Overlapping and Distinct Functions of the Paralogous PagR Regulators of Bacillus anthracis

ABSTRACT

The Bacillus anthracis pagA gene, encoding the protective antigen component of anthrax toxin, is part of a bicistronic operon on pXO1 that codes for its own repressor, PagR1. In addition to the pagAR1 operon, PagR1 regulates sap and eag, two chromosome genes encoding components of the surface layer, a mounting structure for surface proteins involved in virulence. Genomic studies have revealed a PagR1 paralog, PagR2, encoded by a gene on pXO2. The amino acid sequences of the paralogues are 71% identical and show similarity to the ArsR family of transcription regulators. We determined that the expression of either rPagR1 or rPagR2 in a ΔpagR1 pXO1⁺/pXO2^– (PagR1-PagR2) background repressed the expression of pagA, sap, eag, and a newly discovered target, atxA, encoding virulence activator AtxA. Despite the redundancy in PagR1 and PagR2 function, we determined that purified rPagR1 bound DNA corresponding to the control regions of all four target genes and existed as a dimer in cell lysates, whereas rPagR2 exhibited weak binding to the DNA of the pagA and atxA promoters, did not bind sap or eag promoter DNA, and did not appear as a dimer in cell lysates. A single amino acid change in PagR2, S81Y, designed to match the native Y81 of PagR1, allowed for DNA-binding to the sap and eag promoters. Moreover, the S81Y mutation allowed for the detection of PagR2 homomultimers in coaffinity purification experiments. Our results expand our knowledge of the roles of the paralogues in B. anthracis gene expression and provide a potential mechanistic basis for differences in the functions of these repressors.

IMPORTANCE The protective antigen component of the anthrax toxin is essential for the delivery of the enzymatic components of the toxin into host target cells. The toxin genes and other virulence genes of B. anthracis are regulated by multiple trans-acting regulators that respond to a variety of host-related signals. PagR1, one such trans-acting regulator, connects the regulation of plasmid-encoded and chromosome-encoded virulence genes by controlling both protective antigen and surface layer protein expression. Whether PagR2, a paralog of PagR1, also functions as a trans-acting regulator was unknown. This work advances our knowledge of the complex model of virulence regulation in B. anthracis and furthers our understanding of the intriguing evolution of this pathogen.

INTRODUCTION

Microbial pathogens have evolved complex regulatory pathways by which to coordinate virulence gene expression. The evolution of complex pathways is sometimes driven by the emergence of new paralogous regulatory genes in the genome. Regulatory paralogs in bacteria may arise from gene duplication events within the native genome or from horizontal gene transfers among different strains. Some paralogs retain similar regulatory functionality, while others may develop diverse regulatory functions that are driven by differences in primary amino acid sequences or differences in gene expression among paralogous genes (1, 2).

An example of regulatory paralogs with largely overlapping functions can be found in the pathogen Pseudomonas aeruginosa, whose genome contains two stringent response regulators, DksA1 and DksA2 (3). These paralogues are 41% identical and have been reported to be mostly interchangeable for the regulation of virulence and biofilm formation. The Staphylococcus aureus genome contains three paralogs (CspA, CspB, and CspC) of cold shock regulatory proteins, which bind mRNA to control the expression of cold shock response proteins (4). Despite being over 70% identical, these paralogs appear to have divergent functions. CspA controls the expression of a pigment involved in oxidative stress, whereas CspB is the only paralog significantly induced during cold stress, and CspC is induced in the presence of various antibiotics (4). Another well-documented example of regulatory network expansion bythe emergence of paralogous regulators is the Rap-Phr system in the nonpathogen Bacillus subtilis (5). There are eight paralogous rap-phr loci and three rap genes with no cognate phr in B. subtilis strain 168 (5). Rap proteins, whose functions are modulated by secreted Phr polypeptides, control the master regulators of sporulation, competence, flagellum synthesis, and biofilm formation; expansion of the Rap-Phr regulatory repertoire leads to phenotype heterogeneity within a B. subtilis population (5). This system has also been documented in the Bt8741 strain of the insect pathogen B. thuringiensis, whose genome also contains eight rap-phr paralogous loci (6). A specific subset of these paralogs has evolved to regulate extracellular proteases in Bacillus thuringiensis, serving to adapt this system to virulence gene regulation (6).

Mounting evidence suggests the evolution of paralogous regulators to expand control of virulence in the Bacillus cereus sensu lato species, a group of closely related pathogenic Bacillus species. In addition to B. thuringiensis, which harbors the Rap-Phr system, this group includes Bacillus cereus 14579, which contains the paralogous transcriptional regulators PlcR and PlcRa (7). PlcR is a well-characterized DNA-binding protein that positively controls the expression of a wide range of enterotoxins, hemolysins, and proteases (8). PlcRa is 29% identical to PlcR and has been shown to positively regulate genes involved in oxidative stress resistance (7). Bacillus anthracis, another member of the Bacillus cereu sensu lato group and the causative agent of anthrax, has also evolved regulatory paralogs to control virulence-related genes. The genome of this pathogen contains a 5.2 Mb circular chromosome and two virulence plasmids, the 182 kb pXO1 and the 94 kb pXO2 (9, 10). The master virulence regulator AtxA, encoded by a gene on pXO1, controls the expression of three anthrax toxin genes that are located on the same plasmid and other genes located throughout the genome (11–13). AtxA has two paralogs found on pXO2: AcpA and AcpB. The amino acid sequences of AcpA and AcpB share 40% identity to each other, and both are approximately 26% identical to AtxA (12). AcpA and AcpB, which are regulated by AtxA, control the synthesis of the pXO2-encoded B. anthracis capsule, and, like AtxA, have additional target genes (13–15). A RNA-seq analysis identified several loci that are controlled by two or three of these paralogs, suggesting some functional overlap between these regulators (12).

While AtxA, AcpA, and AcpB are the best-characterized examples of paralogous regulators controlling virulence loci in B. anthracis, another set of paralogs with regulatory function, PagR1 and PagR2, is less well studied. PagR1 is part of a bicistronic operon on pXO1 that includes the toxin gene pagA (16). The gene pagA encodes protective antigen (PA), the major player in B. anthracis toxin activity, as it mediates the entry of the enzymatic toxin components (lethal factor [LF] encoded by lef and edema factor [EF] encoded by cya) into the host cell (17–20). The pagAR1 promoter is directly bound and regulated by AtxA (13). In a previous report, deletion of pagR1 led to a 2-fold increase in secreted PA protein levels (21). PagR1 has been shown to also regulate the surface layer (S-layer) genes sap and eag, which encode the components of a proteinaceous layer between the capsule and the peptidoglycan cell wall of B. anthracis (22, 23). The S-layer serves as a mounting structure for S-layer-associated proteins with roles in B. anthracis virulence. During growth in rich media, Sap (encoded by sap) is primarily produced during exponential-phase, while EA1 (encoded by eag) is primarily produced during stationary-phase (24). When cultures are grown in the presence of atmospheric CO₂, however, only EA1 is primarily expressed (22). PagR1 participates in the control of S-layer composition by repressing sap and facilitating the activation of eag (22).

Previous genomic studies revealed the presence of a second pagR1-like gene in pXO2 (GBAA_pXO2_0069), which is highly identical to the pXO1 pagR1. A recent report revealed that the attenuated vaccine strain Pasteur II, which produces low levels of toxin relative to fully virulent strains, carries a deletion in the promoter region of pagR2 (25). The deletion was associated with reduced pagR2 expression; replacing the native Pasteur II pagR2 promoter with that of a fully virulent strain resulted in increased pagR2 transcription (25). The recombinant strain expressing higher pagR2 levels exhibited increased PA toxin secretion as well as increased expression of pagA mRNA (25). Interestingly, this recombinant strain also showed increased atxA mRNA expression, which in turn led to increased lef and cya transcripts (25). The direct binding of PagR2 to the promoters of regulated genes was not explored, and whether the increased pagR2 expression in this strain had any effects on sap and eag genes was not tested.

The apparent PagR2-mediated control of PA toxin levels suggests that PagR1 and PagR2 may have related functions. Moreover, our previously published RNA-seq data show that pagR2 is significantly upregulated by AtxA, as well as by its paralogs AcpA and AcpB (12). PagR1 is predicted to be a part of the ArsR-family of metal-dependent transcriptional regulators (26). The PagR1 crystal structure contains a winged helix-turn-helix motif that is typical of members of this family and is associated with DNA binding (27, 28). Fittingly, PagR1 has been shown to directly bind the promoters of pagA, sap, and eag in DNase footprinting experiments (22). We hypothesized that PagR1 and PagR2 are paralogs that act in concert to regulate toxin and S-layer expression.

In this work, we compared the regulatory functions of PagR1 and PagR2 in the pXO1⁺/pXO2^– ANR-1 strain. We found that ovthe erexpression of either PagR1 or PagR2 in a strain devoid of the pagR genes results in highly reduced pagA, sap, and eag transcript levels. We also observed the PagR1-mediated and PagR2-mediated control of atxA transcript. We compared the binding of the PagR proteins to the promoters of their regulated targets and observed differences in DNA-binding affinity between these proteins in vitro. Using site-directed mutagenesis, we determined that a single amino acid difference at the residue 81, which is a tyrosine in PagR1 and a serine in PagR2, partially mediates their observed differences in DNA-binding ability. We also compared the protein-protein interactions of PagR1 and PagR2. This work furthers the collective understanding of virulence gene regulation and paralog-mediated crosstalk between genetic elements in this human pathogen.

RESULTS

Amino acid sequence and predicted similarity of PagR1 and PagR2.

The pagR1 locus is located on the virulence plasmid pXO1, as the second gene of a bicistronic operon with the toxin gene pagA. As such, pagR1 is a part of the pXO1 pathogenicity island that also contains atxA and the other toxin genes, lef and cya. The pagR2 locus (GBAA_pXO2_0069) is located on pXO2, the virulence plasmid which contains the capsule biosynthesis operon capBCADE and the two atxA-like paralogs, acpA and acpB. The pagR2 gene is located 1,449 nt downstream and in the opposite direction of capB, the first gene of the capsule biosynthesis operon. Immediately flanking pagR2 are GBAA_pXO2_0068, a gene encoding a hypothetical protein of 43 amino acids, and an IS231-related transposon region. Both of these flanking genetic features are positioned in the opposite direction of pagR2, suggesting that pagR2 transcription is monocistronic. Previous RNA-seq studies, in which a ΔatxAΔacpAΔacpB mutant was complemented with atxA, acpA, and acpB individually, indicate that while pagR1 is regulated only by AtxA, pagR2 expression responded to complementation with AtxA, AcpA, and AcpB (12).

The pagR paralogs are 300 nt long, encoding proteins of 99 aa. The amino acid sequence alignment of PagR1 and PagR2 (Fig. 1) revealed a high degree of amino acid conservation between the two proteins (Fig. 1A); the PagR1 and PagR2 amino acid sequences are 71% identical and 89% similar. The published crystal structure of PagR1 indicates the presence of five α-helix domains and a β-sheet hairpin (26). We modeled the predicted secondary structure of the PagR2 monomer using PyMOL v2.5 and found that the predicted PagR2 structure closely matches the PagR1 crystal structure, with both proteins sharing the same basic topology (Fig. 1B).

FIG 1.

Comparison of PagR1 and PagR2 amino acid sequence. (A) Amino acid sequence alignment of PagR1 and PagR2. Both proteins are 99 amino acids long. Amino acid residues shaded in black boxes indicate conserved identical residues between the two proteins, while those shaded in gray boxes indicate similar residues belonging to the same group. Unshaded amino acids are completely different between the two proteins and are therefore not conserved. Amino acids predicted to participate in DNA binding, according to the published PagR1 crystal structure (26), and in the conservation of ArsR-family protein residues are labeled in red squares. Regions of secondary structure that were elucidated from the PagR1 crystal structure are shown above the aligned sequences. The sequence alignment was generated using the T-Coffee and Boxshade webservers. (B) Comparison of PagR1 and PagR2 secondary structure, as modeled using PyMOL 2.5 software. The predicted PagR2 secondary structure closely matches the published PagR1 crystal structure (26). Predicted secondary structure features are labeled as follows: α-helix 1 (H1: dark blue), α-helix 2 (H2: light blue), α-helix 3 (H3: green), α-helix 4 (H4: yellow), two β-sheets forming a β-sheet hairpin (β1 and β2: orange), and α-helix 5 (H5: pink).

Zhao et al. proposed several amino acid residues of PagR1 that could be involved in DNA binding and protein-protein interactions, based on their locations within the crystal structure of the protein (26). All of these residues are highly conserved among the ArsR-family of transcription regulators (26). These residues are shown in red boxes in Fig. 1A. When comparing these predicted key residues of PagR1 to their PagR2 counterparts, we found that most were conserved between the two proteins. Of these highly conserved residues, the PagR1 Y81 is the only amino acid not conserved in PagR2; the PagR2 sequence contains a serine residue at position 81. Thus, the in silico analysis of the PagR1 and PagR2 amino acid sequences suggests that these proteins are likely to have overlapping functions and that any distinct functions may be mediated by specific key amino acid differences.

Effect of PagR1 and PagR2 on virulence gene expression in Bacillus anthracis.

We compared the regulatory effects of PagR1 and PagR2 on the expression of the known PagR1 targets pagA, sap, and eag. Additionally, we tested for the PagR1-mediated and PagR2-mediated control of atxA. We first constructed a ΔpagR1 mutant in the ANR-1 (pXO1⁺, pXO2^–) parent background. Because ANR-1 is devoid of pXO2, it does not carry pagR2. We then individually expressed either His-tagged PagR1 or PagR2 in trans from an IPTG-inducible promoter in the pagR1/pagR2-null mutant, termed UTA428. We grew the ANR-1 parent, ΔpagR1 mutant, ΔpagR1 p(PagR1-His), and the ΔpagR1 p(PagR2-His) strains in toxin-inducing conditions (CA-CO₂ at 37°C). We induced the expression of the His-tagged recombinant PagR1 and PagR2 proteins at early exponential-phase (OD_{600 m} = 0.2 to 0.3) and collected the cells at late exponential-phase (OD_{600 nm} = 0.8 to 1.0). The growth of all four strains was comparable under these conditions (Fig. S1A). The induction of PagR1 and PagR2 with 100 μM and 1 mM IPTG, respectively, resulted in comparable levels of protein expression (Fig. S1B and C).

TaqMan-based qRT-PCR assays were used to assess the effect of the induction of the recombinant PagR proteins on virulence gene expression (Fig. 2). The transcript levels of pagA increased ~5-fold in the ΔpagR1 mutant compared to the parent strain (Fig. 2A). The induction of PagR1 resulted in the complete loss of detectable pagA transcript, in concordance with previous reports of the PagR1-mediated repression of pagA (22). The induction of PagR2 in the ΔpagR1 mutant also resulted in a significant reduction of pagA transcript. The expression of atxA mirrored the results obtained for pagA transcript levels (Fig. 2B). The transcript level of atxA increased ~2-fold in the ΔpagR1 mutant, and the induction of PagR1 led to a marked decrease in atxA expression. PagR2 induction had a similar effect on atxA transcript levels. Overall, these data indicate that pagA and atxA expression responds to both PagR1 and PagR2, which act to repress the expression of the transcripts.

FIG 2.

Effect of PagR1 and PagR2 on virulence gene expression. Total RNA isolated from cell lysates was used to probe for (A) pagA, (B) atxA, (C) sap, and (D) eag expression using TaqMan-based qRT-PCR assays. The mRNA levels in each strain were normalized to levels of the gyrB reference mRNA. Data are presented as averages of three biological replicates with the standard deviations. An analysis of variance (ANOVA) followed by Tukey’s multiple-comparison test was used to determine statistical significance. Asterisks directly above the bars indicate a statistically significant difference between that bar and the “Parent” bar. Additional comparisons are indicated with brackets, with the respective significance shown by asterisks above the bracket. * indicates P-value < 0.05, ** indicates < 0.01, *** indicates < 0.001, and **** indicates < 0.0001.

PagR1 has been shown to repress sap transcript expression by binding the sap promoter (22). In agreement with the PagR1 repression of sap, the expression of sap increased ~3-fold to ~4-fold in the ΔpagR1 mutant (Fig. 2C). The induction of PagR1 led to a total loss of detectable sap transcript levels. The induction of PagR2 had a similar effect on sap transcript. The proposed mechanism of PagR1-mediated control of eag transcript is more complex. In a previously proposed model, PagR1 binds and represses the expression of one of the two eag promoters (22). A yet uncharacterized pXO1-encoded factor is then able to preferentially drive the expression of the unbound promoter, resulting in a PagR1-assisted increase in eag transcript. We did not observe a significant difference in eag transcript levels in the parent and ΔpagR1 mutant (Fig. 2D). However, we did observe the expected reduction in eag transcript when PagR1 was induced in the ΔpagR1 mutant. The transcript levels of eag were lower than those observed for the parent strain. The expression of PagR2 also led to a significant decrease in relative eag transcript. These data suggest that PagR2 is a regulatory protein that has functional overlap with PagR1.

Promoter DNA binding activity and specificity of PagR1 and PagR2.

The results of previously reported DNase protection studies suggest that PagR1 directly binds the promoters of pagA, sap, and eag, protecting them from DNase-mediated degradation (22). PagR1 protects a region encompassing both the P1 and P2 promoters of pagA, ranging from nucleotide −52 to −3 with respect to the P2 transcriptional start site. For the sap promoter, PagR1 protects a region that encompassed nucleotides −63 to −15 with respect to the transcriptional start site. The transcription of eag is controlled by two promoters, a σ^H-dependent promoter distal to the eag gene and a σ^A-dependent promoter that is proximal to the eag gene. PagR1 protects a region encompassing nucleotides +45 to +94 with respect to the distal promoter transcriptional start site (which was −142 nucleotides away from the proximal promoter). Given that PagR1 appears to control the regulation of these targets via direct promoter binding, we hypothesized that PagR2 would also bind promoter DNA. We additionally hypothesized that both PagR1 and PagR2 would directly bind the promoter of atxA, given that we determined that PagR1 and PagR2 regulated atxA expression.

We purified PagR1-His and PagR2-His from the ΔpagR1 strain after a 2 h induction of the recombinant proteins in BHI medium. Increasing concentrations of the purified proteins were mixed with biotinylated DNA probes obtained via PCR amplification of the previously described DNase-protected promoter regions (22). In addition, we amplified DNA probes that corresponded to the distal promoter of eag, the P1 promoter of atxA, and the promoter of the housekeeping gene dnaA. The biotinylated probes (PpagA, PatxA, Psap, Peag distal, Peag proximal, PdnaA) were incubated at a constant concentration with each of the purified recombinant proteins before performing EMSAs (Fig. 3). Unlabeled competitor DNA probes were added to additional binding reactions as controls. The retarded mobility of DNA probes was indicative of protein binding. The specific binding to DNA probes was assessed by comparing migration relative to PdnaA, the negative-control probe.

FIG 3.

Specific promoter binding by PagR1 and PagR2 proteins. Purified recombinant PagR1, PagR2, and PagR2 S81Y proteins were incubated at the indicated, increasing concentrations with 0.2 nM biotin-labeled probes. The probes were made via the PCR amplification of the promoters of pagA, atxA, sap, and eag (both the distal and the proximal promoters). The promoter of dnaA was used as a negative-control for the no binding interaction. Unlabeled probes were added to the indicated binding reactions at a final concentration of either 0.2 or 20 nM. The blots shown are representative images of three replicates.

As expected, PagR1 bound the pagA, sap, and eag distal promoters. We additionally observed binding of the atxA and eag proximal promoters and no binding of the dnaA control promoter. The addition of an unlabeled competitor at a 100-fold excess concentration (20 nM) led to the loss of a detectable upshift for bound promoters, further demonstrating specific binding. Interestingly, despite PagR2-mediated effects on the transcript levels of target genes, purified PagR2-His bound weakly to the pagA and atxA promoters and exhibited no detectable binding to the sap promoter or to the eag promoters. Importantly, purified PagR2-His did not bind the dnaA promoter control. The addition of an unlabeled competitor DNA probe at a 100-fold excess (20 nM) titrated the weak PagR2-mediated binding of pagA and atxA promoters. These data suggest that while PagR1 regulates the gene expression of all known PagR1-controlled genes by direct binding to promoter DNA sequences, these genes may be regulated indirectly by PagR2 or may require additional factors in concert with PagR2 for full regulation.

We calculated the binding affinities (Kds) of PagR1 for its DNA targets. We expanded the range of purified protein concentration and repeated the binding experiments (Table 1). Overall, the PagR1 binding affinities for the five targets tested were comparable and in the μM range. PagR1 had the highest affinity for the eag proximal promoter (~1.8 μM) and the lowest affinity for the sap promoter (~4.3 μM). We attempted to calculate the binding affinities of PagR2 for the pagA and atxA promoters; however, we were unable to detect a sufficient shift of the probes to calculate Kds, further supporting a model for the indirect PagR2-mediated regulation of these targets.

TABLE 1.

PagR1 and PagR2 S81Y DNA-binding affinities

Promoter	PagR1a	PagR2 S81Ya
pagA	3.2 μM (2.8 to 3.8)	NA
atxA	2.1 μM (1.9 to 2.3)	NA
sap	4.3 μM (3.5 to 6.6)	2.4 μM (1.6 to 5.4)
eag proximal	1.8 μM (1.6 to 2.1)	0.5 μM (0.4 to 0.8)
eag distal	2.3 μM (1.8 to 3.9)	NA

^aBinding affinities to the test promoters are reported as dissociation constants (Kds) with 95% confidence intervals (CI) in parentheses. NA indicates either no binding detected or an insufficient shift to calculate the Kd.

Effect of S81Y mutation on PagR2 DNA binding.

Although the amino acid sequences of PagR1 and PagR2 are highly identical, and although both proteins affected the expression of virulence-related genes, only purified PagR1 interacted with the promoters of these genes with a detectable affinity. A comparison of the PagR1 and PagR2 amino acid sequence revealed residue 81 to be a potentially key difference between these two proteins (Fig. 1A). Residue 81 is a tyrosine (Y) in PagR1 and a serine (S) in PagR2. We decided to test whether this amino acid position contributes to differences in DNA binding between PagR1 and PagR2. We created a construct expressing recombinant PagR2 bearing a S to Y substitution at position 81. The resulting mutant is referred to as PagR2 S81Y.

We tested the ability of PagR2 S81Y to bind the promoters of pagA, atxA, sap, and eag. Recombinant PagR2 S81Y was purified and utilized in EMSAs as described above. While PagR2 appeared to weakly bind the pagA and atxA promoters, we did not detect PagR2 S81Y-mediated binding of these promoters (Fig. 3). On the other hand, whereas native PagR2 did not bind the promoters of the surface layer genes sap and eag in this assay, PagR2 S81Y readily bound the sap promoter and the proximal promoter of eag (Fig. 3). We were also able to detect weak binding to the distal promoter of eag. We expanded the concentration range of purified PagR2 S81Y incubated with the DNA probes and calculated the Kds for the PagR2 S81Y-bound promoters. We were unable to calculate a Kd for the weakly bound eag distal promoter. However, we found that PagR2 S81Y binds the sap and eag proximal promoters with Kds similar to those observed for PagR1 (Fig. 4 and Table 1). The data suggest that the Y81 amino acid of PagR1 facilitates binding to S-layer gene promoters and that the difference in DNA binding by PagR1 and PagR2 is at least partially explained by amino acid differences at this residue location.

FIG 4.

Comparison of PagR1 and PagR2 Kds. Purified recombinant PagR1 and PagR2 were incubated at increasing concentrations with 0.2 nM biotin-labeled probes. The probe binding of (A) Psap and (B) Peag proximal were calculated as a fraction of bound promoter DNA over the total promoter DNA. Increasing binding with increasing protein concentration was plotted, and Kds were calculated using the Hill equation. Kds are reported with 95% confidence intervals.

Multimerization of PagR1 and PagR2 proteins.

Zhao et al. predicted that the Y81 residue of PagR1 may mediate protein-protein interactions (26). Y81 is found within the β-sheet 2 of the β-sheet hairpin. This residue appears to form part of the hydrophobic core of the protein, and its orientation toward adjacent α-helices of the protein suggests a role in the stabilization of protein integrity. PagR1 forms multimers in the crystal lattice, suggesting that multimerization may play a role in the function of this protein. The protein crystallized primarily as a dimer, similarly to other members of the ArsR-family of regulators. Given that the residue 81 of PagR2 is a serine instead of a tyrosine and that this difference appears to have effects on DNA binding, we decided to compare the dimerization of PagR1, PagR2, and PagR2 S81Y using coaffinity purification (Fig. 5).

FIG 5.

Homomultimerization of recombinant PagR proteins. Multimerization of recombinant proteins was explored using coaffinity purification. Cell lysates from ΔpagR1 strains expressing either His- or FLAG-tagged PagR proteins were coincubated as indicated and subjected to affinity purification using Ni-NTA resin. Coincubation with GFP-FLAG was performed as a negative-control. Proteins present prior to (Input) and after (Eluate) purification were detected using Western blotting with α-His and α-FLAG antibodies. Detection of FLAG-tagged proteins post Ni-NTA purification was indicative of interaction between the His- and FLAG-tagged coincubated proteins. (A) His- and FLAG-tagged PagR1. (B) His- and FLAG-tagged PagR2. (C) His- and FLAG-tagged PagR2 S81Y mutant.

In this experiment, PagR proteins with distinct epitope tags (His-tag or FLAG-tag) were expressed individually in separate ΔpagR1 cultures. PagR-His and PagR-FLAG culture samples were pooled after the induction of the recombinant proteins, followed by cell lysis. The lysates were then passed through an NTA-Ni resin column which immobilized the His-tagged variant. Protein-protein interactions between the differentially tagged proteins are indicated by the presence of both the His- and FLAG-tagged variants in eluate samples. As a negative-control for nonspecific interactions, cultures expressing tagged PagR proteins were pooled with cultures expressing GFP-FLAG.

First, we tested whether FLAG-tagged PagR1 would co-elute with His-tagged PagR1 after pulldown with an NTA-Ni resin column (Fig. 5A). PagR1 crystallizes as a dimer. Therefore, we expected to detect protein-protein interactions between the His-tagged and FLAG-tagged variants. We grew separate ΔpagR1 cultures expressing either PagR1-His, PagR1-FLAG, or GFP-FLAG in CA-CO₂. We pooled the culture samples in the combinations indicated in Fig. 5A (PagR1-His with PagR1-FLAG, PagR1-His with GFP-FLAG, PagR1-FLAG with GFP-FLAG) and collected samples prepurification (input) and postpurification (eluate). As expected, PagR1-FLAG coeluted with PagR1-His after passing the pooled lysates through the NTA-Ni resin column. We then tested the protein-protein interactions of His-tagged and FLAG-tagged PagR2. Interestingly, PagR2-FLAG did not coelute with PagR2-His (Fig. 5B). These data suggest that, unlike its dimeric paralog PagR1, PagR2 may exist as a mostly monomeric protein.

We then tested whether the S81Y mutation would facilitate the detection of protein-protein interactions of His-tagged and FLAG-tagged PagR2 proteins. We repeated the coaffinity purification experiment, pooling together cell cultures expressing PagR2 S81Y-His and PagR2 S81Y-FLAG (Fig. 5C). Unlike the results of the experiments using PagR2 with the native S81 residue, we were able to detect the coelution of PagR2 S81Y-FLAG with PagR2 S81Y-His, suggesting that the S81Y mutation is sufficient to stabilize multimeric forms of PagR2.

Given their high degree of amino acid identity, we hypothesized that PagR1 and PagR2 proteins may interact to form heteromultimers. To test this hypothesis, we pooled cultures expressing PagR2-His and PagR1-FLAG and passed the pooled samples through an NTA-Ni resin column. Interestingly, we did not observe the coelution of PagR1-FLAG with PagR2-His, suggesting a lack of interaction between the paralogs (Fig. 6A). On the other hand, we did observe the coelution of PagR1-FLAG with PagR2 S81Y-His (Fig. 6B). These data further implicate the presence of a Y residue at position 81 in mediating protein-protein interactions of PagR proteins. These data suggest that a single amino acid change between PagR1 and PagR2 led to a divergence in terms of DNA-binding and protein-protein complex ability.

FIG 6.

Heteromultimerization of recombinant PagR proteins. Cell lysates from ΔpagR1 strains expressing either PagR1-FLAG or PagR2-His were coincubated as indicated and subjected to affinity purification using Ni-NTA resin. Proteins present prior to (Input) and after (Eluate) purification were detected using Western blotting with α-His and α-FLAG antibodies. (A) PagR2-His and PagR1-FLAG. (B) PagR2 S81Y-His mutant and PagR1-FLAG.

DISCUSSION

In this work, we assessed the function of the regulatory proteins PagR1 and PagR2 and compared their involvement in virulence gene regulation in B. anthracis. Figure 7 summarizes our current model for the paralog-mediated regulation of virulence loci. The trans-acting virulence activator AtxA positively regulates the pagAR1 locus in pXO1. PagR1 directly represses the expression of its own promoter to regulate pagA levels. As shown for the first time in this work, PagR1 also directly represses atxA, leading to further autogenous regulation. PagR1 also represses the single sap promoter and at least one of the two eag promoters, allowing for preferential transcription initiation at the alternative promoter by a still unknown pXO1 factor. AtxA and its pXO2 paralogs, AcpA and AcpB, all positively regulate the pagR2 locus in pXO2. Like PagR1, PagR2 binds the pagA and atxA promoters, albeit weakly. Thus, PagR2 may autoregulate its own transcript levels by repressing atxA. Despite the apparent PagR2-mediated repression of sap and eag in our qRT-PCR assays, purified PagR2 did not bind these promoters in vitro, suggesting indirect regulation. The differences in DNA binding affinity between the paralogs is partially explained by one key amino acid difference. The mutation of PagR2 S81 to Y, which mimics the PagR1 sequence, allowed the purified protein to bind the sap and eag promoters. Thus, the PagR1 Y81 residue may mediate the recognition of and the binding to S-layer promoters. Furthermore, Y81 may play an important role in facilitating dimerization. While PagR1 is known to crystalize as a dimer, previous work did not explore whether multimerization was required for DNA-binding activity. Native PagR2 did not form detectable protein-protein interactions and displayed limited DNA-binding activity. On the other hand, the PagR2 S81Y mutant formed readily detectable homomultimers and bound the S-layer promoters with an affinity comparable to that of PagR1. Protein-protein interactions between native PagR1 and PagR2 S81Y further demonstrate the role of Y81 in mediating dimerization.

FIG 7.

Model of virulence gene regulation by trans-acting paralogs. The pagR1 and pagR2 loci are shown in red bolded letters on the pXO1 and pXO2 plasmids, respectively. The master virulence regulator of B. anthracis, AtxA, directly activates the expression of the pagAR1 locus. PagR1 strongly represses pagA, atxA, and sap. PagR1 also represses one of the two eag promoters, allowing for preferential transcription initiation by an unknown pXO1 factor at the alternative promoter. AtxA and its pXO2-encoded paralogs, AcpA and AcpB, positively control the pagR2 locus. Purified PagR2 weakly bound the pagA and atxA promoters, as shown with solid arrows. The PagR-mediated control of atxA suggests autogenous regulation of the pagR loci. PagR2 appears to indirectly affect sap and eag expression, as shown with dashed arrows. PagR1 is shown as a dimer, while PagR2 is shown as a monomer. The surface-exposed PagR1 Y81 may mediate protein-protein interactions that facilitate PagR1 dimerization and influence DNA-binding activity. PagR2 S81 residue appears to be buried in the secondary structure. The mutation of PagR2 S81 to Y facilitates protein-protein interactions and the DNA-binding activity of PagR2.

The PagR paralogs are members of the ArsR-family of transcriptional regulators, which typically responds to metal ions and represses gene expression (27, 28). Both PagR1 and PagR2 contain the canonical winged helix-turn-helix motif and the structural fold of ArsR-family regulators, as evidenced by the PagR1 crystal structure and the PagR2 modeled structure. However, there are some features of the proteins that are not shared with well-characterized ArsR proteins. For example, PagR1 and PagR2 lack discernible metal-sensing motifs. Also, as shown in this work, PagR1 and PagR2 regulate virulence-associated genes, which is not usually the case for classical ArsR-family regulators. Mounting evidence suggests the existence of an atypical group of ArsR-family regulators that lack metal-sensing motifs and regulate virulence genes instead of metal ion resistance loci. For example, PyeR P. aeruginosa regulates genes involved in biofilm formation but lacks metal-sensing motifs and does not respond to the addition of a variety of metals to the growth medium (29). Like PyeR, PagR1 and PagR2 of B. anthracis are potential members of this emerging class of atypical ArsR proteins.

PagR1-mediated repression of the toxin gene pagA is well-documented in a toxigenic, noncapsulated Sterne-like strain (pXO1⁺ pXO2^–) (13, 16, 21, 30). The pagAR1 locus is 100% conserved across B. anthracis strains. The pagR2 coding sequence is also 100% conserved between strains. However, unlike other B. anthracis strains, the Pasteur II strain (pXO1⁺, pXO2⁺) harbors a 5cnt deletion in the pagR2 promoter region (25). This strain is attenuated and, in some countries, has been used as a live vaccine, but the molecular basis for this attenuation is unknown. The pagR2 promoter mutation in Pasteur II leads to the complete loss of pagR2 expression in this strain. Exchanging the native Pasteur II pagR2 promoter with that of the archetype Ames (pXO1⁺, pXO2⁺) strain leads to detectable pagR2 transcript and, interestingly, increased pagA transcript and PA protein levels (25). These data suggest opposing effects of the two paralogs on pagA and serve as the initial rationale to further compare the regulatory roles of these highly identical proteins. We created a ΔpagR1 mutant in the Ames-derived ANR-1 background (pXO1⁺, pXO2^–) and expressed either pagR1 or pagR2 in trans. The expression of either PagR1 or PagR2 at comparable levels severely repressed the pagA transcript. Thus, the PagR2-mediated effect on the pagA transcript in our experiments differs from the effect reported by Liang et al. There are several key differences in the two studies. In our work, we investigated PagR2 function in the ANR-1 strain, utilizing the overexpression of recombinant pagR2 from an inducible promoter. Paralog expression levels in the Ames strain, as obtained from our previous RNA-seq studies, were 5,230 FPKM for pagR1 and 119 FPKM for pagR2, indicating an approximately 44-fold difference in transcript expression under toxin-inducing conditions (12). These data suggest the robust overexpression of recombinant PagR2 in our qRT-PCR experiments. The overexpression of PagR2 allows this protein to complement the ΔpagR1 mutant, resulting in the repression of pagA that was observed. However, purified PagR2 weakly binds the pagA promoter in vitro, suggesting that PagR1 may be the main repressor of pagA in conditions of native PagR expression. Liang et al.’s studies in the Pasteur II strain, in which the expression of pagR2 was restored in a pagR1⁺ strain, suggest the apparent disruption of the PagR1-mediated repression of pagA (25). Thus, the expression of pagR2 in the Pasteur II pagR1⁺ background may result in increased pagA transcript due to the disruption of PagR1 function by competitive interference or by some other unknown mechanism.

Zhao et al. identified several PagR1 amino acids which are highly conserved in other ArsR-family members, including Y81, which is part of the interface between the β-sheet hairpin and the α-helices of the protein (26). As we show here, this residue appears to impact PagR2-mediated DNA binding. There are three potential mechanisms by which Y81 could facilitate the recognition of DNA targets, and they are not necessarily mutually exclusive. First, Y81 may be subject to kinase-mediated phosphorylation. There are some examples of the tyrosine phosphorylation of regulatory proteins in bacteria. In B. subtilis, the FatR transcriptional regulator, which represses an operon involved in the metabolism of polyunsaturated fatty acids, is phosphorylated at the Y45 residue by a bacterial tyrosine kinase called PtkA (31). Another B. subtilis transcriptional regulator SalA, which represses a repressor of a particular exopeptidase, is also phosphorylated at Y327 by PtkA (31). Modeling of residue 81 on the PagR1 and PagR2 proteins in silico predicts the hydroxyl group of the PagR1 Y81 aromatic side chain to be surface-exposed, while the PagR2 S81 appears to be buried within the hydrophobic core (Fig. S2). Although serine phosphorylation has also been observed in bacteria (31), the occluded side chain of PagR2 S81 seems to be an unlikely target for phosphorylation.

A second potential mechanism for the Y81-mediated recognition of DNA is that Y81 participates in base-stacking interactions with DNA bases. The aromatic ring of amino acids, such as tyrosine, forms a π-electron cloud above and below the planar benzene ring (32). This type of noncovalent interaction is often referred to as π-stacking, and it can occur between aromatic residues and the rings of nucleic acid bases (33). Such interactions may facilitate nucleic acid recognition by regulatory proteins. For example, bacterial cold shock proteins use a highly conserved aromatic residue patch to interact with mRNA to control the expression of cold shock response proteins. Mutating the B. subtilis cold shock protein CspB aromatic patch to nonaromatic amino acids disrupts nucleic acid binding (34). The side chain of a serine residue, such as S81 in PagR2, lacks the necessary aromatic structure to participate in such interactions.

Finally, Y81 may be involved in protein-protein stabilizing interactions that facilitate the dimerization of PagR1. Noncovalent interactions, such as the previously described π-stacking, hydrogen bonding, and hydrophobic interactions, influence protein-protein stabilization. Mutation of S81Y allowed PagR2 to form readily detectable homomultimers in coaffinity purification experiments, potentially signifying stronger dimerization interactions. While the PagR2 S81 could potentially form stabilizing hydrogen bonds, this residue would be unable to participate in π-stacking interactions. The dimerization of AtxA, which is also critical for function, is mediated by a histidine (H379), another amino acid capable of π-stacking interactions via its five-carbon ring (35, 36). Mutation of H379 abrogates AtxA dimerization (35, 36). We predict that the mutation of Y81 would similarly disrupt the dimerization of PagR1.

Although PagR1 binds specific promoters, a PagR1 consensus DNA sequence for promoter recognition is not apparent. We utilized the MEME webserver to analyze the promoter regions of pagA, atxA, sap, and eag; however, no consensus sequences were identified. A previous model suggested that the elongated and curved shape of the PagR1 dimers may recognize the intrinsic DNA curvature in target promoters rather than specific sequences (37). Our coaffinity purification experiments suggest that native PagR2 is mostly monomeric and is unlikely to form the same curved dimer as PagR1. The mutation of S81Y likely allows PagR2 to form a dimer similar to that of PagR1, thus restoring DNA curvature recognition, at least for the S-layer promoters. Whether the PagR1 promoter targets display different degrees of curvature that may aid in fine-tuning regulation remains to be tested.

There are other examples in the literature of one single amino acid difference altering paralog function. For the S. aureus cold shock proteins CspA and CspC, residue 58 leads to distinct target recognition (38). The P58 of CspA is critical for the regulation of the pigment staphyloxanthin, and the mutation of the CspC E58 to P allows this paralog to control pigment expression (38). Another example of a single amino acid having drastic effects on the functions of related paralogs is found in the E. coli CueR and ZntR paralogs, which are part of the MerR-family of metal-sensing regulators (39–41). A single amino acid difference appears to play a crucial role in the recognition of +1 and +2 metal ions. The CueR S77 residue is highly conserved across other MerR-family regulators that respond to Cu⁺ ions, while ZntR C79 is similarly highly conserved across MerR-family regulators that bind Zn²⁺ ions (39–41). It is possible that specific amino acid differences allow PagR2 to directly regulate targets other than those bound by PagR1. The identification of targets regulated by PagR2 alone awaits discovery.

The PagR proteins, particularly PagR1, regulate virulence-associated loci. The pagA gene is one of the three structural genes of the tripartite anthrax toxin. AtxA is a trans-acting virulence regulator that controls both the toxin and the capsule, in addition to the loci related to amino acid metabolism (12, 42). The S-layer-associated proteins with roles in virulence serve as a mounting structure for S-layer-associated proteins with roles in B. anthracis virulence. These include the adhesin BslA, which is crucial for attachment to endothelial cells and has been implicated in the infiltration of the blood-brain barrier, and murein hydrolase BslO, which controls the elongation of B. anthracis cells growing in connected chains (43, 44). The chaining phenotype of B. anthracis is thought to protect cells from phagocytosis by the host’s immune cells. The promoter of one other putative S-layer-associated protein, GBAA_pXO1_0124, is directly bound and controlled by AtxA, further demonstrating the role of the S-layer as an important virulence factor of this human pathogen (13). The exact role of the PagR-mediated regulation of these loci in virulence remains to be elucidated.

In conclusion, our work corroborates the role of PagR1 as a direct regulator of toxin and S-layer gene expression. Our DNA binding experiments also demonstrate, for the first time, PagR1-mediated binding of the atxA promoter. While high overexpression of PagR2 appears to complement for the loss of PagR1 in a ΔpagR1 mutant, purified PagR2 displayed minimal ability to bind DNA in our assays. The S81Y mutation allowed PagR2 to bind some DNA promoters and form easily detectable homomultimers, like native PagR1. Future studies should establish whether the simultaneous expression of PagR1 and PagR2 results in competition for shared promoter targets in vivo in order to confirm the mechanism of PagR2-mediated disruption of PagR1 function. Additionally, it would be interesting to determine whether any other amino acid differences between PagR1 and PagR2 are associated with the distinct functionalities of these paralogs.

MATERIALS AND METHODS

Growth conditions.

The Escherichia coli strains TG1 and GM2163 (dam-) used for cloning were grown in 5 mL of Luria-Bertani (LB) (45, 46) (Invitrogen, Carlsbad, CA) medium at 37°C or 30°C with shaking. Samples for the Western blot analysis and the RNA extraction were obtained from Bacillus anthracis parent and mutant strains cultured in CA-CO₂ (47, 48). Briefly, strains were grown overnight in brain heart infusion (BHI, Becton, Dickson and Company, Franklin Lakes, NJ) media in air at 30°C. Strains were then subcultured at an OD_{600 nm} of 0.08 in 25 mL of CA (Casamino Acids medium supplemented with 0.8% [wt/vol] sodium bicarbonate and 0.1% [wt/vol] of glucose) and grown at 37°C with shaking and 5% atmospheric CO₂. For protein purification, strains expressing recombinant PagR proteins were subcultured in 500 mL of fresh BHI and were grown at 37°C with shaking in air. Under all growth conditions, PagR1 and PagR2 protein expression was induced with 100 μM and 1 mM IPTG, respectively. All cultures were incubated with shaking at 200 rpm. LB agar was used for plating both E. coli and B. anthracis. Liquid and solid media contained antibiotics when appropriate: carbenicillin (100 μg mL⁻¹), kanamycin (50 μg mL⁻¹), tetracycline (5 μg mL⁻¹), and spectinomycin (100 μg mL⁻¹).

Strain construction.

The strains and plasmids used in this study are listed in Table 2. The primers used in the polymerase chain reactions (PCR) are listed in Table S1. Recombinant PagR1 and PagR2 proteins were expressed in the ΔpagR1 mutant, which was constructed from the ANR-1 (pXO1⁺, pXO2^–) parent strain using gene disruption with a kanamycin cassette. PCR was used to amplify DNA fragments corresponding to sequences one kb directly upstream and downstream of the pagR1 gene locus on pXO1. Restriction enzyme digestion and ligation were performed to incorporate these sequences into a pUTE29 vector that contained the kanamycin resistance gene. pUTE29 also contains a tetracycline resistance marker outside the kanamycin cassette to screen for the loss of the plasmid upon a double-crossover event that results in the incorporation of the kanamycin cassette into the targeted locus. The resulting plasmid, pUTE560, was electroporated into B. anthracis. Plasmid-containing isolates were grown in LB supplemented with kanamycin at 37°C with shaking. After three passages in LB containing kanamycin, the cultures were subcultured in LB with no antibiotics and were grown again at 37°C with shaking. After three passages with no antibiotic in the medium, cultures were plated on LB agar plates, followed by replica plating on LB-kanamycin and LB-tetracycline agar plates. Resistance to kanamycin and sensitivity to tetracycline were indicative of incorporation of the kanamycin cassette into the pagR1 locus. Disruption of the pagR1 gene was confirmed using PCR and Sanger sequencing (GENEWIZ, South Plainfield, NJ).

TABLE 2.

B. anthracis strains and plasmids

Strain or Plasmid	Relevant characteristicsa	Source
Strains
ANR-1	B. anthracis (pXO1+, pXO2-)	53
UT428	ANR-1 derivative, pagR1::kanR	This work
Plasmids
pUTE29	Shuttle vector containing tetracycline resistance marker and multiple cloning site, Tetr	54
pUTE560	pUTE29-derived vector containing 2 kb total of pagR1 upstream and downstream sequences flanking Kanr marker; Kanr	This work
pUTE1133	pUTE657-derived expression vector containing pagR1-His sequence under the control of the lac operon promoter and atxA RBS; Spcr	This work
pUTE1135	pUTE657-derived expression vector containing pagR2-His sequence under the control of the lac operon promoter and atxA RBS; Spcr	This work
pUTE1209	pUTE657-derived expression vector containing pagR1-FLAG sequence under the control of the lac operon promoter and atxA RBS; Spcr	This work
pUTE1210	pUTE657-derived expression vector containing pagR2-FLAG sequence under the control of the lac operon promoter and atxA RBS; Spcr	This work
pUTE1211	pUTE657-derived expression vector containing pagR2-His sequence with S81Y mutation under the control of the lac operon promoter; Spcr	This work
pUTE1212	pUTE657-derived expression vector containing pagR2-FLAG sequence with S81Y mutation under the control of the lac operon promoter; Spcr	This work
pUTE1013	pUTE657-derived expression vector containing gfp-FLAG sequence under the control of the lac operon promoter; Spcr	52

^aKan^r, kanamycin resistant; Spc^r, spectinomycin resistant; Tet^r, tetracycline resistant.

Genes encoding recombinant PagR1 and PagR2 were placed under the control of the lac operon in the pUTE657 expression vector (49). PCR primers were used to amplify the pagR1 and pagR2 loci. To drive the translation of the proteins, the well-characterized atxA ribosomal binding site (RBS) was added immediately upstream of the start codon. For protein detection, either a His or FLAG C-terminal tag was also added. Constructs were ligated into pUTE657, followed by electroporation into the ΔpagR1 strain. Successful electroporation was confirmed using PCR with pUTE657-specific primers.

Site-directed mutagenesis of the plasmids containing recombinant PagR2 (pUTE1135 for PagR2-His, pUTE1210 for PagR2-FLAG) was used to create the PagR2 S81Y mutant. Briefly, PCR primers bearing the S81Y mutation were used to amplify pUTE1135 and pUTE1210. After the DpnI (New England Biolabs, Ipswich, MA) digestion of the template plasmid, the PCR product was transformed directly into E. coli. Sanger sequencing (GENEWIZ, South Plainfield, NJ) of the transformants was used to confirm the mutation. The resulting plasmids pUTE1211 (for PagR2 S81Y-His) and pUTE1212 (for PagR2 S81Y-FLAG) were then electroporated into the ΔpagR1 strain.

All strains were prepared as spore stocks for long-term storage as previously described (50). Briefly, clones were inoculated in 25 mL of phage assay (PA) medium that contained the appropriate antibiotics per strain and were grown at 30°C for 72 h. Spores were centrifuged and washed with sterile water before incubation in a 65°C water bath for 1 h. Finally, spore preparations were resuspended in fresh sterile water, and strain name designations were assigned according to Table 2.

RNA isolation and qRT-PCR.

Total RNA was prepared from cell cultures grown in CA-CO₂ until the indicated time points as described previously (51). Briefly, 10 mL of cell cultures were pelleted by centrifugation at 10,000 × g for 5 min at 4°C. The pellets were resuspended in 500 μL of 1× PBS and mixed with an equal volume of saturated phenol, pH 4.3 (Fisher Bioreagents, Fair Lawn, NJ), at 65°C. Samples were transferred to screw cap tubes containing 400 μL of 0.1 mm zirconia/silica beads (BioSpec Products, Bartlesville, OK). Cells were mechanically lysed by bead beating with a Mini Beadbeater (BioSpec Products, Bartlesville, OK) for two 1 min intervals with a 5 min incubation at 65°C in between. After centrifugation, the aqueous phase was removed and transferred into fresh tubes containing 500 μL of saturated phenol, pH 4.3, at 65°C. The samples were thoroughly mixed, incubated at room temperature for 5 min, and then centrifuged. Once again, the aqueous phase was removed and transferred to fresh tubes containing 0.3 volumes of chloroform. After a 10 minute incubation, samples were centrifuged, and the aqueous phase was finally removed and transferred into new tubes. To precipitate RNA, samples were incubated with 20 ng of glycogen, 0.1 volumes of 3 M sodium acetate, and 3 volumes of ice-cold 100% ethanol. RNA was precipitated for 1 h at −20°C before centrifugation. The RNA pellet was washed with 75% ice-cold ethanol and resuspended in DEPC-treated water. The RNA yield was quantified using a Nanodrop Spectrophotometer ND-1000.

For the qRT-PCR, 10 μg of RNA were treated with 20 units of DNase I (New England Biolabs, Ipswich, MA) for 30 min at 37°C to remove potential DNA contamination. DNase-treated RNA was recovered using the RNA Clean and Concentrator Kit (Zymo Research, Irvine, CA), according to kit instructions. The complete removal of DNA was confirmed by PCR, using primers specific to the gyrB control gene. cDNA synthesis was performed using random hexamers and the Super Script III Reverse Transcriptase Kit (Invitrogen, Carlsbad, CA), according to kit instructions. The resulting cDNA was recovered from the synthesis reaction using the DNA Clean and Concentrator Kit (Zymo Research, Irvine, CA). TaqMan-based qPCR assays were designed to detect pagA, atxA, sap, and eag target genes, as well as the gyrB reference gene. The assays for atxA (assay ID AII1OHI), sap (assay ID APDJXYV), and gyrB (assay ID AIHSQBA) were designed by the Applied Biosystems Custom TaqMan Assay Design team (Thermo Scientific, Waltham, MA). The assays for pagA and eag were designed in-house, and the primer/probe sequences are listed on Table S1. The cDNA and TaqMan primer/probe mixes were combined with SsoAdvanced Universal Probes Supermix (Bio-Rad, Hercules, CA) in duplicate per experiment. No reverse transcriptase (NRT) and no template (NTC) controls were also included. The qPCR CT values were recorded using a CFX96 Real-Time System C1000 Touch Thermal Cycler (Bio-Rad, Hercules, CA). The CT value for each gene was the average of duplicate reactions. The ΔCT values were calculated by subtracting the internal control gene gyrB CT values from those of the experimental genes of interest. The log₁₀ relative expression (2[–ΔCT]) was calculated for each gene of interest. The final relative expression is reported as the average of three biological replicates.

Western blotting.

Cell lysates for Western blotting were prepared as previously described (52). Briefly, culture volumes were pelleted by centrifugation at 16,000 × g for 5 min at 4°C. The pellets were resuspended in 550 μL of KTE Buffer (10 mM Tri-HCl pH 8.0, 10 0 mM KCl, 10% ethylene glycol) supplemented with EDTA-free Pierce Protease Inhibitor cocktail (Thermo Scientific, Waltham, MA). The samples were transferred to screw cap bead-beating tubes containing 400 μL of 0.1 mm zirconia/silica beads (BioSpec Products, Bartlesville, OK). Samples were lysed by bead beating for 2 min with a Mini Beadbeater (BioSpec Products, Bartlesville, OK). Clear cell lysates were mixed with SDS Loading Buffer (0.05% bromophenol blue, 0.1 M DTT, 10% glycerol, 2% SDS, 5 mM Tris-Cl pH 6.8) and boiled for 5 min. Proteins were subjected to SDS-PAGE and transferred to 0.45 μM PDVF membranes (Thermo Scientific, Waltham, MA) in CAPS Buffer (10 mM CAPS pH 11.0, 10% methanol). Membranes were blocked with a 5% skim milk solution in TBS-T (1× TBS pH X, 0.1% Tween 20).

All antibody dilutions were prepared in TBS-T. To detect His-tagged and FLAG-tagged proteins, membranes were incubated with a 1:5,000 dilution of α-His antibody (Genscript, Piscataway, NJ) and α-FLAG antibody (Genscript, Piscataway, NJ), respectively. To detect the RNA polymerase subunit β load control, the membranes were incubated with a 1:3,000 dilution of α-RNApolβ antibody (Thermo Scientific, Waltham, MA). All membranes were subsequently incubated in a 1:10,000 dilution of the appropriate secondary antibodies (α-mouse or α-rabbit) (Bio-Rad, Hercules, CA), followed by washing and detection with SuperSignal West Dura chemiluminescent substrate (Thermo Scientific, Waltham, MA). ImageJ software was used for the densitometry calculations.

Purification of PagR proteins.

Recombinant PagR1-His, PagR2-His, and PagR2-His S81Y were purified from B. anthracis cultures using affinity purification as described previously (12). Briefly, the ΔpagR1 strain harboring plasmids pUTE1133 (PagR1-His), pUTE1135 (PagR2-His), or pUTE1211 (PagR2-His S81Y) were grown in 500 mL of BHI media at 37°C with shaking. After 2 h of growth, expression was induced with 100 μM and 1 mM IPTG for the PagR1 and PagR2 recombinant proteins, respectively. After a 2 h induction, the cultures were pelleted by centrifugation. The pellets were resuspended in 10 mL of Binding Buffer (5 mM Imidazole pH 7.9, 0.5 M NaCl, 20 mM Tris pH 7.15, 5 mM β-mercaptoethanol) supplemented with EDTA-free protease inhibitor cocktail (Thermo Scientific, Waltham, MA), 40 units of DNase I (New England Biolabs, Ipswich, MA), and 10 mM MgCl₂. Cells were lysed by sonication (20% amplitude for 5 min, cycling between a 5 sec sonication and 5 sec pause on ice), and soluble lysates were collected by centrifugation. Cell lysates were incubated with preequilibrated 4 mL Ni-NTA resin (Qiagen, Hilden, Germany) for 60 min with shaking at 4°C. The resin was pelleted by centrifugation and washed twice with 4 mL of Binding Buffer, followed by one 8 mL wash with Wash Buffer 1 (40 mM Imidazole pH 7.9, 1 M NaCl, 20 mM Tris pH 7.15, 5 mM β-mercaptoethanol) in batch. The resin was centrifuged and resuspended one last time in 8 mL of Wash Buffer 1 before being loaded onto Poly-Prep chromatography columns (Bio-Rad, Hercules, CA). The loaded resin was washed with 4 mL of Wash Buffer High Salt (40 mM Imidazole pH 7.9, 1.5 M NaCl, 20 mM Tris pH 7.15, 5 mM β-mercaptoethanol), again with Wash Buffer 1, and then with 4 mL of Wash Buffer 2 (75 mM Imidazole pH 7.9, 1 M NaCl, 20 mM Tris pH 7.15, 5 mM β-mercaptoethanol). After a final wash with Wash Buffer 1, the bound proteins were eluted using Elution Buffer (20 mM Tris pH 7.15, 800 mM Imidazole pH 7.9, 500 mM NaCl, 5 mM β-mercaptoethanol) into tubes containing prechilled Collection Buffer (20 mM Tris pH 7.15, 150 mM NaCl, 10% Glycerol, 5 mM EDTA, 5 mM β-mercaptoethanol). The purification of proteins was confirmed using SDS-PAGE and Coomasie staining. Protein-containing fractions were dialyzed in Dialysis Buffer (150 mM KCl, 5% Glycerol, 20 mM Tris pH 7.5, 5 mM DTT). After 3 passages at 4°C, proteins were recovered, and concentrations were determined using Bradford assay reagent (Bio-Rad, Hercules, CA).

Electrophoretic mobility shift assays (EMSAs).

EMSAs were performed with biotinylated DNA probes that were amplified using PCR. The labeled DNA probes correspond to promoters of the pagA, atxA, sap, and eag genes. As a negative-control, the promoter of the housekeeping gene dnaA was utilized. For eag, both the distal and the proximal promoters were tested. Each probe (0.2 nm) was incubated with increasing concentrations of purified recombinant PagR1-His, PagR2-His, and PagR2-His S81Y proteins in 1× Reaction Buffer (100 mM Tris pH 7.5, 780 mM KCl, 1 mM DTT, 50% glycerol, 0.1 mg/mL BSA) supplemented with 1 μg of poly(dI-dC) (Thermo Scientific, Waltham, MA) synthetic DNA substrate as a competitor for nonspecific interactions. Unlabeled DNA probes (0.2 nm and 20 nm) were also added to two of the reactions. The samples were incubated at room temperature for 30 min before being subjected to electrophoresis in a native 5% polyacrylamide gel in 1× TBE buffer at 4°C. DNA:protein complexes were transferred to Amersham Hybond-N nylon membranes (GE Healthcare, Little Chalfront, UK) in 1× TBE. DNA was cross-linked to the membranes via UV light exposure, and biotin-labeled DNA probes were detected using the North2South Chemiluminescent Detection Kit (Thermo Scientific, Waltham, MA) according to the manufacturer’s instructions. To calculate binding Kds, EMSAs were repeated with an expanded range of increasing protein concentration. ImageJ software was used for the densitometry calculations. Probe binding was calculated as a fraction of the bound promoter DNA over the total promoter DNA added to each reaction. The fraction of promoter DNA bound by the proteins was plotted over the increasing protein concentration, and Kds were calculated using the Hill equation. Kds were reported with 95% confidence intervals.

Coaffinity purification.

Coaffinity purification was performed as described previously with His-tagged and FLAG-tagged proteins (12). The ΔpagR1 strains expressing epitope-tagged PagR1, PagR2, and the GFP control were grown in CA-CO₂ at 37°C with shaking. After 2 h of growth, the expression of epitope-tagged proteins was induced with 100 μM, 1 mM, and 25 μM IPTG for PagR1, PagR2, and GFP, respectively. After a 2 h induction, the cultures were pooled in the combinations indicated in Fig. 5 and 6 and then pelleted by centrifugation. The pellets were washed with 10 mL of Binding Buffer supplemented with EDTA-free Pierce Protease Inhibitor cocktail (Thermo Scientific, Waltham, MA) before being resuspended in 2 mL of Binding Buffer. Clear cell lysates were prepared by bead beating as described above. An aliquot of each combination sample was saved and labeled as an Input sample. The remaining volume was incubated at 37°C in 5% CO₂ for 20 min to facilitate protein interactions. A volume of 70 μL of NTA-Ni agarose beads preequilibrated with Binding Buffer was added to each sample, followed by incubation at 4°C for 2 h with light shaking. The resin was washed twice with 500 μL of Binding Buffer and loaded onto Pierce Spin Snap Cap Columns (Thermo Scientific, Waltham, MA). The resin was thoroughly washed, according to the protein purification protocol described above, using 500 μL volumes of wash buffers and centrifugation at 1,000 × g at 4°C to remove unbound proteins. His-tagged proteins and any interacting partners were eluted with Elution Buffer and labeled as Eluate samples. Both the Input and the Eluate samples were mixed with SDS Loading Buffer and subjected to SDS-PAGE. Western blotting as described above was used to detect His-tagged and coeluting FLAG-tagged proteins.

Statistical analysis.

All statistical analyses were performed using GraphPad Prism software version 9. A Student's t test was used to compare PagR1-His and PagR2-His levels post IPTG induction. One-way analyses of variance (ANOVAs) followed by Tukey’s multiple-comparison tests were used to compare differences in relative mRNA levels. A nonlinear regression model of specific binding was fit to the DNA promoter binding data. The Hill equation was used to calculate Kds and confidence intervals.