural aberrations such as coiled-coil structures and bent cells were observed in the BAS ΔprpN strain. Interestingly, these variations were more prominent at stationary growth phase synchronizing with PrpN expression level (Fig 3A-lower middle panel). We then visualized these cells at ultra-structural level by scanning electron microscopy to get a better insight of this phenotype. Strains lacking prpN showed structural deformities and coiled vegetative bacilli mostly at the chain ends and septal regions, while the complemented strain bacilli were similar to BAS WT strain ( Fig 3B). Since, accurate cell septation is crucial for bacterial growth and cell division, we also looked for septation pattern using FM4-64 membrane staining dye in an agarose pad setup (details in material and methods section). Confocal micrographs showed a high population of multi-septa cells (18.2%)  Interestingly, morphological variations and a similar multi-septation phenotype were reported by our group in the absence of the ser/thr kinase PrkC in B. anthracis [41]. This indicates that ser/thr phosphorylation is pivotal for normal vegetative bacterial cell morphology and septation in B. anthracis.

PrpN is necessary for sporulation process
B. anthracis spores are the infectious form of anthrax that can infect the host via three different routes: inhalation, gastrointestinal or cutaneous. These anthrax spores were even used as potential biowarfare agents posing a serious threat to not just cattle but even humans [1,57,58]. Thus, understanding the process of sporulation is crucial in the context of anthrax pathogenesis. While the role of bacterial STKs and STPs in B. subtilis sporulation is an extensively researched and reviewed topic [59][60][61][62][63], there are few studies done in B. anthracis. Our results above indicated maximal expression of PrpN protein in stationary growth phase suggesting its strain. Equal amounts of protein lysates prepared from different growth phases (as indicated in Fig 2A and 2D) were loaded and probed using anti-PrpN and anti-GroEL. Ladder-PageRuler Prestained Protein Ladder, Thermo-Scientific (Cat. No. 26616). (F) Densitometer analyses were done using Amersham Imager600 software and the corresponding PrpN/GroEL ratios were plotted using GraphPad Prism. Densitometer readings calculated from three experiments executed independently are shown in the bar graph. https://doi.org/10.1371/journal.ppat.1010729.g002

PLOS PATHOGENS
PrpN regulates toxin synthesis in B. anthracis possible role during the sporulation process. To examine this hypothesis, we first microscopically monitored BAS WT and BAS ΔprpN bacterial cells in agarose pad prepared with sporulation media. Images were captured at five time points-24, 48, 72, 96 and 144 hours ( Fig 4A). BAS WT showed initiation of sporulation process at 72 hours and release of matured spores at 96 hours (Fig 4A-above panel), while in the BAS ΔprpN strain severe defects in sporulation process with multiple coiled-coil vegetative bacilli were observed until 96 hours (Fig 4Abelow panel). To quantitate this result, sporulation efficiencies and total spore counts of both the strains along with the complemented strain were determined. A severe sporulation defect in the absence of prpN was confirmed while this defect was rescued in the complemented strain (Fig 4B and 4C). These results confirmed a requirement for PrpN in B. anthracis sporulation.
Sporulation is a highly regulated process that involves asymmetric septation of the vegetative cells resulting in the generation of two distinct compartments-mother cell and forespore, forespore engulfment by mother cell, spore maturation and finally the release of mature spore following mother cell lysis [64][65][66]. As sporulation defects were evident in our null mutant strain, we examined BAS WT, BAS ΔprpN and BAS ΔprpN::prpN strains grown in sporulation medium at the ultrastructural level by transmission electron microscopy (TEM). To visualize For this, we examined the viability of the BAS ΔprpN strain for four days after complete sporulation was observed in BAS WT strain. Bacterial cells were harvested and CFU were determined without heat treatment to detect both vegetative bacilli and any viable spores (See Material and Methods). Absence of colonies post 2-4 days confirmed the death of vegetative bacilli and also ruled out the possibility of delayed sporulation in the BAS ΔprpN strain ( Fig  4F). These results signify the indispensable role of PrpN during sporulation process in B. anthracis.

PrpN is required for efficient toxin production
Since, the expression of anthrax exotoxins is crucial for the pathogenic cycle of B. anthracis, we next examined the role of PrpN in toxin synthesis as well as secretion. For this, BAS WT, BAS ΔprpN and BAS ΔprpN::prpN strains were grown in NBY medium containing 1% NaHCO 3 for toxin proteins synthesis [67][68][69]. Whole cell lysates and protein obtained from culture supernatant fractions were probed using PA and LF specific antibodies to examine the synthesis and secretion of toxin proteins. Whole cell lysates immunoblot were stripped and probed again using GroEL antibody for normalization. A drastic decrease in both toxin synthesis (~80% reduction-PA and LF) and secretion (~75% reduction-PA and~70% reduction-LF) was evident in the BAS ΔprpN strain, while an episomal copy of prpN in the mutant strain background (BAS ΔprpN::prpN) was able to rescue this defect and toxin production was comparable to BAS WT (Fig 5A, 5B, 5C and 5D). These results clearly indicate the vital role of PrpN in B. anthracis toxin proteins expression.
Since, defective PA and LF expression was evident in the prpN null mutant strain, we checked for the expression of anthrax toxin activator protein-AtxA (master regulator of anthrax virulence), a prerequisite for anthrax toxins and capsule protein synthesis [20,22,27]. More recently, it was reported that AtxA binds specifically to the promoter region of pagA (encoding PA) and thereby positively regulates its expression [24]. Immunoblot analysis using AtxA specific antibody showed a significant decrease (~60% reduction) in AtxA protein expression in the BAS ΔprpN strain with respect to control strain (BAS WT) ( Fig 5E). The complemented strain (BAS ΔprpN::prpN) showed an AtxA protein level similar to that in the BAS WT strain (Fig 5E), confirming the role of PrpN in regulation of AtxA protein level.  . The blots were stripped and probed using anti-GroEL. Densitometer analyses were done using Amersham Imager600 or ImageLab software and the corresponding PA, LF, and AtxA ratios to GroEL were plotted using GraphPad Prism. Average values and standard deviations calculated from minimum three independent experiments are shown in the bar graphs. Ladder-BlueRAY Prestained Protein Ladder, GeneDirex (Cat. No. PM006-0500). Statistical Analysis: Asterisks indicate statistical significance of the data set calculated using twotailed Student's t test. � corresponds to p<0.05; �� corresponds to p<0.01; ��� corresponds to p<0.001 and ���� corresponds to p<0.0001. https://doi.org/10.1371/journal.ppat.1010729.g005

PLOS PATHOGENS
PrpN regulates toxin synthesis in B. anthracis Taken together, these results suggest that PrpN positively regulates toxin synthesis in B. anthracis at least in part by maintaining the normal amount of AtxA.

CodY is a target of PrpN and PrkC
The data presented so far demonstrates the important role of PrpN in several life stages of B. anthracis, but the underlying PrpN-mediated dephosphorylation events remained elusive. To examine this question, CodY, a global transcriptional regulator reported to be important for growth, sporulation and toxin synthesis in B. anthracis was selected for our study [30,[70][71][72][73]. In B. subtilis, CodY phosphorylation was detected at a serine residue (Ser215) located in its highly conserved DNA-binding domain (helix-turn-helix motif) [74][75][76]. To examine CodY as a potential PrpN target, we cloned the codY gene with a hexa-histidine tag in pYS5 shuttle vector and electroporated the resulting over-expression plasmid in BAS WT and BAS ΔprpN strains. The recombinant CodY protein was purified from both the strains by affinity purification and resolved on SDS-PAGE ( Fig 6A). Next, the relative phosphorylation status of CodY protein was estimated by anti-phosphoserine antibody after normalization to total protein amount. Densitometer analysis showed that CodY from BAS ΔprpN strain contained 3-fold more phosphate than CodY from BAS WT strain (Fig 6B and 6C). Additionally, PrpN mediated dephosphorylation of CodY protein purified from the BAS ΔprpN strain was estimated by measuring the generated inorganic phosphate in the reaction, thus confirming CodY as a target of PrpN ( Fig 6D).
Since, protein phosphorylation at ser/thr residues is regulated by ser/thr kinases and ser/thr phosphatases, we next explored ser/thr kinase mediated phosphorylation of CodY. In view of this, an in-vitro kinase assay was performed using the catalytic (cat) domain of B. anthracis ser/thr kinase, PrkC (GST tag) and CodY (His 6 tag). CodY alone and CodY incubated with ATP were used as controls and the phosphorylation signal was detected using anti-phosphoserine antibody and Pro-Q diamond stain.

Phosphorylation-mediated regulation of CodY activity (at serine215) by PrpN
CodY is a pleiotropic transcriptional regulator that modulates the expression of various genes at exponential as well as stationary growth phase and shows an unaltered expression profile throughout the bacterial growth [72] (S6 Fig). The logical next step was to identify effects resulting from PrpN-mediated CodY dephosphorylation. Since, the serine215 phosphorylation site in CodY protein is located in its DNA binding region, we suspected this modification

PLOS PATHOGENS
PrpN regulates toxin synthesis in B. anthracis

PLOS PATHOGENS
PrpN regulates toxin synthesis in B. anthracis could modulate its DNA-binding affinity to its target genes [75]. To test this conjecture, we generated a CodY phosphomimetic protein by mutating serine215 to aspartate residue (CodYS215E) and a CodY phosphoablative by mutating serine215 to alanine residue (CodYS215A) (S7A Fig). The recombinant proteins were purified by affinity chromatography and the secondary structures were examined by circular dichroism (S7B Fig). Superimposed spectra of native CodY, CodYS215A and CodYS215E showed similar patterns, thus negating the possibility of mutation induced misfolding of CodY proteins (S7B Fig-Right panel). Additionally, to validate in-vivo phosphorylation of CodY at serine215 residue, we generated CodY phosphoablative (CodYS215A with His 6 tag) overexpression strains using pYS5 shuttle vector in the background of BAS WT (BAS WT::codYS215AHis 6 ) and BAS ΔprpN strains (BAS ΔprpN::codYS215AHis 6 ). CodYS215A-His 6 purified from these strains was probed using antiphosphoserine. Auto-phosphorylating catalytic domain of PrkC (GST-tag) and CodY (His 6tag) purified from E. coli BL21(DE3) strain were used as positive and negative controls, respectively. The immunoblot probed using anti-phosphoserine antibody showed a band only in lane 4 (positive control), while no phosphorylation was detected in lanes 2 and 3 (CodY-S215A-His 6 produced in bacillus strains) and lane 5 (negative control) confirming CodY phosphorylation at this specific residue ( Fig 7A). Additionally, the mobility pattern of native CodY and CodYS215A produced in BAS WT and BAS ΔprpN strains was examined using Phos-tag gel system for separation of phosphorylated and unphosphorylated forms of CodY (Fig 7B-upper image). These proteins were also resolved on SDS-PAGE to compare the mobility pattern in the absence of Phos-tag (Fig 7B- The promoter region of AtxA, previously validated as a CodY target in B. anthracis and downregulated in our study, was selected for DNA binding assays [29]. In B. anthracis, CodY positively regulates the accumulation of AtxA protein and toxin proteins and is reported to be crucial for virulence in animal model of anthrax infection [30,73]. Electrophoretic mobility shift assay (EMSA) was performed using an increasing concentration of the CodY proteins (CodYS215E, CodYS215A and native CodY) and a fixed amount of atxA promoter as probe. A prominent DNA shift was evident in the case of CodY phosphoablative mutant (CodYS215A) and native CodY, while CodYS215E mutant showed no DNA binding with atxA promoter region even at higher protein concentration (Fig 7C). Interestingly, CodYS215A showed higher binding affinity with atxA as compared to native CodY suggesting a charged based interaction of CodY protein with its target DNA sequences (Fig 7C). We also examined CodY binding pattern in the presence of GTP and BCAA (branched chain amino acids), previously reported as positive regulators of CodY binding with its the target genes including atxA [29,[77][78][79]. As expected, native CodY and CodYS215A binding affinity was significantly enhanced in the presence of these effector molecules, while CodYS215E still showed no DNA binding (S7C Fig). Next, since the expression of AtxA protein was significantly reduced in the BAS diamond phospho specific gel stain to examine the phosphorylation status (upper image) and Coomassie stain to confirm loading pattern (lower image). Lane 1: CodY-His 6 produced in E. coli using pETDuet without PrkCcat; Lane 2: CodY-His 6 produced using pETDuet with PrkCcat; Lane 3 and 4: CodY-His 6 produced using pProEXHTc.

PLOS PATHOGENS
PrpN regulates toxin synthesis in B. anthracis ΔprpN strain (Fig 5E), we checked the transcript level of atxA mRNA in the BAS WT and BAS ΔprpN strains. Consistent with the AtxA protein levels, atxA mRNA expression was also downregulated in the BAS ΔprpN strain (Fig 7D). Finally, a CodY phosphoablative (CodYS215A) overexpression strain was generated in the BAS ΔprpN strain background (BAS ΔprpN::codYS215A) to mimic the in-vivo CodY phosphorylation level under native conditions (BAS WT). Toxins (PA and LF) and AtxA protein synthesis was examined and compared to BAS WT and BAS ΔprpN strains by immunoblot analysis (Fig 8A, 8B and 8C). The results indicated a partial reversion in toxins and AtxA protein level in the BAS ΔprpN::codYS215A strain, confirming unphosphorylated CodY as a positive regulator of AtxA and successive toxins synthesis in B. anthracis.

Discussion
Reversible protein phosphorylation at specific ser/thr residues is an important post-translational modification by which physiological signals are transmitted and thereby regulate multiple cellular functions [80][81][82][83]. Ser/thr phosphorylation is catalyzed by ser/thr kinases by the transfer of phosphate group from ATP, while removal of phosphate group requires another class of enzyme namely, ser/thr protein phosphatases. Ser/thr protein phosphatases are further classified into three groups based on their protein structure and catalytic mechanism. These are metal-dependent protein phosphatases (PPM), phosphoprotein phosphatases (PPP) and haloacid dehalogenase (HAD). Amongst these, PPM is the most highly conserved STP family amongst all kingdoms of life [84]. Further, the role of STPs is extensively reviewed by several groups and linked to crucial cellular pathways such as cellular division, protein translation, immune response and pathogenesis in prokaryotes and eukaryotes [85][86][87][88][89][90][91]. For instance, M. tuberculosis encodes 11 STKs and a lone essential STP-PstP, which is reported to be vital for growth and virulence [92]. In B. anthracis too, three STKs (PrkC, PrkD and PrkG) and only one non-essential STP, PrpC have been functionally characterized [43,45]. This was the driving thought of our study leading to the functional characterization of another ser/thr protein phosphatase-PrpN in B. anthracis. Sequence analysis of this putative ser/thr protein phosphatase revealed the presence of PPM family domain, also known as PP2C phosphatases (Fig 1A) [84]. Biochemical characterization of PrpN was done using synthetic ser/thr phosphopeptides to validate it as a ser/thr protein phosphatase (Fig 1D).  Apart from these, deletion of prpN showed defects in biofilm formation, while this defect was complemented in the BAS ΔprpN::prpN strain (S8 Fig). These were very interesting findings, but how is one ser/thr protein phosphatase regulating all the above-mentioned cellular pathways? To investigate and answer this important question, we checked the expression profile of PrpN protein and observed its expression level peaking at the onset of stationary growth phase (Fig 2E and 2F). In the model organism B. subtilis, the precise temporal expression of different STKs/STPs is linked to the regulation of specific cellular and developmental pathways such as biofilm formation, sporulation, and germination [59,62,93]. The reported data and our results suggested possible role of PrpN in transition and stationary phase genes regulation. To gain further insight, we focused the next part of our study on toxin synthesis as it is one of the most studied transition phase phenomena crucial for bacterial survival and virulence. While in the recent years considerable progress has been made in the host cellular pathways that are triggered by anthrax toxins and its mechanism of action resulting in the disease progression inside host, far less is known about the toxin synthesis paradigm in the life cycle of B. anthracis. In the present study, we have tried to explore this aspect and presented a novel ser/ thr protein phosphatase mediated regulation of anthrax toxin expression via a global transcription regulator.
Considering the widespread ecological niches in which Bacillus is found, bacterial adaptation in response to environmental stimuli becomes imperative for its survival and growth. Modulation of bacterial transcriptome by global transcriptional regulators is one of the major ways by which it adapts to environmental fluctuations. Bacterial transcriptional regulators are often regulated by protein phosphorylation resulting in differential binding to the promoter regions and thereby expression/repression of various target genes [94][95][96][97]. CodY, a GTP binding global transcriptional regulator of over 500 genes including various stationary and virulence genes in B. anthracis was selected for our study [29]. Previous reports also indicated it as a potential target of ser/thr phosphorylation machinery and identified phosphorylation site at a serine residue (Ser215) in B. subtilis and B. anthracis [74,98]. Since the expression of CodY is constant throughout the bacterial growth phase (S6 Fig), additional regulatory mechanisms such as protein phosphorylation might contribute to the modulation of CodY cellular activity [72]. This hypothesis was indeed reflected in our study as hyperphosphorylated CodY protein level (3 times) was detected in the prpN null mutant strain (Fig 6B and 6C). In-vitro ser/thr phosphatase assays of CodY purified from BAS ΔprpN strain incubated with PrpN further validated CodY as a target of PrpN (Fig 6D). Since protein phosphorylation at ser/thr residues is a reversible modification requiring specific ser/thr kinases and ser/thr phosphatases, we examined phosphorylation of CodY by the most well studied ser/thr kinase, PrkC in B. anthracis and our results confirmed it as a target of PrkC (Figs 6E, 6F and S5A). Though, the present study is limited to regulation of CodY phosphorylation by PrkC and PrpN, other ser/thr kinases (PrkD and PrkG) and ser/thr phosphatase (PrpC) might also contribute to this phenomenon. This assumption, however requires experimental evidence and further validation. Moving forward, our results also confirmed in-vivo phosphorylation of CodY specifically at serine 215 residue (Fig 7A and 7B). DNA binding assays of CodY phosphorylation site mutants revealed that CodY phosphorylation impedes its DNA binding activity with the atxA promoter gene (Figs 7C and S7C). A similar aberration in DNA binding activity is previously reported for the transition state transcriptional regulator-AbrB by phosphorylation at a serine residue in B. subtilis [99]. In B. anthracis, absence of codY resulted in severe AtxA and toxin protein synthesis defect thus indicating it as an activator of toxin synthesis, while AbrB is reported as a repressor of toxin synthesis [20,[28][29][30]. This indicates that a multifactorial regulatory network drives the expression of anthrax toxins and probably that is why toxin synthesis was only partially regained on overexpression of CodYS215A in the BAS ΔprpN strain (Fig 8).

PrpN regulates toxin synthesis in B. anthracis
Based on these studies and our results we conclude that not just CodY presence but the in-vivo CodY phosphorylation stoichiometry mediated by PrpN is also crucial for the activation of anthrax toxins (Fig 9 Graphical Abstract). Further, a recent report in B. cereus showed CodY mediated regulation of clhAB2 operon involved in cell shape and bacterial chaining [100]. To study the functional implication of in-vivo CodY phosphorylation status in these processes, we examined the morphology of BAS ΔprpN::codYS215A strain and compared it with BAS WT and BAS ΔprpN strains. Morphological aberrations such as twisting and bending in the bacterial chains were considerably reduced in the BAS ΔprpN::codYS215A strain at both exponential and stationary growth phase (S9 Fig). This indicates that the regulation of CodY phosphorylation status by PrpN is critical for rod-shape maintenance of metabolically active vegetative bacterial cells in B. anthracis. Moreover, though the mechanistic aspect and PrpN targets in context of sporulation were not explored in the present study, a previous study in B. anthracis

PLOS PATHOGENS
PrpN regulates toxin synthesis in B. anthracis demonstrated CodY as a repressor of sporulation [72]. This suggests that the effect of PrpN on sporulation depends on other regulators or factors downstream of CodY in the sporulation cascade.
Further, since CodY HTH motif is highly conserved among CodY homologs and a direct role of CodY in virulence and toxin synthesis is widely reported in various Gram-positive pathogens, the present study can be extended to other bacterial strains as well and might suggest a similar phosphorylation dependent modulation of CodY DNA binding ability and thereby its target gene regulation [75,101,102]. Most importantly, as lower toxin gene expression accounts for lower virulence, the strain used in this study might also be attenuated for virulence. Thus, it can be further explored for the generation of novel vaccine strain against anthrax and help in overcoming the shortcomings of the currently available vaccines by providing the required balanced level of virulence attenuation along with retention of protective ability.

Ethics statement
All the animal experiments were conducted in accordance with the guidelines mentioned by the Institutional Animal Ethics Committee (IAEC) of Department of Zoology, University of Delhi, India. Balb/c mice were used for raising polyclonal antibodies used in the manuscript and animal handling was done in accordance with the IAEC approved letter with the protocol number-DU/ZOOL/IAECR/2019.

Bacterial strains and growth conditions
Escherichia coli (E. coli) strains DH5α was used for cloning and BL21(DE3) was used for the recombinant proteins production. For electroporation in bacillus cells, the plasmids were further passed through E. coli SCS110 (dam -/dcm -). Luria Bertani (LB) broth (Difco) and LB Agar (Difco) was used for growing E. coli strains. B. anthracis Sterne (BAS) strains were either grown in LB broth/Agar (Difco) or Nutrient Broth Yeast (NBY) Extract medium (0.8% nutrient broth, 0.3% yeast, and 0.5% glucose based on the experimental requirement. Ampicillin and kanamycin antibiotics were used for selections in E. coli and B. anthracis at a final concentration of 100 μg/mL and 25 μg/mL, respectively. Cultures were grown at 37˚C with proper aeration (1:5 head space) and shaking at 200 rpm. For toxin synthesis, strains were grown in NBY medium containing 1% sodium bicarbonate (NaHCO 3 ) for 4 hours at 37˚C. To examine the growth kinetics, glycerol stocks of bacillus strains were streaked on LB agar plate and an isolated colony was used for primary inoculum. Log phase bacillus cultures were used to initiate secondary cultures at a starting A 600nm of 0.02. A 600nm was monitored at 2-hour interval until 14 hours. For stress experiments, either 1 M NaCl (salt stress) or 2.5 mM H 2 O 2 (oxidative stress) was introduced when the cultures attained an A 600nm of approximately 0.1 (lag to log phase transition). Description of plasmids and strains used in the study are provided in S2 and S3 Tables, respectively.

Operon prediction
Genomic arrangement of prpN gene (BAS0539) was fetched from NCBI Database and two intergenic primer sets were synthesized. rpoB and prpN gene specific primers were also used to confirm successful cDNA preparation. PCR reactions were set in a reaction volume of 20 μL containing 2 μL template, 4 μL 5X HF buffer, 2 μL dNTP mix (2.5 mM stock), 2 μL primer mix (10 μM stock) and 0.2 U Phusion DNA Polymerase (Thermo Scientific). DNA Amplification

PLOS PATHOGENS
PrpN regulates toxin synthesis in B. anthracis was performed in a thermocycler (Bio-Rad, T100) with the following parameters: initial denaturation at 95˚C for 10 mins, followed by 35 cycles of denaturation at 95˚C for 30 s, annealing at 54˚C for 1 min, extension at 72˚C for 1 min, and a final extension at 72˚C for 10 mins. The amplified products were resolved on an agarose gel and stained using EtBr. Primers details are listed in S1 Table.

PrpN knockout and complemented strain generation
The single gene knockout strain of prpN in Bacillus anthracis Sterne (BAS ΔprpN) strain was generated using the method described earlier [47]. Briefly, two single crossover plasmids (pSC) derived from the temperature sensitive shuttle vector pHY304 were used in a sequential order for the insertion of loxP sites in the surrounding region of prpN gene (BAS0539). Another, temperature sensitive plasmid pCrePAS2, expressing Cre recombinase was then used for the subsequent excision of the DNA region between the two inserted loxP sites, hence deleting the prpN gene from the BAS genome. Positive colonies were confirmed using prpN gene flanking and internal primers. For complemented strain (BAS ΔprpN::prpN), prpN gene along with 400 bp upstream region was amplified. The amplified product containing KpnI and SmaI sites was digested and ligated in pYS5 shuttle vector [103]. Positive clones were confirmed by sequencing and unmethylated plasmid was electroporated in BAS WT strain and the transformants were selected on LB agar plates supplemented with kanamycin (25 μg/mL). For confirmation, colony PCR was done using pYS5 specific primers. Primers details are listed in S1 Table.

Site-Directed Mutagenesis (SDM) for CodY mutants preparation
CodY Serine215 site was mutated to alanine residue (CodYS215A) and glutamate residue (CodYS215E) following the instructions provided by QuikChange II XL Site-Directed Mutagenesis Kit (Agilent). CodY cloned in pPROEXHTc vector was used as a template for the SDM reactions. The positive clones are named as pPROEXHTc-codYS215A and pPROEXHTc-codYS215E in the study. For pYS5-codYS215A, pYS5-codY plasmid served as a template and a hexa-histidine tag was introduced in the reverse primer for protein purification. All the clones were confirmed by sequencing and primer details are provided in S1 Table.

Cloning, expression, and protein purification
The genes encoding prpN (BAS0539), prpC (BAS3714), atxA (GBAA_pXO1_0146), and codY (BAS3679) were cloned in pPROEXHTc vector, which produces proteins having an N-terminal hexa-histidine (His 6 ) tag. E. coli BL21 (DE3) cells were transformed with these recombinant plasmids for protein production and the recombinant His 6 -tagged proteins were purified using the method described earlier [41]. For purification of PrkCcat-GST tagged protein, the gene encoding the catalytic domain of prkC was cloned in pGEX5x3 vector and the recombinant protein was purified using the method described earlier [104]. For purification of CodY protein from bacillus strains, the genes encoding codY and codYS215A were cloned in pYS5 vector. A His 6 tag was incorporated in the reverse primer for the production of CodY-His 6 tag protein. Bacillus strains harboring this plasmid (pYS5-codY His6 or pYS5-codYS215A His6 ) were grown at 37˚C at 200rpm. The cultures were harvested and the cell pellets were washed thrice with 1X PBS followed by resuspension in lysis buffer [50 mM Tris-HCl (pH 8.0), 300 mM NaCl, 1 mM PMSF, 10% glycerol, 1X protease inhibitor cocktail (Roche Applied Science) and 1X phosphatase inhibitor cocktail (Sigma-Aldrich)]. The cells were sonicated (25% amplitude, 30:20 ON:OFF, total time 8 mins) and the cell lysates were centrifuged at 20,000 × g at 4˚C for 30 mins. The supernatant containing recombinant proteins was incubated with Ni-NTA

Circular dichroism (CD) spectroscopy
CD spectroscopy was performed to detect the structural integrity and mutation induced folding properties of CodY proteins used in the present study (CodY native, CodYS215E and CodYS215A). The CD spectrum of purified proteins was recorded on JASCO, J-815 CD spectrophotometer and the data was plotted in a graph to examine structural properties of these proteins.

Polyclonal antibody generation
30 μg of PrpN, AtxA and CodY proteins were emulsified in Freund's complete adjuvant (Sigma-Aldrich) and injected subcutaneously in three BALB/c mice. Two boosters of 15 μg each (emulsified in Freund's incomplete adjuvant) were given post 15 days for the stimulation of antibody production. All mice were bled 15 days post the 2 nd booster injection and serum was collected. Antibody titer was determined by indirect ELISA.

Bacillus whole cell lysates preparation and immunoblot analysis
Harvested bacterial cells were resuspended in a lysis buffer containing 50 mM Tris-HCl (pH 8.0), 1 mM EDTA, 100 mM NaCl, 1 mM PMSF, 10% glycerol, and 1X protease inhibitor cocktail (Roche Applied Science). The cells were sonicated at 25% amplitude, 30:20 ON:OFF, total time 8 mins and protein was estimated using BCA (Thermo Fisher Scientific). For immunoblots, equal amounts of whole cell lysates (20-40 μg) were resolved on 10% or 12% SDS-PAGE along with a prestained protein ladder and transferred to a NC membrane (Millipore). Following this, the membrane was blocked for 1 hour in 5% BSA prepared in PBST (phosphate buffer saline with 0.1% Tween 20). Membrane was washed thrice with PBST and incubated with a primary antibody [anti-PrpN (1:20,000) or anti-CodY (1:50,000) or anti-AtxA (1:50,000)] for 1 hour. To prevent non-specific antibody binding 1% BSA was included in the primary antibody dilution. The blots were washed 5 times with PBST and incubated with anti-mouse IgG antibody conjugated with horseradish peroxidase (1:20,000 -Cell Signaling Technology, Cat. No. 7076S) for another 1 hour. The blot was again washed 5 times with PBST and finally developed using SuperSignal West Pico PLUS/Femto PLUS Chemiluminescent substrate (Thermo Fisher Scientific) and quantified with the luminescent image analyzer (Amersham Imager600 or Ima-geLab6.0.1). The same blot was stripped and probed similarly using anti-GroEL (1:200,000), used as a loading control. Detection of phosphorylated serine residues was done using commercial anti-phosphoserine antibody (1:2000, Abcam, Cat. No. ab9332) and anti-rabbit IgG conjugated with horseradish peroxidase (1:20,000 -Cell Signaling Technology, Cat. No. 7074S) was used for secondary antibody. For phosphoprotein immunoblots, PBST was replaced by tris buffer saline (TBS) containing 0.05% Tween 20 (TBST). The data were plotted using GraphPad Prism.

PLOS PATHOGENS
PrpN regulates toxin synthesis in B. anthracis

Phos-tag western blotting to examine in-vivo stoichiometry of CodY protein phosphorylation at serine215 residue
Native CodY and CodYS215A with a C-terminus hexa-histidine tag were produced in BAS WT and BAS ΔprpN strains using pYS5 shuttle vector. 2 μg of purified proteins was resolved on SuperSep Phos-tag (50 μmol/L), 12.5% gel (Cat. No. 195-17991) for separation of phosphorylated and unphosphorylated CodY using the method described earlier [105]. These proteins were also resolved on 12.5% SDS-PAGE to examine the mobility pattern of CodY in the absence of phos-tag. The gels were transferred onto a NC membrane and probed using anti-CodY antibody following the protocol mentioned above.

In-vitro kinase assay
In-vitro kinase assay was performed using the method described earlier with slight modifications [38]. Briefly, 100 ng GST tagged PrkC catalytic domain (PrkCcat-GST) was first preactivated with 1 mM ATP. 2 μg CodY-His 6 was then incubated with autophosphorylated PrkCcat-GST in kinase buffer [20 mM HEPES (pH 7.2), 1 mM DTT, 10 mM MgCl 2 and 10 mM MnCl 2 ) containing 1 mM ATP at 25˚C for 30 minutes. Reactions were terminated by 1X SDS sample buffer followed by boiling at 95˚C for 10 minutes. CodY alone and CodY with ATP were used as controls and the protein samples were resolved on 12.5% SDS-PAGE. The gel was transferred onto a NC membrane and probed using anti-phosphoserine antibody to analyze the phosphorylation status of CodY. Additionally, PrkC mediated phosphorylation of CodY was also examined by staining the gel with Multiplexed Proteomics Phosphoprotein Gel Stain Kit #2 containing Pro-Q Diamond and SYPRO Ruby gel stains (Invitrogen, Thermo-Fisher Scientific). Images were captured using Bio-Rad Imager.

Analysis of anthrax toxin proteins-PA and LF
Bacillus strains were grown in NBY broth and the cells were harvested at 12,000g, resuspended in lysis buffer and sonicated (as mentioned in above section). The whole cell lysates were estimated by BCA (Thermo-Fisher Scientific) and used for examining toxin proteins synthesis, while the supernatant was filtered through 0.22 μm filter assembly and the protein was precipitated using 20% trichloroacetic acid (TCA). The precipitated protein was resuspended in a resuspension buffer [1% SDS, 10% glycerol, 10 mM Tris-HCl (pH 6.8), 2 mM EDTA, 100 mM DTT and 1X protease inhibitor cocktail (Roche Applied Science)] and used further for assessing the secretion of PA and LF proteins. Anti-PA (1:50,000) and Anti-LF (1:50,000) raised in rabbit were used for immunoblot analyses.

Spore preparation and quantification
Bacillus spores were prepared using the method described earlier [106]. Half of the sample was heated at 65˚C to kill any residual vegetative cells. Serial dilutions of untreated and heat-treated samples were plated on LB agar plates. The plates were incubated overnight at 37˚C and colonies were counted (CFU) for estimation of sporulation efficiency and total spore count.

PLOS PATHOGENS
PrpN regulates toxin synthesis in B. anthracis

Phosphatase assay
Phosphatase activity of purified PrpN protein was determined spectrophotometrically using para-nitrophenyl phosphate-pNPP (Sigma-Aldrich) as substrate in a 96-well plate. The hydrolysis of pNPP to para-nitro phenolate was measured at 405 nm. The experiment was performed using different pH buffers and time intervals for optimizing the assay conditions. Also, since PPM phosphatases requires metal ions (magnesium or manganese) for phosphatase activity, the experiment was carried out in the presence of different metal ions [5 mM magnesium (Mn +2 ), 5 mM manganese (Mg +2 ), 5 mM zinc (Zn + ) and 5 mM calcium (Ca +2 )]. B. anthracis, ser/thr protein phosphatase-PrpC was taken as a positive control and various negative controls were taken (only buffer, only protein, only substrate and EDTA/EGTA). Once all the required standardization was done, an equal amount of PrpN and PrpC protein (1 μg) was added in the wells containing assay buffer [50 mM Tris-HCl (pH 8.0), 5 mM MnCl 2 , 0.01% β-mercaptoethanol] and 5 mM pNPP. Absorbance was recorded at 405nm after 30 minutes using Epoch multi-plate reader (Biotek) and the calculated PrpN phosphatase activity was plotted using GraphPad Prism. Following this, PrpN ser/thr specific phosphatase activity was measured following the protocol provided by Serine/Threonine Phosphatase Assay System kit (Promega). Briefly, PrpN protein was added in a reaction volume containing assay buffer and synthetic ser/thr phosphopeptides as substrate. The free phosphate in the reaction was measured at 600nm using Epoch multi-plate reader (Biotek). PrpN phosphatase activity was calculated using a standard phosphate curve and the values were plotted in a Michaelis-Menten curve to determine the enzyme kinetics parameters. Additionally, the synthetic ser/thr phosphopeptide used in the above assay was replaced with CodY protein purified from BAS ΔprpN strain and the generated phosphate in the reaction was estimated.

Phase contrast microscopy
Bacterial cells at different growth phase were harvested and processed using the method described earlier [41]. The cells were visualized under Zeiss Axio Imager Z2 upright microscope (100x/1.4 oil DIC objective) and images were procured using Axiocam 506 color camera attached to the microscope. All the images were processed using ZEN 2 Pro software. Sporulation kinetics of BAS WT and BAS ΔprpN strains were studied by live-cell phase-contrast microscopy using agarose pads supplemented with sporulation medium [8g of LB broth/L supplemented with 85.5 mM NaCl, 0.025 mM ZnSO 4 , 0.6 mM CaCl 2 , 0.3 mM MnSO 4 , 0.8 mM MgSO 4 , and 0.02 mM CuSO 4 ] and pH was set to 6.0. Agarose pads were prepared using the AB gene frame on frosted glass slides (Corning Micro slide Frosted; 75 � 25 mm) [41,92]. Low melting agarose (3%) supplemented with sporulation medium was evenly poured on the glass slides and left for solidification. Exponentially growing cultures of both the strains were diluted to an A 600nm = 0.035 and 2 μL of this suspension was spread evenly on the agarose pads. These were then incubated at 30˚C and images were captured at 24, 48, 72, 96 and 144 hours using an inverted microscope.

Electron microscopy (Scanning and transmission)
Exponentially growing vegetative cells of BAS WT, BAS ΔprpN and BAS ΔprpN::prpN strains were processed for scanning electron microscopy using the protocol described earlier [41]. Cells were visualized under Zeiss Scanning Electron Microscope EVO LS15 at 20 KV and analyzed using Smart SEM software. Ultrastructural details of BAS WT, BAS ΔprpN and BAS ΔprpN::prpN spores were examined using transmission electron microscopy. Samples were processed by two different methods to visualize vegetative sporulating cells population using the protocols described earlier

PLOS PATHOGENS
PrpN regulates toxin synthesis in B. anthracis [41,107]. Briefly, for pure spore suspension, sporulation medium was removed after sporulation completion and the harvested spores were kept in water for three days to ensure vegetative cell lysis. These were again washed thrice with water followed by resuspension in water to remove vegetative cell debris and fixed for TEM analysis. To visualize intact cells in the sporulation medium, the bacterial cells were harvested and washed thrice with 0.85% saline, resuspended in the same and fixed. Sectioning was done using Leica UC6 ultra-cut and the sections were then observed in FEI Tecnai G2 Spirit at 200 KV.

Confocal microscopy
Morphological characteristics and septal analysis of BAS WT and BAS ΔprpN strains were done by labeling the cells with the membrane stain FM4-64 (Thermo Fisher Scientific) and observing the live cells by fluorescence microscopy. 2 μL of BAS WT and BAS ΔprpN exponentially growing culture diluted to A 600nm = 0.035 were evenly spread on LB agarose pads supplemented with 1μg/mL FM4-64 dye for staining the cell membrane and incubated at 37˚C overnight. Image acquisition was done using Leica TCS SP8 confocal laser scanning microscope (63X oil immersion objective).

RNA extraction and quantitative real time PCR
BAS WT and BAS ΔprpN strains were grown in triplicates up to the early stationary phase and RNA was extracted by hot lysis method described previously [41]. The purified RNA samples were used to prepare the cDNA using a first-strand cDNA synthesis kit (Thermo-Fisher Scientific). To analyze the expression of atxA in BAS WT and BAS ΔprpN strains, the corresponding cDNA was used along with atxA gene-specific primers and SYBR Green master mix (Roche Life Science), following the instructions provided in the manual. Reactions were prepared in triplicates along and run in a LightCycler 480 Instrument II (Roche Life Science). A no template control was also included to check gDNA contamination and data normalization was done with respect to rpoB ((DNA-directed RNA polymerase β subunit) expression level [41]. Primer details are listed in S1 Table. Electrophoretic mobility shift assay 4 nM atxA promoter gene was used as probe and incubated with increasing amounts of CodY proteins (1 μM-12 μM) in a 20 μL reaction volume containing 1X binding buffer [20 mM Tris-HCl (pH 7.2), 50 mM KCl, 1.5 mM MgCl 2 , 5% glycerol, 0.5 mM EDTA, 1 mM DTT, 0.05% Nonidet P40]. The reaction was incubated at 30˚C for an hour and resolved by electrophoresis using a 6% Native-PAGE. The same protocol was followed in the presence of 1 mM GTP and 1 mM BCAA (Isoleucine, Leucine and Valine). A fluorescence-based EMSA Kit was used for sequential detection of DNA by SYBR Green and protein by SYPRO Ruby in the same gel using the instructions provided in the manual (Invitrogen, Thermo-Fisher Scientific). Images were captured using Bio-Rad Imager.

Biofilm formation
Biofilm formation and efficiency was measured using the method described earlier with some modifications [108]. Briefly, BAS WT, BAS ΔprpN and BAS ΔprpN strains were grown overnight in LB medium and secondary cultures for each strain (1.5% vol/vol) were initiated in 6-well plates containing 5mL LB medium. The plates were incubated at 37˚C for 72 hours without shaking for biofilm formation and quantified by crystal violet assay as described previously [109].

PLOS PATHOGENS
PrpN regulates toxin synthesis in B. anthracis

Statistical analysis
Unless mentioned otherwise, a minimum of three independent experiments (biological) repeated thrice (technical) were done to ensure data reproducibility. GraphPad Software (Prism 6) was used to plot the data and significance of the results was analyzed using two-tailed Student's t test or one-way ANOVA followed by a post hoc test (Tukey test). P values < 0.05 were considered as statistically significant.

Densitometer analysis
Unless mentioned otherwise, all immunoblots densitometer analysis were done using Amersham Imager600.

Fig preparation
All illustrations/schematics were prepared using BioRender and figures were prepared using ImageJ and Adobe Illustrator.