Upstream sequence-dependent suppression and AtxA-dependent activation of protective antigens in Bacillus anthracis

Bacterial strain and plasmids construction

Genomic DNA was extracted from overnight culture of B. anthracis str. CZC5 (Ohnishi et al., 2014) using a PowerFecal DNA isolation kit (MO BIO, Carlsbad, CA, USA) according to the manufacturer’s protocol. The atxA gene was amplified by polymerase chain reaction (PCR) using two primers with a restriction enzyme site, namely KpnI-AtxA-F and NheI-AtxA-R, and was then cloned into the pGEM-T-easy plasmid (Promega, Madison, WI, USA) according to manufacturer’s protocol (Table S1). After digestion by KpnI-HF (NEB, Ipswich, MA, USA) and NheI-HF (NEB, Ipswich, MA, USA), the full length of the atxA gene was inserted into a bait plasmid, pB1H2-ω2 (Addgene plasmid # 18038) (Noyes et al., 2008) and digested by KpnI-HF and XbaI (NEB, Ipswich, MA, USA); this construct expresses AtxA with the omega subunit of bacterial RNA polymerase. For the control transcription factor, pB1H2-ωL-Prd (Addgene plasmid # 18040), a bait plasmid expressing Paired, a well-characterized transcription factor, was used (Jun & Desplan, 1996). To construct a bait vector lacking the omega factor, the pB1H2-ωL vector was digested by EcoRI-HF (NEB, Ipswich, MA, USA) and XbaI, blunted using a DNA blunting kit (Takara, Kusatsu, Japan), and then ligated. For expression of AtxA in trans, where AtxA is expressed without the omega factor of RNA polymerase, the pSTV28 vector (Takara, Kusatsu, Japan) was used. The fragment of the atxA gene with the restriction enzyme sites was amplified by PCR using two primers from the pB1H2-ω2-AtxA vector, namely ERI-pB1H2-TF-F and BHI-AtxA-R. After digestion by EcoRI-HF and BamHI-HF (NEB, Ipswich, MA, USA), the amplicon was inserted into the expression vector pSTV28 and digested by the same enzymes.

To analyze the AtxA–DNA interaction, plasmids with the upstream sequence of pagA, here defined as Ppag, were constructed as prey. The primers used to amplify the upstream regions are shown in Table S1. After digestion by NotI-HF (NEB, Ipswich, MA, USA) and EcoRI-HF, the sequence was inserted into the prey vector pH3U3 (Addgene plasmid # 12609) (Meng, Brodsky & Wolfe, 2005). A total of 1,004 bp DNA fragment of lambda phage genome DNA was used as a control sequence. The DNA fragments with lambda phage DNA (sequence homology with the region from 17,817 to 18,798 bp in J02459.1, corresponding to partial sequence of host specificity protein J) were purified from one Kb Plus DNA Ladder (Invitrogen, Carlsbad, CA, USA) and introduced into the pH3U3 plasmid digested by EcoRI-HF after treatment with a DNA blunting kit. The full DNA sequence used as the control is available in Data S1. There was no significant similarity between Ppag and control DNA on BLAST alignment (Zhang et al., 2000). For construction of prey plasmids with partial Ppag sequences, pH3U3 plasmids with intact Ppag were amplified by the primers listed in Table S1, followed by digestion with EcoRI-HF (NEB, Ipswich, MA, USA), self-ligation with Ligation High ver2 (TOYOBO,Osaka, Japan), and transformation into Escherichia coli 10 β (NEB, Ipswich, MA, USA) with the heat shock method described elsewhere.

Detection of protein–DNA interaction by transcriptional fusion with an auxotrophic marker

The E. coli strain US0 hisB pyrF rpoZ (Addgene # 18049) was transformed with bait pB1H2-ω2-AtxA, pB1H2-ωL-Prd, pB1H2(-), pSTV28-AtxA, or pSTV28 by the calcium chloride method (Meng, Brodsky & Wolfe, 2005; Meng & Wolfe, 2006; Noyes et al., 2008). The E. coli strain harboring bait plasmids was then treated with 10% glycerol for electroporation. Prey plasmids were subsequently introduced by electroporation (1,750 V, 25 μF, 200 Ω, one mm gap). The transformants were incubated for 60 min at 37 °C in one ml of Super Optimal broth with Catabolite repression (SOC) medium. Incubated cells were centrifuged and then resuspended in NM medium (Meng, Brodsky & Wolfe, 2005) supplemented with 200 μM uracil and 0.1% histidine. The cells were washed with NM medium four times and suspended in one ml of NM medium. From this suspension, five μl drops were spotted on NM plates without histidine after 10-fold serial dilution and incubated for 36 h at 37 °C. Transformation efficiency was determined using 2xYT plates, which include histidine, after overnight incubation at 37 °C. Given that the nutrient composition differs between histidine-deficient and histidine-rich conditions, colony size is larger in a histidine-rich condition using 2xYT plates even though the incubation period is shorter. For experiments using the pB1H2 vector, carbenicillin, and kanamycin were used as selection antibiotics for both 2xYT plates and NM plates. For experiments using the pSTV28 vector, chloramphenicol, and kanamycin were used as selection antibiotics for both 2xYT plates and NM plates. All antibiotics except for chloramphenicol were dissolved in water; chloramphenicol was dissolved in ethanol.

Purification of HIS-FLAG-AtxA-HIS recombinant protein

To examine the interaction between AtxA and DNA sequences in vitro, the atxA gene was cloned into the expression vector pET 28-b (Novagen, Madison, WI, USA) with N- and C-terminal 6xHIS-tags and an internal FLAG-tag. The BL21 (DE3) expression strain of E. coli was then transformed with an AtxA expression vector. Cells harboring the AtxA expression vector were cultured at 37 °C in LB medium containing 50 μg/ml kanamycin. After overnight growth, cells were diluted to an OD600 of 0.05 in 200 ml of LB medium containing kanamycin and cultured to an OD600 of 0.5. Next, the cells were induced by adding one mM IPTG and cultured at 25 °C for 16 h. After incubation, the cells were collected by centrifugation at 10,000×g for 10 min and stored at −80 °C. Thawed cells were suspended in eight ml binding buffer (50 mM NaH₂PO₄, 300 mM NaCl, 0.5% Triton X-100, 10 mM imidazole, one mM PMSF, pH 8.0) and disrupted by sonication. After centrifugation for 15 min at 15,000×g, recombinant AtxA protein was purified by Ni-NTA (Qiagen, Hilden, Germany). The extract was then desalted by HiPrep 26/10 Desalting (GE, Boston, MA, USA) and purified by RESOURCE S, one ml (GE, Boston, MA, USA) using 25 mM MES, pH 6.5. The final concentration of recombinant AtxA was determined by a Bio-Rad Protein Assay (Bio-Rad, Hercules, CA, USA).

In vitro gel mobility shift assay

Purified recombinant AtxA protein was serially diluted twofold from 200 μg/ml to 12.5 μg/ml. A total of 200 μg/ml of BSA was used as a control protein. Next, 6.6 μl protein solution was mixed to a final concentration of 20 ng/μl of Ppag or lambda phage genome DNA in 10 μl. For each DNA fragment, both the linear form prepared by PCR using HU100 and OK181v2 and the circular form cloned into pH3U3 were used. After 30 min at 37 °C, the mixture was electrophoresed using 1% agarose gel and 1xTAE buffer (Hellman & Fried, 2007). Ethidium bromide staining was used to detect DNA.

In silico analysis

The guanine and cytosine (GC) contents in the DNA fragments were analyzed using GENETYX-MAC software (Network version 19.0.1; GENETYX Co., Tokyo, Japan). The average span for the analysis was set at 20 bp and the step was set at 1. DNA fragments were compared by dot plot analysis using dotmatcher (Rice, Longden & Bleasby, 2000). The window size was set at 10 and the threshold was set at 25.

Transcription of pagA is suppressed by the upstream region and released by interaction with AtxA

To test the interaction between the upstream sequence of pagA (Ppag) and AtxA, we constructed a transcriptional fusion system with an auxotrophic marker, which was originally used in the Bacterial One-Hybrid (B1H) system (Fig. 1A) (Meng, Brodsky & Wolfe, 2005; Meng & Wolfe, 2006; Noyes et al., 2008). Although the B1H system was originally developed as a method for selection of sequences to which a eukaryotic transcription factor binds, it can also be used for selection of prokaryotic transcription factors that bind to specific sequences (Guo et al., 2009). The system consists of two vectors: (i) an expression plasmid harboring a gene encoding a bait protein and (ii) a reporter plasmid harboring a DNA region that works as prey. The bait protein is a transcription factor fused with a FLAG-tag and the omega subunit of a bacterial RNA polymerase. The prey is a DNA sequence that is a candidate of the binding site of the bait protein. The gene connected downstream of the prey sequence is the auxotrophic marker HIS3 gene, whose product is required for survival in the condition lacking histidine. When the bait protein stably binds to the prey sequence, the fused omega subunit recruits the whole bacterial RNA polymerase, resulting in the induction of HIS3 expression. Although the basal level of HIS3 expression without the bait protein is sufficient for growth in the condition without histidine, the higher level of HIS3 expression induced by the bait protein is required for survival when the culture that lacks histidine is treated with 3-amino-triazole (3-AT), the HIS3 competitive inhibitor (Fig. 1A). By using 3-AT, the system can select the combination of bait and prey that interact above a certain threshold, resulting in higher expression of HIS3 and survival under the chosen concentration of 3-AT. The system can also be utilized for measuring the interaction between a specific combination of bait protein and prey sequence by measuring the survival rate of strains introduced by plasmids encoding the bait and prey under conditions with or without histidine and 3-AT.

We adopted this system to investigate the interaction between the upstream sequence of the pagA gene and AtxA. AtxA fused with a FLAG-tag and the omega subunit of bacterial RNA polymerase was constructed using the expression plasmid as a bait, while Ppag was fused upstream of HIS3 gene on the reporter plasmid as prey. The strain harboring the expression plasmid was transformed with the reporter plasmid with Ppag and the transformation efficiency ratio, which is the transformation efficiency without histidine divided by that with histidine, was calculated to test the expression of HIS3. The expected transformation efficiency ratio was observed when the strain harboring the expression plasmid encoding AtxA was transformed with the reporter plasmid with Ppag. However, a significantly lower transformation ratio was noted when the strain without AtxA bait was transformed with the reporter plasmid with Ppag (Figs. 1B and 1C). These observations indicate that, without AtxA, the expression level of HIS3 transcriptionally fused with Ppag was lower than the basal level and insufficient for survival in the condition without histidine. We hypothesized that Ppag suppressed the expression of the downstream gene when AtxA was absent, and that suppression was released by the interaction between Ppag and AtxA. Suppression without bait was not observed as expected when we used Prd as the bait and the Prd binding motif as the prey, which is a known combination of bait and prey with a strong interaction (Meng & Wolfe, 2006) (Figs. 1B and 1C).

To validate our system, we also tested the growth of the strains under the condition with 3-AT but in the absence of histidine. Although the combination of Prd and the Prd binding motif showed growth under the condition with 3-AT, the strain expressing AtxA and harboring Ppag did not show any growth with 3-AT (Fig. 1B). Therefore, the results suggested that AtxA could recover the growth of the strain harboring Ppag but that AtxA could not induce HIS3 expression to a level that would allow bacteria to grow on the plate with 3-AT. The reporter plasmid with the Prd binding motif showed the same transformation efficiency when transformed to strains with and without Prd as bait. In addition, the introduction of Prd as bait did not release suppression by Ppag, suggesting that AtxA-specific interactions with Ppag are required to release suppression (Figs. 1B and 1C).

To test if suppression was caused by any upstream sequence, two reporter plasmids were constructed as additional controls: one was a reporter plasmid without any additional sequence insertion upstream of the marker, and the other was a plasmid inserted with a partial sequence of the lambda phage. Those plasmids were transformed into strains with either Prd or AtxA as the bait and the strain without bait, but they showed no differences in transformation efficiency (Figs. 1B and 1C). This supports our hypothesis that suppression was specifically caused by Ppag. Suppression by Ppag might be partly explained by the AT-rich profile of Ppag relative to that of the lambda phage (GC contents of 24.5% for Ppag vs. 54% for lambda phage DNA, see also Fig. S1), as early studies suggested that AT-rich regions negatively affect transcription in E. coli (Navarre et al., 2006). This is consistent with an earlier report suggesting a functional relationship between AtxA and AT-rich regions of DNA (Hadjifrangiskou & Koehler, 2008).

Binding of AtxA to upstream regions was also assessed using a vector expressing AtxA without the omega factor to determine whether AtxA-dependent growth recovery required the omega factor fused to AtxA. A strain with an expression plasmid expressing FLAG-AtxA, in which AtxA is fused with a FLAG-tag but without the omega factor, was constructed. The transformation assay for this strain using reporter plasmids with and without Ppag showed that FLAG-AtxA significantly recovered growth inhibition (Fig. 2; Fig. S2). This result indicated that, together with the recovery effect of the omega-AtxA fusion protein and omega-Prd fusion protein (Figs. 1B and 1C), AtxA and the omega factor likely recover transcription in an additive manner.

Hadjifrangiskou & Koehler (2008) predicted that the minimal functional promoter region for AtxA-dependent transcriptional regulation of pagA transcription is a 178 bp sequence of the pagA promoter region (Fig. 3A). We first tested if this predicted minimal functional promoter, here defined as Ppag_916-1093, contributed to suppression by Ppag. Surprisingly, Ppag_916-1093 as well as Ppag_1-915, which lacks the minimal functional promoter region, did not affect transformation efficiency (Fig. 3; Fig. S3). This suggests that coexistence of the minimal functional promoter region and the other region of Ppag are required for the suppression observed with whole length Ppag. To identify the region in the minimal functional promoter responsible for the suppression by Ppag, we then constructed a reporter plasmid with the Ppag sequence with stepwise 30 bp deletions from the 3′ end in the minimal functional promoter region (Fig. 3A). The results indicated that deletion of 30 bp from the 3′ end of Ppag was sufficient to eliminate suppression (Fig. 3B; Fig. S3), suggesting that the 30 bp end is one of the core regions contributing to suppression by the whole Ppag sequence.

In vitro analysis of AtxA–DNA interaction

To determine whether recovery of AtxA depends on the specificity of the AtxA–DNA interaction, we tested the binding specificity of AtxA to Ppag in vitro. Recombinant full-length AtxA was expressed in E. coli as a HIS-tagged protein and purified using Ni-NTA agarose and a Resource S column (Fig. S4). Recombinant AtxA was incubated with DNA fragments with the Ppag sequence or the partial lambda phage sequence as a control, and then the interaction was tested using an electrophoretic mobility shift assay (EMSA). To detect the interaction of AtxA with long DNA, we used agarose gel for EMSA (Hellman & Fried, 2007). For both Ppag and lambda phage DNA, we observed DNA bands stuck in the wells with recombinant AtxA at concentrations of 50–100 μg/ml, which is suggestive of AtxA-DNA binding (Fig. 4). DNA substrates in both the plasmid form and linear form were tested and showed similar results, suggesting that the form of the DNA substrate does not affect the interaction with AtxA. Considering the low sequence similarity between Ppag and lambda phage DNA (see Materials and Methods and Fig. S5), the similar binding profile between Ppag and lambda phage DNA suggests that the binding of recombinant AtxA to DNA fragments is not likely to be affected by the DNA sequence. Despite positive results obtained in in vitro binding assay, we observed no growth improvement in the in vivo B1H experiment when lambda phage DNA was used as prey and AtxA as a bait (Figs. 1B and 1C). In this study, we used recombinant AtxA with N- and C-terminal 6xHIS-tags and an internal FLAG-tag to analyze AtxA–DNA interaction. While both 6xHIS-tags and FLAG-tag are not likely to affect the function of AtxA, further study using a tag-free AtxA protein will be needed for further understanding the AtxA–DNA interaction.