Formation of Intermediate Transcription Initiation Complexes at pfliD and pflgM by ς²⁸ RNA Polymerase

Abstract

The ς subunit of prokaryotic RNA polymerase is an important factor in the control of transcription initiation. Primary ς factors are essential for growth, while alternative ς factors are activated in response to various stimuli. Expression of class 3 genes during flagellum biosynthesis in Salmonella enterica serovar Typhimurium is dependent on the alternative ς factor ς²⁸. Previously, a novel mechanism of transcription initiation at the fliC promoter by ς²⁸ holoenzyme was proposed. Here, we have characterized the mechanism of transcription initiation by a holoenzyme carrying ς²⁸ at the fliD and flgM promoters to determine if the mechanism of initiation observed at pfliC is a general phenomenon for all ς²⁸-dependent promoters. Temperature-dependent footprinting demonstrated that promoter binding properties and low-temperature open complex formation are similar for pfliC, pfliD, and pflgM. However, certain aspects of DNA strand separation and complex stability are promoter dependent. Open complexes form in a concerted manner at pflgM, while a sequential pattern of open complex formation occurs at pfliD. Open and initiated complexes formed by holoenzyme carrying ς²⁸ are generally unstable to heparin challenge, with the exception of initiated complexes at pflgM, which are stable in the presence of nucleoside triphosphates.

The ς subunit of bacterial RNA polymerase (RNAP) is an important factor in the positive control of transcription initiation. Several alternative ς factors can be used interchangeably to regulate which genes will be transcribed under a particular set of conditions. The core (E) form of RNAP is a multisubunit enzyme consisting of the α₂, β, β′, and ω subunits (7). Core RNAP is responsible for transcript elongation, but is insufficient to facilitate promoter-driven transcription (60), which requires a dissociable ς subunit (38, 60). The use of different ς factors allows RNAP to recognize a variety of promoters and to rapidly respond to environmental changes (22). The ς factors are divided into two groups based upon their homology to ς⁷⁰, the primary ς factor from Escherichia coli, or to ς⁵⁴, which is involved in transcription of genes in the nitrogen regulon (37, 41). The ς⁷⁰ family has been further subdivided into primary and alternative ς factors (25, 57). Primary ς factors are essential proteins required for the expression of housekeeping genes, while alternative ς factors are activated in response to various stimuli (22, 25, 37). Analysis of Eς⁷⁰-dependent promoters reveals several important elements for recognition: the conserved hexamers at −10 (TATAAT) and −35 (TTGACA) upstream from the transcription start site (23, 24), the 17-bp spacer DNA separating the hexamers (2, 42, 56), and, at some promoters, an AT-rich region between −40 and −60 called the UP element (14).

Amino acid sequence comparison of ς⁷⁰ family members led to the identification of four highly conserved regions that have each been further subdivided into between two and four subregions (25, 37). Region 1.1 is found only in primary ς factors (37) and prevents the ς subunit from binding to DNA in the absence of the core subunits (15). Deletion of region 1.1 of ς⁷⁰ results in defective promoter melting and transcription initiation at the λP_R promoter (65), while single amino acid substitutions have been shown to affect initial DNA binding by holoenzyme (3) and the stability of ς⁷⁰ (4). Region 1.2, while highly conserved and found in almost all ς factors, has not yet been assigned a function (37). Deletion of regions 1.1 and 1.2 of ς⁷⁰ results in transcriptional arrest after initial binding of RNAP to the promoter and an inability to form open complexes (65). Region 4.2 recognizes the −35 consensus sequence, while the −10 consensus hexamer is recognized by region 2.4 (15, 18, 53, 63, 68). Region 2.3 is involved in DNA melting (29, 30, 32, 48), and regions 2.1, 2.2, and 3.2 are important for core binding (31, 36, 67). Almost all ς factors contain the highly conserved regions 1.2, 2, 3, and 4, but variability of composition exists, especially among the alternative ς factors (37).

Late gene expression during flagellum biosynthesis in Salmonella enterica serovar Typhimurium is dependent upon an alternative ς factor, ς²⁸. This ς factor is unusual because it lacks both amino-terminal regions 1.1 and 1.2, as well as the nonconserved spacer region between regions 1 and 2, making it less than half the size of ς⁷⁰ (11). This notable difference may be at least partially responsible for promoter recognition and binding properties that differ from those of other holoenzymes (50). The flagellar operon is divided into three hierarchical transcriptional classes of genes (1–3): early, middle, and late (11, 34). The fliA gene, which encodes the alternative sigma factor, ς²⁸, is transcribed from a ς⁷⁰-dependent class 2 promoter (11, 34). The class 3, or late gene, promoters are specific for ς²⁸ RNAP and include fliC, fliD, and flgM, encoding flagellin (43), the flagellar cap protein (11, 35), and an anti-ς factor, respectively (11, 19, 44)

The process of transcription initiation has been well characterized for E. coli Eς⁷⁰ at the λP_R promoter (13, 14, 26, 46). First, RNAP binds to the promoter to form an initial closed complex, RP_c1. This complex forms a “short” DNase I footprint that protects the DNA from −55 to −5 with sequence-specific recognition at the −35 hexamer (46). RP_c1 undergoes rearrangement to a kinetically significant intermediate called I₁, where contacts with the DNA are extended (13, 14). The rate-limiting step is the conversion of I₁ to I₂, where conformational changes occur that are characterized by burial of nonpolar groups (26). The I₁ and I₂ complexes are sometimes referred to collectively as RP_c2. RP_c2 complexes maintain the upstream contacts established in RP_c1, but also exhibit interactions that extend toward +20 (−55 to +20) (26). Strand opening follows the formation of an open complex, RP_o1, where the length of the footprint is unchanged, but strand separation occurs in the area of −11 to −1 (14, 46). Completion of strand opening, defined by formation of RP_o2, is dependent upon the presence of Mg²⁺and is characterized by DNA melting from −12 to +2 (14, 46). In the presence of initiating nucleoside triphosphates (NTPs), an initiated complex, RP_init, produces small abortive transcripts until promoter clearance (46), when the ς factor is released and RNAP enters the elongation phase of transcription. Intermediate complexes that form during transcription initiation can accumulate and be visualized though the use of lower temperatures. Temperature-dependent steps do not always strictly correspond to time-dependent events, but chemical and enzymatic footprinting as a function of temperature can be used to gain an understanding of structural changes within intermediate complexes and to make relative comparisons between promoters (12, 33, 39).

Previously, a novel mechanism of transcription initiation was proposed for Eς²⁸ (50). Temperature-dependent intermediate complexes formed during initiation at the flagellin (fliC) promoter were distinct from those previously identified for Eς⁷⁰ or Eς³². At 0°C, Eς²⁸ forms a short closed complex with pfliC that protects the promoter from DNase I digestion in the region from −65 to −19. This protection does not include the −10 element, and thus the footprint is much shorter than those typically observed for Eς⁷⁰ or Eς³² complexes. Initial binding to the promoter does not require the −10 element, indicating that binding of Eς²⁸ initially occurs mainly through interaction in the −35 region (50). Eς²⁸ then isomerizes to make additional contacts that extend to +20, while relinquishing upstream contacts between −65 and −46. One major difference between Eς²⁸ and Eς⁷⁰ intermediates is the absence of a detectable RP_c2-like complex for Eς²⁸. Instead, Eς²⁸ appears to form a single short closed complex (RP_c1) and then progresses directly to the open complex (RP_o).

Here we have characterized the mechanism of transcription initiation by Eς²⁸ at two additional promoters. Temperature-dependent intermediate complexes were analyzed in detail at pfliD and pflgM to determine if the mechanism of initiation observed at pfliC is a general phenomenon for all ς²⁸-dependent promoters or is unique for the flagellin promoter. The results demonstrate that the promoter binding properties and low-temperature open complex formation are similar between pfliC, pfliD, and pflgM, but certain details of DNA strand separation and complex stability are promoter dependent.

MATERIALS AND METHODS

Overproduction and purification of ς²⁸ and reconstitution of holoenzyme.

^. The fliA gene, encoding ς²⁸, was inserted into plasmid pET15b (Novagen, Inc.) to generate pKH439 (a gift from K. Hughes), which resulted in the addition of six histidines at the amino terminus. Hexahistidine-tagged ς²⁸ was overproduced and purified by the method described by Wilson and Dombroski (65). Holoenzyme was reconstituted by adding 1.0 pmol of E. coli core RNAP (8) to 7.0 pmol of ς²⁸ in protein dilution buffer (10 mM Tris-HCl [pH 8.0], 10 mM β-mercaptoethanol, 1 mM EDTA, 0.4 mg of bovine serum albumin per ml, 0.1% Triton X-100) and incubated on ice for at least 15 min. The high degree of similarity between the core subunits of E. coli and Salmonella enterica serovar Typhimurium justified the use of a heterologous system, as reported by others (1, 9, 27, 50, 52).

Generation of promoter fragments.

³²P-5′-end-labeled primers were generated for use in synthesizing labeled promoter DNA containing the fliD or flgM promoter (15). Oligonucleotide primers were obtained from Genosys Biotechnologies, Inc., or Integrated DNA Technologies, Inc. Plasmids pJK284 and pKG12 (gifts from K. Hughes) were used as template DNA to generate the fliD and flgM promoter fragments. Radiolabeled fliD and flgM promoters were synthesized by using PCR to generate 200 (pfliD)- and 176 (pflgM)-bp-long fragments. The PCRs were performed as described previously (50) in a Perkin-Elmer Thermocycler, with the exception that step-down annealing temperatures were set at 55, 53, 51, and 48°C for 8 cycles each. The products were purified with the Qiaquick PCR DNA purification kit (Qiagen, Inc.).

DNase I footprinting.

DNase I footprinting was performed as described previously (28, 50). ³²P-end-labeled DNA promoter fragments and Eς²⁸ were incubated in DNase I buffer (20 mM Na HEPES [pH 7.5], 10 mM MgCl₂, 100 mM NaCl, 0.1 mM EDTA, 1 mM dithiothreitol, 200 μg of bovine serum albumin per ml) in a total volume of 50 μl. DNA-alone controls were performed at each temperature with equivalent patterns of digestion, except at 0°C, at which a region resistant to digestion was observed as described in Results. Protection of the DNA by RNAP was determined by comparing the experiments containing DNA alone to those containing RNAP by visual inspection. Variations of DNase I footprinting are described below.

Temperature variation.

The Eς²⁸-promoter complexes were allowed to form at 0, 6, 16, 23, and 37°C for at least 15 min. The complexes formed at 0 and 6°C were subjected to DNase I digestion (3 U of DNase I) for 24 and 11.5 min, respectively. Complexes formed at 16°C were digested for 7.5 min (1.5 U of DNase I), and samples at 25 and 37°C were digested for 4 and 1.5 min, respectively (1.0 U of DNase I). The samples were then processed as described above (28, 50).

Heparin competition.

The Eς²⁸-promoter complexes were allowed to form at 37°C for 15 min. In some cases, the first three NTPs were added to a final concentration of 0.2 mM for 1 min prior to digestion with DNase I or the addition of heparin. In some cases, heparin was added (final concentration, 25 to 50 μg/ml) followed by incubation for 1 min prior to the addition of heparin or digestion with DNase I. The complexes were treated with DNase I (1.0 U) for 1 min. The samples were processed as described previously (27, 48).

KMnO₄ footprinting

Potassium permanganate (KMnO₄) footprinting was performed as described in references 28 and 50, except Eς²⁸ and ³²P-end-labeled DNA promoter fragments were incubated in KMnO₄ buffer (20 mM Na HEPES [pH 7.5], 10 mM MgCl₂, 100 mM NaCl, 0.1 mM EDTA, 0.2 mM dithiothreitol, 200 μg of bovine serum albumin per ml) at 0, 6, 16, 23, or 37°C for at least 15 min prior to treatment with 2.5 μl of 50 mM KMnO₄ for 2 min.

Nucleotide stabilization assay in vitro (nitrocellulose filter binding).

The nucleotide stabilization assay was performed as described previously (50, 65).

Construction of hybrid promoters.

pfliC-flgM and pflgM-flliC were constructed by a combination of PCR mutagenesis and recombinatory PCR as outlined by Skinner and Jones (54). The nucleotides from the −8 position to the −2 position of pflgM replace those of pfliC in pfliC-flgM. Similarly, pflgM-fliC contains the −8 to −2 region of pfliC. The resulting DNA fragments were ligated into PST-Blue-1 and transformed into Novablue Singles competent cells from the Perfectly Blunt cloning kit (Novagen). These plasmids were used as templates for PCR to synthesize radiolabeled DNA for DNase I footprinting.

RESULTS

Promoter binding and intermediate complexes formed as a function of temperature.

Previously, we characterized promoter binding and open complex formation at the promoter for the flagellin gene, pfliC (50). Here, we chose to explore RNAP-DNA interactions at two other ς²⁸-dependent genes with class 3 promoters to determine whether the mechanism observed for pfliC would also apply to these promoters. We examined initial promoter binding, open complex formation, and complex stability by using the fliD and flgM promoters (Fig. 1).

FIG. 1.

Comparison of promoter sequences used in this study. The ς²⁸ consensus sequence as proposed by Ide et al. (27) is shown at the top. The −10 and −35 regions for each promoter are indicated in boldface type, and the transcription start sites are underlined. The sequences extend from −60 to +8 relative to the transcription start site.

Intermediate preinitiation complexes can be visualized by performing DNase I footprinting experiments over a range of temperatures, with the rationale that temperature-dependent intermediates may represent time-dependent events (12, 33, 39) and can minimally provide an indication of structural changes that occur on the pathway to open complex formation (66). A sequential mechanism of promoter binding as a function of temperature was previously observed for Eς²⁸ at pfliC (50). The prevalence of this mechanism for promoter recognition by Eς²⁸ was tested by examining temperature-dependent complex formation at pfliD and pflgM.

A 200-bp DNA fragment containing the fliD promoter was radioactively labeled on the template strand. This fragment was incubated with or without Eς²⁸ at various temperatures prior to digestion with DNase I. Partial protection of the DNA by Eς²⁸ was observed from −72 to −25 at 0°C (Fig. 2). Some areas outside of this region appear to be protected (+15 to +23), but this is an artifact of DNase I digestion of this promoter at 0°C, because these bands are also absent in control experiments in which the DNA alone is digested at 0°C (data not shown). At 6 and 16°C, some of the upstream contacts (−72 to −58) were progressively lost as protection extended downstream. Thus, the region from −57 to +3 was the primary segment protected from DNase I digestion at 6 and 16°C. By 23°C, Eς²⁸ fully occupied the promoter DNA from −40 to +17, with partial protection extending from −57 to −40. At 37°C, the upstream contacts from −57 to −39 disappeared, but strong protection from −40 to +17 was observed. This pattern of temperature-dependent contacts is similar to the pattern for Eς²⁸ at pfliC (50).

FIG. 2.

DNase I footprinting of Eς²⁸-pfliD complexes as a function of temperature. Radiolabeled pfliD (template strand) was incubated with Eς²⁸ at the temperature indicated above each lane for 15 min before the complexes were treated with DNase I as described in Materials and Methods. DNA alone was digested at 37°C. A similar, but not identical, pattern of digestion was obtained at 0°C with the differences noted in the text (data not shown). Samples were analyzed by electrophoresis on a denaturing 8% polyacrylamide gel.

Eς²⁸ DNA complexes were also characterized at the pflgM promoter. In this case, the results for DNase I footprinting on the nontemplate strand are shown in Fig. 3. (Similar results were observed with the template strand [data not shown].) At 0, 6, and 16°C, protection was observed from −41 to −35 and from −31 to −24. When the temperature was increased to 23 or 37°C, downstream protection extended to +20. The flgM promoter is initially not as well protected by Eς²⁸ in the region upstream of the −35 consensus as pfliD and pfliC. At low temperatures, protection from DNase I digestion is only observed to −41 at pflgM compared to −65 or −72 for pfliC and pfliD, and these upstream contacts are eventually lost at higher temperatures. At pflgM, Eς²⁸ either does not make extended upstream contacts and thus binds in a more optimal position at pflgM, so loss of the upstream contacts is not necessary, or the upstream contacts are too unstable to be detected by DNase I footprinting.

FIG. 3.

DNase I footprinting of Eς²⁸-pflgM complexes as a function of temperature. Radiolabeled pflgM nontemplate strand (template strand gave similar results [data not shown]) was incubated with Eς²⁸ at the temperature indicated above each lane for 15 min before the complexes were treated with DNase I. DNA alone was digested at 37°C. Samples were analyzed by electrophoresis on a denaturing 8% polyacrylamide gel.

Despite a few minor differences, a general pattern of promoter binding emerged. At each promoter examined, Eς²⁸ exhibited initial binding in the upstream region without making significant interactions in the −10 region, but eventually established strong contact between approximately −40 and +15. Overall, these footprints are significantly shorter in the upstream region at 23 and 37°C than those observed for Eς⁷⁰ or Eς³², which extend to about −60 under similar conditions.

Open complex formation.

Sensitivity to potassium permanganate (KMnO₄) can be used to determine which Eς²⁸-DNA complexes are open complexes. KMnO₄ oxidizes unpaired thymine bases that become accessible in the strand-separated open complex. Modified bases are then susceptible to cleavage by piperidine. We used this method to follow the process of DNA melting during open complex formation for Eς²⁸ under the same conditions used in the DNase I temperature series. Open complexes at pfliD were allowed to form for 15 min at 0, 6, 16, 23, and 37°C on radiolabeled DNA and then subjected to KMnO₄ treatment and piperidine cleavage. KMnO₄ reactivity at 0°C was observed at positions −8 and −9 (nontemplate strand) of pfliD, indicating that strand separation occurred in the −10 region, but not in the +1 region (Fig. 4). At 6°C, the degree of KMnO₄ reactivity at −8 and −9 increased. At 16°C, new bands appeared at +1 and +2 as expected for a fully open transcription bubble. At 23 and 37°C, the KMnO₄ reactivity increased in the +1 region. Thus, Eς²⁸ at pfliD displays a sequential pathway of open complex formation as a function of temperature. As the temperature increases, DNA melting propagates downstream in a unidirectional manner. This is not a general property of ς²⁸-dependent promoters, because both pfliC (50) and pflgM (see below) appear to open in a concerted manner.

FIG. 4.

KMnO₄ footprinting of Eς²⁸-pfliD complexes as a function of temperature. Radiolabeled pfliD (template strand) was incubated with Eς²⁸ at the temperature indicated above each lane for 15 min before the complexes were treated with KMnO₄ and piperidine. DNA alone was treated at 37°C. Arrows indicate the positions of thymine residues. Samples were analyzed by electrophoresis on a denaturing 8% polyacrylamide gel. A phosphoimager was used to quantify the extent of cleavage relative to the 37°C lane. In the −10 region, the relative reactivities were 1.2, 0.8, 0.8, and 0.3 at 23, 16, 6, and 0°C, respectively. In the +1 region, the relative reactivities were 0.8, 0.4, 0.1, and 0.0 at 23, 16, 6, and 0°C, respectively.

Eς²⁸-pflgM complexes were also tested for KMnO₄ sensitivity over a range of temperatures (Fig. 5). From 0 to 6°C, the DNA remained base paired. KMnO₄ reactivity was first observed at 16°C at −8, −6, −4, +1, and +2. The sensitive sites remained the same at 23 and 37°C. Therefore, open complex formation for Eς²⁸-pflgM appears to occur in a concerted manner at the temperatures utilized, similar to Eς²⁸-pfliC. In fact, open complexes at both pfliC (50) and pflgM both are first apparent at 16°C. Typically, Eς⁷⁰ and Eς³² form primarily extended closed complexes (RP_c2) under low-temperature conditions (12, 16, 33, 39, 46, 55).

FIG. 5.

KMnO₄ footprinting of Eς²⁸-pflgM complexes as a function of temperature. Radiolabeled pflgM (template strand) was incubated with Eς²⁸ at the temperature indicated above each lane for 15 min before the complexes were treated with KMnO₄ and piperidine. DNA alone was treated at 37°C. Arrows indicate the positions of thymine residues. Samples were analyzed by electrophoresis on a denaturing 8% polyacrylamide gel. A phosphoimager was used to quantify the extent of cleavage relative to the 37°C lane. The relative reactivities were 0.6, 0.2, 0.1, and 0.08 at 23, 16, 6, and 0°C, respectively.

The KMnO₄ reactivity for Eς²⁸ at pfliD and pflgM was also tested in both the presence and absence of Mg²⁺. The presence of Mg²⁺ typically stimulates open complex formation and is required for the transition from RP_o1 (+12 to +1) to RP_o2 (−12 to +2) for the ς⁷⁰-dependent λP_R promoter (13, 46, 57). The presence of Mg²⁺ did not increase the length of the strand-separated region for either pfliD or pflgM (data not shown). However, at all temperatures, the presence of Mg²⁺ increased the magnitude of KMnO₄ reactivity at all susceptible positions. Thus, we did not observe an Mg²⁺-dependent step in the process of open complex formation at these promoters.

Stability of Eς²⁸-promoter complexes to heparin challenge.

Challenge with the polyanionic competitor heparin can be used as a tool to assess the relative stability of open complexes. Eς⁷⁰ typically forms a heparin stable open complex (5, 16, 40, 47), with exceptions such as the ribosomal operon promoter, rrnB P1, which requires the addition of initiating NTPs to confer heparin stability (20). Eς²⁸-pfliC open complexes have been shown to be unstable to low levels of heparin even in the presence of initiating NTPs (50). Heparin was used here to test the stability of Eς²⁸-DNA complexes at pfliD and pflgM. The binding of Eς²⁸-pfliD was sensitive to 25 μg of heparin per ml, even in the presence of the first three initiating NTPs (Fig. 6), as seen by the lack of protection in the presence of heparin. Thus, Eς²⁸-pfliD open complexes (RP_o and RP_init), like Eς²⁸-pfliC open complexes, appear generally less stable than Eς⁷⁰-promoter open complexes. RNAP with the primary ς factor from Bacillus subtilis, ς^A, also forms heparin-sensitive open complexes even in the presence of NTPs (64).

FIG. 6.

Challenge of Eς²⁸-pfliD complexes with heparin to determine stability. Radiolabeled pfliD (template strand) was incubated with Eς²⁸ for 15 min at 37°C, and the complexes were subsequently treated with DNase I. All samples were analyzed by electrophoresis on a denaturing 8% polyacrylamide gel. First lane, DNA alone; second lane, Eς²⁸; third lane, Eς²⁸ plus 0.2 mM ATP, CTP, and GTP (ACG); fourth lane, Eς²⁸ plus 25 μg of heparin (Hep) per ml; fifth lane, Eς²⁸ plus 0.2 mM ATP, CTP, and GTP plus 25 μg of heparin per ml.

The Eς²⁸-pflgM open complexes were also tested for heparin stability. Like the other ς²⁸-dependent promoters tested, the binding of Eς²⁸ to pflgM was sensitive to 25 μg of heparin per ml in the absence of NTPs (Fig. 7). Unexpectedly, the presence of the first three NTPs (ATP, CTP, and GTP) allowed the Eς²⁸-pflgM complex to remain stable during a challenge with 50 μg of heparin per ml (Fig. 7, lanes 6 to 8). The presence of initiating NTPs increased the length of the DNase I footprint (template strand) by approximately 5 nucleotides (Fig. 7, lanes 2 to 4).

FIG. 7.

Challenge of Eς²⁸-pflgM complexes with heparin to determine stability. Radiolabeled pflgM (template strand) was incubated with Eς²⁸ for 15 min at 37°C, and the complexes were subsequently treated with DNase I. All samples were analyzed by electrophoresis on a denaturing 8% polyacrylamide gel. First lane, DNA alone; second lane, Eς²⁸; third and fourth lanes, Eς²⁸ plus 0. 2 mM ATP, CTP, and GTP (ACG); fourth and fifth lanes, Eς²⁸ plus 25 μg of heparin (Hep) per ml; seventh lane, Eς²⁸ plus 0.2 mM ATP, CTP, and GTP plus 25 μg of heparin per ml; eighth lane, same as lane 7, except 50 μg of heparin per ml was used.

Role of the discriminator region in conferring heparin stability.

The composition of the base pairs between the −10 element and +1 can affect the stability of open complexes to heparin and is inversely related to the G/C content in this discriminator region for the E. coli ς⁷⁰-dependent tyrT tRNA promoter (45). We hypothesized that one explanation for the stability of Eς²⁸ complexes at pflgM to heparin challenge in the presence of initiating NTPs is that it contains 6 A/T bp out of 8 between −7 and +1. In contrast, the pfliC and pfliD promoters, which form unstable complexes in the presence or absence of NTPs, contain 5 A/T bp out of 8, including 3 adjacent G/C bp, and 2 A/T bp out of 7, respectively. To test this idea, we constructed two hybrid promoters. The base pairs between −8 and +1 for pflgM were substituted for the corresponding region of pfliC and vice versa. DNase I footprinting was performed in the presence and absence of the first three NTPs and heparin. The results showed that the discriminator region of pflgM is not solely responsible for conferring resistance to heparin, because the Eς²⁸-pfliC-flgM open complexes were no more stable than Eς²⁸-pfliC open complexes (data not shown). Additionally, Eς²⁸ is not destabilized at the flgM promoter during a heparin challenge by the replacement of its discriminator region with that of pfliC (data not shown).

Stability of Eς²⁸-promoter complexes to high salt.

The initiated open complex, RP_init, for Eς⁷⁰ is stable to a high-salt wash during nitrocellulose filter binding, while RP_o complexes are unstable (47, 59, 65). Resistance to a challenge with a high-salt wash (0.8 M NaCl) in the presence of NTPs was used to test the stability of RP_init complexes in the present study. Eς²⁸-pfliD or -pflgM open complexes were formed in the presence or absence of NTPs and then filtered through nitrocellulose. The bound complexes were subjected to a 0.1 or 0.8 M NaCl wash. After a 0.1 M NaCl wash in the presence of NTPs, 41% of Eς²⁸-pfliD complexes were retained on the filter, while only 13% of the Eς²⁸-pfliD complexes were retained after the high-salt wash. In the absence of NTPs, 17% of Eς²⁸-pfliD complexes remained bound after a 0.8 M NaCl wash. Thus, the presence of initiating NTPs did not stabilize Eς²⁸-pfliD complexes to either a challenge with heparin or a high-salt wash. RP_init complexes were also examined for Eς²⁸-pflgM in the presence of NTPs. Sixty percent of Eς²⁸-pflgM complexes remained bound to the filter after a 0.1 M NaCl wash, and 45% remained bound after a 0.8 M NaCl wash. In the absence of NTPs, only 6% of complexes were retained after a 0.8 M NaCl wash. Therefore, Eς²⁸-pflgM RP_init complexes, like Eς²⁸-pfliC and some Eς⁷⁰-DNA complexes, are stabilized in response to a high-salt wash in the presence of NTPs.

DISCUSSION

Previously, a mechanism was proposed for transcription initiation by holoenzyme carrying the alternative ς factor ς²⁸ at the fliC promoter (50). Eς²⁸ initially binds through interaction in the vicinity of the −35 consensus sequence without any detectable contact in the −10 region. Eς²⁸ then isomerizes to make additional contacts that extend to the −10 region, the start site, and downstream sequences with strong contacts between approximately −40 and +15. Extension of the −10 contacts is accompanied by release of upstream contacts between −72 and −57. Overall, the footprints are shorter in the upstream region at 23 and 37°C than those observed for Eς⁷⁰ and Eς³², which extend to about −60 under similar conditions.

We have characterized the mechanism of transcription initiation by Eς²⁸ at the fliD and flgM promoters to determine if this is a general phenomenon for all ς²⁸-dependent promoters or a unique characteristic of the fliC promoter. Temperature-dependent footprinting analysis demonstrated that promoter binding can be characterized by a similar sequence of events for all three ς²⁸-dependent promoters examined. Thus, initial binding in the −35 region and loss of upstream contacts appear to be characteristic of Eς²⁸ promoter complexes. In contrast, low-temperature footprinting has shown that Eς⁷⁰ initially binds the λP_R promoter to form the closed complex RP_c1, which protects the DNA from about −55 to −5. At higher temperatures, a second closed complex is observed with contacts from −55 to +20 (13, 26, 46). DNA sequential binding with initial contacts in the −35 region is not a novel characteristic of Eς²⁸-promoter complexes. Recently, rapid time-resolved laser UV irradiation was used to show that recognition of the lacUV5 promoter by Eς⁷⁰ occurs initially in the −35 region of the promoter in a short-lived complex (6). Otherwise, Eς²⁸ is the only polymerase known to exhibit sequential binding within the much slower time frame of DNase I footprinting.

The contacts appearing between −60 and −40 in the DNase I experiments may be indicative of specific interactions between the α subunit of RNA polymerase and the DNA (21). These interactions are typically dependent upon the presence of an A/T-rich sequence known as the UP element (22). We previously tested the response of the fliC promoter to RNAP lacking the carboxyl-terminal domain (CTD) of the α subunit, which is required for UP element interactions. We found no evidence for a UP element-mediated effect on transcription (49). While we cannot rule out the possibility that the α subunit makes specific interactions in the upstream regions of the fliD and flgM promoters, neither shows strong similarity to the UP element consensus sequence.

KMnO₄ sensitivity was used to monitor strand melting during open complex formation by Eς²⁸. At many E. coli promoters, the transcription bubble forms in a concerted manner extending from −11 or −10 to +2 or +3 (10, 26), but in some cases, discrete intermediates can be visualized by controlling the reaction temperature or by omitting Mg²⁺ (26, 58). Eς²⁸ at pfliD exhibits an unusual sequential pattern of strand separation. At low temperatures (0 and 6°C), DNA distortion or strand separation occurs in the −10 region only. As the temperature increases, strand separation becomes apparent in the +1 region as well. Eς^D from B. subtilis at the flagellin promoter also displays sequential strand separation (10, 26). At 0°C, Eς^D establishes a partially open complex with KMnO₄ reactivity localized between −11 and −4. Then, as the temperature is increased to 20°C, the transcription bubble extends to near −1. Finally, at higher temperatures (40°C) and in the presence of Mg²⁺, the transcription bubble extends to +3 (10, 26).

Sequential open complex formation is not a general property of ς²⁸-dependent promoters, because both pfliC (50) and pflgM open in a concerted manner. However, strand separation occurring at low temperatures appears to be a general and somewhat novel property of ς²⁸-dependent promoters. Eς²⁸ open complexes appear at pfliC and pflgM at 16°C, as well as possibly lower temperatures. Typically, temperatures above 16°C are required for significant open complex formation at Eς⁷⁰ and Eς³² (12, 33, 46). However, a β subunit mutant of RNAP has been shown to allow open complex formation by Eς⁷⁰ at temperatures as low at −20°C at the T7 A2 promoter (51). Additionally, Eς²⁸ and Eς^D both generate strand-separated promoter regions at low temperatures, but at 16°C, Eς²⁸-pfliD displays strand separation in both −10 and +1 regions, while Eς^D does not complete open complex formation without higher temperatures (10).

Closer examination of the sequence between −7 and +1 (Fig. 1) shows that pfliD is more G/C rich than either pflgM or pfliC. This region has been implicated in affecting open complex formation as well as complex stability (61). Analysis of the lifetime of preinitiation complexes as a function of this sequence has revealed a direct relationship between the A/T content of the DNA-melting region and the stability of the complex for Eς⁷⁰ at the tyrT promoter (45). Thus, the G/C-rich stretch in pfliD between −10 and +1 may be imposing a kinetic block to formation of the open complex by providing a barrier to DNA untwisting or DNA melting (45, 62) and may at least partially responsible for the pattern of strand separation at pfliD versus pflgM. At low temperatures, there may be a kinetic restraint on DNA melting at pfliD that is relieved at higher temperatures, resulting in a sequential pattern of melting.

A challenge with the polyanionic competitor heparin has been used as a tool to assess the relative stability of open (RP_o) and initiated (RP_init) complexes. Most Eς⁷⁰ open complexes are stable to a heparin challenge even in the absence of NTPs (5, 16, 40, 42, 47). Exceptions include the rRNA and tRNA promoters, which form preinitiation complexes that are only stable in the presence of NTPs (20, 45). The flgM promoter is the only Eς²⁸ promoter examined that could be stabilized to a challenge with heparin by NTPs. In runoff transcription experiments, discriminator mutants that have significantly lower G/C content are more stable in the absence of initiating NTPs (17, 45). pfliC (50) and pfliD both have a higher G/C content in their discriminator regions than pflgM and were susceptible to heparin challenge even in the presence of NTPs. We directly tested whether altering the G/C content between the −10 element and +1 of the discriminator region would affect the stability of open complexes by exchanging the base pairs in this region between pflgM and pfliC. Our results do not support the notion that the G/C content of this region alone is causing pflgM to develop heparin resistance in the presence of NTPs. Thus, the general instability of ς²⁸-promoter complexes appears to involve a more complex set of issues and will require a more extensive investigation.

pfliD, the promoter with the most G/C-rich discriminator, was also unstable in the nucleotide stabilization assay, while Eς²⁸-pfliC (50) and pflgM complexes were stable to a 0.8 M NaCl wash in the presence of NTPs. The fairly G/C-rich discriminator regions of pfliC and pfliD may explain their decreased stability compared to that of pflgM. Nonetheless, overall Eς²⁸-DNA complexes appear to be less stable than most Eς⁷⁰-DNA complexes.

Variability in open complex stability seems to be a function of DNA sequence or structure for Eς²⁸-promoter complexes. Promoter binding, however, is similar for the different ς²⁸-dependent promoters examined here, but varies from typical Eς⁷⁰ or Eς³² binding patterns. It is possible that these differences may be related to predicted structural variation in the N-terminal region of ς²⁸. Region 1.2 is typically present in all ς factors, including the flagellar biosynthesis ς factor from B. subtilis, ς^D, which retains homology to region 1.2. ς²⁸ is an unusual member of the ς⁷⁰ family of proteins, since it lacks any homology to region 1.2. The structural differences within ς²⁸ may affect the nature of the RNAP-DNA complexes that form during transcription initiation. In summary, this study reveals that Eς²⁸ binding and low-temperature open complex formation are similar between pfliC, pfliD, and pflgM, but certain details of DNA strand separation and complex stability are promoter dependent.