The Role of the Pro Sequence of Bacillus subtilisςK in Controlling Activity in Transcription Initiation*

The sigma (ς) subunit of prokaryotic RNA polymerase is required for specific recognition of promoter DNA sequences and transcription initiation. Regulation of gene expression can therefore be achieved by modulating the activity of the ς subunit. In Bacillus subtilis the mother cell-specific sporulation sigma factor, ςK, is synthesized as a precursor protein, pro-ςK, with a 20-amino acid pro sequence. This pro sequence renders ςK inactive for directing transcription of ςK-dependent genesin vivo until the pro sequence is proteolytically removed. To understand the role of the pro sequence in controlling ςK activity, we have constructed NH2-terminal truncations of pro-ςK and characterized their behaviorin vitro at the gerE promoter. In this report we show that the pro sequence inactivates ςK by interfering with the ability of ςK to associate with the core subunits of polymerase and also influences the interactions between holoenzyme and promoter DNA. Additionally, removal of as few as 6 amino acids (pro-ςKΔ6) is sufficient to activate pro-ςK for DNA binding and transcription initiation. Surprisingly, pro-ςKΔ6 binds to DNA with higher affinity and stimulates transcription 30-fold more efficiently than ςK, under certain conditions.

The sigma () subunit of prokaryotic RNA polymerase is required for specific recognition of promoter DNA sequences and transcription initiation. Regulation of gene expression can therefore be achieved by modulating the activity of the subunit. In Bacillus subtilis the mother cell-specific sporulation sigma factor, K , is synthesized as a precursor protein, pro-K , with a 20-amino acid pro sequence. This pro sequence renders K inactive for directing transcription of K -dependent genes in vivo until the pro sequence is proteolytically removed. To understand the role of the pro sequence in controlling K activity, we have constructed NH 2 -terminal truncations of pro-K and characterized their behavior in vitro at the gerE promoter. In this report we show that the pro sequence inactivates K by interfering with the ability of K to associate with the core subunits of polymerase and also influences the interactions between holoenzyme and promoter DNA. Additionally, removal of as few as 6 amino acids (pro-K ⌬6) is sufficient to activate pro-K for DNA binding and transcription initiation. Surprisingly, pro-K ⌬6 binds to DNA with higher affinity and stimulates transcription 30-fold more efficiently than K , under certain conditions.
The subunit of bacterial RNA polymerase is a DNA-binding protein that confers promoter specificity to the core subunits (␣ 2 ␤␤Ј) (1,2). factors are classified into two main groups: the 70 family and the 54 family (3). The 70 family can be further divided into two major groups, the primary and alternative factors. Primary factors are essential proteins that are responsible for directing transcription of genes important for vegetative growth and the housekeeping functions of the cell, whereas alternative factors are responsible for development and adaptive functions (3)(4)(5).
K is an alternative factor from Bacillus subtilis and is a mother cell-specific member of the factor cascade that controls sporulation (6 -8). K is synthesized as an inactive precursor protein that is processed by proteolysis about 4 h after the onset of sporulation. Removal of the 20-amino acid pro sequence ( Fig. 1) is necessary to observe K -dependent transcription activity in vivo as well as in vitro (9 -12). In an analysis of the effects of the amino terminus of various factors on DNA binding in the absence of the core subunits, pro-K was found to bind to DNA with 10-fold lower affinity than K in vitro (13). To understand better the role of the pro sequence in controlling K activity in DNA binding and transcription initiation, we have created several NH 2 -terminally truncated pro-K derivatives.
In this report, we describe the behavior of these K derivatives in DNA binding and holoenzyme containing the K derivatives in transcription initiation, both at the K -dependent gerE promoter in vitro. We show that the pro sequence affects transcriptional activity of holoenzyme by interfering with core association by pro-K and by influencing DNA interactions, including initial binding and promoter DNA melting. Surprisingly, we discovered that a 6-amino acid deletion of the pro sequence is sufficient to activate pro-K completely, to an even greater extent than removal of the entire pro sequence, primarily through altered interactions of holoenzyme with the DNA.

EXPERIMENTAL PROCEDURES
Materials-Restriction endonucleases, T4 DNA ligase, and T4 polynucleotide kinase were obtained from New England Biolabs. Calf intestinal alkaline phosphatase was from Boehringer Mannheim, and Taq DNA polymerase was from Fisher Scientific. Thrombin (T6759) was from Sigma. NTPs were from Pharmacia Biotech Inc., and [␣-32 P]GTP (3,000 Ci/mmol), [␣-32 P]CTP (3,000 Ci/mmol), and [␥-32 P]ATP (3,000 Ci/mmol) were from Amersham. Purified Escherichia coli core RNA polymerase was from Epicentre Technologies (Madison, WI). Nitrocellulose filters were from Millipore. Buffer and gel components were from Sigma or Fisher Scientific. Oligonucleotides were synthesized by BioServe Biotechnologies (Laurel, MD) or Life Technologies, Inc. Primer sequences are available upon request.
Plasmid Constructions-All K derivatives were constructed by placing portions of the sigK gene downstream of the gene encoding glutathione S-transferase (GST) 1 in pGEX-2T (14). Oligonucleotides that incorporated EcoRI and BamHI restriction sites at the 5Ј-and 3Ј-end, respectively, were used to amplify fragments of the sigK gene from pSK5 (gift of L. Kroos) using the polymerase chain reaction (PCR). Fragments containing full-length sigK (pro-K ) and three successive NH 2 -terminal deletions of sigK, removing 6 (pro-K ⌬6), 10 (pro-K ⌬10), and 20 ( K ) amino acids, were generated. PCR methods were described previously (13). Amplified DNA was inserted in-frame into the pGEX-2T plasmid using EcoRI and BamHI restriction sites. All plasmid constructs were transformed into E. coli BL21 (Novagen, Inc.) for overproduction of the GST fusion proteins.
Overproduction and Purification of Proteins-GST-fusions were induced by adding isopropyl ␤-D-thiogalactopyranoside to a final concentration of 0.2 mM to a liter of cells growing exponentially at 37°C in LB plus 100 g/ml ampicillin. Cells were harvested after 3 h of induction by centrifugation and the resulting pellet stored at Ϫ20°C. Purification of GST-fusions was performed as described by Dombroski et al. (15) except that the GST tag was removed as follows. The fusion proteins, while bound to glutathione agarose beads, were incubated with thrombin at 0.4 -1.2 units/0.5 ml of beads for 2 h at 25°C. This removes the GST tag by cleaving at a thrombin site between GST and factor. Two additional amino acids, Gly and Ser, remained on the NH 2 * This work was supported in part by Grants 94G-263 from the Texas Affiliate of the American Heart Association and NP-902 from the American Cancer Society (both to A. J. D.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Nitrocellulose Filter Binding Assays-DNA carrying the K -dependent gerE promoter was labeled with [␥-32 P]ATP and used as the substrate for filter binding assays. Oligonucleotide primers corresponding to the flanking regions of the gerE promoter were used to amplify a 224-base pair DNA fragment from pSC146 (gift of L. Kroos) which spanned 96 base pairs upstream and 131 base pairs downstream of the ϩ1 start site. Primer labeling and PCR conditions were described previously (15), except the annealing step was at 45°C. DNA binding assays with the K derivatives, in the absence of the core subunits, were performed as described previously (13,15) except buffer conditions were modified such that binding buffer contained either 3.5 or 14 mM potassium acetate in 25 mM Tris-HCl (pH 7.5), 0.1 mM EDTA, 1 mM dithiothreitol, 0.01% Triton X-100, and 100 g/ml bovine serum albumin.
Reconstitution of RNA Polymerase and Transcription in Vitro-Holoenzyme was reconstituted by mixing the purified core with a 6 -20fold molar excess of each factor. gerE DNA template was synthesized using PCR as above without the 5Ј-end labeling. The mixtures were incubated on ice for 15 min. Optimal transcriptional activity of each factor was determined using increasing ratios of factor to core.
Runoff transcription assays were performed by first incubating reconstituted holoenzyme (1 pmol) with PCR-amplified gerE DNA template (0.1 pmol) in transcription buffer (100 or 250 mM KCl, 40 mM Tris-HCl (pH 7.5), 5 mM MgCl 2 , 0.1 mM dithiothreitol) at 37°C for 10 min. This allowed the polymerase to bind the promoter and form an open complex. Transcription was started upon the addition of 1.2 Ci of [␣-32 P]GTP (3,000 Ci/mmol) and NTPs (final concentrations, 2 mM ATP, CTP, UTP and 0.2 mM GTP). Reactions were incubated for 30 min at 37°C and terminated by adding 5 l of 0.04% formamide stop solution (0.04% bromphenol blue, 0.04% xylene cyanol, in formamide). RNA products were analyzed on 8% polyacrylamide, 7 M urea gels, and the bands were quantified on a Packard InstantImager. Abortive transcriptions were performed similarly to the runoff transcription assay except that the dinucleotide GpA (2 mM), CTP (0.2 mM), and [␣-32 P]CTP (3,000 Ci/mmol) were added to initiate the reactions instead of all four NTPs. This limited transcription to a three-nucleotide transcript (GpApC).
DNase I Footprinting Analysis and KMnO 4 Modification-Labeled DNA for DNase I cleavage assays was prepared by first digesting pSC146 plasmid with HindIII and KpnI to generate a 312-base pair fragment carrying the gerE promoter. Digested samples were incubated with calf alkaline phosphatase (according to manufacturer's instructions) at 37°C for 2 h. The phosphatase and restriction enzymes were removed using Micropure-EZ enzyme remover (Amicon). The fragment was labeled with [␥-32 P]ATP (3,000 Ci/mol) by adding T4 polynucleotide kinase (30 units) and incubating the DNA at 37°C for 45 min. Samples were then digested with BamHI to remove the 3Ј-end label, producing a 281-base pair 5Ј-32 P-labeled gerE promoter DNA fragment. This fragment was isolated from 2% agarose gels and purified with the Qiaquick gel extraction kit (Qiagen). 5Ј-End-labeled gerE DNA for KMnO 4 experiments was synthesized using the PCR amplification of gerE from pSC146 plasmid as described for nitrocellulose filter binding assays.
KMnO 4 modification footprinting was performed as described previously (16, 17) except reactions were incubated for 30 min at 37°C in KMnO 4 buffer with either 100 or 250 mM KCl. Radioactive bands were quantified using a Packard InstantImager.
Core Binding Assay-KMnO 4 footprinting was used to assess the ability of the various K derivatives to bind to core RNA polymerase (16). KMnO 4 reactions were performed as above except that 32 P-gerE DNA was present at a 1.5-fold molar excess over RNA polymerase holoenzyme. Core RNA polymerase (1.0 pmol) was incubated with either pro-K ⌬6 (6 pmol), equimolar pro-K and pro-K ⌬6 (6 pmol each), or equimolar K and pro-K ⌬6 (6 pmol each). Cleavage products were quantified using a Packard InstantImager.
Nucleotide Binding Stabilization Assay-The stability of RNA polymerase-DNA initiated open complexes was examined using the nucleotide binding assay described previously (16,18). K holoenzyme was formed as described above. 32 P-gerE promoter DNA in binding buffer (40 mM Tris-HCl (pH 7.5), 10 mM MgCl 2 , 100 mM KCl, 1 mM dithiothreitol, 100 g/ml bovine serum albumin) was added to holoenzyme and incubated for 30 min at 37°C. Mixtures were filtered through nitrocellulose filters (Millipore) then washed with 0.5 ml of wash buffer (10 mM Tris-HCl (pH 7.5), 0.1 mM EDTA) containing 0.1 M or 0.8 M NaCl. The DNA retained was determined using liquid scintillation counting and Ultima Gold Mixture (Packard).

RESULTS
The Pro Sequence Modulates DNA Binding by K -Previous work revealed that pro-K , in the absence of the core subunits of RNA polymerase, binds the K -dependent promoter, gerE, with 10-fold lower affinity than K (13). Here we generated two successive deletions from the amino terminus of the pro sequence to determine more precisely the effect of the pro sequence on DNA binding by K . Deletions of the first 6 (pro-K ⌬6) and 10 (pro-K ⌬10) amino acids of pro-K (Fig. 1B) were constructed as GST fusion proteins and purified as described under "Experimental Procedures." In this work, unlike in previous studies, we removed the GST affinity tag to eliminate any influence that GST might have on K activity.
All four K derivatives were tested for DNA binding ability using nitrocellulose filter retention assays (Fig. 2) (13,15). Pro-K bound to the gerE promoter in a manner similar to that observed in previous work (13). Pro-K ⌬6 and pro-K ⌬10 displayed higher affinity for pgerE than pro-K by approximately 100-fold and 10-fold, respectively (Fig. 2). This suggests that the first 6 amino acids of the pro sequence are important for modulating DNA binding. Surprisingly, we observed that K retained only 20% of the input gerE promoter DNA even at the lowest salt concentrations used (3.5 mM sodium acetate). The discrepancy between this observation and data obtained previously using GST-K (13) may be attributed to the removal of the GST tag in the current experiments. We do not understand how the GST tag might have improved filter retention for K in the past, but it is most likely due to increased nonspecific interaction between the protein and the nitrocellulose filter rather than any change in the interactions between the protein and the DNA. As shown below, K is capable of effectively directing promoter binding and transcription initiation by holoenzyme.
Deletion of Only 6 Amino Acids of the Pro Sequence Activates Pro-K -Transcription from K -dependent promoters by K holoenzyme (E K ) requires the removal of the pro sequence, since it has been shown that pro-K is unable to direct transcription (10,11). Because pro-K alone can interact with promoter DNA as shown and because deleting as few as 6 amino acids of the pro sequence increases binding affinity, we hypothesized that Epro-K ⌬6 and Epro-K ⌬10 may be active in transcription from the K -dependent promoter gerE.
Holoenzymes containing each of the K derivatives were reconstituted in vitro using E. coli core RNA polymerase and were tested for their ability to synthesize runoff transcripts. The core subunits are very highly conserved, and others have used such heterologous holoenzymes successfully to examine transcription properties in vitro (19 -21). The synthesis of a 131-nucleotide transcript from the K -dependent gerE promoter on a linear DNA fragment, was monitored. Initially, we analyzed runoff transcription for all four E K derivatives at low salt (100 mM KCl). We found that pro-K ⌬6 was active in directing transcription by holoenzyme and that the level of transcription was similar to E K (data not shown).
The same analysis was performed at high salt (250 mM KCl), where Epro-K synthesized an undetectable level of runoff transcript (Fig. 3A). This is consistent with the lack of significant activity reported for this enzyme in the past (10). Surprisingly, we found that Epro-K ⌬6 activity was optimal at high salt concentrations, resulting in an approximately 34-fold higher level of product than that observed for E K under the same conditions (Fig. 3A). Epro-K ⌬10 was also active, generating transcripts at a level similar to E K . Transcription initiation has been described as a multistep process. Initially, factor associates with the core subunits of RNA polymerase to form holoenzyme (R) and then directs holoenzyme binding to the promoter (P) to form a closed complex (RP c ). DNA promoter melting near the Ϫ10 recognition site, presumably by isomerization of holoenzyme, forms the open complex (RP o ). Polymerase then incorporates nucleotide triphosphates to form an initiated complex (RP init ), which is capable of abortive RNA synthesis, followed by promoter escape and elongation of the transcript (22).
During initiation, RNA polymerase can remain at the promoter and generate small (2-10 nucleotides) abortive transcripts. An abortive transcription assay has been developed which exploits this ability and allows assessment of events that precede promoter clearance (23,24). We used this assay to determine if the differences observed for the K derivatives at 250 mM KCl in the runoff assay occurred before or after promoter clearance. The conditions were identical to the runoff assays except that a dinucleotide (GpA) primer and [␣-32 P]CTP were used, rather than all four NTPs, to limit transcription to a three-nucleotide product (GpApC). All of the holoenzymes were capable of synthesizing abortive transcripts even at 250 mM KCl (Fig. 3B). The fact that Epro-K and Epro-K ⌬10 generated abortive transcripts more efficiently than runoff transcripts (compare Fig. 3A with Fig. 3B) implies a defect in promoter clearance under the conditions used.
Analysis of Holoenzyme-Promoter Complexes Using DNase I Footprinting-The ability of holoenzyme to bind to DNA, and the boundaries of the complexes on the DNA, can be assessed using DNase I footprinting (25,26). We used this technique to analyze the E K derivatives for interactions with the gerE promoter.
All four E K derivatives were able to produce specific protein-promoter DNA complexes (Fig. 4A). Protection of the gerE promoter extended from ϩ20 to Ϫ60 relative to the ϩ1 start site. The length of the footprint suggested that these complexes  (22). Epro-K ⌬6 showed a stronger affinity for the DNA than either Epro-K or E K , at 100 mM KCl, given the overall better protection of the gerE promoter. Also, Epro-K ⌬6 created hypersensitive bands between the Ϫ10 and Ϫ35 promoter recognition sites (Ϫ21 and Ϫ22), which were not visible in the Epro-K and E K footprint, indicating that pro-K ⌬6 may be distorting the DNA in that region. Epro-K ⌬10 protected the gerE promoter similarly to Epro-K ⌬6, but with a reduction in the intensity of the hypersensitive sites.
Many E 70 -DNA open (RP o ) complexes are stable to challenge by heparin, a polyanionic competitor, whereas closed (RP c ) complexes are not (18,(27)(28)(29). We used heparin as a tool to assess the stability of E K -gerE complexes in DNase I footprinting. All four E K derivatives formed complexes that were sensitive to even a low level of heparin (25 g/ml), indicating an unstable open complex (RP o ) or complexes that are closed (RP c ) rather than open (RP o ) (data not shown). Because these experiments were conducted in the presence of initiating nucleotides (GTP, ATP, CTP), the presumed formation of initiated complexes (RP init ) does not appear to improve the stability to heparin challenge, as has been observed for the E. coli rrnBP1 promoter (17).
We also examined the binding of the E K derivatives to pgerE DNA at four concentrations of salt because of the dramatic increase in transcription observed for Epro-K ⌬6 at 250 mM KCl. The DNase I protection assay was used but with varied concentrations of KCl (60, 100, 150, and 200 mM). Data are shown only for Epro-K ⌬6 (Fig. 4B). The extent of protection for all of the holoenzymes was generally from about Ϫ60 to ϩ20. Three of the holoenzymes, Epro-K , Epro-K ⌬10, and E K , showed decreased DNA protection as the salt concentra-tion was increased. Epro-K appeared to retain some binding at the highest salt (250 mM KCl) since the Ϫ35 region was better protected compared with E K under the same conditions. Interestingly, Epro-K ⌬6 bound to pgerE very well at high salt concentrations in agreement with its high transcriptional activity under these conditions. The hypersensitive band at Ϫ22 decreased in intensity as salt concentration increased, implying that the structure of the protein-DNA complex may be changing as a function of salt concentration. We observed that the Epro-K ⌬6 footprint, at 150 and 200 mM KCl, extended slightly beyond ϩ20 and shorter than Ϫ60 compared with 60 and 100 mM KCl, making the footprint appear to "shift" slightly forward. This shift in the footprint, from greater protection in the upstream region at low salt, to greater protection of the downstream region at high salt, may help explain the improved performance of Epro-K ⌬6 in transcription under high salt conditions. We were unable to detect any DNA cleavage due to KMnO 4 sensitivity for Epro-K -pgerE complexes at either salt concentration (100 mM or 250 mM), suggesting that the Epro-K complexes observed in DNase I footprinting at low salt are primarily extended closed complexes (Fig. 5). For E K , Epro-K ⌬6, and Epro-K ⌬10, we observed that approximately equal amounts of DNA were susceptible (50%) at low salt, indicating that these three proteins form RP o equivalently (Fig. 5A). Three bands were resolved in these experiments, corresponding to thymines Ϫ9, Ϫ6, and Ϫ4 of the gerE promoter. A band appearing at cytosine Ϫ1 was visible but at much lower levels. As the salt concentration was increased, E K open complexes decreased (8% cleaved) (Fig. 5B)  because cleavage under high salt conditions was similar to that observed at low salt (60 and 40%). We found that for both Epro-K ⌬6 and Epro-K ⌬10, another cleavage site appeared at cytosine Ϫ21. This band corresponds to the hypersensitive site found in the DNase I footprint experiments and indicates that base pairing in this region has been disturbed.
Characterization of RP init -Previous work has shown that formation of the open complex by E 70 on the P R promoter in the presence of a subset of initiating nucleotides, was stable to a 0.8 M NaCl wash (16,18). This stabilization is caused by the formation of RP init , which has incorporated a few nucleotides but has not left the promoter and is thus more resistant to 0.8 M NaCl than RP o . This 0.8 M NaCl challenge assay can be used to determine the relative extents of RP o and RP init formation. The E K derivatives were incubated with 32 P-labeled gerE promoter with and without NTPs (GTP, ATP, and CTP). The mixtures were filtered through nitrocellulose, and the filters were washed with a buffer containing either 0.1 or 0.8 M NaCl.
All four E K -DNA complexes were resistant to a 0.1 M NaCl wash both with and without the addition of NTPs (Table I). As expected, all of the complexes were destabilized when the filters were washed with 0.8 M NaCl. In the presence of NTPs, all four E K -DNA complexes became more resistant to the 0.8 M NaCl wash, indicating that they are forming RP init complexes (Table I) although with varying efficiency. Because RP o is in rapid equilibrium with RP init , a retention of approximately 50% of the input DNA is expected for those polymerases forming RP init efficiently. The resistance to 0.8 M NaCl was weakest for Epro-K -DNA complexes and strongest for Epro-K ⌬6-DNA complexes. About half of the Epro-K ⌬6-gerE complexes were resistant to the high salt wash, indicating equal distribution between RP o (not resistant) and RP init (resistant). Only 25% of the E K and Epro-K ⌬10 complexes were resistant to the 0.8 M NaCl wash, implying that the equilibrium favors RP o over RP init . For Epro-K , less than 10% of the complexes were resistant to the high salt wash. The results of the NTP stabilization assays are in good agreement with the transcriptional behavior of these enzymes.
The Pro Sequence Inhibits Core Binding by Pro-K -One explanation for weak protein-DNA interactions by pro-K is reduced ability to associate with the core subunits of RNA polymerase. To test this, we used the KMnO 4 footprinting assay at 250 mM KCl. We found that Epro-K ⌬6 forms KMnO 4sensitive complexes very well under these conditions, whereas Epro-K and E K do not. We reasoned that if pro-K or K could compete with pro-K ⌬6 for core association, then the amount of cleaved single-stranded DNA in a KMnO 4 footprint would decrease when either pro-K or K was added. On the other hand, if both failed to compete for core binding then cleavage of the DNA would be the same as when no competitor was added. Similar methodology has been used previously to assess core binding (16). Equimolar amounts of either K or pro-K were mixed with pro-K ⌬6 and then incubated with the core subunits. Holoenzyme mixtures were then added to 32 P-gerE promoter DNA and incubated for 30 min at 37°C. Samples were then subjected to KMnO 4 footprinting as described above. K was able to compete efficiently with pro-K ⌬6 for core binding as shown by a reduction in the amount of cleaved DNA by half compared with the amounts of cleaved DNA in Epro-K ⌬6 only samples (Fig.  6). This suggests that K and pro-K ⌬6 have a similar affinity for core. On the other hand, we found that pro-K was unable to compete with pro-K ⌬6. Cleavage of DNA in these competition assays was the same as that observed for pro-K ⌬6 alone. A concentration series with increasing ratios of pro-K to pro-K ⌬6 was performed to determine relative affinity of these sigmas for core. We found that pro-K was unable to compete with pro-K ⌬6 even when the ratio of pro-K to pro-K ⌬6 was 10:1 (data not shown).

DISCUSSION
Analysis of the mechanism of transcription initiation by E. coli RNA polymerase containing 70 has identified multiple intermediate steps in the process of open complex formation (22). The subunit has been implicated in recognition of the Ϫ10 and Ϫ35 promoter elements (3,5) as well as in isomerization and promoter melting (16, 30 -32). Homology among both the primary and alternative factors in regions 2 and 4, which recognize the Ϫ10 and Ϫ35 elements, suggests that their modes of DNA interaction may be similar (3). However, the amino terminus of 70 , which inhibits DNA binding by factor alone and is required for efficient open complex formation during initiation (13,16), is poorly conserved or absent in many alternative factors implying fundamental differences in some  6. KMnO 4 core binding competition assay. Equal amounts pro-K ⌬6 and either pro-K or K were mixed and allowed to compete for binding to a fixed amount of core RNA polymerase. Complexes formed on linear pgerE DNA and were probed for strand melting using KMnO 4 modification and piperidine cleavage at 250 mM KCl. The positions of cleavage within the promoter region are indicated by arrows. The bar plots indicate the fraction of the total input DNA cleaved. aspect of factor function and hinting at a role for the NH 2 terminus in specifying behavior that is characteristic of a particular factor. In this report, we find that the NH 2 -terminal pro sequence of K from B. subtilis controls core binding by pro-K and affects DNA interactions by Epro-K .
Characterization of Transcription Initiation in Vitro by E K -DNase I footprinting of E K on the gerE promoter demonstrated protection extending from Ϫ60 to ϩ20 relative to the start site of transcription. The extent of this footprint is similar for polymerases carrying other members of the 70 family (25,26,33). For E 70 on many promoters, these footprints are indicative of heparin-stable open complexes. The E K -gerE complexes were also open complexes as demonstrated by KMnO 4 footprinting, however they were unstable to low levels of heparin even in the presence of initiating nucleotides. Additionally, only 25% of E K -initiated complexes (RP init ) were stable to 0.8 M NaCl, whereas 50% of E 70 -P R complexes are typically stable (16,18). Taken together, our evidence suggests that at least two different E K -DNA complexes (RP o and RP init ) are generally less stable than their E 70 counterparts. Because K lacks region 1.1, its structure may be distinct from sigma factors that possess region 1.1, and this may alter the type or strength of interactions that K makes with the core subunits and with DNA during initiation. Because K -dependent transcription is utilized only transiently during sporulation, perhaps stability of the complexes is not crucial for adequate levels of mRNA synthesis. Dissociable open and initiated complexes have also been observed for B. subtilis E A at some promoters (34).
Role of the Pro Sequence of K in Controlling Activity-B. subtilis has evolved pro sequences to regulate the activity of two different factors, E and K . Both pro-E and pro-K are regulated developmentally by requiring proteolytic removal of the pro sequences for activation. In the case of E , the pro sequence is also required for accumulation of pro-E in vivo (35). This is not the case for pro-K , since K can be expressed stably without pro, and it retains full activity in vivo (10,12). This suggests that pro is not necessary for proper folding of K (35). Previous work had indicated that the pro sequence decreases the affinity of K for DNA (13). However, it was still unclear whether pro-K was able to bind to core, but the resulting holoenzyme was unable to recognize the promoter or whether the pro sequence was preventing core association.
Here we observed that the pro sequence affects at least three aspects of K activity: core binding, DNA binding, and DNA melting. Pro-K was unable to compete effectively with pro-K ⌬6 for binding to core RNA polymerase even when present in 10-fold excess. In the control experiment, K lacking the pro sequence was able to compete equally with pro-K ⌬6, despite slightly lower affinity of E K for DNA. Thus, we believe that this competition assay does reflect accurately the differential affinity of the factors for core. This corresponds with recent analysis of the distribution of pro-K and K in vivo, where the majority of pro-K fractionates with the cell membrane and not with the core subunits of polymerase, and the majority of K is associated with core. This work proposes that the pro sequence masks the core binding activity of K and targets pro-K to the membrane to interact with the processing machinery. 2 This is in agreement with the results presented here.
The pro sequence is known to negatively affect DNA binding by K in the absence of core (13). Here we see that whereas Epro-K and E K appear to have similar DNA binding affinity in DNase I footprinting experiments, Epro-K is defective in melting the DNA, even in the presence of NTPs, as judged by KMnO 4 susceptibility. Thus, although pro-K binds more weakly to core, once holoenzyme is formed it can bind to the promoter, but the equilibrium is shifted toward the closed complex.
The regions of factor implicated in core binding include region 2.1 of 70 (36), region 2.2 of 32 (37), and region 3.2 of 32 (38). Because K lacks region 3.2 (Fig. 1A), it seems likely that region 2 plays an important role in mediating core binding. One model to explain the function of the pro sequence of K in down-regulating K activity in vitro is that the pro sequence participates in intramolecular interactions with another region of the subunit. The hydrophobic nature of the pro sequence would suggest that it is not solvent-exposed and might interact with another hydrophobic region of K , which could mask the core binding domain and/or the DNA binding domain or more generally perturb the structure of region 2 to affect both core binding and DNA interactions. The pro sequence also appears to interact peripherally with some component of the membrane in vivo, and removal of pro may cause a conformational change that permits K to interact with core and prevents it from interacting with the membrane. 2 Pro-K ⌬6 and Pro-K ⌬10 -We constructed two K derivatives that retained only a portion of the pro sequence, pro-K ⌬6 and pro-K ⌬10. These deletion derivatives displayed some novel properties in DNA binding and transcription initiation. Most notably pro-K ⌬6 was hyperactive for DNA binding in the absence of the core subunits and showed transcriptional behavior similar to K at lower salt concentrations. This suggests that the entire pro sequence is necessary for inhibition of K activity and that removal of only a few amino acids is all that is required for activating pro-K . Epro-K ⌬6 and Epro-K ⌬10 both bind more tightly to DNA than Epro-K . This implies that either the lack of the first 6 -10 amino acids or the presence of the remaining 14 residues of the pro sequence can modify the interactions of holoenzyme with the promoter.
The altered interactions between Epro-K ⌬6 and the gerE promoter were most evident in transcription, where at 250 mM KCl a 34-fold increase in runoff transcripts was observed compared with E K . The nature of the interactions contributing to this effect was easily observed in DNase I footprinting. Pro-K ⌬6 appears to bind very tightly to the promoter, particularly in the upstream regions, at low concentrations of salt, and this may hinder initiation, which is consistent with the lower level of transcription observed at 100 mM KCl. As the salt concentration is increased, the Epro-K ⌬6-DNA footprint shifts forward toward ϩ20 relative to the ϩ1 start site, relinquishing upstream contacts and establishing downstream contacts. This repositioning of Epro-K ⌬6 on the promoter may permit more efficient transcription initiation, perhaps through an improvement in Ϫ10 interaction or an increase in the rate of promoter escape. Interestingly, Epro-K ⌬6 and Epro-K ⌬10 both caused a noticeable distortion in the promoter DNA between the Ϫ10 and Ϫ35 element, which is susceptible to KMnO 4 modification. Thus, the binding of these polymerases is strongly influencing the DNA structure in the promoter region. A similar distortion has been observed for E 70 at the rrnBP1 promoter (17).
In summary, B. subtilis has evolved a 20-amino acid pro sequence that functions to regulate K activity during the cascade of events leading to spore formation. From our experiments it is clear that removal of as few as 6 amino acids can activate K , yet the level of activity obtained for Epro-K ⌬6 may be inappropriate for the level of gene expression needed at this point in development. In the process of spore formation, the level and timing of gene expression are critical for spore maturation. From an evolutionary standpoint then, it appears 2 B. Zhang and L. Kroos, manuscript in preparation. that cleavage of pro-K to remove 20 amino acids results in an appropriately active version of K .