Structural and Functional Analysis of Toxin and Small RNA Gene Promoter Regions in Bacillus anthracis

ABSTRACT

It was previously demonstrated that anthrax toxin activator (AtxA) binds directly to the σ^A-like promoter region of pagA (encoding protective antigen, PA) immediately upstream of the RNA polymerase binding site. In this study, using electrophoretic mobility shift assays and in vivo analyses, we identified AtxA-binding sites in the promoter regions of the lef and cya genes (encoding lethal and edema factors, respectively) and of two Bacillus anthracis small RNAs (XrrA and XrrB). Activities of all four newly studied promoters were enhanced in the presence of CO₂/bicarbonate and AtxA, as previously seen for the pagA promoter. Notably, the cya promoter was less activated by AtxA and CO₂/bicarbonate conditions. The putative promoter of a recently described third small RNA, XrrC, showed a negligible response to AtxA and CO₂/bicarbonate. RNA polymerase binding sites of the newly studied promoters show no consensus and differ from the σ^A-like promoter region of pagA. In silico analysis of the probable AtxA binding sites in the studied promoters revealed several palindromes. All the analyzed palindromes showed very little overlap with the σ^A-like pagA promoter. It remains unclear as to how AtxA and DNA-dependent RNA-polymerase identify such diverse DNA-sequences and differentially regulate promoter activation of the studied genes.

IMPORTANCE Anthrax toxin activator (AtxA) is the major virulence regulator of Bacillus anthracis, the causative agent of anthrax. Understanding AtxA’s mechanism of regulation could facilitate the development of therapeutics for B. anthracis infection. We provide evidence that AtxA binds to the promoters of the cya, lef, xrrA, and xrrB genes. In vivo assays confirmed the activities of all four promoters were enhanced in the presence of AtxA and CO₂/bicarbonate, as previously seen for the pagA promoter. The cya and lef genes encode important toxin components. The xrrA and xrrB genes encode sRNAs with a suggested function as cell physiology regulators. Our data provides further evidence for the direct regulatory role of AtxA that was previously shown with the pagA promoter.

INTRODUCTION

Interaction of transcription factors (TF) with their regulated DNA elements is an integral part of gene regulatory networks. Determining how TFs recognize specific DNA sequences is important for understanding mechanisms of gene regulation. Anthrax toxin activator (AtxA) is a TF and the master virulence gene regulator of Bacillus anthracis, the causative agent of anthrax (1, 2). AtxA regulates transcription of genes located principally on the B. anthracis virulence plasmids, pXO1 and pXO2 (2). The pXO1 plasmid encodes the toxin genes (pagA, cya, and lef), as well as AtxA, while the pXO2 plasmid encodes the genes for the antiphagocytic poly-γ-d-glutamic acid capsule. Besides these targets, AtxA also upregulates two small RNAs, XrrA and XrrB, which were shown to have regulatory functions (3). Expression of these sRNA and toxin genes is dependent not only on AtxA, but also the presence of CO₂ in the environment, although the mechanism of this synergy is still unclear (2).

Earlier, we reported AtxA-meditated regulation of pagA (protective antigen gene) transcription is by direct binding to the promoter region upstream of the RNA polymerase binding site (4). In silico, in vitro, and in vivo analyses showed a putative stem-loop with a palindromic sequence, located upstream of the RNA polymerase binding site in the pagA promoter region, is important for AtxA binding and pagA regulation. This finding is consistent with the view that homodimeric helix-turn-helix (HTH) transcription factors like AtxA (5) usually bind to palindromes (6).

Importantly, the AtxA-binding site was found to be near the target of the housekeeping sigma factor A (σ^A), which was previously identified by Dai and others in the pagA promoter region (7). This finding would suggest other promoters regulated by AtxA have binding sites for σ^A RNA-polymerase sigma subunits. However, Hadjifrangiskou and Koehler could not identify DNA-binding sites for σ^A or any other known RNA-polymerase sigma subunits in the promoter regions of the lethal factor (lef) and edema factor (cya) genes (8). They suggested the presence of high intrinsic DNA curvature may play a role in recognition of the promoters by the transcription machinery and/or transacting regulators such as AtxA. Because many genomic regions and promoters not regulated by AtxA also show a high level of intrinsic curvature, it is difficult to conclude that curvature represents the only factor important for AtxA-binding (9).

Based on our recent analysis of AtxA-dependent pagA promoter activation, we extended our analyses to the promoters of lef and cya as well as to the promoters of the B. anthracis small RNAs (sRNAs) xrrA and xrrB, the transcription of which is also regulated by the presence of CO₂/bicarbonate and AtxA in B. anthracis (2, 3). We also include analysis of a third, newly identified sRNA, XrrC (9). Our interest in these promoters was provoked by the very high transcriptional efficacy of these sRNAs and the absence of information on the structure and function of the corresponding promoters. We also sought to clarify the role of a predicted upstream element (UP) of the pagA promoter on AtxA activation. The UP-element binds the C-terminal domain of the RNA polymerase α-subunit (10). We found that deletion of this element from the pagA promoter results in a complete loss of AtxA-dependent pagA promoter activation, indicating an importance of AtxA interaction with the α-subunit of the RNA polymerase. This would suggest AtxA acts as a class I activator (11).

Understanding the specifics of gene regulatory systems also requires knowledge of cis-regulatory elements. Given that it is well known CO₂/bicarbonate acts in synergy with AtxA for promoter activation, all our experiments were performed both with and without CO₂/bicarbonate in the growth media (2, 4).

RESULTS

Identification of xrrA and xrrB sRNAs promoter regions.

Our previous RNA-seq experiments (2) did not determine the direction of transcription for these two sRNAs. To identify the orientations and locations of these promoters, we cloned potential promoter regions from both DNA strands (see Materials and Methods section) and created plasmids pRPxrrAD, pRPxrrARC, pRPxrrBD, and pRPxrrBRC (Table 1). The addition of a ribosome-binding site (RBS) upstream of the translation start codon of gfp allowed us to monitor fluorescence and identify the active promoters of the two sRNAs. Accordingly, BH500 (pKA2) strains separately containing these plasmids were analyzed by fluorometric assay. The analyses showed plasmids pRPxrrAD (from sense strand) and pRPxrrBRC (from reverse strand) had significantly higher levels of activation in the presence of CO₂/bicarbonate than their respective alternates and thus, contained active promoters regulating the sRNAs transcription (Table 1; Fig. 1A). Nucleotide sequences of the promoters regulating transcription of the sRNAs are represented in Fig. S3 (xrrA) and S4 (xrrB).

TABLE 1.

Plasmids used in this study

Plasmid	Relevant characteristic(s)	Source or reference
pKA2	Plasmid containing pUB110 replicon, KmR for selection both in E. coli and B. anthracis, and functionally active atxA	4
pMR	pXO1 minireplicon, Spr in B. anthracis, Apr in E. coli	21
pMRC4	pMR encoding DasherC4 GFP under control of full-length pagA promoter region	4
pRSP	pMRC4 plasmid with upstream portion of full-length pagA promoter region deleted	4
pRSPΔUP	pRSP plasmid with UP element of RNA-polymerase sigma binding-site deleted	This study
pRPL	pMR encoding DasherC4 GFP under control of a 333-bp lef promoter region	This study
pRPE	pMR encoding DasherC4 GFP under control of a 318-bp pagA promoter region	This study
pRPxrrAD	pMR encoding DasherC4 GFP under control of a 396-bp of xrrA sRNA 5′-region located on direct DNA strand	This study
pRPxrrARC	pMR encoding DasherC4 GFP under control of a 305-bp of xrrA sRNA 5′-region located on complement DNA strand	This study
pRPxrrBD	pMR encoding DasherC4 GFP under control of a 234-bp of xrrB sRNA 5′-region located on direct DNA strand	This study
pRPxrrBRC	pMR encoding DasherC4 GFP under control of a 241-bp of xrrB sRNA 5′-region located on complement DNA strand	This study
pRPxrrC	pMR encoding DasherC4 GFP under control of a 164-bp of xrrC sRNA 5′-region located on direct DNA strand	This study
pRPL L	pRPL plasmid with 159 bp of lef gene promoter truncated from 5′-end	This study
pRPL M	pRPL plasmid with 112 bp of lef gene promoter truncated from 5′-end	This study
pRPL S	pRPL plasmid with 77 bp of lef gene promoter truncated from 5′-end	This study
pRPE L	pRPE plasmid with 191 bp of cya gene promoter truncated from 5′-end	This study
pRPE M	pRPE plasmid with 133 bp of cya gene promoter truncated from 5′-end	This study
pRPE S	pRPE plasmid with 101 bp of cya gene promoter truncated from 5′-end	This study
pRPE SΔ	pRPE plasmid with 54 bp of cya gene promoter truncated from 5′-end	This study
pRPxrrA L	pRPxrrAD plasmid with 159 bp of xrrA sRNA promoter truncated from 5′-end	This study
pRPxrrA M	pRPxrrAD plasmid with 93 bp of xrrA sRNA promoter truncated from 5′-end	This study
pRPxrrA S	pRPxrrAD plasmid with 64 bp of xrrA sRNA promoter truncated from 5′-end	This study
pRPxrrB L	pRPxrrBRC plasmid with 160 bp of xrrB sRNA promoter truncated from 5′-end	This study
pRPxrrB M	pRPxrrBRC plasmid with 111 bp of xrrB sRNA promoter truncated from 5′-end	This study
pRPxrrB S	pRPxrrBRC plasmid with 68 bp of xrrB sRNA promoter truncated from 5′-end	This study

FIG 1.

Promoters of sRNAs are activated by CO₂/bicarbonate and are bound by AtxA. (A) Identification of the B. anthracis sRNA promoters. The sense (xrrAD, d-direct) and the anti-sense (xrrBRC, R-reverse, C-complement) strands of the promoter regions were cloned ahead of GFP in pRP type plasmids and transformed into B. anthracis BH500 containing the pKA2 plasmid encoding AtxA. GFP reporter activities for xrrA and xrrB are dependent on CO₂/bicarbonate conditions. Differences were analyzed by Welch’s t test. Nucleotide sequences of these promoters are represented in Fig. S3 and S4. (B) EMSA using infrared dye-labeled fragments of the XrrA (top) and XrrB (bottom) promoters. The labeled probes (1.7 nM) were incubated with the indicated concentrations of AtxA protein and samples were run on 2% agarose in EABE buffer. (C, D) Competitive EMSA with 750 nM AtxA and labeled xrrA and xrrB promoter fragments (1.7 nM) mixed with unlabeled promoter fragments (competitor probes) in 10×, 60×, and 250× molar ratios. Arrows indicate shifted fragments.

GFP reporter analysis and AtxA binding of lef, cya, xrrA, and xrrB promoters.

We previously reported that transcription of all three B. anthracis toxin genes and these two sRNAs are regulated by AtxA in the presence of CO₂/bicarbonate (2). Later, we identified a region in the pagA promoter (SLII) that is required for AtxA binding and activity using electrophoretic mobility shift assays (EMSA) and a GFP reporter plasmid (4). Here, we employed a similar strategy to understand the regulation of the lef, cya, and sRNA genes.

We used EMSAs to assess direct binding of AtxA to the promoters. For the sRNA promoters, EMSAs showed purified AtxA could bind to both, with the xrrA promoter appearing to have a stronger affinity (Fig. 1B). Confirming the specificity of AtxA binding to the promoters, unlabeled promoter fragments added in excess blocked the gel shift of the dye-labeled promoter DNA of xrrA and xrrB (Fig. 1C and D). An EMSA performed with the xrrA promoter and an excess of unlabeled Bacillus cereus genomic DNA further confirmed AtxA’s specificity, as the B. cereus DNA was unable to block the binding of AtxA to the dye-labeled xrrA promoter (Fig. S6). No binding was found for the alternative strands of the xrrA and xrrB promoters (data not shown).

To understand the regulation of the lef and cya promoters we created two GFP reporter plasmids, pRPL and pRPE, for the respective genes (Table 1). Nucleotide sequences of the lef and cya promoters in the pRPL and pRPE plasmids are shown in Fig. S1A and S2A. Activities of the promoters were measured by fluorescence of GFP in B. anthracis BH500 strains. Measurements were done in strains with or without the AtxA-expressing pKA2 plasmid and grown in the presence or absence of CO₂/bicarbonate. As expected, the lef and cya promoters were activated in the presence of pKA2 (AtxA) and CO₂/bicarbonate, though to a slightly lower level than the pagA promoter (Fig. 2A). Absence of either pKA2 (AtxA) or CO₂/bicarbonate significantly reduced the lef and cya promoter activity (Fig. 2A). With EMSAs, we identified the direct binding of AtxA to the lef and cya promoter fragments with an apparent stronger affinity for the lef promoter (Fig. 2B). As shown in Fig. 2B, as low as 600 nM AtxA could bind to 1.7 nM the lef promoter fragments to produce a full shift in the EMSA, whereas approximately 1,000 nM purified AtxA was required to generate a similar shift with 1.7 nM the cya promoter fragment. Competitive EMSAs with an excess of unlabeled promoter fragments showed the binding to be specific (Fig. 2C).

FIG 2.

AtxA and CO₂/bicarbonate-dependent transcriptional activation of lef and cya promoters correlated with direct binding of AtxA to the promoters. (A) GFP fluorescence levels due to transcriptional activation of the pagA, cya, and lef promoters in pRSP, Prpe, and pRPL plasmids, respectively. High fluorescence was seen only in AtxA-expressing strains (+pKA2 plasmid) grown in CO₂/bicarbonate containing medium. Differences were analyzed by Welch’s t test. (B) EMSA with infrared dye-labeled fragments of the cya and lef promoters and the indicated concentrations of AtxA protein. Samples were run on 2% agarose in EABE buffer. (C) Competitive EMSA with 750 nM AtxA and labeled cya and lef promoter fragments (1.7 nM) mixed with unlabeled promoter fragments (competitor probes) in 10×, 60×, and 250× molar ratios. Arrows indicate shifted fragments.

To compare the levels of promoter activity, estimation of GFP expression in vivo under the control of the pagA, lef, cya, sRNA xrrA, and sRNA xrrB promoters was performed again in a single experiment (Fig. 3). Paradoxically, in comparison with high levels of sRNA transcription (2), production of GFP by the sRNA promoters was unexpectedly low compared to amounts from the toxin gene promoters.

FIG 3.

Comparison of GFP reporter activity controlled by the toxin gene and sRNA natural promoters in the pRSP, pRPL, pRPE, pRPxrrBRC, and pRPxrrAD plasmids, respectively. All determinations were made at the same time and under the same conditions. Promoter activity is dependent on CO₂/bicarbonate growth conditions. Statistical analysis by Welch’s t test of GFP fluorescence levels due to transcriptional activation by the promoters is included.

Analysis of lef, cya, and sRNA promoters for the regions responsible for AtxA activation and expression of GFP gene.

Although we established that the lef, cya, xrrA, and xrrB promoters need both AtxA and CO₂/bicarbonate to produce GFP, our work thus far did not identify the promoter sequences to which AtxA might bind, nor what promoter sequences might be required for the activating effects of CO₂/bicarbonate. As noted, in silico sequence comparisons have not identified consensus sequences in the promoter regions of AtxA-activated genes (8). Previously, we found that AtxA binding to the pagA promoter depends on a sequence near the −35 site to which RNA polymerase binds (4). For this reason, shortened promoter fragments (created using G-blocks from Integrated DNA Technologies [IDT, Inc., Coralville, IA]) of lef, cya, xrrA, and xrrB were synthesized to identify regions required for AtxA binding. Truncated promoters of three different lengths (referred to as long, medium, and short; shortened to L, M, and S) were used for this purpose. Each truncated promoter was cloned into the GFP reporter plasmid to allow for the quantification of activating ability.

The final nucleotide sequences of all shortened promoters are shown in Fig. S1A to S4A in comparison with the originally cloned promoters. Fluorometric analyses with shortened promoter fragments (Fig. 4; Fig. S1 to S4) showed that all the long fragment promoters retain maximal ability to activate GFP production in the presence of AtxA and CO₂/bicarbonate. On the other hand, none of the short fragment promoters were able to activate transcription even in the presence of CO₂/bicarbonate. The medium size promoter fragments had different levels of ability to activate production of GFP, but in general were similar to the long fragment promoters. The medium cya promoter was somewhat less active (Fig. 4B; Fig. S2), while the medium sRNA promoters-maintained potency at a level almost equal with the longest promoters (Fig. 4; Fig. S3 to S4). These results indicate that sequences necessary for AtxA-binding are in regions near the end of the medium sized promoters while those for the sRNA promoters are largely within the medium sized fragments.

FIG 4.

Transcriptional activation and AtxA binding by promoters of different lengths. Abilities of long (L), medium (M), and short (S) promoters (left images, panels A to C, respectively) to activate GFP expression in B. anthracis. Statistical analysis was by Welch’s t test of GFP fluorescence levels. EMSA to assess binding of AtxA to the same promoters (panels A to C, right hand images). The labeled probes (final concentration 1.7 nM) were added to each reaction mixture and incubated with the indicated concentration of AtxA protein. Samples were run on 2% agarose in EABE buffer.

We further analyzed these promoter regions to identify potential sequence or structural features that may play a role in AtxA binding. Two potential stem-loops palindromes (SL) were identified in both the lef and cya promoters (SLI, SLII; Fig. S1 and S2). The xrrA promoter region contains a single SL (Fig. S3A), as does the xrrB promoter region (Fig. S4A). Comparison of sequences of these putative regulatory signals did not identify strong homologies. A possible exception is the weak similarity in the sequences of the SLs of small RNAs xrrA and xrrB (Fig. S4). These interrupted palindromes have nearly identical shared sequences of 12 bp that occur infrequently, about every 18,000 bp. The fact they are both located 90 to 100 bp upstream of the transcriptional start sites suggests they could be part of the targets of factors regulating transcription. With this possible exception, there remain no clear sequence features that could explain recognition of all the promoters responsive to AtxA.

Consistent with the activation of the GFP reporter, EMSA analysis showed a gradual decrease in AtxA-binding ability for all promoter fragments from the large to medium size promoter regions (Fig. 4). No binding by any of the short promoters was observed (Fig. 4C), consistent with the lack of activation of GFP production by the shortest fragments. An interesting exception was found for the cya promoter fragments in that even the shortest one could still activate some GFP production (Fig. 4C).

For analysis of XrrC, we constructed a single plasmid using the positive-strand promoter region identified by Furuta at al. (9). In contrast to that work, we found no activating ability of AtxA, and the level of transcriptional activity was much lower than for the other promoters studied here (Fig. S7). Interestingly, there did seem to be a small decrease in promoter activation upon the removal of sodium bicarbonate/CO₂ when AtxA was present.

Deletion of predicted UP element from pagA promoter leads to the loss of AtxA activating ability.

Earlier reports on bacterial promoters identified the important role of an AT-rich upstream region (UP element) in bacterial gene regulation (10). Based on standard UP-element features, we identified a similar sequence in the pagA promoter, as depicted in Fig. 5A. Deletion of the 21-bp sequence from the pagA promoter (boxed in Fig. 5A) juxtaposed the AtxA and σ^A binding sites and resulted in a complete loss of AtxA-dependent pagA promoter activation, even in CO₂/bicarbonate conditions (Fig. 5B). This deletion did not include the putative AtxA binding site in the pRSPΔUP sequence (Fig. 5A), as was confirmed by showing with EMSA that the promoter with the deleted sequence could still bind AtxA (Fig. 5C).

FIG 5.

Deletion of the pagA promoter predicted UP-element leads to the loss of AtxA activating ability although its DNA-binding effect was saved. (A) Deletion of the predicted UP-element (boxed) could bring together the AtxA (uppercase and bold) and σ^A (lowercase and bold) binding sites. The predicted DNA elements are represented as: transcription start site (2) in uppercase, underlined and bold; conserved Shine–Dalgarno sequence in uppercase, bold and italic; translation start codon shown as uppercase bold and gray. (B) Transcriptional activity measured as GFP fluorescence in BH500 strains containing pKA2 (AtxA) and either pRSP (pagA promoter) or pRSPΔUP (pagA promoter deleted of UP), grown in CO₂/bicarbonate conditions. Statistical analysis was by Welch’s t test of GFP fluorescence levels. (C) EMSA for binding activity of AtxA to the intact or UP-deleted pagA promoter (pRSP, and pRSPΔUP, respectively). The labeled probe (final concentration of probe was dependent on respective affinity for AtxA and was: 3.5 nM for pRSP and 5 nM for pRSPΔUP) was added to each reaction mixture and incubated with the indicated concentrations of the AtxA protein. Samples were run on 2% agarose in EABE buffer.

DISCUSSION

In this study, we show the transcriptional regulator AtxA of B. anthracis (1), which tightly controls transcription of pXO1-encoded toxin genes and two small RNAs (2), exhibits DNA-binding specificity in vitro for the promoters of all these genes and requires CO₂/bicarbonate conditions in vivo to activate production of GFP directed by promoters of these toxin genes and sRNAs.

To identify the promoters of the XrrA and XrrB sRNAs we paired a conserved Bacillus anthracis toxin RBS and gfp gene with the DNA regions immediately upstream and downstream of the sRNAs in a reporter plasmid. Functional analysis of these DNA fragments allowed us to identify the promoters (Fig. 1A) and showed the direction of sRNA transcription, confirming the results obtained by Corsi et al. (3) and Furuta et al. (9). As mentioned before, in comparison with the high levels of sRNA transcripts in the Ames35 strain grown to A₆₀₀ 2.1 (2), production of GFP controlled by the same promoters in the BH500 strain grown to A₆₀₀ 1.0 was much lower than that from the toxin gene promoters (Fig. 3). However, the low levels of GFP fluorescence under the control of sRNA promoters (9.9 ± 0.9 × 10³ a.u. for XrrB and 7.2 ± 0.2 × 10³ a.u. for XrrA), determined at A₆₀₀ 1.0 (late exponential phase of growth, according to Fig. S8), did not essentially differ from those (9.8 ± 0.5 × 10³ a.u. for XrrB and 7.5 ± 0.3 × 10³ a.u. for XrrA) obtained at A₆₀₀ 2.1 (the early stationary phase, according to Fig. S8).

All cloned promoters not only activated production of GFP in B. anthracis BH500 (pKA2) strains grown in CO₂/bicarbonate conditions, but also bound AtxA in vitro, although there are no obvious similarities in the promoter regions of these AtxA-controlled genes. Moreover, consensus sequences for recognition by RNAP sigma factors are not apparent for any of the studied AtxA-dependent transcription start sites, excluding the promoter of the pagA gene in which the primary σ^A binding site was identified (4).

Bacteria contain several different sigma factors to modulate gene expression in response to various conditions. While the primary σ^A factor is responsible for the transcription of housekeeping genes, alternative sigma factors are needed for the expression of genes induced in response to environmental stress or during development processes (12). In view of this, it is possible that lef and cya transcription are regulated by alternative sigma factors which act with AtxA in response to CO₂/bicarbonate conditions, although we do not know of any such alternative sigma factors in B. anthracis. To conserve cellular resources, production of these factors could be activated by alternative sigma factors at an appropriate time and to an ideal level in a different way than PA.

Unfortunately, in silico sequence analyses do not identify exact binding sites for any known factors in either lef or cya, nor the sRNAs promoters. Further work is needed to determine the identity and binding sites of the sigma factors involved with these promoters. However, our experiments with the shortened versions of the promoters clearly showed regions important for activity of these promoters. Ability to activate production of GFP by toxin genes promoters is located mostly within the long sections and partly within the medium sections of our truncated toxin promoters. For the sRNA promoters this activating ability seems to be found mostly in the medium sections. Interestingly, Furuta et al. (9) found that cya was not directly regulated by AtxA under the same culture conditions as ours, or that regulation may be dependent on weak binding of AtxA. Based on examination of the toxin gene promoters, a cause for a weaker bond between AtxA and the cya promoter could be small changes in SL nucleotide composition. For example, the putative SLII in the pagA promoter, which appears to be critical for AtxA binding, contains a stem composed of seven AT nucleotides. This feature is also found in the putative SLI of the lef promoter, but not in either of the cya promoter SLs. In any case, there are several other differences between the proposed SLs of these promoters, and it is difficult to explain this discrepancy in the observed interaction between AtxA and the cya promoter. Although our data appear to show weaker binding between the cya promoter and AtxA, it is difficult to conclude this from EMSA results alone. Further quantitative work is needed to characterize AtxA’s binding affinity for the toxin and sRNA promoters.

Activation by AtxA for the sRNA promoters was retained in the medium length truncated sequences. The M fragments for xrrA (93 bp) and xrrB (111 bp) are shorter than the optimal L fragments of the toxin genes promoters. The smaller size of these sRNA M regions could facilitate identification of perfect AtxA and RNAP binding sites. The structures of the proposed xrrA and xrrB SLs are not perfect palindromes and differ from the SLs of the toxin gene promoters. This may explain the lower activating ability of these promoters (<10 times than toxin gene promoters) under the same conditions. While these sRNA promoters seem relatively weak, the sRNA transcripts accumulate to high concentrations (2), possibly stabilized by the palindromes internal to the transcripts, which provide stability even in the absence of Hfq (9).

The short-truncated sequences of all studied promoters are not able to bind to or be activated by AtxA, even if they have RNAP binding sites as suggested for xrrA and xrrB by Corsi et al. (3). Therefore, the region of xrrA overlapping 29 bp in the M block minus the sequence of the S block, and the region of xrrB overlapping 43 bp in the M block minus the sequence of the S block appear necessary for AtxA binding. Despite showing no activating ability, the S area of each studied promoter includes a canonical −10 σ^A binding site (TATAAT): TATAAT for the pagA gene, TTTTAT for lef, TAAAAT for cya, TATATT for xrrA, and TATTAT for xrrB. However, only the pagA promoter contains the TTGAAA sequence close to the canonical −35 σ^A binding site (20 bp upstream of canonical −10 sequence) (4). Both the lef and cya promoters contain a TTTCTA sequence close upstream of the probable −10 regions (nine bp for lef and eight bp for cya). It is not clear if an alternative σ-factor binds here, as we could not find any homology to known alternative σ-factors. Nevertheless, it is reasonable to assume that these promoters could contain a binding site for a σ-factor other than σ^A (13, 14). Similarly, the sRNA promoters contain the same sequence ATAAAAGTA upstream of the probable −10 regions (32 bp for xrrA and 33 bp for xrrB regions). However, this sequence already overlaps the M area for both promoters (Fig. S1 to S4). In view of these data, we suggest that the hypothetical RNA-polymerase binding sites for the lef and cya promoters are in the S region, and for xrrA and xrrB in the M region.

It is still not clear why the 154-bp long xrrC promoter could not activate production of GFP in our BH500 transcription/translation system, even though Furuta et al. (9) described XrrC as a member of the direct regulon of AtxA. There could be several explanations for this phenomenon: either pXO1 contains some additional factor for this activation or the selected xrrC promoter is too short to produce this activation.

It is interesting to compare the activities of the truncated lef and cya promoters with data published previously (8). Our analyses showed decreases in activity for the cya promoter similar to the previous data. However, we could not identify the previously reported sharp break in activity near the −70 region of the lef promoter. Instead, we saw a constant decrease in activating ability of the lef promoter dependent on the promoter length from the L to S regions.

Finally, in reexamining our previous experiments with the pagA promoter (4) we noted that AtxA binding depends on a sequence immediately upstream of a 25-bp predicted UP-element. This would suggest AtxA acts as a class I activator (11). Based on its position relative to the transcription start site and sigma factor recognition site, this predicted-UP element may interact with the C-terminal domain of the RNAP α subunit (αCTD) (10, 15, 16). The αCTD of the σ^A-directed RNA polymerase (which we believe controls the transcription of pagA in B. anthracis) binds to the UP element of the promoter region (11). Deletion of this predicted element led to complete loss of activating ability for the promoter (Fig. 5B). This loss of activating ability could indicate a possible topological problem in the interaction between the RNAP, σ^A, and AtxA. Theoretically, the deletion of the proposed-UP element could cause overcrowding, where AtxA could still bind, but αCTD could not. It is possible that without an AtxA-αCTD interaction, transcription could not proceed. Support for this suggestion is found in a recently published paper showing that the global transcriptional regulator Spx of Gram-positive bacteria engages RNAP by simultaneously interacting with the C-terminal domain of the RNAP alpha subunit (αCTD) and σ^A, thereby stabilizing the Spx–transcription activation complex (17). This result offers an example of the type of role that AtxA may play in activating B. anthracis promoters.

MATERIALS AND METHODS

Construction of pRPL and pRPE plasmids with lef and cya promoters controlling expression of gfp gene.

The primer pair lefprF and lefprR was used to amplify the lef gene promoter region. Primer pair efprF and efprR (Table S1) was used for the cya gene promoter. The amplified promoters were cloned into the pRSP plasmid by replacing the pagA promoter (Table 1) according to the previously described protocol (4). The finished plasmids contained an ampicillin resistance gene and a spectinomycin resistance gene.

Construction of plasmids for identification of xrrA and xrrB promoters.

To identify potential sRNA promoters, both DNA strands of the sRNA loci were used for amplification. Primer pairs xrrAF/xrrAR and xrrBF/xrrBR were used for amplification of the direct strands while xrrARCF/xrrARCR and xrrBRCF/xrrBRCR were used for amplification of the complement (RC) strands (Table S1). The reverse primers contained a TCTCCT sequence corresponding to the coding strand 5′-AGGAGA-3′ sequence that represents a prokaryotic RBS, also called Shine-Dalgarno (SD) sequence (18). The amplified DNA fragments were cloned into the pRSP plasmid by replacing the pagA promoter upstream of the gfp gene (Table 1). The finished plasmids contained an ampicillin resistance gene and a spectinomycin resistance gene.

Construction of plasmids with shortened lef, cya, and sRNAs promoters.

For each of the promoters for lef, cya, and the sRNAs, DNA fragments encoding three variants of different truncated lengths were synthesized as gBlocks (Fig. S1 to S4) by IDT. These shortened promoter variants are identified in the text as L, M, and S, with S being the shortest variant compared to the natural promoter. All gBlocks contained terminal EcoNI and NcoI endonuclease restriction sites which were used for their insertion into corresponding plasmids in place of full-length versions of the promoters (Table 1). All plasmids, including pRSPΔUP, pRPL, and pRPE, were transformed into Escherichia coli One Shot TOP10 competent cells (Invitrogen Corporation) and then into E. coli SCS110 competent cells (Agilent Technologies, Santa Clara, CA), following the manufacturers’ protocols for each. When appropriate, media were supplemented with spectinomycin to 150 μg/mL and ampicillin to 100 μg/mL. Plasmids were isolated using the QIAprep Spin miniprep kit (Qiagen). Restriction enzymes for plasmid constructions were obtained from New England BioLabs, Ipswich, MA. Plasmid pRPE SΔ containing a shortened form of the cya S DNA was a result of a spontaneous deletion.

Construction of pRSPΔUP plasmid with pagA promoter deleted for predicted-UP element.

The Q5 site-directed mutagenesis kit (New England Biolabs) was used following the manufacturer’s protocol to delete the predicted-UP element of the promoter region of the pagA gene in the pRSP plasmid. The UPdel_F and UPdel_R primers (Table S1) were designed using the NEBaseChanger v1.2.7 program (https://nebasechanger.neb.com/) and were synthesized by IDT.

Structural analysis of selected promoter regions.

The promoter DNA sequences were examined for structural motifs using the mfold Web server (http://rna.urmc.rochester.edu/RNAstructureWeb/Servers/Predict1/Predict1.html) (19). Selected stem-loop structures are shown in supplemental materials (Fig. S5).

Multiple sequence alignment of selected promoter regions.

The Clustal Omega program (https://www.ebi.ac.uk/Tools/msa/clustalo/) was used for comparing structural motifs in identified promoter regions.

Fluorometric assays of promoter activity for studied genes.

The plasmid pRSP was used as a source of the green fluorescent protein (DasherC4 GFP) gene (4). To analyze the activity of pRSP containing different variants of the pagA, lef, cya, and sRNA promoters, recombinant plasmids were transformed into the B. anthracis BH500 strain (20) containing the pKA2 plasmid encoding AtxA. BH500 was transformed as previously described (21) and plated onto selective LB agar containing 15 μg/mL kanamycin to maintain pKA2 and 150 μg/mL spectinomycin to maintain the pRSP recombinant plasmids (containing the lef, cya, xrrA, xrrB and xrrC promoters). All kanamycin-resistant (Km^r) and spectinomycin-resistant (Sp^r) isolated strains were verified for the presence of intact plasmids by sequencing.

To assess the role of AtxA in promoter-dependent expression of GFP, overnight cultures of plasmid-containing B. anthracis were diluted to an A₆₀₀ of 0.1 in fresh NBY medium supplemented with 0.8% sodium bicarbonate and corresponding antibiotics and grown at 37°C in air with 15% CO₂ (4). The NBY medium used here contains 8 g of Difco Nutrient Broth (3 g of Beef Extract, 5 g of Peptone) and 3 g of Bacto Yeast Extract per L. Fluorometric measurements of bacteria grown to approximately A₆₀₀ 1.0 were performed in a Victor3 reader (PerkinElmer) using a GFP filter set (excitation at 485 nm and emission at 535 nm), as previously described (22). Fluorescence intensities were compared with the fluorescence of the BH500 strain containing only the pRSP-derivative plasmids. This control strain was grown under the same conditions in medium containing only 150 μg/mL spectinomycin. As a rule, the same experiments were performed for the samples grown in air atmosphere in NBY medium without sodium bicarbonate. Fluorometric measurements were normalized to bacterial density measured as A₆₀₀.

Preparation of DNA for electrophoretic mobility shift assays.

IRDye 700-labeled DNA fragments were prepared using 5′-end-labeled IRDye 700 primers (IDT) listed in Table S1. Previously, for analysis of the pagA promoter, we used both forward and reverse labeled primers (PAGSL_R700 and PAGSL_F700) (Table S1) (4). These primers were used again to amplify fragments from pRSP-derivative plasmids for analysis in binding experiments. Unlabeled excess competitor fragments were prepared using primers with the same sequences. PCRs using OneTaq Hot Start 2× mastermix with standard buffer (New England BioLabs) and pRPE, pRPL pRPxrrAD, pRPxrrBRC, and pRPxrrC as the templates were performed under the following conditions: 30 s at 94°C, 30 cycles of 30 s at 94°C, 30 s at 57°C, and 30 s at 68°C, followed by 5 min at 68°C. Infrared-labeled PCR products and unlabeled PCR products were purified using the DNA Clean & Concentrator kit (Zymo Research, Irvine, CA), following the manufacturer’s protocol.

Electrophoretic mobility shift assays.

Infrared dye end-labeled probe was incubated with AtxA and reagents from the Odyssey EMSA buffer kit (Li-Cor Biosciences, Lincoln, NE) in 20-μL reaction mixtures (4). Final reactions contained various concentrations of IRDye 700-labeled DNA probe (generally 1.67 nM to 2.5 nM, depending on the affinity of the respective promoter for AtxA), 50 μg/mL poly(dI-dC), 500 μg/mL bovine serum albumin (New England BioLabs), 10 mM Tris-HCl, 50 mM KCl, 3.5 mM dithiothreitol (DTT), 10 mM EDTA, 0.25% Tween 20 (pH 7.5), and various concentrations of AtxA proteins, typically 50 to 750 nM. For each competitive EMSA a specific concentration of AtxA was used, dependent on the particular promoter’s binding affinity for AtxA. The reaction mixtures were incubated at room temperature in the dark for 30 min. Then, 1× orange loading dye was added, and the samples were run on 2% agarose gels in 0.5× EABE buffer (90 mM ethanolamine, 90 mM boric acid, 1 mM EDTA [pH 8.0]). The gels and buffer were previously chilled in an ice-water bath and then run in the dark (still in the ice-water bath) at 100 V for 15 min, followed by 80 V for 45 min, and imaged on an Odyssey infrared imaging system (Li-Cor Biosciences).

Preparation of nonspecific competitor DNA.

Genomic DNA was extracted from B. cereus strain 569 (23) using the Wizard Genomic DNA purification kit (Promega Corporation, Wisconsin, WI), following the manufacturer’s protocol. B. cereus was selected as a source of potential competitor DNA due to its close relation to B. anthracis (24). The DNA (10 μg) was sonicated in a Diagenode Bioruptor 300 (Diagenode Inc., Denville, NJ) for 40 cycles with a total sonication time of 20 min. The sonicated DNA was purified using the DNA Clean & Concentrator kit (Zymo Research, Irvine, CA), following the manufacturer’s protocol.