|Home | About | Journals | Submit | Contact Us | Français|
Bordetella pertussis BvgA is a global response regulator that activates virulence genes, including adhesin-encoding fim3 and fhaB. At the fhaB promoter, PfhaB, a BvgA binding site lies immediately upstream of the −35 promoter element recognized by Region 4 of the σ subunit of RNA polymerase. We demonstrate that σ Region 4 is required for BvgA-activation of PfhaB, a hallmark of Class II activation. In contrast, the promoter proximal BvgA binding site at Pfim3 includes the −35 region, which is composed of a tract of cytosines that lacks specific sequence information. We demonstrate that σ Region 4 is not required for BvgA activation at Pfim3. Nonetheless, Region 4 mutations that impair its typical interactions with core and with the −35 DNA affect Pfim3 transcription. Hydroxyl radical cleavage using RNA polymerase with σD581C-FeBABE positions Region 4 near the −35 region of Pfim3; cleavage using RNA polymerase with α276C–FeBABE or α302C–FeBABE also positions an αCTD within the −35 region, on a different helical face from the promoter proximal BvgA~P dimer. Our results suggest that the − 35 region of Pfim3 accommodates a BvgA~P dimer, an αCTD, and σ Region 4. Molecular modeling suggests how BvgA, σ Region 4, and α might coexist within this DNA in a conformation that suggests a novel mechanism of activation.
Pertussis is a reemerging disease in developed nations and a continued presence in the developing world. Although vaccines are available, they do not provide lifetime coverage; outbreaks of pertussis have occurred in populations that were vaccinated a few years earlier 1; 2; 3. Bordetella pertussis, the causative agent of pertussis, employs a simple yet elegant system to regulate its virulence genes. In the environment of the human host, the BvgS-BvgA two-component system is active, resulting in the phosphorylation of BvgA (BvgA~P) (reviewed in 4). BvgA~P is a global activator, which binds to promoters of virulence genes with varied stoichiometry and affinity. Thus, upon entry into an inducing environment, the increasing intracellular concentration of BvgA~P orchestrates a temporal program of virulence gene expression 5. Genes that are activated early include those encoding adhesive proteins, such as fhaB (filamentous hemagglutinin) and fim2 and fim3 (fimbrial subunits). Late-activated genes include those encoding toxic virulence factors such as adenylate cyclase and pertussis toxin 6. The temporal program of gene expression is thought to be crucial for the virulence of the pathogen, as demonstrated in a murine model of B. pertussis respiratory infection 7. Thus, understanding the molecular mechanism of B. pertussis virulence gene expression and its regulation by BvgS-BvgA will help to develop better strategies to fight the disease.
As in all bacteria, a multisubunit RNA Polymerase (RNAP), consisting of a core of subunits (α1 α2, β, β’, ω) and a specificity factor, σ, transcribes the B. pertussis genome. Promoter recognition involves multiple interactions between regions of σ and the promoter DNA. For primary σ factors, such as σBp in B. pertussis or σ70 in Escherichia coli, these contacts can include Region 2 with the −10 element, Region 3 with the sequence −15TG−14 (extended −10 element), and Region 4 with the −35 element (reviewed in 11). As good matches to two out of three of these elements are typically sufficient for promoter activity, many promoters can be classified as belonging to the −35/−10, TGn/−10 (extended −10), or −35/TGn class. In addition, the C-terminal domains (CTDs) of the α subunits of RNAP contribute to promoter binding through their interaction with DNA sequences (UP elements) upstream of position −40 (reviewed in 12).
Regulated promoters require additional factors besides RNAP, usually to compensate for weak σ/DNA element interactions. In Class I activation, the activator may bind to several locations upstream of the −35 element and interact directly with the αCTDs; in Class II activation, the activator binds to a site adjacent to or overlapping the −35 element and contacts an αCTD and/or σ Region 4 (reviewed in 13; 14). Class I and Class II activation may also function in combination with each other, but in each case, σ Region 4 is thought to be required in order to maintain contacts with the −35 region of the DNA.
σ appropriation is a different type of activation, used by bacteriophage T4, in which the binding site for the T4 activator MotA includes the −35 region of the DNA (reviewed in 15). Thus, MotA and σ Region 4 are essentially competing for the DNA. In this case, the T4-encoded co-activator AsiA binds to Region 4, remodeling its structure and preventing its interaction with the −35 portion of the promoter. However, σ Region 4 is still required because of its needed interactions with AsiA and MotA.
Most of the virulence genes in B. pertussis have an organization that resembles that seen at ClassI/II activated promoters 16; 17. For example, the fhaB gene is activated by three dimers of BvgA~P, which bind from −95 to −38 relative to the transcription start site 18 (Fig. 1a). In contrast, at the promoters for fim2 (Pfim2) and fim3 (Pfim3; Fig. 1a), the promoter proximal BvgA~P site fully includes the −35 region of the DNA. In addition, Pfim2 and Pfim3 both contain a tract of poly cytosines (C), which is located upstream of the −10 element and includes the −35 region 20. The length of this C-tract regulates BvgA~P activation 24. We have recently shown that the C-tract is needed to position BvgA~P correctly relative to the bound RNAP; however, there is little, if any, specific sequence information imparted by the tract 20. The second unusual feature of Pfim3 is that it contains both the −15TG−14 motif, which is recognized by σ Region 3, and a good match (−12TATTCT−7) to the −10 element, which is recognized by σ Region 2. Thus, Pfim3 appears to be a TGn/−10 (extended −10) promoter, and we would predict that it would exhibit some activity with RNAP alone 25; 26; 27. However, Pfim3 displays very little activity in the absence of BvgA~P 20.
In this study, we have investigated BvgA~P activation at PfhaB and Pfim3. We show that σ Region 4 is required for BvgA~P activation at PfhaB, providing further evidence that the promoter proximal BvgA~P functions as a Class II activator at this promoter. In contrast, BvgA~P activation at Pfim3 is different. Although we find that it does not require σ Region 4, we also find that Region 4 is still located close to the Pfim3 −35 DNA. Furthermore, using RNAP reconstituted with FeBABE-conjugated α subunits, we find that each α contacts the same region of Pfim3 as BvgA, but on a different helical face. Taken together, we speculate that within the activated complex at Pfim3, σ Region 4, BvgA~P, and an αCTD are all located within the −35 region C-tract of the DNA.
The C-terminal portion of primary σ factors, such as E. coli σ70, is composed of five alpha helices (H1–H5) with a turn (T) between H3 and H4 (Fig. 1b) 28; 29; 30. Residues within σ70 Region 4 (H1–4) and H5 perform several important functions, including interaction with the β-flap of core polymerase, recognition of −35 element promoter DNA, and interactions with regulators that are needed to activate promoters by Class II activation and by σ appropriation (Fig. 1b) (reviewed in 15; 31; 32). The interaction with the β-flap positions Region 4 so that it is located close to the −35 DNA, which then allows residues within the second helix of the helix-turn-helix (H3-T-H4) to make specific contact with base determinants within the −35 DNA (Fig. 1b) 28; 33; 34; 35; 36; 37.
To investigate the importance of various portions and functions of Region 4 in BvgA~P activation at PfhaB and Pfim3, we employed several σ mutants, detailed below. The behavior of each mutant σ was evaluated using various promoters (Fig. 1a): the −35/−10 promoters PRNAI 38 and unactivated PuvsX 39, which require an interaction between Region 4 and the −35 DNA; the extended −10 promoter PRE# 25, which is much less dependent on an interaction with Region 4; and activated PuvsX 39, which uses contacts between Region 4 and the T4 activator MotA and the T4 co-activator AsiA (Fig. 1b). We used transcription conditions that recapitulate BvgA~P regulation of PfhaB and Pfim3; specifically, for fim3 transcription a C-tract of 13 C’s is inactive while a tract of 14 or 15 C’s is active (Fig. 2a, lanes 13; Fig S1 20). We also limited transcription to a single round in order to observe how protein or DNA mutations affect the formation of transcriptionally competent complexes.
We expected that two of the σ mutants, σ4.1+4.2 ala and σ580-3ala, might impair recognition of the −35 region of the promoter DNA. σ4.1+4.2 ala contains alanine patches at residues 551–555 and 584–588, substitutions that replace residues that directly contact the −35 element or that interact with the β-flap (Fig. 1b). In addition, the Region 4.2 alanine mutations alone have previously been shown to eliminate σ appropriation at PuvsX because they disrupt important interactions between the T4 co-activator AsiA and Region 4 (40 (Fig. 1b)). As expected, transcription from the −35/−10 promoter PRNAI (Fig. 2a, lanes 7–9) and unactivated and activated transcription from T4 PuvsX (Fig. S2, lanes 1–4) is essentially abolished when using RNAP reconstituted with σ4.1+4.2 ala. In addition, transcription from the extended −10 promoter PRE# is also reduced (Fig. 2a, lanes 7–9), consistent with other work showing that non-specific contacts between the −35 region of an extended −10 promoter and Region 4 are beneficial 25; 41; 42. The σ580-3ala mutant contains an alanine patch within the turn of the DNA binding motif, H3-T-H4. As seen in Fig. 1b, these mutations might impart some impairment in the interaction of the binding motif with the −35 DNA or with the interaction with AsiA. Transcription from the −35/−10 promoters PRNAI (Fig. 2a, lanes 10–12) and unactivated PuvsX (Fig. S2, lane 9 vs. 13) is indeed impaired. However, MotA/AsiA activation of PuvsX is similar to that seen with RNAP reconstituted with σ70 (Fig. S2, lane 12 vs. 15).
The σHyb4 mutant protein is a chimeric σ factor in which Region 4.2 and H5 of σ70 (residues 565 to 613) have been replaced with the corresponding residues (280–330) of σ38, the stationary phase σ factor of E. coli (Fig. 1b). Thus, the protein retains residues that contact the −35 element despite many amino acid changes throughout Region 4; in addition, it has been shown that σ38 Region 4 interacts well with the β-flap 37. However, some of the substitutions present in σ38 include Region 4 residues involved in the MotA/σ70 interaction (Fig. 1b), so they should render RNAP reconstituted with σHyb4 significantly impaired at MotA/AsiA-mediated σ appropriation. Experiments with σHyb4 thus sought to address whether BvgA activation resembles σ appropriation. As seen in Fig. 2a, lanes 4–6, and Fig. S2, lane 5, this mutant σ recognizes the PRNAI, PRE#, and unactivated PuvsX promoters, although the overall activity of the σHyb4 is lower than wt. In contrast, it is significantly impaired for MotA/AsiA activation (Fig. S2, lane 8).
Finally, we tested mutants of E. coli σ70 and the B. pertussis primary σ (σBp) that lack all of Region 4, residues 530–613 in σ70 and residues 675–760 in σBp, respectively. Region 4 of E. coli and Region 4 of B. pertussis are 73% identical (Fig. 1b). Transcription from known promoters using RNA polymerase reconstituted with either E. coli σ70 or σBp has been shown to be similar 40; 43 and σBp, like σ70, is competent for MotA/AsiA mediated σ appropriation 40. Thus, it is thought that Region 4 of each σ behaves similarly. As expected, RNA polymerase with σ70ΔR4 had practically no activity at PRNAI, and like σ4.1+4.2 ala, its activity with PRE# was significantly reduced although not as much as at PRNAI (Fig. 2b, lanes 4–6 and 16–18). As expected, neither σ70ΔR4 nor σBpΔR4 were active at PuvsX with or without MotA/AsiA (data not shown). Surprisingly, σBpΔR4 retained some activity at PRNAI (Fig. 2b, lanes 10–12 and 22–24), suggesting that there are differences in how the E. coli and the B. pertussis primary σ factors use this promoter. However, we have not examined this difference at PRNAI further.
The positions of the BvgA~P binding sites within PfhaB are suggestive of a combination of Class I and Class II activation 16; 17. In such an arrangement, σ Region 4 is typically required for activation, providing a direct contact with the promoter proximal activator (reviewed in 14). However, previous work has shown that at BvgA~P activated PfhaB, the αCTDs are located on the same sequence but an alternate face from a BvgA~P dimer 18. This position differs from other Class I and Class II activators, in which the αCTDs are located adjacent to the activator 44; 45; 46. Consequently, it is important to ask directly whether σ70 Region 4 is needed for BvgA~P activation at PfhaB.
We find that Region 4 is indeed needed for BvgA~P activation at PfhaB. RNAP reconstituted with σ4.1+4.2 ala, σ70ΔR4, or σBpΔR4 is not functional at this promoter (Fig. 2a, lanes 7–9 and 2B, lanes 4–6 and 10–12). In contrast, transcription with RNAP containing σHyb4 or σ580-3 ala, which retain contacts with the −35 DNA and the β-flap, is much less affected (Fig. 2a, lanes 4–6 and 10–12). Taken together, these results are consistent with the idea that BvgA~P activation at PfhaB relies on Region 4 contacts with the −35 region of the DNA, a hallmark of Class II activation.
Unlike at PfhaB, the promoter proximal BvgA~P binding site at Pfim3 completely overlaps the −35 portion of the promoter (Fig. 1a). This positioning suggests several possibilities for the importance of Region 4 at Pfim3. First, this BvgA~P site at Pfim3 is in a similar location to the PhoB site at the promoter PpstS 23. PhoB is known to contact a portion of Region 4 that is not typically used by Class II activators (47; 48 (Fig. 1b)), and as expected, deletion of σ70 Region 4 eliminates PhoB activation at PpstS 41. However, another activator whose binding site overlaps the −35 region is bacteriophage T4 MotA, which uses σ appropriation. Thus, if similar to MotA, BvgA~P activation at Pfim3 might require a repositioning of Region 4 to allow the downstream BvgA~P dimer to interact with its binding site. Finally, BvgA~P might need an unrecognized reconfiguration or accommodation of Region 4 in order to activate at Pfim3. Besides these considerations, the fim3 promoter is also unusual in that its −35 region contains a C-tract, whose length determines whether the promoter is inactive (<13 or >15 C’s) or active (14 or 15 C’s) (20 (Fig. S1)). We have previously shown that the C-tract appears to function simply as a spatial unit that does not impart significant sequence specificity 20. In fact, although both BvgA and σ Region 4 have a “C” base within their respective consensus binding sites, mutation of each C throughout the C-tract is inconsequential 20. This finding would be consistent with either a model in which Region 4 is removed from its normal position on the DNA by the presence of BvgA~P or a model in which Region 4 stays in its typical position through protein-protein contacts, despite the absence of specific contacts with the DNA.
As a first step in determining which of these models are possible at Pfim3, we assayed the effect of the σ mutants on BvgA~P activation at this promoter. In contrast to the results with PfhaB, our transcription analyses with Pfim3 indicated that none of the tested σ mutants eliminated BvgA~P activation (Fig. 2a and b). In particular, RNAP reconstituted with σ4.1+4.2 ala, which was essentially nonfunctional with every other tested promoter, yielded 60% of the BvgA~P activated transcription seen with σ70, (Fig. 2a, lanes 7–9) and RNAP containing either σ70Δ4 or σBpΔ4 still retained some activity for BvgA~P activation at Pfim3 (Fig. 2b, lanes 16–18 and 22–24).
To analyze the role of σ Region 4 by a separate and independent method, we asked how the dramatic remodeling of Region 4 imposed by AsiA (29; reviewed in 15; 49) affects BvgA~P activation at Pfim3. The binding of AsiA to Region 4 changes the structure of Region 4, eliminating its ability to interact with either the −35 DNA or the β-flap of core polymerase. In these transcriptions, we added NTPs and heparin to the transcription complexes after only 1 min. Thus, the formation of stable complexes was performed under stringent conditions. Consistent with our previous results, we found that AsiA-modified RNAP transcribes from Pfim3 in a BvgA~P-dependent manner but is inactive at PfhaB (Fig. 3, lanes 4–6). Although it is formally possible that BvgA~P displaces AsiA from σ70, this seems highly unlikely. Other work has shown that the interaction between AsiA and σ Region 4 is extremely tight; urea is used to disrupt the association 50.
Taken together, our results demonstrate that BvgA~P activation at Pfim3 does not require σ Region 4 and thus, it differs from Class II activation, PhoB activation, or MotA/AsiA mediated σ appropriation. Two simple possibilities then emerge: 1) BvgA~P somehow replaces σ Region 4, making its presence superfluous or 2) Region 4 retains its typical position within RNAP, which would position it close to the −35 region of the DNA, but this position neither interferes with BvgA~P nor is required for BvgA~P activation.
The −15TG−14 sequence (extended −10 motif) provides an important contact for RNAP through σ70 Region 3 and has been identified in nearly one-quarter of E. coli promoters (reviewed in 11). The presence of this motif together with a good −10 element has been shown to be sufficient for transcription with or without Region 4 25. However, even though Pfim3 contains the −15TG−14 sequence, Pfim3 is dependent on BvgA~P for a high level of activity in vitro and in vivo (Fig. 2a; 20). To investigate in vitro whether the Region 3/−15TG−14 contact was required for this activation, we tested Pfim3 mutant promoters in which the −15TG−14 was changed to either TA (Pfim3−TA) or AC (Pfim3−AC). In the presence of BvgA~P, both mutant promoters yielded a modest (~2 to 3-fold) decrease in the level of transcription, indicating a preference for the −15TG−14 sequence over TA or AC (Fig 4a, lanes 3–6; Fig. 4b, lanes 4–9).
To test the importance of the −15TG−14 sequence in vivo, we took advantage of transcriptional gene fusions with the luxCDABE operon of Photorhabdus luminescens, encoding a bacterial luciferase enzyme and other enzymes required for luciferase substrate synthesis. B. pertussis strains were constructed in which the Pfim3 promoter, harboring specific substitutions within −15TG−14, was fused with the lux operon. The resulting strains were tested for light output (Fig. 5a left panel, with quantitation in Fig. 5b). As previously shown 20; 24, Pfim3−13C, containing a shortened (C)-tract of 13 Cs, had zero promoter activity, while the presence of the longer (C)-tract in Pfim3−15C rendered the promoter active. As in vitro, substitution of the −15TG−14 sequence of Pfim3−15C with TA yielded only a modest (~2-fold) decrease in promoter activity. However, other substitutions were significantly more deleterious, although a low level of activity (~10%) was still observed with the CG and AC mutations. The discrepancy between the in vitro and in vivo effects of the AC mutation may be attributed to the conditions of the in vitro system, in which there is excess RNAP holoenzyme and no competition with endogenous promoters. As expected, all of the promoters were inactive in the presence of the virulence modulator MgSO4 (Fig. 5a right panel; Fig. 5b). This compound is known to down regulate the BvgS-BvgA two-component system and, in doing so, inhibit virulence gene activation by BvgA~P 51; 52. This confirms that the active Pfim3 derivatives are still regulated by BvgA~P. Taken with our in vitro data, we conclude that the extended −10 motif of Pfim3 contributes to BvgA~P activation. Thus, even though Pfim3 does not behave like a “classic” extended −10 promoter, which is active in the presence of RNAP alone, it does appear to use the extended −10 element for optimal activity with BvgA~P.
We next sought to understand how BvgA~P activation functions when both Region 3 contacts with the DNA and Region 4 contacts with the −35 DNA and core polymerase are missing or weakened. To this end, we used Pfim3, Pfim3−TA, and Pfim3−AC and RNAP reconstituted with σ70, σ4.1+4.2 ala, or σΔR4. With the RNAPs reconstituted with the mutant σ proteins, we observed a strong dependence on the extended −10 motif within Pfim3 (Fig. 4a, lanes 7–12; Fig. 4b, lanes 10–18).
Because the 4.1+4.2 ala mutations impair the ability of Region 4 to correctly interact with the −35 DNA and with the β-flap, our results above suggested that during BvgA~P activated transcription from Pfim3−TA and Pfim3−AC, σ Region 4 retains its typical position relative to the −35 DNA and core polymerase, and that an interaction of Region 4 with the −35 DNA might compensate for the loss of the Region 3/−15TG−14 contact. If this were also true at the wt Pfim3 promoter, then our results would imply that the position of Region 4 within the Pfim3/BvgA~P/RNAP complex might resemble its position at a typical −35/ −10 promoter. To investigate this possibility, we examined the abortive initiation products from Pfim3 using RNAP reconstituted with either σ70 or σ4.1+4.2 ala. The short RNA abortive transcripts are synthesized and released from RNAP before the RNAP/promoter complex transitions from an initiating to elongating complex (reviewed in 53). Previous work has indicated that mutations within σ70 Region 4 that reduce its interaction with the −35 DNA element and the β-flap facilitate promoter clearance, thus decreasing the level of abortive products relative to full length RNA 42; 53; 54; 55. When using Pfim3 DNA, we observed that the level of abortive products made by RNAP reconstituted with wt σ70 is much greater than the level observed using RNAP reconstituted with σ4.1+4.2 ala (Fig. 6). These results are inconsistent with a model in which the presence of promoter proximal BvgA~P simply releases Region 4 from its typical place within RNAP. Instead these results suggest that σ Region 4 either retains its typical contacts with core polymerase and thus, is positioned to potentially interact with the −35 C-tract DNA or that the presence of BvgA~P has moved Region 4 to a new location where it can now make other contacts with polymerase and/or a portion of the DNA.
To ask directly if σ Region 4 is located near its typical position, we used RNA polymerase reconstituted with a mutant σ70 (σ C132S/C291S/ C295S/ D581C), which has a single cysteine residue at position 581. Conjugation with the reactive moiety FeBABE generated σD581C-FeBABE. Since residue 581 of σ70 lies at the C-terminal end of H3 of Region 4 (Fig. 1b), it should be located near the −35 region of promoter DNA. Multiple studies 56; 57; 58; 59; 60; 61 have demonstrated that upon generation of the cleavage reaction, a bacterial σ conjugated at this location cleaves promoter DNA in the vicinity of the −35 region. Specifically, for the template strand of galP1, major cleavage sites are observed between −38 to −41, while weaker sites are seen at −28/−2959; our control reactions with galP1 produced similar results (Fig S3). When using Pfim3−15C end-labeled on the template strand, we found that RNAP containing σD581C-FeBABE resulted in cleavage around positions−41 and −30 (Fig. 7a). End-labeled Pfim3−14C DNA also was cleaved around position −41 but the cleavage bands were less intense (Fig. S4). These cleavage sites are consistent with Region 4 being near the −35 DNA. Taken with the previous results that indicate that Region 4 is indeed needed for BvgA~P activation of Pfim3 that lacks the −15TG−14 element, we speculate that Region 4 retains its typical position in the presence of BvgA~P. However, we cannot eliminate the formal possibility that the conjugated RNAP is cleaving a subset of the complexes that differ from the transcriptionally competent species.
At PfhaB, the αCTDs and BvgA proteins have been mapped to the same region, but different helical faces, of the DNA 18. Given the differences in the requirement for σ70 Region 4 at PfhaB and Pfim3, we wondered if the αCTDs are also located in a similar position relative to BvgA~P at Pfim3−14C. (A C-tract of either 14 or 15 is an active configuration for this promoter (Fig. S1, 20; 24).) To investigate this question, we used an RNAP that had been reconstituted with α subunits in which an FeBABE moiety had been conjugated to residue 276 or 302, the same conjugates previously used to map the position of the αCTDs at PfhaB 18. Hydroxyl radical cleavage of Pfim3−14C DNA, after initiation of the Fenton reaction, generated two major sets of cuts for each conjugate. α302-FeBABE yielded cleavage at −61 and −39 on the template strand and −56 and −33 on the non-template strand; α276-FeBABE yielded cleavage at −53 and −31 on the template strand and −49 and −25 on the non-template strand (Fig. 7b). These cleavage sites are consistent with each αCTD contacting the DNA in the minor groove on a DNA face different from that occupied by the corresponding BvgA~P dimer, i. e. with the same relative orientation as that at PfhaB 18. However, when compared to PfhaB, the cleavage sites for the promoter proximal αCTD at Pfim3 are shifted slightly downstream (approximately 4–5 bp) relative to BvgA~P.
Within RNAP, both the α and σ subunits play crucial roles in promoter binding and regulation of transcription. Residues within the αCTD can interact with promoter DNA upstream of −40 and with various transcriptional activators (reviewed in 12; 45). σ Regions 2, 3, and 4 interact with promoter elements within the −10, the −15TG−14, and the −35 element, respectively (reviewed in 11). In addition, σ Region 4 residues can also interact with various binding partners, including the β-flap of core polymerase 37; 62; 63, and activators needed for Class II activation and σ appropriation (discussed in 14; 31; 64; Fig. 1b).
BvgA~P, the master regulator of B. pertussis virulence genes, binds to multiple sites upstream of each of its targeted promoters. At most promoters that have been analyzed, weaker BvgA~P binding sites are located downstream of a strong binding site and rely on cooperativity of BvgA~P binding to activate transcription (65; 66, reviewed in 67). Because the affinity of each site for BvgA~P varies, there is a hierarchy of promoter activity as the concentration of BvgA~P increases, which in turn leads to a temporal pattern of regulation. Adhesin and adhesin-related genes such as fhaB, fim2, and fim3 are expressed early as virulence is being established. It is crucial to understand the mechanism of activation of these adhesin genes, both because the gene products are involved in early steps of pathogenesis and because they are components of the acellular vaccine. However, in order to understand mechanism, one must first determine how the various players (BvgA~P and subunits of RNAP) are positioned within the transcriptionally competent complexes of PfhaB and Pfim3. For PfhaB, a strong BvgA~P site is located at positions −95 to −82, while two weaker sites lie at regular helical turns downstream (−73 to −60 and −51 to −38; Fig. 1a). Previous work 18 has shown that activation of PfhaB is dependent on the αCTDs. We show here that this activation is also dependent on σ Region 4. This dependence and the PfhaB organization imply that BvgA~P uses a combination Class I/II mechanism of activation at this promoter, as has been seen for GcvA activation at PgcvT 68, UhpA activation at PuhpT 69; 70, and synthetic promoters activated by tandem CRP molecules 44; 71. In such a mechanism, contacts between the promoter proximal activator and/or αCTD and σ70 Region 4 facilitate the interaction of RNAP with non-ideal promoter elements (Fig. 8a and b), aiding both in RNAP recruitment and in the formation of the stable, transcriptionally competent open complex (reviewed in 13). However, unlike these classic systems, the positions of the αCTD binding sites within PfhaB 18 are located on a different helical face from the BvgA~P binding site rather than immediately upstream or downstream as is seen for CRP 72 (Fig. 8a and b). Thus, BvgA~P at PfhaB represents a different flavor of this type of activation.
In contrast to PfhaB, the promoters for the fimbrial genes fim2 and fim3 display architectures in which the promoter proximal BvgA~P binding site fully overlaps the −35 region of the promoter (Fig. 1a). While such an arrangement is dissimilar from many well-characterized Class II activators, including CRP, it is similar to the architecture of λcII at λPRE 21; 57; 73; 74, E. coli PhoB at PpstS 2,3; 41; 47; 75, Enterococcus faecium VanRB, a regulator of vancomycin resistance, at PRB 22; and bacteriophage T4 MotA at T4 middle promoters (reviewed in 15). Based on the cII/DNA structure 21, it is likely that in these cases except MotA, the activator dimer uses each H-T-H motif to interact with the major groove of the DNA on either side of the σ Region 4 interaction with the major groove of the −35 DNA (Fig. 8c). Additional contacts at PpstS between PhoB and σ Region 4, such as D570 and E575 47; 48 (Fig. 1b), or between the promoter proximal αCTD and σ Region 4 at cII-activated λ PRE 21 are thought to help to “lock” Region 4 onto the DNA. Indeed, the presence of Region 4 is required for PhoB activity 41 and the interaction of Region 4 with the −35 element of PRE is not changed by the presence of cII 57. Furthermore, like Pfim3, λPRE and E. faecium PRB both contain the −15TG−14 motif that is recognized by σ Region 3. These similarities suggest that the architecture of BvgA~P at Pfim3 might share features with these activators.
Despite its similarities with the cII, VanRB, and PhoB systems, BvgA~P at Pfim3 has significant differences. First, the −35 region of Pfim3 (and Pfim2) includes the C-tract, whose length varies stochastically during replication. The length of the C-tract determines promoter activity 24, by positioning BvgA~P relative to the promoter 20. The C-tract appears to function simply as a spatial unit that does not impart significant sequence specificity. In fact, although both BvgA and σ Region 4 have a “C” base within their respective consensus binding sites, mutations of each C throughout the C-tract are inconsequential 20. Consequently, it appears that neither BvgA~P nor σ Region 4 have any specific recognition determinants within this portion of the DNA. Thus, it is perhaps not particularly surprising that BvgA~P activation at Pfim3 does not require σ Region 4. However, it is surprising that the presence of Region 4 contributes to maximal Pfim3 activity, that disrupting the typical contacts between Region 4 and core polymerase and the −35 DNA significantly decreases the synthesis of Pfim3 abortive transcripts, and that FeBABE cleavage analysis suggests that Region 4 is located close to the −35 DNA. We suggest then that Region 4 retains its typical position within the BvgA~P activated Pfim3 complex and thus, is positioned to potentially interact with the −35 C-tract DNA. Furthermore, the promoter proximal αCTD at Pfim3 is also positioned to interact with a portion of the C-tract, even though its typical binding site involves A/T rich sequences 12. A major question then is what is the “glue” that holds these various protein/nonspecific DNA interactions together?
Our results suggest the BvgA~P/RNAP/Pfim3 complex must accommodate BvgA~P, an αCTD, and σ Region 4, all within the −35 region of the promoter (Fig. 8c). To ask whether such a simultaneous engagement is possible, we have generated a speculative molecular model, which combines the structures of T. aquaticus σ Region 4 bound to −35 element DNA 28 (PDB 1KU7) and αCTD bound to UP element DNA 72 (PDB 1LB2) with a structure-based model of the BvgA~P DNA binding domain (BvgA residues 148–209) bound to its optimal binding site 18. (The BvgA~P/DNA interaction is based on the structure of the NarL/DNA complex 76 (PDB 1JE8), since BvgA is a member of the NarL family of response regulators.) In the model shown in Fig. 8d, the downstream BvgA~P dimer binds within the major groove along one face of the DNA, from −61 to −26 relative to the transcription start site, while σ Region 4 binds in the major groove on the opposite face of the DNA double helix, positioned between the two BvgA~P monomers. The αCTDs are centered at −55 (promoter distal) and at −32 (promoter proximal), based on the FeBABE analyses of Fig. 7. Such a position is approximately 4 to 5 bp farther downstream than that predicted from the α302-FeBABE analysis of the cII/RNAP/PRE complex 73. In the generated model at Pfim3, there is little, if any, predicted interaction between BvgA~P and σ Region 4. This model is consistent with our inability to demonstrate an interaction between BvgA or BvgA~P and σ using native gel shifts or bacterial two-hybrid assays (data not shown). A prediction of the model is that αCTD is required for activation, either through an interaction with BvgA~P or σ Region 4. Interactions between the upstream and downstream BvgA dimers may also help stabilize the various protein/nonspecific DNA interactions within the C-tract DNA. Additional experiments are underway to test this model and assemble a detailed map of the transcription activation complex at Pfim3.
The strains used in this study are listed in Table 1. Luria-Bertani (LB) agar or broth 77 and Bordet Gengou (BG) plate 20 were used for the growth of E. coli strains and B. pertussis strains, respectively. To obtain the modulating conditions repressing the bvgAS locus activities, MgSO4 (50mM) was added to the BG plates. Unless indicated elsewhere, the antibiotics used in LB agar were ampicillin (200µg/ml), gentamicin sulfate (5µg/ml) and kanamycin sulfate (10 µg/ml). The antibiotics used in BG agar were gentamicin sulfate (5µg/ml) and streptomycin sulfate (100µg/ml).
Plasmids in this study are listed in Table 1. Plasmids were constructed using standard cloning procedures. Transcription template plasmids pfha 19, pPRE# 20, and pDKT90 (for PuvsX; 39) have been described. Plasmid pSS3447 is a derivative of pTE103 with the multi-cloning site (MCS) replaced with one of our own design and a 100-bp deletion between the MCS and the terminator. Template plasmids pPfim3−15C, Pfim3−TA and Pfim3−AC and pPfim3−14C were constructed by ligating the EcoRI/SalI fragments containing derivatives of the fim3−15C promoter constructed in pSS4159 20 with plasmid pSS3447 that had been cut with the same restriction enzymes. The PfhaB, Pfim3, and Pfim3 extended −10 mutant templates were transcribed as supercoiled plasmids; a transcription termination site located downstream of the promoters yields RNAs of 401 (for PfhaB) and 261 nt (for Pfim3 and Pfim3 mutants). pPRE# was digested with NaeI to yield a linear template that produces a run-off RNA of 296 nt; pDKT90 was digested with BsaAI to yield a linear template that produces a run-off PuvsX RNA of 490 nt. PRNAI was present on each of these pBR-based plasmids and yields a transcript of 108 nt because of an internal transcription terminator 38.
To assess the transcriptional activity of promoters in vivo, we used vector pSS3967 that contains the luxCDABE operon of Photorhabdus luminescens, oriT and 1.8 kb sequence of B. pertussis. Plasmids pSS4159 (Pfim3−13C) and pQC1157 (Pfim3−15C) contain the fim3 promoter region from −130 to +33, relative to the transcription start site. Fragments were cloned into the EcoRI and SalI sites upstream of luxCDABE in pSS3967. The mutagenesis of the extended −10 motif was performed in pQC1157 with the methods described previously 20 to generate plasmids pQC1245 - pQC1250 and pQC1352. E. coli DH5α was used as a host strain for plasmid construction. After sequencing confirmation, the Pfim3 variants in pSS3967 were conjugated via E. coli strain SM10 into B. pertussis strain BP536 as described previously 20.
Wild type σ70 and σBp were produced using pLHN12 (described in 78) and pETsigmaBp 20, respectively. pETσ4.1+4.2ala was constructed by digesting both pETσ551–552, 554–555A and pETσ584–588A 40 with ClaI and EcoRI. The ClaI restriction site is located between the sequence coding for σ70 Regions 4.1 and 4.2. Complementary fragments were then ligated together to yield pETσ4.1+4.2ala, which produces His6-tagged σ4.1+4.2 ala. To construct pETσ580-3A, we used the Quikchange procedure (Stratagene) and primers that introduced codons for alanine at σ70 residues 580–583. pETσΔR4 was derived from a variant of pETσfl, pETσflCF (gift of L. Knipling), in which the XhoI site downstream of the σ70 gene (at bp 2087) had been removed. Digestion of pETσflCF with XhoI and AvrII removed the fragment encoding the C-terminal end of σ70 (residues 528–613). The cut plasmid was then ligated to a XhoI/AvaII linker that recreated the codons for residues 528 and 529 followed by a stop codon. The resulting plasmid produced His6-tagged σΔR4. pETσBpΔR4 was constructed by PCR amplification of the region encoding residues 1–674 from pETsigmaBp (Chen et al., 2010), using primers, which included an NdeI site (upstream) and EcoRI site (downstream). The resulting fragment was ligated to the vector pET28a(+) (Novagen), which had been digested with NdeI and EcoRI. The resulting plasmid expressed His6-tagged σBpΔR4. To construct pETσHyb 4, we first digested pBRα-σHyb4 (gift of S. Dove and A. Hochschild) with XhoI and SalI to isolate the fragment encoding σ70 residues 528–564 followed in frame with σ38 residues 280–330. This fragment was then ligated into pETσflCF that had been previously digested with XhoI/SalI to yield pETσHyb 4, which produced His6-tagged σHyb 4. Sequences of the primers are available from the authors upon request. DNA sequencing (Center for Biologics Evaluation and Research core facility, FDA) throughout the cloned regions confirmed the presence of only the desired sequences.
The Cys(−)rpoD plasmid (gift of C. Meares; described in 79 encodes a cysteine-less σ, in which the three native cysteine residues at positions 132, 291, and 295 were converted to serines. To create pETσD581C, this plasmid was first digested with BfuAI. The fragment containing the cysteine to serine mutations was ligated to BfuAI-digested pETσflCF (described above), resulting in a plasmid encoding His6-tagged σ-Cys(−). We then used the Quikchange procedure (Stratagene) and primers that introduced a codon for cysteine at residue D581. Plasmid pDL934, containing the galP1 promoter region, was the kind gift of D. Lewis and S. Adhya.
Pfim3−15C, Pfim3−14C, and galP1 DNA, 5’-32P-end labeled on the template (bottom) or nontemplate (top) strand, were prepared as described 20 using PCR with appropriate primers and the plasmids pPfim3−15C, pPfim3−14C, and pDL934.
E. coli RNAP core was purchased from Epicentre Technologies. The purification of wild type σ70 80, His6-tagged AsiA 81; 82, MotA 64, and BvgA and σBp 20 have been described. The His6-tagged proteins σHyb 4, σ4.1+4.2 ala, σ580-3ala, and σ C132S/C291S/C295S/D581C (referred to as σD581C) were purified from BL21(DE3)/pLysE 83 cultures containing the desired pETσ plasmid as described 84, by denaturation of inclusion bodies containing the protein, Ni2+ resin affinity chromatography under denaturing conditions, and then a slow renaturation of the protein. The His6-tagged σΔR4 and σBpΔR4 were purified similarly, except that after sonication of the cells, these proteins were found in the 17,400 × g supernatant rather than in inclusion bodies; consequently, they were subjected to Ni2+ resin chromatography under native conditions. Generation of α276C and α302C and isolation of RNAP containing each of these mutants was performed as described 18.
BvgA was phosphorylated by incubation for 30 min at room temperature in a solution containing 20 mM Tris-Cl (pH 8) and 20 mM acetyl phosphate. Transcription reactions (10 µl) containing 0.1 pmol DNA, the indicated amount of BvgA~P, 0.75 pmol reconstituted RNAP (sigma:core of 2.5:1), 60 mM Tris-Cl (pH 8), 25 mM acetyl phosphate, 15 mM NaCl, 10 mM KCl, 4 mM Hepes, 2 mM MgCl2, 3 mM magnesium acetate, 0.4 mM dithiothreitol, 0.3 mM ethylenediaminetetraacetic acid (EDTA), 25 µg/ml BSA, 20% glycerol (v/v), and 0.002% Triton X-100 were incubated at 37o C for 15 min before the addition of heparin to a final concentration of 200 µg/ml and NTPs to a final concentration of 250 µM ATP, GTP, and CTP and 25 µM [α-32P]UTP (4.4×103 dpm/pmol) (Fig. 2, ,3,3, ,4)4) or 250 µM ATP, GTP, and UTP and 50 µM [γ-32P]GTP (2.5×104 dpm/pmol) (Fig. 6). For transcriptions using AsiA-associated RNAP (Fig. 3), σ was incubated with 6-fold excess of AsiA for 10 min at 37° C before RNAP reconstitution. When using PuvsX (Fig. S2), reactions were performed as described 64. Reactions were stopped by the addition of 25 µl gel loading solution (1X TBE, 7 M urea, 0.1% (w/v) bromophenol blue, 0.1% (w/v) xylene cyanol FF), and the resulting solution was heated at 95° C for two minutes before electrophoresis on 7 M urea, 4% or 23% polyacrylamide denaturing gels run in 0.5X TBE. Gels were exposed to either film or a phosphorimaging screen, which was subsequently scanned using a Fujfilm FLA-5100 phosphorimager. Autoradiograms or phosphorimages were then analyzed on a Powerlook 2100XL densitometer and band intensities quantified using Quantity One software from Bio-Rad, Inc.
B. pertussis strains integrated with Pfim3 variants were streaked in sectors on BG plates supplied with gentamicin. After incubation for 48 hours at 37° C, luciferase activity was visualized by imaging with an IVIS−50 instrument (Caliper Life Sciences) and quantified using Living Image software v2.6.1 (Caliper Life Sciences) running in Igor Pro Carbon (WaveMetrics), as described previously 20.
Conjugation of σ C132S/C291S/C295S/D581C (referred to as σD581C) was performed essentially as described 85: σD581C (2 nmol) was dialyzed into storage buffer without DTT (50 mM Tris-Cl (pH 7.9), 50 mM NaCl, 0.1 mM EDTA (pH 8.0), 0.01% Triton X-100, and 50% glycerol) for 1 hr at 4°C to remove DTT. FeBABE ((iron (S)-1-(p-bromoacetamidobenzyl)ethylenediaminetetraacetate, 10 nmol, Dojindo Laboratories) was added and the mixture incubated for 1 hr at 37°C, then dialyzed against storage buffer containing 0.1 mM DTT for 1 hr at 4°C to remove free FeBABE. Conjugation efficiency (50%) was determined by mass spec analysis performed by the NIDDK Mass Spec Core Facility (population of protein with mass equal to σD581C-FeBABE divided by total population of σD581C and σD581C-FeBABE protein). Protein activity of the conjugated σD581C-FeBABE was estimated to be 35% based on its activity in in vitro transcription.
FeBABE cleavage reactions with RNAP that was reconstituted with σD581C-FeBABE were performed using a protocol similar to that of Colland et al 56. Equimolar σD581C-FeBABE and RNAP core were incubated for 10 min at 37° C to reconstitute holoenzyme. Binding reactions consisted of 1.7 pmol reconstituted holoenzyme, 3.2 pmol BvgA~P, and 0.2 pmol 32P-labeled fim3 DNA or 0.3 pmol 32P-labeled galP1 DNA in 150 mM potassium glutamate, 40 mM Tris-acetate (pH 7.9), 34 mM NaCl, 9 mM Tris-Cl (pH 8), 5 mM KCl, 4 mM magnesium acetate, 2 mM acetyl phosphate, 2 mM Hepes-OH (pH 7.4), 1 mM MgCl2, 0.3 mM DTT, 0.2 mM EDTA, 100 µg/ml BSA, 14% glycerol, 0.0005% TritonX-100. Reactions (10 µl total volume) were incubated for 10 min at 37° C. Heparin (0.5 µg in 0.5 µl) was added 1 min prior to the addition of 2 µl 100 mM sodium ascorbate and 2 µl 0.6% H2O2. The cleavage reaction was allowed to proceed for 10 min at 37° C at which point 110 µl quench buffer (72 mM thiourea, 4.5 µg/ml calf thymus DNA, 2.7 mM Tris-Cl (pH 7.9), 0.27 mM EDTA) were added. The end-labeled DNA was also used in G+A sequencing reactions 86 to map cleavage coordinates. DNA was phenol extracted and ethanol precipitated, then separated by gel electrophoresis on an 8% acrylamide denaturing gel in 0.5 X TBE. Gels were exposed to film and analyzed as described above.
Cleavage of 32P end-labeled Pfim3-14C DNA using RNAP with α276C and α302C that had been conjugated with FeBABE was performed essentially as described for PfhaB 18.
Using MolSoft ICM software, the −33 nucleotide from the structure of T. aquaticus σ Region 4 bound to DNA (PDB 1 KU7) 28 was superimposed with the −33 nucleotide from the model of BvgA~P bound to DNA 18 so that the major grooves aligned. A monomer of RNAP αCTD was added to the model by superimposing the −35 region minor groove from the structure of the CAP-αCTD-DNA complex (PDB 1LB2) 72 with the −35 region minor groove of the aforementioned σ Region 4/BvgA model, yielding a molecular model of the ternary complex of σ Region 4, BvgA~P, αCTD, and −35 region DNA.
We are grateful to C. Meares for the Cys(−)rpoD plasmid, to D. Lewis and S. Adhya for pDL934, to S. Dove and A. Hochschild for pBRα-σHyb4, to T. James for the molecular modeling, and to L. Knipling for the construction of pETσflCF and pETσHyb4 and purification of σHyb4. We also thank R. Bonocora, L. Knipling, T. James, A. Boulanger-Castaing, S. Jha, and C. Jones for helpful discussions. K. Decker is a graduate student in the Graduate Partnership Program, Johns Hopkins University-National Institutes of Health. This research was supported in part by the Intramural Research Program of the NIH, National Institute of Diabetes and Digestive and Kidney Diseases.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.