Transcriptomic and phenotypic analysis of paralogous spx gene function in Bacillus anthracis Sterne

Abstract Spx of Bacillus subtilis is a redox-sensitive protein, which, under disulfide stress, interacts with RNA polymerase to activate genes required for maintaining thiol homeostasis. Spx orthologs are highly conserved among low %GC Gram-positive bacteria, and often exist in multiple paralogous forms. In this study, we used B. anthracis Sterne, which harbors two paralogous spx genes, spxA1 and spxA2, to examine the phenotypes of spx null mutations and to identify the genes regulated by each Spx paralog. Cells devoid of spxA1 were sensitive to diamide and hydrogen peroxide, while the spxA1 spoxA2 double mutant was hypersensitive to the thiol-specific oxidant, diamide. Bacillus anthracis Sterne strains expressing spxA1DD or spxA2DD alleles encoding protease-resistant products were used in microarray and quantitative real-time polymerase chain reaction (RT-qPCR) analyses in order to uncover genes under SpxA1, SpxA2, or SpxA1/SpxA2 control. Comparison of transcriptomes identified many genes that were upregulated when either SpxA1DD or SpxA2DD was produced, but several genes were uncovered whose transcript levels increased in only one of the two SpxADD-expression strains, suggesting that each Spx paralog governs a unique regulon. Among genes that were upregulated were those encoding orthologs of proteins that are specifically involved in maintaining intracellular thiol homeostasis or alleviating oxidative stress. Some of these genes have important roles in B. anthracis pathogenesis, and a large number of upregulated hypothetical genes have no homology outside of the B. cereus/thuringiensis group. Microarray and RT-qPCR analyses also unveiled a regulatory link that exists between the two spx paralogous genes. The data indicate that spxA1 and spxA2 are transcriptional regulators involved in relieving disulfide stress but also control a set of genes whose products function in other cellular processes. Bacillus anthracis harbors two paralogs of the global transcriptional regulator of stress response, SpxA. SpxA1 and SpxA2 contribute to disulfide stress tolerance, but only SpxA1 functions in resistance to peroxide. Transcriptome analysis uncovered potential SpxA1 and SpxA2 regulon members, which include genes activated by both paralogs. However, paralog-specific gene activation was also observed. Genes encoding glutamate racemase, CoA disulfide reductase, and products functioning in bacillithiol biosynthesis, are among the genes activated by the SpxA paralogs.


Introduction
Bacillus anthracis is a spore-forming, nonmotile Grampositive bacterium that is the causative agent of the zoonotic infectious disease, anthrax (Beyer and Turnbull 2009). It is an effective pathogen because the infectious agent of anthrax is the metabolically dormant and highly resistant spore. Upon ingestion by a professional phagocytic cell (e.g., activated macrophage), the spore undergoes germination and outgrowth to generate a vegetative cell that is capable of reproduction within the infected host, as it produces plasmid-encoded toxins and protective capsule material for evading immune capture and destruction (Fouet et al. 1999;Koehler 2009;Moayeri and Leppla 2009;Tournier et al. 2009). Germination and outgrowth in the macrophage takes place within a hostile environment, made so by the phagocyte's oxidative burst, which generates a toxic combination of reactive oxygen species (ROS), nitric oxide (NO), and hypochlorous acid (HOCl), as well as phospholipase, and antimicrobial peptides (Piris-Gimenez et al. 2005;Dawson and Liu 2008;Welkos et al. 2011). Successful establishment of infection involves mechanisms of oxidant resistance (Shatalin et al. 2008;Welkos et al. 2011). Such systems in bacteria are activated by encounters with a variety of toxic agents, not only components of the oxidative burst, but antibiotics and other chemical and physical insults (Gusarov et al. 2009;Mols and Abee 2011).
Much of what is known about the oxidative stress response in Bacilli has come from studies of Bacillus subtilis, a nonpathogen, which is a model genetic system used in studying Gram-positive physiology and the bacterial response to harsh conditions. Recent studies of oxidant sensitivity indicated that B. subtilis is more sensitive to the lethal effects of peroxide and superoxide-generating agents than is B. anthracis (Pohl et al. 2011). The findings suggest that more robust processes of oxidant detoxification and tolerance have evolved in the pathogen, which is in keeping with its developmental cycle involving reproduction within phagocytic hosts. Several regulatory proteins govern the oxidative stress response in B. subtilis, including the peroxide sensor PerR (Lee and Helmann 2006), organic hydroperoxide-sensing MarR family protein, OhrR (Fuangthong et al. 2001), HypR, which senses thiol-reactive HOCl stress (Palm et al. 2012), and SpxA (Zuber 2004). All of these proteins have orthologs in B. anthracis (PerR [BA0537], HypR [BA3379], OhrR [BA4699]). Both species possess the general stress response sigma subunit, r B , which controls a large regulon that becomes activated by starvation and reduced energy-generating capability, as well as by stress brought about through encounters with toxic chemical and physical agents (Hecker et al. 2007; van Schaik et al. 2007). Peroxide induces the r B regulon in B. subtilis, but r B is poorly activated by peroxide stress in B. anthracis, which is likely due to the different regulatory architectures that operate in the two organisms (Pohl et al. 2011;Tu et al. 2012b). The response of B. anthracis to superoxide stress, which is likely encountered within the infected macrophage, is the elevated expression of genes within the Fur (Ferric uptake regulator) regulon specifying iron uptake mechanisms, which is also observed in B. subtilis (Mostertz et al. 2004;Passalacqua et al. 2007;Tu et al. 2012a). The response seems maladaptive as it would expose macromolecules to potentially damaging, hydroxyl radicals generated by the Fenton reaction (Liochev and Fridovich 1999;Imlay 2008). Superoxide is known to cause decomposition of enzyme iron centers, which could trigger an iron starvation response through inactivation of the Fur transcriptional regulator and stimulation of the Fur regulon (Mostertz et al. 2004;Passalacqua et al. 2007). However, it has been proposed that superoxide is a germination signal for B. anthracis spores (Fisher and Hanna 2005), and accelerated iron uptake during subsequent outgrowth may assist in coping with the iron-poor environment that characterizes the infected host. In contrast to B. subtilis, the response of B. anthracis to superoxide is limited involving around 40 genes, which might reflect the signaling role of superoxide in B. anthracis infection rather than an agent of general stress generation (Tu et al. 2012b).
Peroxide stress induces over 200 genes in B. anthracis that specify a variety of activities related to detoxification, macromolecular damage repair, and disposal of damaged protein (Pohl et al. 2011). Thus, genes encoding DNA repair enzymes, and other members of the LexA regulon, are induced. Genes functioning in redox homeostasis, including those encoding components of the bacillithiol biosynthesis pathway, are also activated. A significant change in the B. anthracis transcriptome is evident from the observation that several genes encoding regulatory proteins are activated after peroxide treatment. These include genes specifying the PerR and SpxA transcriptional regulators Passalacqua et al. 2007;Pohl et al. 2011).
SpxA is a global regulator of the stress response that is activated upon thiol stress. Over 250 genes, or 144 operons, are controlled by SpxA in B. subtilis (Nakano et al. 2003;Rochat et al. 2012). The protein is a direct transcriptional activator through an interaction with the RNA polymerase alpha subunit in B. subtilis Newberry et al. 2005). Among the genes activated by SpxA are those required to establish the reduced state of thiols in the cytoplasm. Such genes include those that encode thioredoxin (trxA), thioredoxin reductase (trxB), methionine sulfoxide reductase (You et al. 2008), enzymes required for synthesis of the low-molecularweight thiol bacillithiol (Gaballa et al. 2010;Chi et al. 2011;Gaballa and Helmann 2011), and genes that function in cysteine biosynthesis (Nakano et al. 2003;Choi et al. 2006;Zuber et al. 2011;Rochat et al. 2012). A null mutation in spx causes sensitivity to thiol reactive compounds, partial cysteine auxotrophy, and causes disruption of iron uptake control. The spx gene is under complex transcriptional control that is responsive to stress caused by variety of physical and chemical agents Petersohn et al. 2001;Jervis et al. 2007;Leelakriangsak et al. 2007;Eiamphungporn and Helmann 2008). In B. subtilis and in other Gram-positive species, spxA is transcriptionally induced by mechanisms responsive to cell envelope stress. SpxA can undergo stress-induced thiol oxidation of a CxxC disulfide center, which is necessary for its productive interaction with RNA polymerase . The SpxA protein is under proteolytic control that requires the ATP-dependent protease, ClpXP, and a substrate recognition factor, YjbH (Larsson et al. 2007;Garg et al. 2009). SpxA, as might be expected given its role in regulating the oxidative stress response, has been found to be required for virulence in Streptococci and Enterococcus (Kajfasz et al. 2010(Kajfasz et al. , 2012Chen et al. 2012).
Several of the low %GC Gram-positive bacteria possess multiple paralogs of Spx (Veiga et al. 2007;Turlan et al. 2009). The genome of B. anthracis bears two spxA genes, spxA1 and spxA2 (Fig. 1A). The spxA1-linked genes show syntenic similarity to those in B. subtilis, but B. anthracis also contains an additional paralogous spx gene in a part of the genome that shows no synteny with B. subtilis (Fig. 1B). Previous transcriptomic studies have shown that spxA1 is expressed in early log phase of a B. anthracis culture, while spxA2 transcript is detected during stationary phase . The spxA2 gene is one of the most highly induced genes in B. anthracis cells following germination in the host macrophage ). The differences in expression patterns exhibited by the two paralogous genes suggest differences in their roles within the stress response network of B. anthracis.
To uncover the roles of the two paralogous genes of spxA in B. anthracis, a study was conducted to examine the phenotype conferred by null mutations in the genes encoding the two paralogs and to identify the genes that are regulated by SpxA1 and SpxA2. The work reported herein shows that the two paralogous SpxA proteins oversee two large overlapping regulons. While both function in the oxidative stress response, SpxA1 plays an essential role in the bacterium's resistance to peroxide and disulfide stress.

Experimental Procedures
Bacterial strains and growth conditions Bacterial strains and plasmids are listed in Table 1. The B. anthracis Sterne strains used in this study are derivatives of 7702 (pXO1 + pXO2 À ) and, in most cases, were grown in Lysogeny Broth (LB) medium at 37°C or sporulation medium (SM) at 30°C (Barua et al. 2009). The B. subtilis strains used in this study are derivatives of JH642 (trpC2 pheA1) and were grown at 37°C in Difco sporulation medium (DSM) unless otherwise indicated. Escherichia coli DH5a was used for plasmid construction and was grown at 37°C in LB liquid or on LB solid medium containing 1.2% agar (Difco, BD Biosciences, San Jose, CA). Appropriate antibiotics were added: B. subtilis, 75 lg/mL spectinomycin, 1 lg/mL erythromycin/25 lg/ mL lincomycin, 5 lg/mL kanamycin; B. anthracis,   100 lg/mL streptomycin, 100 lg/mL spectinomycin, 20 lg/mL kanamycin; E. coli, 100 lg/mL ampicillin, 100 lg/mL spectinomycin, 20 lg/mL kanamycin.

Construction of B. anthracis spx mutants
Two kilobytes DNA amplicons containing 1 kb up-and 1 kb downstream flanking DNA regions of either spxA1 or spxA2 were generated by overlapping fusion polymerase chain reaction (PCR) using chromosomal DNA from B. anthracis 7702 (see Table S6 for primer sequences). The DNA amplicons were digested with HindIII and KpnI, purified, and ligated with similarly digested and purified pRP1028 using T4 DNA ligase to construct pSB2 (pRP1028::DspxA1) and pSB3 (pRP1028::DspxA2). The plasmids containing in-frame deletion mutations of spxA1 and spxA2 were introduced into B. anthracis 7702 (or a derivative 7702 SR1) using a markerless allelic replacement technique developed by Janes and Stibitz (Janes and Stibitz 2006). Escherichia coli DH5a strains harboring pRP1028:: DspxA1 (pSB2) or pRP1028::DspxA2 (pSB3) were conjugated at room temperature for 24 h with B. anthracis Sterne 7702 in the presence of an E. coli helper strain containing pSS1827 on brain-heart infusion (BHI) plates lacking antibiotic. The transconjugate diploid intermediates (Dspx spx + ) were conjugated for a second time at 37°C with E. coli strains containing pSS4332 (encoding nuclease I-SceI) and pSS1827. Conjugants were screened for the loss of pRP1028 and were subsequently subjected to several rounds of plating on BHI agar with incubation at 37°C to eliminate plasmid pSS4332. Spectinomycin and kanamycin sensitive isolates were obtained and screened by PCR (using primers that anneal outside of the sequence used in the construction of the mutant) and nucleotide sequence analysis of these PCR products was performed to ensure that the final DspxA1 (ORB8170) and DspxA2 (ORB8115 and ORB8438) constructs were correct.

Construction of B. anthracis strains that produce ClpXP-insensitive SpxA1 and SpxA2
The spxA1DD and spxA2DD genes were amplified by PCR from B. anthracis 7702 chromosomal DNA using oMN10-535 and oMN10-536 (for spxA1DD) or oMN10-537 and oMN10-528 (for spxA2DD). The PCR products were digested with HindIII and SphI and ligated into pDR111 (Britton et al. 2002) digested with the same enzymes. The resultant plasmids pMMN818 and pMMN819 contain spxA1DD and spxA2DD under the control of an IPTG (isopropyl b-D-1-thiogalactopyranoside)-inducible P spank (hy) promoter. pMMN818 and pMMN819 were digested with BamHI and were filled-in using T4 DNA polymerase and dNTPs, followed by EcoRI digestion. The released 2.3-kb fragment containing lacI and spxADD was subcloned into pJMA402 (kindly provided by C. Lee and A. D. Grossman), digested with EcoRI and HincII to generate pCSZ35 and pCSZ36. pCSZ35 and pCSZ36 were used to transform B. subtilis JH642 with selection for spectinomycin resistance, resulting in strains ORB7854 and ORB7860, respectively. Donor strains used for ICEBs1mediated conjugation, namely ORB7861 and ORB7862, were generated by transformation of ORB7854 and ORB7860 with chromosomal DNA isolated from JMA475 with selection for kanamycin resistance. ICEBs1-mediated conjugation was carried out as previously described (Auchtung et al. 2005) using ORB7861 or ORB7862 as a donor and 7702 SR1 as a recipient with selection for both streptomycin and spectinomycin resistance. spxA1DD and spxA2DD in B. anthracis conjugants ORB7863 and ORB7864 were verified by sequence analysis of the PCR product amplified from the chromosomal DNA using oligonucleotides spac-up and spac-down (Table S6).

Culture preparation and RNA extraction
Bacillus anthracis Sterne 7702 SR1 and overexpression derivatives bearing P spank(hy) -spxA1DD (ORB7863), P spank (hy) -spxA2DD (ORB7864), and empty vector control strain bearing P spank(hy) -empty (ORB8092), were grown in 60 mL LB 37°C, 200 rpm until mid-exponential phase (OD 600 0.2-0.4) and then split into two equal volumes. One millimoles per liter IPTG was added to one set of cultures and the cultures were incubated for an additional 15 or 45 min, after which the pellets from 10 mL culture samples were harvested by centrifugation (5180g, 4°C, 10 min.), and frozen at À80C until use. For diamidetreated cultures, B. anthracis Sterne 7702 and deletion derivatives, DspxA1 (ORB8170) and DspxA1 DspxA2 (ORB8285), were grown in 40 mL LB at 37°C, 200 rpm until mid-exponential phase (OD 600 0.2-0.4) and split into two equal volumes. One mmol/L diamide (freshly prepared, Sigma-Aldrich, St. Louis, MO) was added to one set of cultures and, following incubation for an additional 20 min, were harvested by centrifugation, as above. RNA was purified as previously described (Igo and Losick 1986).The resulting RNA concentrations were measured by ultraviolet spectrophotometry. RNA quality was assessed by measuring the ratio of absorbance at 260 nm to absorbance at 280 nm, as well as by visualization in agarose gels.
Microarray design, data collection, and data analysis cDNA for microarray experiments were generated by adding 2 lg of total RNA in a mixture containing 6 lg of random hexamers (Life Technologies, Carlsbad, CA), Step-One Plus with Step-One Plus (Life Technologies) Software version 2.0 sequence detection system, an annealing temperature of 58°C, and extension at 72°C for 1 min for 40 cycles. Primer sequences (Table S6) were designed to specifically amplify a 100-250-bp portion of each transcript of interest (see Table S6). The amplification efficiencies were roughly equivalent across all primer sets (E% = 80-110). Control reactions (cDNA reactions lacking RT) were performed to verify that no genomic DNA contamination was present (that is, the threshold cycle [C T ] for detection in the control without RT was above 35). Normalization of C T values was done relative to the signal obtained from reactions amplifying a portion of the gatB/Yqey (BA4533) transcript; the expression level of this gene has been shown previously to remain stable across the entire B. anthracis life cycle (Reiter et al. 2011), and the microarray experiments described in this study suggest that its expression levels change less than twofold when increased amounts of SpxA1 or SpxA2 are present (see Tables S1, S2). Fold-changes for microarray validation studies were obtained by taking the ratio of +IPTG/ ÀIPTG samples after normalization to gatB/Yqey. Transcripts (ng) were directly reported for RT-qPCR assays of diamide-treated cultures following normalization to the gatB/Yqey endogenous control.
Promoter-lacZ strain construction and b-galactosidase assay Amplified promoter regions, generated by primers listed in Table S6, were digested with EcoR1 and BamHI, ligated into pDG793 (harboring the lacZ reporter construct, [Guerout-Fleury et al. 1996]) and propagated by transformation of competent DH5a cells. Plasmids containing the correct insert were introduced by transformation into ORB3834 (B. subtilis JH642 Dspx::neo) with selection for Erm resistance and screening for thrC (threonine auxotrophy). These resulting strains were transformed with pMMN818 (P spank(hy) -spxA1DD) or pMMN819 (P spank(hy) -spxA2DD) (see Table 1) to obtain the final strains (e.g., B. subtilis JH642 Dspx::neo [Neo] P exoA -lacZ [Erm] amyE:: P spank(hy) -spxA1DD [Spec]) used in b-galactosidase assays. Strains bearing various promoters (see Table 1) fused to a lacZ-reporter were grown at 37°C overnight on DSM supplemented with appropriate antibiotics. Assays of b-galactosidase in DSM cultures were performed as previously described (Nakano et al. 1998); activity was calculated in Miller units (Miller 1972

Western blot analysis
Bacillus anthracis Sterne 7702, and derivatives ORB8170 (DspxA1), ORB8438 (DspxA2), and ORB8285 (DspxA1 DspxA2), were grown in 40-60 mL of LB liquid medium lacking antibiotic at 37°C with agitation (200 rpm) and culture samples were taken at various time points during vegetative growth (see Fig. 2B). Cell pellets were harvested by centrifugation (5180g, 4°C, 10 min.) and stored until use at À80°C. Cell lysate was prepared by suspending the cell pellet in 200 lL-1 mL phosphate-buffered saline and vortexing with 0.1 mm glass beads for a total of 10 min, resting on ice every 5 min. After brief centrifugation to pellet the beads, the supernatant was analyzed for protein amount (Coomassie protein assay, Pierce) and an equal amount (25 or 30 lg) of cell lysate was separated on a SDS-polyacrylamide gel (15%) and transferred to nitrocellulose membranes. Immunoblotting was carried out in Tris-buffered saline plus 0.05% (v/v) Tween-20 using preabsorbed anti-Spx derived from B. subtilis purified Spx protein  and alkaline phosphataseanti-rabbit conjugated secondary antibody (Sigma) at appropriate dilutions. A representative immunoblot is shown for each experiment in Figures 1, 2, with each experiment done twice. Densitometry obtained with Ima-geJ (Schneider et al. 2012).

Phenotype testing
A fresh colony grown on LB agar was used to inoculate an overnight LB culture (37°C, 200 rpm). Ten microliter of the overnight cultures were subcultured into 2 mL LB and grown (37°C, 200 rpm) until OD 600 reached around 1.0. Five microliter of 10-fold serial dilutions (in T-base) of the cultures were spotted onto LB agar unsupplemented or supplemented with either 100 lmol/L diamide, 1.5 9 10 À3 % (0.44 mmol/L) hydrogen peroxide, or 0.001% deoxycholate. Sensitivity was checked after overnight incubation (37°C) of the agar plates.

Microscopy
Two microliter aliquots of each spore culture grown for 18 h in SM broth were added to a precleaned glass slide, a coverslip was added, and cells were subsequently examined using a Leica DMIL Inverted Contrasting Microscope (Leica, Buffalo Grove, IL) equipped with a Leica HI Plan I 409 (numerical aperture, 0.5). Images were captured using a Leica DFC295 digital color camera and processed with Leica Software Application Suite V3.8.0.

Results and Discussion
Phenotypic analyses

Construction of Spx protease resistant and Dspx mutants
Low levels of Spx are normally maintained in B. subtilis in part through posttranslational control involving degradation catalyzed by ClpXP and mediated by an adaptor protein, YjbH (Larsson et al. 2007;Garg et al. 2009). This mechanism can be averted in B. subtilis by substituting the last two C-terminal residues of SpxA to DD (see sequence alignment Fig. 1A, [Nakano et al. 2003]). We utilized the same method in B. anthracis. The SpxDDproducing strains used in this study were constructed by interspecies conjugation using a recombinant form of the integrative conjugative element ICEBs1 of B. subtilis ([Auchtung et al. 2005], Experimental Procedures) and a streptomycin-resistant variant of B. anthracis 7702 as recipient (7702 SR1). Mutation of SpxA1 and SpxA2 in this manner (SpxA1DD and SpxA2DD) caused accumulation of either protein compared to levels observed in the parental strain (see below in the case of SpxA1). In-frame deletions of both spx genes, DspxA1 and DspxA2, were also constructed in order to study the phenotypic effects of these two paralogs in B. anthracis. These mutations were constructed using a markerless allelic replacement method developed by Janes and Stibitz ([Janes and Stibitz 2006], Experimental Procedures).
spxA1 is essential for peroxide resistance; both spxA1 and spxA2 participate in disulfide stress tolerance In B. subtilis, a DspxA mutant is sensitive to oxidant stress (Nakano et al. 2003). Therefore, experiments were conducted to determine if a similar phenotype was conferred by B. anthracis spx null mutations. DspxA1 (ORB8170), DspxA2 (ORB8438), and DspxA1 DspxA2 (ORB8285) mutant cells were treated with several oxidants that had been shown to trigger a stress response in related bacteria ([Kristoffersen et al. 2007;Rukmana et al. 2009;McLean et al. 2010;Antelmann and Helmann 2011;Tu et al. 2012b]; Experimental Procedures). Either SpxA1 or SpxA2 can function in disulfide-stress tolerance, as only the DspxA1 DspxA2 (DA1/A2 in Fig. 2A and C) double mutant shows growth impairment upon diamide treatment. However, SpxA1 is essential for peroxide resistance, as DspxA1 (DA1 in Fig. 2A, B, and C) showed significant peroxide sensitivity. Defects in diamide and peroxide resistance could be corrected by introducing either spxA1DD or spxA2DD constructs in trans (+A1DD or +A2DD in Fig. 2C), suggesting that either Spx protein can potentially function in peroxide-induced stress. Bile salt can induce oxidative stress and disulfide formation in bacteria (Rodriguez-Beltran et al. 2012;Yang et al. 2013). DspxA1 and DspxA1 DspxA2, but not DspxA2, mutants were sensitive to the bile salt deoxycholate (Fig. S1).

SpxA1 and SpxA2 differ in transcriptional regulation during disulfide stress
Experiments were conducted to determine if the spxA paralogous genes were induced by disulfide stress, as part of the oxidative stress response in B. anthracis. cDNA was synthesized using RNA harvested from cells that were untreated or treated with 1 mmol/L diamide for 20 min and analyzed by RT-qPCR to determine the relative amount of spxA1 or spxA2 transcript produced. spxA1 or spxA2 transcript amounts were normalized to a portion of the gatB/Yqey (BA4533) transcript, which served as an endogenous control. The gatB/Yqey transcript is stable during the B. anthracis cell cycle (Reiter et al. 2011), and we observed no increase in this transcript in response to elevated SpxA1 or SpxA2 levels (Table 2); however, in the presence of diamide the gatB/Yqey transcript increased more than threefold (1.65 C T ). We also found a similar change using the16S gene transcript when used as an endogenous control (data not shown). Although foldchanges are not absolutely comparable in this experiment because of the severe stress applied to the cell by diamide treatment, we believe that general trends in transcript amount could be discerned for both spx paralogs. The level of spxA2 transcript increased 150-to 350-fold after diamide treatment; conversely, the spxA1 transcript amount remained nearly unchanged or decreased slightly (data not shown), suggesting that the observed repression of spxA1 exerted by SpxA2 that was uncovered from transcriptomic analysis (see below) was ongoing during disulfide stress.
Although spxA1 transcription was not stimulated by diamide treatment, the protein concentration might increase if a posttranslational control mechanism similar to that governing Spx in B. subtilis is operating in B. anthracis. The presence of both yjbH and clpXP orthologs in B. anthracis suggests that SpxA1 and SpxA2 might be subject to a similar mechanism of proteolytic control (Larsson et al. 2007;Garg et al. 2009). Therefore, the amount of SpxA1 protein after diamide-induced stress in B. anthracis was examined. The culture extracts of the B. anthracis parent (7702), DspxA1 (ORB8170), and DspxA2 (ORB8438) grown in LB medium were used in Western blotting experiments with preabsorbed rabbit polyclonal anti-serum raised against B. subtilis SpxA, which crossreacts only with SpxA1. Although very low levels of SpxA1 protein were observed in the parent strain grown without stress, the amount of SpxA1 increased at 20 min after the addition of 1 mmol/L diamide (57-fold), suggesting that posttranslational, rather than transcriptional, control of spxA1 governs SpxA1 concentration in response to oxidative stress (Fig. 3A). Interestingly, SpxA1 protein levels also increased in the DspxA2 (ORB8438) mutant without stress (7.4-fold), but remained unchanged in the DspxA2 mutant after diamide treatment (1.2-fold), suggesting that SpxA2-dependent dampening of spxA1 expression likely occurs at the transcriptional level (Fig. 3A).
One likely scenario for the increase in B. anthracis SpxA1 and SpxA2 concentration may be during germina- tion and outgrowth, where newly formed vegetative cells would presumably encounter higher levels of ROS. This regulation could extend to spore germination and outgrowth in the oxidizing environment of the lysosome within the host macrophage. In line with this argument, previous transcriptomic studies identified increased spxA1 transcript during spore outgrowth, increased spxA2 transcript during sporulation, and both paralogous transcripts increased inside murine macrophage cells . In keeping with the previous transcriptomic results, we found that SpxA1 was produced during exponential growth and that its concentration declined during early stationary phase (Fig. 3B).
The spike in spxA2 transcriptional activity upon diamide treatment does not correlate well with the contribution of its product toward diamide resistance, as this function is served primarily by SpxA1 based on sensitivity tests (Fig. 2). There are other factors, aside from transcriptional control that can figure prominently in the final level of SpxA2 activity, notably protein stability and redox activation of the Spx protein through its redox disulfide center. How these features of Spx control might differ between the two paralogs is currently not known. Differences in affinity for RNA polymerase binding surface that engages Spx protein, or holoenzyme preference exhibited by the two paralogs might also account for the disparate contributions of SpxA1 and A2 to oxidant resistance. Evidence from studies of B. subtilis suggest that the spx paralog, mgsR is transcriptionally activated by the r B form of RNA polymerase, after which the MgsR protein associates with r B holoenzyme (Reder et al. 2012). However, the finding that each of the B. anthracis spx paralogous genes can complement a DspxA mutation in B. subtilis suggests that each product can engage the r A form of RNA polymerase (data not shown).
Together our data suggest that both SpxA1 and spxA2 are upregulated when grown in the presence of the oxidant diamide; but different mechanisms exist for maintaining  the level of each paralog in B. anthracis. SpxA1 is likely governed posttranslationally by a mechanism similar to that in B. subtilis; whereas another unknown control mechanism responsive to oxidative stress functions in elevating spxA2 transcript levels.

Transcriptomic analyses
Transcriptomic changes in the presence of protease-resistant SpxA1DD or SpxA2DD in B. anthracis To identify genes potentially regulated by SpxA1 or SpxA2, we compared gene expression profiles between the parent 7702 strain and its isogenic derivatives expressing protease-resistant forms of either Spx paralog (parent vs. ICEBs1::P spank(hy) -spxA1DD or parent vs. ICEBs1::P spank (hy) -spxA2DD) using microarray hybridization array analysis. The strains were grown in LB broth and induced by the addition of IPTG for 15 or 45 min prior to culture sampling. Figures 5A and B summarize the overall transcriptional trends from the microarray experiments, and lists of all transcript changes can be found in Tables S1, S2.
The expression trends from the microarray transcriptomic analysis underwent validation by (RT-qPCR, Table 2) and by assaying b-galactosidase activity in B. subtilis strains harboring promoter-lacZ transcriptional fusions (Fig. 6). For RT-qPCR, the transcripts were normalized against a portion of the gatB/Yqey (BA4533) transcript (Reiter et al. 2011). The individual transcript values were reported as the averaged fold-change for two biological replicates (Table 2). b-galactosidase assays were performed by constructing lacZ transcriptional fusions to several gene promoters (intergenic regions roughly 200-600 nucleotides upstream of the start codon, extending downstream into the 5′ end of the coding sequence), and integrating these alleles into the chromosome of B. subtilis strains harboring a deletion of the native spxA gene as well as IPTG-inducible versions of B. anthracis spxA1DD or spxA2DD. b-galactosidase activity in the presence or absence of IPTG was then measured in these strains and the ratio of b-galactosidase activity (highest activity/starting time point activity) was plotted for each strain when grown in the presence or absence of IPTG (Fig. 6, see Experimental Procedures).
There were 773 and 699 transcript changes (≥twofold), respectively, when either spxA1DD or spxA2DD was expressed for 15 min. After 15 min of induction, transcript levels of 377 genes were higher and those of 396 genes were lower in the SpxA1DD-producing strain; and transcript levels of 367 genes were higher and 332 genes were lower when SpxA2DD production was induced (Fig. 5A). After 45 min of induction, only 363 transcript changes (≥twofold; 211 up-and 152 downregulated) were noted in SpxA1DD with 306 of those transcripts also found at higher levels after 15 min of induction ( Fig. 5A and B). Far more transcripts changed after prolonged SpxA2DD induction (45 min) with 509 total transcript changes (≥twofold; 280 up-and 229 downregulated), while 358 of these also exhibited a change in expression after 15 min of induction ( Fig. 5A and B). The total number of upregulated genes between SpxA1DD and SpxA2DD after 45 min of induction was 163, with 110 upregulated and 53 downregulated (Fig. 5B). A list of all transcriptomic changes can be found in Tables S1, S2. Due to the large number of transcripts that showed changes in abundance when either SpxA1DD or SpxA2DD was present, we chose to focus on upregulated transcripts after 15 min of SpxA1DD or SpxA2DD induction for further analysis. Lists of all transcript changes ≥twofold can be found in Tables S1, S2; and lists of genes whose transcripts changed by ≥threefold, grouped by function, can be found in Tables S3, S4. Many genes having elevated transcript amounts were uncovered in analysis of spxA1DD and spxA2DD transcriptomes, suggesting some overlap in the composition of SpxA1 and SpxA2 regulons. In total, several upregulated genes unique to either SpxA1 or SpxA2 regulons, 184 and 222, respectively, were also identified (Fig. 5B), supporting the hypothesis that each Spx protein has some unique sequence recognition properties (Tables S1, S2) and/or perhaps holoenzyme specificity. Genes that overlap between the 15-and 45min time points are considered common genes. A list of up-or downregulated transcripts, grouped by cellular function, can be found in Tables S1, S2. Transcripts were grouped by cellular function based on annotations from Genolist (http://genolist. pasteur.fr) and primary literature sources (Moszer et al. 2002;Pohl et al. 2011). Growth conditions for obtaining total cellular RNA and subsequent microarray experiments are presented in Experimental Procedures.

Redox homeostasis
In Bacilli, as in many bacteria, low-molecular-weight thiols are important for maintaining redox homeostasis and alleviating stress caused by toxic oxidants (Newton et al. 1996(Newton et al. , 2009. As regulated by orthologous spx in B. subtilis, trxA, trxB, and thioredoxin family protein, ytpP, among others, were also upregulated in the spxA1DD and spxA2DD B. anthracis strains (Nakano et al. 2003;Zuber et al. 2011;Rochat et al. 2012). RT-qPCR and b-galactosidase assays further confirmed a role for both paralogs in the transcriptional control of trxB (Table 2 and Fig. 6).
In several organisms that do not produce glutathione (GSH), coenzyme A (CoASH) can become an important molecule involved in intracellular redox functions (Delcardayre and Davies 1998;Delcardayre et al. 1998). In the presence of SpxA1DD or SpxA2DD, the transcripts specifying two CoASH disulfide reductases, BA1263 (CoADR) and BA0774, which encodes an isoform of CoADR that bears a rhodanese domain (CoADR-RHD), increased >threefold, suggesting that both genes are under the control of the spx paralogs in B. anthracis (Table 2). By RT-qPCR, we observed an increase in BA1263 and BA0774 transcripts with either SpxA1DD or SpxA2DD induction (Table 2); however, in b-galactosidase assays only SpxA2 could activate P BA0774 -lacZ, while either paralog activated P BA1263 -lacZ (Fig. 6). It may be possible that an activator (missing in B. subtilis) is necessary, in conjunction with SpxA1, to stimulate BA0774 transcription. B. subtilis lacks CoADR and CoADR-RHD isoforms, perhaps utilizing bacillithiol and cysteine instead of CoA in the cytoplasmic redox buffer (Helmann 2011).
The synthesis of bacillithiol has been linked to diamide stress and the spx regulon (Gaballa et al. 2010). Similarly, we found several orthologous genes encoding proteins functioning in the sequential synthesis of UDP-GlcNAc to bacillithiol that are upregulated upon induction of SpxA1DD and SpxA2DD ( [Gaballa et al. 2010]; Tables S1, S2). In particular, the transcript for the glycosyltransferase gene, bshA, increased 4.3-fold and 2.5-fold in the SpxA1DD-and SpxA2DD-producing strains, respectively (Tables S1, S2); and both orthologous genes encoding Figure 6. SpxA1DD and SpxA2DD activate several Bacillus anthracis genes in B. subtilis. Promoters of selected genes (horizontal axis) controlled by either SpxA1DD or SpxA2DD, as identified by microarray, were transcriptionally fused to a promoterless lacZ and integrated into the chromosome at the thrC locus of a B. subtilis JH642 Dspx strain harboring an IPTG-inducible copy of either spxA1DD or spxA2DD integrated at the amyE locus. b-galactosidase activity was measured in these strains at 30 min. intervals after the addition of 1 mmol/L IPTG. The maximal activity, which was usually observed at 1-2 h after IPTG induction, was divided by the b-galactosidase activity when IPTG was added; except for P spxA1 -lacZ, where the minimal activity was divided by the b-galactosidase activity when IPTG was added (SpxA1DD: minus IPTG 0.64 AE 0.16, plus IPTG 0.49 AE 0.2; SpxA2DD: minus IPTG 0.52 AE 0.27, plus IPTG 0.44 AE 0.18). The ratio was shown as the average of three biological triplicates with standard deviation. Significance was determined by a two-tailed T-test comparing plus IPTG to minus IPTG ratios; *P < 0.05 and **P N-acetylhydrolases, bshB1 and yojG, as well as the Cysadding enzyme, yllA, were also upregulated >twofold in the presence of either SpxA1DD or SpxA2DD (Tables S1, S2), suggesting expression that is governed by SpxA1 and SpxA2 resulting in increased production of bacillithiol in B. anthracis. Additionally, an ortholog of a B. subtilis gene putatively involved in the reduction of oxidized bacillithiol, namely ypdA, encoding a pyridine nucleotide-dependent disulfide oxidoreductase, was upregulated >twofold by either SpxA1DD or SpxA2DD (Tables S1, S2). Two putative bacilliredoxin genes, yphP and ytxJ, also showed elevated transcript levels after SpxA1DD or SpxA2DD production. ytxJ transcript level was increased 2.5-fold by SpxA1DD and yphP 2.3-fold by SpxA2DD (Tables S1,  S2).
Bacillus anthracis Spx proteins also stimulated the transcription of genes whose products function in overcoming DNA damage, which studies have shown is a direct consequence of oxidative stress brought about by encounters with oxidizing agents such as H 2 O 2 . We found that SpxA2DD upregulated BA3868, the ortholog of B. subtilis exoA, an apurinic-apyrimidinic endonuclease important in base excision repair (Ibarra et al. 2008) and genes encoding components of the ABC excinuclease (uvrC and uvrD) (Tables S1, S2) are also upregulated by SpxA1DD and SpxA2DD. Both Spx paralogs were able to activate a P BA3868 -lacZ reporter construct in B. subtilis, suggesting that BA3868 is a member of SpxA1 and SpxA2 regulons (Fig. 6). ExoA Bsu is an important component of spore outgrowth and germination when oxidative DNA damage occurs, presumably when the once dormant spore is suddenly exposed to oxygen upon germinated spore hydration (Ibarra et al. 2008). Bacillus subtilis cells deficient in exoA were significantly delayed in spore outgrowth and increasingly sensitive to H 2 O 2 (Ibarra et al. 2008).
Several genes encoding hypothetical products, many having no homology outside of the B. cereus/B. thuringiensis group, were also upregulated when SpxA1DD or SpxA2DD were produced (Tables S1, S2). We focused on one of these genes, a putative oxidoreductase (BA1951), which in the microarray analysis was induced 32.6-fold in cells expressing SpxA1DD. Cells expressing SpxA2DD had a smaller increase in BA1951 transcript (2.9-fold) ( Table 2). The microarray trends were mirrored in b-galactosidase (activation of P BA1951 -lacZ by SpxA1DD [P = 0.036] or SpxA2DD [P = 0.032]) and SpxA1DD, assays, although the transcript changes determined by RT-qPCR were lower than transcript changes seen by microarray (Table 2).
Several genes functioning in NAD(P)H-dependent interconversions were also highly upregulated by SpxA1 and SpxA2. These include genes encoding putative alcohol, aldehyde, and quinone dehydrogenases (BA0838, BA2647, and BA3438) for which orthologs exist in B. subtilis (YogA and YdeQ). Interestingly, the B. subtilis orthologous genes are not members of the Spx regulon, suggesting that the SpxA1 and SpxA2 regulons of B. anthracis include members with different metabolic capabilities than those belonging to the B. subtilis Spx regulon. The products of the Spx-activated genes could function in detoxification or generation of reduced NAD(P)H for coping with oxidative stress (Rochat et al. 2012). Taken together these results reinforce the view that SpxA1 and SpxA2 act as transcriptional regulators to activate genes involved in the reduction of oxidized thiols in B. anthracis.

SpxA2 repression of spxA1
Increasing the amount of SpxA2DD led to a 3.5-fold decrease in the spxA1 transcript, suggesting a relationship between SpxA2 and the transcriptional control of spxA1 expression (Table 2). By microarray, the induction of spxA2 was not dependent on SpxA1DD, suggesting that another regulatory mechanism functions in elevating spxA2 transcript levels during oxidative stress.
The spxA1 repression by SpxA2DD was confirmed in three ways: (1) a roughly fourfold decrease in transcript amount was found by RT-qPCR when SpxA2DD was present (Table 2); (2) a roughly twofold increase in spxA1 transcript was detected in a DspxA2 background (data not shown); and (3) expression of a P spxA1 -lacZ reporter fusion was repressed when SpxA2DD was produced in B. subtilis (Fig. 6). These transcriptomic results parallel the expression results reported above, in which an increase in the SpxA1 protein amount was detected in the DspxA2 mutant (Fig. 3A). In conjunction with the large increase in the spxA2 transcript in the presence of diamide stress, it may be that the muted transcriptional response of spxA1 in B. anthracis Sterne 7702 is a consequence of SpxA2-dependent repression, which seems to persist even in the presence of disulfide stress. However, as shown by Western blot experiments, the SpxA1 protein amount increased during diamide stress (Fig. 3), suggesting that the oxidative stress response in B. anthracis includes a posttranscriptional mechanism that oversees SpxA1 accumulation instead of a surge in transcriptional activity that characterizes spxA2 induction. Together, these experiments indicate that SpxA2 negatively affects spxA1 expression.

Cell wall biosynthesis and sporulation
The involvement of SpxA orthologs in control of genes that function in cell wall metabolism has been observed in other Gram-positive bacteria (Prudhomme et al. 2006;Veiga et al. 2007;Suntharalingam et al. 2009;Turlan et al. 2009;Eldholm et al. 2010;Kajfasz et al. 2010). The microarray data suggest that both B. anthracis paralogs can increase expression of genes involved in cell wall or membrane synthesis, remodeling, and maintenance (Tables S1, S2).
Two two-component signal transduction systems specified by the yycFG and BA1119-BA1118 (yvrHG) operons were upregulated in response to increased amounts of SpxA1DD, but not SpxA2DD. Transcription of yycFG increased more than twofold (2.2-fold yycF, 3.5-fold yycG), while yvrGH, showed an even greater transcript increase (17.9-fold yvrG, 12.9-fold yvrH). Upon induction of SpxA1DD, P BA1119 -lacZ, encompassing the promoter for the gene encoding the YvrH DNA-binding protein, was activated to nearly significant levels (P = 0.055; Fig. 6). Similarly in RT-qPCR experiments, SpxA1DD increased the BA1118 (yvrG) transcript (2.9-fold), although to lesser amounts than found by microarray. Neither BA1119 promoter activation (Fig. 6) or BA1118 transcript increase (1.4-fold, Table 2) were detected after SpxA2DD induction, suggesting that the yvrHG operon is regulated by SpxA1DD only. Both YycFG and YvrGH are responsible for peptidoglycan biosynthesis and cell wall homeostasis in B. subtilis (Serizawa et al. 2005;Szurmant et al. 2007a). The essential YycFG system of B. subtilis also has two accessory proteins YycH and YycI, which are transcribed as part of the same operon (Szurmant et al. 2005(Szurmant et al. , 2007b and both yycHI orthologous transcripts in B. anthracis were elevated only when SpxA1DD was produced (3.3-fold yycH, 2.1-fold yycI).
Bacillus anthracis possesses two glutamate racemases, which catalyze the conversion of L-glu to D-glu for the synthesis of poly-c-glutamate capsule and peptidoglycan (Dodd et al. 2007). One of these glutamate racemaseencoding genes, BA0847 (racE-1), was upregulated in the SpxA2DD-producing strain almost 16-fold by microarray analysis, while its expression did not change when SpxA1DD was overproduced. However, using RT-qPCR, we were able to detect an elevated BA0847 (racE-1) transcript concentration in the SpxA1DD-producing strain (4.6-fold) as well as the expected transcript increase (7.3fold) by SpxA2DD ( Table 2), suggesting that there may be an overestimation of genes uniquely controlled by only one paralogous Spx, as indicated by microarray. A P BA0847 -lacZ transcriptional fusion was also activated in the presence of SpxA2DD or SpxA1DD (Fig. 6), confirming our RT-qPCR results.
Production of SpxA2DD also increased the level of transcripts encoded by three sporulation genes, spoIIID, spoIIIE, and rsfA, whose gene products are involved in sporulation-specific transcriptional regulation and DNA translocation (Kunkel et al. 1989;Bath et al. 2000;Wang et al. 2006). As the DspxA2 mutant does not seem to have altered sporulation efficiency or outgrowth patterns, the increased transcript amount of these sporulation-specific genes after SpxA2DD induction may be an indirect effect. A more detailed study of spxA2 is underway.

SpxA-activated gene expression during disulfide stress
We chose to further characterize two genes regulated by both Spx paralogs and putatively involved in thiol homeostasis, BA5387 (trxB) and BA1951 (a putative oxidoreductase), by measuring their transcript levels after diamide stress (1 mmol/L) for 20 min ( Fig. 7; Experimental Procedures). The levels of the two transcripts were normalized to the gatB/Yqey (BA4533) transcript as described above. Without diamide treatment, gatB/Yqey transcript levels remained unchanged. However, in the presence of diamide the gatB/Yqey transcript increased more than threefold in the parent and roughly eightfold in the DspxA1 DspxA2 mutant. As BA1951 and BA5387 transcripts increased by at least 80-fold when diamide was applied, we believe that general comparisons between untreated and treated parent and DspxA1 DspxA2 transcripts could reasonably be made.
In the absence of oxidative stress, BA1951 and BA5387 transcript levels remained low, but in the presence of diamide, both transcripts increased dramatically in the parent (7702), suggesting a role for both genes in SpxA-dependent disulfide stress tolerance (Fig. 7). In the double spxA mutant, BA1951 transcript levels decreased, but not to untreated levels, signifying that BA1951 is also regulated by a Spx-independent mechanism during oxidative stress (Fig. 7). The level of BA5837 (trxB) transcript in the double spxA mutant was only slightly induced by diamide, indicating a requirement for SpxA1 and/or A2 in activating BA5387 transcription during diamide-induced stress (Fig. 7). Taken together, we believe that the B. anthracis (cereus/thuringiensis)-specific BA1951 gene product is an oxidoreductase, and along with BA5387, have potential roles in oxidative stress defense.
In microarray experiments, the transcript levels of the B.subtilis yjbH ortholog, BA1208, was elevated when either SpxA1DD or SpxA2DD was present (Table 2,  Tables S1, S2) stress where the transcript amount of BA1208 increased in the parent (7702) but not in the DspxA1 DspxA2 mutant (Table 2 and Fig. 7). Elevating the concentrations of YjbH and ClpX through a Spx-dependent mechanism of control might ensure that the means of Spx elimination is in place for when thiol homeostasis is restored following stress. Recent evidence from chromatin immune precipitation and transcription analysis indicated that the clpX gene is under positive transcriptional control by SpxA in B. subtilis (Rochat et al. 2012). However, we were unable to detect upregulation of the B. anthracis clpX (BA4074) gene in SpxA1DD-or SpxA2DDproducing cells.
In summary, four lines of evidence indicate that SpxA proteins of B. anthracis participate in the control of the oxidative stress response: (1) The spxA1 null mutant shows heightened sensitivity to H 2 O 2 treatment, while the spxA1/spxA2 double mutant shows a severe growth defect when cells are treated with diamide; (2) SpxA1 protein is elevated upon diamide-induced stress, in parallel with elevated SpxA-dependent transcriptional induction; (3) SpxA1 or SpxA2, in protease-resistant form, activate the transcription of many genes that have been implicated in disulfide stress resistance and thiol homeostasis; (4) Genes likely involved in thiol homeostasis are induced by diamide treatment through an SpxA-dependent mechanism. Similar to the role of spx in B. subtilis, the B. anthracis spx paralogs are likely to be important transcriptional regulators during disulfide stress. Each paralog may differentially regulate genes involved in other stress-alleviating processes. Additional experiments are underway to further examine the function and regulation of these two paralogous genes and their products. enhancement screen of a regulatory spx mutant unveils a role for the ytpQ gene in the control of iron homeostasis. PLoS One 6:e25066.