Article No. jmbi.1999.2704 available online at http://www.idealibrary.com on
J. Mol. Biol. (1999) 288, 539±553
Sequences in s N Determining Holoenzyme Formation and Properties MarõÂa-Trinidad Gallegos and Martin Buck* Department of Biology, Imperial College of Science Technology and Medicine, Sir Alexander Fleming Building, Imperial College Road, London SW7 2AZ, UK
Sigma subunits of bacterial RNA polymerases are closely involved in many steps of promoter-speci®c transcription initiation. Holoenzyme formed with the specialised minor sigma-N (sN) protein binds rare promoters in a transcriptionally inactive state and functions in enhancerdependent transcription. Using competition and dissociation assays, we show that sN-holoenzyme has a stability comparable to the major s70holoenzyme. Puri®ed partial sequences of sN were prepared and assayed for retention of core RNA polymerase binding activity. Two discrete fragments of sN which both bind the core but with signi®cantly different af®nities were identi®ed, demonstrating that the sN interface with core RNA polymerase is extensive. The low af®nity segment of sN included region I sequences, an amino terminal domain which mediates activator responsiveness and formation of open promoter complexes. The higher af®nity site lies within a 95 residue fragment of region III. We propose that the core to region I contact mediates properties of the sN-holoenzyme important for enhancer responsiveness. Heparin is shown to dissociate sN and core, indicating that disruption of the holoenzyme is involved in the heparin sensitivity of the sN closed complex. # 1999 Academic Press
*Corresponding author
Keywords: sigma 54; region I; RNA polymerase binding; heparin; sigma 70
Introduction Gene expression in prokaryotes is primarily controlled at the transcriptional level; in particular, RNA polymerase (RNAP) activity and its recognition of target promoter sites are strictly regulated. Bacterial RNA polymerases (a2bb0 ) initiate transcription speci®cally when they are in their holoenzyme form (a2bb0 s). The sigma factor (s) is central to the function of the RNA polymerase holoenzyme and the reversible binding of alternative s factors allows formation of different holoenzymes able to distinguish groups of promoters required for different cellular functions. In addition to double-stranded DNA promoter recognition and binding, s proteins are closely involved in promoter melting, inhibit non-speci®c initiation, are targets for activators, and control early transcription through promoter clearance and release from RNAP (Gribskov & Burgess, 1986; Gross et al., Abbreviations used: E, core RNA polymerase; Es, holoenzyme; RNAP, RNA polymerase; NTP, nucleoside triphosphate. E-mail address of the corresponding author:
[email protected] 0022-2836/99/190539±15 $30.00/0
1992; Helmann & Chamberlin, 1988; Lonetto et al., 1992). Most s factors are members of the s70 family of proteins (Lonetto et al., 1992). The only known sigma factor that is not a member of this family is sN. Its amino acid sequence has no obvious resemblance to that of s70 and the sN-holoenzyme requires enhancer-binding activator proteins and nucleoside triphosphate (NTP) hydrolysis for transcription activity (Austin & Dixon, 1992; Gralla, 1991; Kustu et al., 1989; Weiss et al., 1991). Despite the lack of any signi®cant sequence similarity, both types of sigma bind the same core RNAP, but produce holoenzymes with different properties. Relatively little is known about the sigma-core interface for either sigma type, but points of contact to the core are unlikely to be static; on the contrary, changing sigma interactions with the core will occur as each step of initiation is carried out. Competition occurs amongst sigma factors for a limiting amount of core RNAP, may provide the cell with a mechanism to regulate gene expression and may also impose limitations upon expression levels. Several studies have addressed the relative binding constants of alternative and primary sigma factors for core RNA polymerase (Chelm et al., # 1999 Academic Press
540 1982; Gill et al., 1991; Joo et al., 1997; Kusano et al., 1996; Williams et al., 1989), but not sN (Lesley et al., 1991). In principle, sN and s70 could share the same binding site on core RNA polymerase, have overlapping sites or utilise additional independent binding sites. In the ®rst two cases, sN and s70 would be expected to compete with each other for core binding. We have now assayed the af®nity of sN for core, individually and in the presence of s70, by directly measuring holoenzyme formation in vitro. The determinants of core binding in sN are of particular interest, since its binding to core produces a holoenzyme which binds promoters in a transcriptionally inactive state, unlike the s70holoenzyme. Domain analysis of sN from Klebsiella pneumoniae and Escherichia coli has demonstrated that core binding and DNA binding are separable activities residing primarily in different sequences
N Core Binding Sequences
(Cannon et al., 1995, 1997; Wong et al., 1994). The minimal binding fragment liberated by proteolysis (residues 107-303, Cannon et al., 1995) contained a short sequence (amino acid residues 175-182) with some similarity to a sequence in E. coli s70 (amino acid residues 380-387) implicated in core binding (Cannon et al., 1995; Malhotra et al., 1996; Tintut & Gralla, 1995; see Figure 1). Also, localised mutagenesis of E. coli sN and loss of core binding function analysis suggests that the similarity sequence contributes to core binding (Tintut et al., 1994; Tintut & Gralla, 1995). Protease footprints of K. pneumoniae sN have demonstrated that non-contiguous sequences outside the similarity region are protected by core, raising the possibility of extended contact and/or conformational change in sN upon holoenzyme formation (Casaz & Buck, 1997, 1999). One aim of our work was to use deletion analysis to determine if the core binding
Figure 1. K. pneumoniae sN (1-477) can be divided into three regions (I-III, Merrick, 1993). DNA binding functions and associated motifs, e.g. DNA cross-linking region (Cannon et al., 1994), helix-turn-helix motif (Merrick & Chambers, 1992) and the RpoN box (Taylor et al., 1996), reside in the C terminus, adjacent to sequences that modulate DNA binding (180-306) (Cannon et al., 1997). Activator responsiveness involves region I sequences (Sasse-Dwight & Gralla, 1990; Syed & Gralla, 1998; Cannon et al., 1999), and region II is variable. The sequence of the core binding fragment 70-324 (Cannon et al., 1995) is shown together with the secondary structure prediction (Rost & Sander, 1993) for that region ( & , helix; ^ ^ ^ , b-strand: Ð , loop; ., not assigned. The sN conserved residues are indicated in black (identical in at least 16/20 of the aligned sN sequences) or grey (similar in at least 16/20 sequences). The s70 380-387 residues similar to 175-182 sN sequence are also shown (Tintut et al., 1994).
N Core Binding Sequences
determinants of sN could be localised to one or more small fragments, addressing whether there are one or more local interacting surfaces that contribute to the binding of sN to core. Based on multiple alignment of known sN sequences and secondary structure predictions (see Figure 1), we have overproduced partial sN amino acid sequences, puri®ed them as native soluble peptides, and assayed them for binding to core in vitro. sN-holoenzyme binds to its promoters in a stable conformation to form a closed complex in the absence of other factors. The complex is disrupted by treatment with the polyanion heparin (Ninfa et al., 1987; Popham et al., 1989), which has a higher af®nity than does DNA for polymerase binding (Chamberlin, 1976a,b; Walter et al., 1967). Formation of heparin-resistant sN-holoenzyme promoter complexes (stable open complexes) is dependent upon activator and its associated NTP hydrolysis (Popham et al., 1989). We have examined the basis of the heparin sensitivity of the sNholoenzyme closed complex. The results of our investigations into three aspects of sN-holoenzyme formation show that (i) the stability of the sN-holoenzyme is similar to that of s70-holoenzyme and (ii) several non-contiguous sN sequences ¯anking a s70 similarity sequence (Tintut et al., 1994; Tintut & Gralla, 1995) contribute to the af®nity of sN for core. Notably region I sequences, which mediate activation and maintain the sN-holoenzyme in an inactive state (Cannon et al., 1999; Wang et al., 1997), interact with core. We conclude that the core interface of sN appears complex in terms of function: some sequences in sN primarily direct holoenzyme assembly, and others interact with core so as to modify its activity and cause enhancer dependence. Finally, (iii) we ®nd that the heparin instability of the sN-holoenzyme and closed promoter complexes can be accounted for by dissociation of sN from the core, implying that the heparin stability of the open promoter complex re¯ects a changed sN-core-DNA interaction.
541 band corresponding to the holoenzyme (H; Figure 2). Results show that the band corresponding to the sN-holoenzyme reaches a maximum intensity at a core to sN ratio of 1:1 (Figure 2(a)). Further additions of core resulted in the appearance of free core and additional bands of slower mobility (M). M bands are mainly free core, likely
Results We conducted experiments to measure sN-core interactions, determine the sequence requirements in sN for core binding and to examine the stability of the sN-holoenzyme in relation to the heparin resistance shown by activated holoenzyme promoter complexes but sensitivity of closed complexes. Binding of s N and s 70 to core RNAP Initially, we determined the fraction of sN and s able to bind to core by titrating increasing amounts of core with a ®xed amount of sigma. Sigma, core and holoenzyme were resolved using a native gel electrophoresis assay (Cannon et al., 1997). Each sigma bound the core RNAP as judged by depletion of the bands corresponding to core (E), and the appearance of a new faster migrating 70
Figure 2. Binding of sN and s70 to core RNAP (E). Native gel holoenzyme assembly assays were used to detect complexes forming between core and sN or s70 based on the different mobility of free core versus holoenzyme. (a) In order to titrate sN and s70 with core RNAP, a constant amount of sN or s70 (250 nM) was combined with increasing amounts of core RNAP (50, 125, 250, 500 and 1250 nM) at ratios 0.2:1, 0.5:1, 1:1, 2:1 and 5:1 core to sigma. (b) Titrations of core RNAP with sN and s70 were carried out using 250 nM core RNAP and increasing concentrations of sN or s70 (50, 125, 250, 500 and 1250 nM) at ratios 1:0.2, 1:0.5, 1:1, 1:2 and 1:5 core to sigma. H, holoenzyme, M, core-core and/or core-holoenzyme multimers, s, free sigma.
542 dimerised (Polyakov et al., 1995) but also include a small fraction of sN-holoenzyme (detected using [32P]sN, not shown). For s70, titration with core resulted in maximum formation of the holoenzyme at a ratio of 1:1 core to s70; further additions of core depleted the holoenzyme band, and slow running species formed (marked M). These likely include core and holoenzyme-core multimers (R. Burgess, personal communication). Titration of core with increasing amounts of sN and s70 (Figure 2(b)) showed that both sigma proteins formed a maximum amount of holoenzyme at a ratio of 1:2 core to sigma. At lower ratios (1:0.2, 1:0.5 and 1:1) sigma proteins did not saturate the core, and higher ratios (1:5) lead to the appearance of free sigma. We conclude that both sigmas are fully competent for core binding. Comparison of the sN and s70 titrations with core indicates that holoenzymes behave differently in the presence of excess core, given that the sN-holoenzyme appears less susceptible to association with free core than does s70holoenzyme. The sN and s70-holoenzymes appear to have different physical properties as judged by their interactions with core. Competition between s N and s 70 for binding core RNAP sN recognises rare promoters (probably less than 20 in the E. coli genome) and is a minor sigma factor (about 100 molecules per E. coli cell compared to 600-700 of s70; Jishage et al., 1996). The competition between sN and the major s70 for core was assessed to determine how effectively sN would form a holoenzyme. The Kd for the s70-core RNAP interaction is estimated at around 2 nM (Gill et al., 1991) and the intracellular concentration of sN is around 180 nM (Jishage et al., 1996). If the Kd of the sN-core complex was 100 nM, the binding of sN would be 50-fold weaker than that of s70, and the addition of s70 at equal concentration would largely prevent association of sN with core. We used 32P-end-labelled sN prepared using heart muscle kinase and g-ATP (Casaz & Buck, 1997) to detect complexes forming between core and sN, in the presence and absence of s70. Since puri®ed core RNAP is a weak substrate for heart muscle kinase (Severinova et al., 1996 and data not shown), holoenzyme containing 32P-end-labelled sN was excised from the native gel and analysed by SDS-PAGE. Only sN was detectably labelled (data not shown). Subsequently, a competition core binding assay was carried out by adding an equimolar mixture of s70 (5-40 nM) and sN (5-40 nM) to core (5-40 nM) to give a twofold molar excess of sigmas over core. We found that a substantial fraction of sN bound to core, and that the fraction bound was not diminished signi®cantly by s70 (Figure 3). It appears that the binding of sN to core is not especially weaker than that of s70; our inference that the Kd for the sN-core complex is in the low nM range is in accordance with estimations of
N Core Binding Sequences
Figure 3. Competition between sN and s70 for binding core RNAP. Native gel holoenzyme assembly assays were used to detect complexes forming between core and 32P-labelled sN-HMK based on the different mobility of free sN versus sN-holoenzyme. Complexes were formed by mixing 5-50 nM core RNAP and 5-50 nM 32 P-labelled sN-HMK (1:1) in the absence (&) and in the presence of 5-50 nM of s70 (*). Data from three different experiments were averaged and plotted to obtain the graph with the errors bars shown. PSL, photo stimulated light units from phosphorimager quanti®cation.
the promoter Kb for sN-holoenzyme (Popham et al., 1991). To estimate the stability of the sN-core complex with time, we conducted experiments in which unlabelled sN was mixed with core to allow holoenzyme formation, and then labelled sN was added to capture free core arising from dissociation of the cold complex. We detected little formation of the [32P]sN-holoenzyme, indicating that the sN-core complex dissociated slowly (Figure 4). Consistent with a stable association, when we prepared [32P]sN-holoenzyme and then added unlabelled sN, the labelled and unlabelled sN did not appear to exchange rapidly (data not shown). As control reaction, core was mixed with [32P]sN to make holoenzyme. The [32P]sN-holoenzyme formation slowly increased during the timescale of the assay, with a modest dissociation occurring in the ®rst ten minutes. The stability of the sN-core complex was not decreased by adding an equimolar amount of a 650 bp Sinorhizobium meliloti nifH promoter DNA fragment, but we were unable to determine whether the stability was further increased (data not shown). We found that there was a slightly more rapid formation of sNholoenzyme when a preformed s70-core complex was challenged with labelled sN (Figure 4).
N Core Binding Sequences
Figure 4. Stability of sN and s70-holoenzymes. To estimate the stability of the sN-holoenzyme complex, unlabelled sN (100 nM) was mixed with core (50 nM) to allow holoenzyme formation, and then labelled sN (50 nM) was added. Samples were taken at different times (0-50 minutes after the addition of labelled sN) and run in a native gel to measure 32P-labelled sNholoenzyme (*). To estimate the stability of the s70holoenzyme complex, an assay was conducted as above, but unlabelled s70 (100 nM) was mixed with core (50 nM) to allow holoenzyme formation, and then labelled sN (50 nM) was added ( & ). As a control, the same procedure was followed but mixing 50 nM core with 50 nM 32P-labelled sN (~). The graph shows the average percentage of labelled sN remaining in the complex with time from ®ve different experiments.
s N sequences that bind to core RNAP To identify the sN sequences important for interaction with core RNAP, a series of partial sN sequences were puri®ed and tested for retention of core binding activity. The deletion end points were chosen by inspection of sN conserved residues and secondary structure predictions. In the native gel holoenzyme binding assay, a constant amount of core RNAP was combined with increasing amounts of sN fragment to determine the approximate af®nity of the sequence for core RNAP. Representative results are shown in Figure 5 and summarised in Figure 6. Region III sequences bind to core RNAP We compared a series of C-terminally truncated fragments, with amino terminal His-tags, for core binding. The sN and His-tagged sN behaved similarly: both formed substantial amounts of holoenzyme at a ratio of 1:1 (core: sN), and completely titrated 250 nM core at a ratio of 1:2 (Figure 5(a)). The C-terminal deleted sN His-tag fragments 1-175, 1-324 and 1-306 ef®ciently formed holoen-
543 zyme, demonstrating that the C-terminal DNAbinding domain (amino acid residues 329-477) was not necessary for binding of sN to core (Figures 5(a) and 6). We were unable to detect core binding by Cterminal sequences 365-477 or 329-477 (not shown). Internal sN sequences 70-324, 70-268, 70-215, 70208 and 70-192 (Cannon et al., 1997 and this work) differing at their C-terminal ends, bound core with similar ef®ciencies. They titrated all the core at a ratio of 1:5, whereas 70-184, 70-180 and 70-175 did not fully titrate core even at a 1:20 ratio. In order to verify that the weak core RNAP binding of the 70-175 peptide (the shortest peptide of this series) did not depend on the His-tag, the tag was cleavaged with thrombin and the 17 residue tag eliminated. The 70-175 peptide lacking the His-tag behaved in the same manner as the tagged sequence. Clearly, sequences between residues 184 and 192 appear important for core binding and contribute positively to binding and they may account for the difference in binding seen when fragments 1-175 and 1-306 are compared (see Figure 6). To delineate a minimum fragment of sN able to bind to core RNAP, a further series of truncated proteins were generated starting at amino acid residues 52, 86, 112 and 120. The 52-324 fragment bound best and titrated all the core at a ratio of 1:2 (Figure 5(a)). The better binding of 52-324 versus 52-477 is discussed below in relation to region I core binding. The 86-324 peptide behaved similarly to the 70-192, 70-208 or 70-324 fragments and bound core very well. Unexpectedly, the 112-324 fragment bound core RNAP with a low degree of af®nity and at a ratio of 1:20 only a small fraction of the core RNAP was found in the holoenzyme complex (Figure 6). It is possible that the truncation to 112 results in a fragment conformation unfavourable for core binding, potentially related to promoting dissociation of sN from the holoenzyme during the sigma cycle. The 120-215 fragment was signi®cantly more effective for binding core and forming species with a mobility similar to the wild-type holoenzyme than the 120-324 sequence (Figure 5(a)). The 120-324 fragment predominantly formed a species with core of intermediate mobility (* in Figure 5(a)). The 120-184, 120-192 and 120-208 fragments detectably bound core RNAP only when added at ratios greater than 1:10. Sequences between 208 and 215 appeared to be important for core binding when sequences 70120 are missing. Removing sequences between 70 and 120 also seem to be associated with determining that the core forms a complex with a sigma fragment to give mobility * rather than H. We con®rmed that sigma fragments 120-215 or 120-324 were present in the complex noted * in Figure 5(a) by Western blot using an anti-His-tag antibody. The immunoblot showed that core RNAP did not react to the antibody, but that the bands marked * did (Figure 5(b)). Clearly, the 120215 and 120-324 fragments interact with core to produce two fast running species, the faster of
544
N Core Binding Sequences
Figure 5(a-c) (legend overleaf)
N Core Binding Sequences
545
Figure 5. (a) Holoenzyme gel assembly assay with region III sN fragments. A constant amount of core RNAP (250 nM) was combined with increasing amounts of sN fragment to determine the approximate af®nity of the sequence for core RNAP. The assay depends upon the native gel mobility difference between core RNAP along (E), sN-holoenzyme complexes (H) and free sN fragments. When increasing amounts of sN fragments are added to core RNAP, the bands corresponding to the core RNAP diminish and complexes with higher mobility, corresponding to sN-holoenzyme form. (b) Western blotting. A core binding assay was conducted as above, but the native gel was transferred onto a membrane. sN fragments were detected (using an anti-His-tag antibody) in two complexes (H and *) with different mobility. The unbound 70-215 and 120-215 fragments are not seen because they ran off the gel (see also (a)). (c) Binding of region IsN fragments to core RNAP. (d) Af®nity chromatography of core RNAP and sNholoenzyme on immobilised His6-sN fragments. Fragments 1-56, 1-71, 1-71(21-27) and 120-215 were bound to Ni(NTA) agarose spin columns. Core RNAP (E) or sN-holoenzyme (EsN) were applied, the ¯ow-through was collected, the column was washed twice, and bound proteins were eluted with Tris-NaCl buffer containing 200 mM imidazole. Core and sN peptides were detected after SDS-PAGE with Coomasie blue stain. A protein band (MW 66 kDa) appeared in several lanes (2, 6, 7, 8, 9), fainter in the starting material (lane 1), and may re¯ect break down of the core RNAP during the procedure. Controls with unloaded columns were run (lanes 2 and 3). sN-holoenzyme, 1-71, M, 120-215 (lanes 1, 12, 13 and 14, respectively) were used as markers. The 1-56 fragment ran off the gel.
which required higher fragment concentrations to be formed. Possibly more than one site on core binds 120-215 conformers to account for the concentration dependent formation of complexes H and * (Figure 5(a) and (b)). Region I sequences bind to core RNAP The 1-324 fragment had a higher af®nity for core than did the 70-324 (Figure 5(a)), as did 1-175 compared to 70-175 (Figure 6), indicating that the amino terminal sequences of sN between 1-69 could be involved in binding core. To study the interaction of sN amino terminal sequences with core, Histagged fragments comprising amino acid residues 1-51, 1-56, 1-63, 1-70, 1-71 and 1-71(21-27) (fragment 1-71 with an internal deletion between residues 21 and 27) were puri®ed. Each fragment was tested in two different core interaction assays: a native gel holoenzyme assembly assay and a nickel column af®nity method in which the His-tagged sigma fragment was immobilised, core passed over, and any interacting core co-eluted with the sigma fragment. In the native gel assay fragments 1-51, 1-56, 1-63, 1-70, 1-71 and 1-71(21-27) all changed
the mobility of the core. The 1-56 fragment bound best (Figure 5(c)). Because of the tendency of some of the N-terminal fragments to aggregate and remain in the well (see Figure 5(c) and 6), the observed binding to core probably re¯ects a lower limit of binding af®nity. Removal of sequences 2127 did not prevent core binding, suggesting that other determinants operate in core binding (compare 1-71 and 1-71(21-27), Figure 5(c)). The differences in core complex formation between the 1-71 and its internal 21-27 deletion variant was unexpected (compare amounts of * and H species, Figure 5(c)), and may be due to differences in fragment folding. However, we infer that residues 21-27 per se are not necessary for core binding by the 1-71 fragment. The spin column assay con®rmed that sequences 1-51, 1-56, 1-63, 1-70, 1-71 and 1-71(21-27) were able to bind to the core RNAP (Figure 5(d) and data not shown). In order to further analyse the role of the amino terminus of sN in core binding, we assayed sN fragments 52-324 and 52-477, which lack amino terminal sequences. The fragment 52-324 bound core as ef®ciently as fragment 1-324, titrating all
546
N Core Binding Sequences
Figure 6. Core binding data for the sN fragments. Large elements of sequence can be removed from either end of sN without destroying its ability to bind the core RNAP. The dominant binding determinants are in a 95 residue sequence (120-215) within region III. Amino acid residues near 180-190 make an important positive contribution to core binding activity, and several adjacent sequences further contribute. Fragments fully titrated core at ratios of E: sN 1:2 ( ), 1:5 ( ), 1:10 ( ); fragments which bound poorly at ratios >1:10 () and >1:20 () are also indicated. (1)We scored the 120-324 and 120-215 fragments equally for core binding, although the 120-324 produces a complex of different mobility to that of the 120-215 (see Figure 5(a) and (b)). (2)The binding af®nity of the amino terminal fragments 1-51, 1-63, 1-70, 1-71 and 1-71(21-27) was underestimated given their tendency to aggregate. & -, His6-tag.
the core at ratio 1:2 (Figure 5(a)). This is consistent with the weak binding observed with region I-containing fragments (Figure 5(c)). However, fragment 52-477 bound less well compared to fragment 1477 (Figure 6), likely re¯ecting a negative in¯uence of the C-terminal sequences upon core binding in the absence of region I. We have shown that the conformation of the DNA-binding domain (residues 329-477) in holoenzyme is in¯uenced by region I (Casaz & Buck, 1999). It is clear that removal of sequences 1-51 from some larger sigma fragments diminishes their binding to core, indicating a positive contribution of region I towards core binding, consistent with the direct binding of 1-56 sequence to core detected (Figure 5(c). Some interactions of region I sequences with core may be determined by the C-terminal DNA binding domain of sN. Using the af®nity spin column method we determined that amino terminal sequences of sN and
the 120-215 fragment interacted preferentially with core rather than holoenzyme (Figure 5(d)). The nickel column did not retain core or holoenzyme, but when preloaded with fragments 1-56, 1-71, 171(21-27) or 120-215, signi®cantly more core than holoenzyme was detected when proteins were eluted from the column, as judged by the presence of b, b0 and a subunits (Figure 5(d), compare EsN versus E lanes). These results imply that the sigma fragments tested preferentially interact with core rather than holoenzyme, indicating speci®city for core binding. Figure 6 summarises the core binding data for the sN fragments. Large elements of sequence can be removed from either end of sN without destroying its ability to bind the core RNAP and the dominant binding determinant (which includes residues that directly contact core) resides within a 95 amino acid sequence (120-215) of region III. The 120-215 amino acid sequence probably contains the
N Core Binding Sequences
547
Figure 7. Herparin challenge assay. (a) Samples of 250 nM core RNAP, 250 nM sN-holoenzyme (250 nM E: 500 nM sN) or 1.25 mM sN was challenged for ®ve minutes with heparin prior to gel loading nd stained with Coomassie blue stain. The band corresponding to the holoenzyme complex (H) was absent, and a higher mobility band (E-hep) and another corresponding to free sN were detected instead. M, core multimers. (b) Heparin challenge assay using 32P-endlabelled sN. 50 nM sN and 50 nM holoenzyme (50 nM E: 50 nM sN) containing end-labelled sN (EsN-32P) were incubated in the absence (lanes 1-4) or presence of 50 nM 88 mer homoduplex (lanes 5-10), challenged with heparin (100 mg/ml) for ®ve minutes where indicated and loaded onto native gels. PspFHTH (4 mM) and GTP (4 mM) were added for activation (lanes 7 and 8). H, holoenzyme; HDNA, holoenzyme-DNA complex. s, free sN. (c) Heparin challenge assay using labelled DNA. A 16 nM 88-mer homoduplex nifH promoter DNA was incubated alone (lanes 1 and 2) or with 250 nM core RNAP (lanes 3 and 4), 250 nM holoenzyme (250 nM E: 500 nM sN; lanes 5 and 6), 250 nM holoenzyme, 4 mM PspFHTH and 4 mM GTP (lanes 7 and 8), 500 nM sN (lanes 9 and 10) and 2 mM sN (lanes 11 and 12). A 2 mM sample of sN was used to detect binding to homoduplex DNA given the low af®nity of sN for linear DNA (Cannon et al., 1999). Reactions were challenged with heparin (100 mg/ml) for ®ve minutes where indicated, and loaded onto native gels. E-DNA, core RNAP-DNA complex; s-DNA, sN-DNA complex; H-DNA, holoenzyme-DNA complex; F, free DNA.
critical fold recognised by core. Amino acid residues between 184 and 192 make an important positive contribution to core binding activity, and several adjacent sequences further contribute to
core binding. Because the native gel mobility of the core-fragment complexes described above was similar to the sN-core complex, we infer that the 95 amino acid residue fragment includes a major
548 determinant in sN that modi®es core conformation upon holoenzyme formation. Heparin sensitivity of the s N-holoenzyme Resistance to heparin is widely used to distinguish stable open promoter complexes from unstable closed complexes. Early experiments demonstrated that s70-holoenzyme closed complexes and core-DNA complexes were sensitive to disruption by polyanions, including heparin, which can bind to the polynucleotide binding site(s) in the enzyme and inhibit its activity. Heparin has a higher af®nity for core RNAP and holoenzyme than does DNA. However, open promoter complexes were relatively resistant to attack by polymerase inhibitors (Chamberlin, 1976a,b; Walter et al., 1967; Zillig et al., 1971). The sN-holoenzyme closed complex is disrupted by treatment with heparin (Popham et al., 1989), even though sN binds to promoter DNA in the presence of heparin (Guo & Gralla, 1998; Oguiza & Buck, 1997; see also Figure 7). Formation of heparin resistant stable holoenzyme-open promoter complexes requires the presence of an activator and NTP hydrolysis in addition to sN-holoenzyme (Popham et al., 1989). In order to study the effect of heparin on the sNholoenzyme, holoenzyme was formed and then challenged with heparin prior to native gel electrophoresis, and proteins were detected with Coomassie stain (Figure 7(a)). We found that in the presence of heparin, core produced faster running species (E-hep; compare lanes 1 and 2). Also, the band corresponding to the sN-holoenzyme complex was lost, and a higher mobility band (E-hep) and one other corresponding to free sN were detected instead (lanes 3 and 4). Since the addition of heparin provoked the appearance of free sN, it was assumed that the new band of higher mobility was core bound to or modi®ed by heparin. Mobility of sN was not obviously heparin sensitive (compare lanes 5 and 6). To further explore the effect of heparin, we repeated the heparin challenge assay using 32P-end-labelled sN (Figure 7(b)). The addition of heparin to the 32P-end-labelled sNholoenzyme complex provoked the release of sN without affecting its mobility (compare lanes 3, 4 and 1, 2), consistent with the view that the new band (Figure 7(a), lanes 2 and 4) was a core-heparin complex. The presence of a 88-mer DNA fragment containing the R. meliloti nifH promoter sequence did not prevent release of [32P] sN from the holoenzyme by heparin (Figure 7(b), compare lanes 3, 4 and 5, 6) and the presence of activator (PspFHTH) and NTP was required to direct the formation of a heparin resistant complex Figure 7(b) and (c), compare lanes 5, 6 and 7, 8). We demonstrated directly that the binding of sigma to promoter DNA is heparin insensitive (Figure 7(c), lanes 6, 9-12) as suggested before (Guo & Gralla, 1998; Oguiza & Buck, 1997). We conclude that the heparin instability of sN-holoenzyme closed promoter complexes closely involves
N Core Binding Sequences
dissociation of sN from the core RNAP. The sNholoenzyme may be the most heparin sensitive component of the closed complex.
Discussion There are some studies on sequences in core that interact with sigma (Arthur & Burgess, 1998; Greiner et al., 1996; Owens et al., 1998), but relatively little is known about the sequences in sigma that constitute the interface with core RNAP. This interface represents an interacting surface through which s exerts many of its in¯uences upon RNAP function and activity. For sN, the interface is likely to be intimately involved in the activation mechanism, since sN converts core RNAP into a transcriptionally inactive enzyme able to occupy a sNdependent promoter without initiation occurring unless activated by specialised enhancer binding proteins. An extended core interface Our results allow us to propose that the core interface of sN comprises at least two functionally distinct sequences. One, within region III, is a 95 amino acid residue sequence (120-215) which binds core strongly and is likely to include most of the sN fold critical for forming a stable holoenzyme. The second sequence is within the ®rst 56 amino acid residues of sN and binds more weakly (see Figure 6). It is likely that one role of region II, located between region I and III, is to ensure that region I interacts appropriately with core to exert its effect upon holoenzyme function. One function of region I is to inhibit polymerase isomerisation (Cannon et al., 1999), a second is in allowing or mediating activation (Cannon et al., 1995; SasseDwight & Gralla, 1990; Syed & Gralla, 1998; Wang et al., 1997). In contrast, the higher af®nity 120-215 binding-determinant primarily directs formation of the holoenzyme (Sasse-Dwight & Gralla, 1990; Tintut & Gralla, 1995). Core binding by region I sequences 1-56 is consistent with sequence alignments, indicating that it is a discrete domain (i.e. that binding is not spurious as a result of ends of other domains being attached), and with protease footprints which showed that core protects residues 36 to 100 (Casaz & Buck, 1997), suggesting that sequences contacting core may be located between residues 36-56. The relatively low af®nity of region I for core compared to the 120-215 sequences is consistent with a need for a changing region I-core interaction during transcription activation, as suggested by protein footprints which detect changes in region I when closed and activated complexes are compared (Casaz & Buck, 1997). We favour a model in which some residues within 1-56 interact with core to regulate properties of the holoenzyme, and residues within 120-215 primarily anchor the core and sN together, at least in the initial closed complex. The failure of sigma sequences 160-290 to show a core-dependent foot-
N Core Binding Sequences
print has been suggested to be due to their being refractory to attack by the footprint reagent, but possibly some core contact is made to sequences 120-160. Conformational changes in sigma upon core binding may account for the protease footprint between residues 304 and 334, but additional secondary contacts are also possible (Casaz & Buck, 1999). Several lines of evidence suggest that the binding of the 120-215 fragment to core is via a determinant that operates in the intact sN. Firstly, the fragment includes sequences (amino acid residues 154-193) that when mutated, diminish holoenzyme formation in a closed promoter complex assay and result in a non-functional sigma in vivo (Tintut et al., 1994; Tintut & Gralla, 1995). There is a strong correlation between our positive binding data and the ``knockout'' mutant data. Secondly, some lager sigma fragments that have greater potential for non-speci®c binding generally failed to bind as well to core as do further truncated versions (see Figure 6). Thirdly, our minimal fragments of sigma that bind core lack an excess of acidic residues, a potential source of non-speci®c binding between sigma and core. Finally, immobilisation experiments (Figure 5(d)) showed that the sigma fragment bound more core than holoenzyme. The delineation of a minimal core binding fragment to residues 120-215 ®ts well with the prior demonstration that residues 216-306 are more closely associated with DNA binding than with core binding (Cannon et al., 1997). By dividing the K. pneumoniae sN at amino acid residue 215 it appears that core-binding associated interactions are located in the amino half, and DNA-binding associated interactions are located in the carboxy half. The direct association of region I, which mediates activation, with a core binding activity implies that activation of the sN-RNAP is likely achieved in part through modifying the region Icore RNAP interaction to allow holoenzyme to make the productive contact with the template DNA strand necessary for phosphodiester bond formation. The polymerase isomerisation envisaged as part of the activation process may also involve changes in those parts of sN associated with DNA contacts, but to date only evidence for a change occurring in region I has been obtained (Cannon et al., 1999; Casaz & Buck, 1997). Core binding sequences Previously, a proteolytic fragment of sN (amino acid residues 107-303) that bound the core and which included the most proteolytic resistant sN domain (amino acid residues 180-303) was identi®ed (Cannon et al., 1995; Casaz & Buck, 1997). The surface-exposed element around amino acid residue 180 has been studied by generating loss-of-function sN mutants defective in binding to core RNAP (Tintut & Gralla, 1995). Amongst our series of sN partial sequences, those that tightly bound core RNAP contained the 180 element, those that bound
549 weakly had this element at the C-terminal extreme (for example, 70-175, 70-180), and those that lacked detectable binding lacked the 180 element (365-477, 329-477, see also Cannon et al., 1997). We conclude that the sequences around residue 180 contribute to core binding of sN. However, deletion of sequences N and C terminal to the 180 element in¯uences binding either positively or negatively, indicating that adjacent sequences modulate core binding, possibly important for controlling holoenzyme assembly and its dissociation in the transcription cycle. Although sequences 120-184 and 120-192 bind core, they do so less well than the 120-215 fragment. The contribution of sequences 180-192 for core binding deduced from genetic experiments (Tintut & Gralla, 1995) may be dependent upon their being within, rather than at the end, a fragment. Binding of the 1-175 fragment to core likely involves region I sequences 1-56 and possibly the 36-140 sequence footprinted by core (Casaz & Buck, 1999). The acidic region II may also assist binding of this fragment (Cannon et al., 1995). Sequence comparisons of s70 and sN do not reveal any statistically signi®cant similarity between the two protein classes. However the short s70 similarity sequence of sN found between residues 175 and 189 and implicated in core binding by a loss of function analysis (Tintut & Gralla, 1995) lies within our minimal 120-215 core binding fragment. Whether the same fold is adopted at this point in the two sigma structures is dif®cult to predict, since the helices that contribute to the helix bundle that includes the putative s70 core binding surface are far apart in the primary sequence, being separated by a large non-conserved sequence (Malhotra et al., 1996). The sN core binding 120-215 fragment is within a conserved segment of sN separated from region I sequences by a non-conserved region II (Merrick, 1993). Region I could be in close physical proximity to the 120-215 sequence and could be part of an extended but physically localised core-interacting surface within the holoenzyme. Although the precise arrangement of structural features in sN that bind core cannot be predicted, it is clear that they conspire to form a binding surface that affords a binding energy which is not especially lower than that of s70 in in vitro assays. Differences exist between our results and those reported by Lesley et al. (1991) who did not detect stable sN-holoenzyme formation. Independent experiments have shown that the stability of the sN-holoenzyme is greatly increased by glycerol (D. J. Scott et al., unpublished results) so, the previous failure to detect stable binding may be due to the in vitro reaction and immunoprecipitation conditions. The dissociation experiments showed that sN is released slowly from core once bound. Therefore, the measured af®nity of sN for core is presumably suf®ciently high to enable the minor sN to effectively form holoenzyme in competition with other sigmas. After initial association of sN and core, a slow conformational change may serve to
550 stabilise sN-holoenzyme. Evidence for this has been obtained (D. J. Scott et al., unpublished results). It is possible that the slow dissociation of sN then re¯ects an unchanging interaction with the core that favours the closed complex. Closed complexes Two extreme states of the holoenzyme exist: that in the closed complex and that in the open complex, distinguished in part by the isomerisation of the polymerase that precedes or accompanies strand separation (Cannon et al., 1999). One interpretation of our demonstration that the sN-core interaction is heparin sensitive is that the heparin binding site present in the core RNAP is in some way in communication with a surface of sN that binds to core RNAP. Inaccessibility of the heparin site on core within open complexes would explain the apparent resistance of the open complex to heparin. The insensitivity of the sN-DNA complex towards heparin is consistent with the core being one target for heparin. Challenging a closed complex with heparin does not produce a new species corresponding to a core DNA complex, implying that a core-heparin interaction stops the core binding to DNA. In contrast, a sigma-DNA complex is seen when assays of binding are done with a heparin addition. However, our experiments do not address the order of disruption of the closed complex or whether a sN-heparin interaction is involved; only that two interactions are heparin unstable, the coresigma and core-DNA interactions, whereas the sigma-DNA interaction is relatively heparin stable. Interestingly, during the in vitro sigma cycle, the sN can remain bound at the promoter as core dissociates (Tintut et al., 1995). Overall, our results indicate that the core interface of sN appears extensive and likely contributes to the specialised properties of the inactive holoenzyme. The contacts of core with sN are not static, but during activation are modi®ed to allow the successive steps in the initiation pathway to occur (Casaz & Buck, 1997, 1999). An early conformational change is likely to allow polymerase isomerisation, and probably this step involves the region I interface with core (Cannon et al., 1999). Other changes may be those facilitating strand opening (Guo & Gralla, 1998) and maintenance of the opened DNA. Also, the sN interface is possibly in communication with sites in the core RNA polymerase that bind heparin, sites likely to be closely associated with the interaction that the core makes with DNA and nucleotides during initiation of transcription and formation of the ®rst few phosphodiester bonds in the nascent RNA.
Materials and Methods Plasmid constructions Deletion derivatives of the partial rpoN sequence in pMB28b:core (which directs the synthesis of the amino-
N Core Binding Sequences terminal 6-His-tagged 30 kDa core binding peptide (70324) of K. pneumoniae sN) were constructed as described (Cannon et al., 1997). rpoN sequences were ampli®ed using a T7 primer (50 -TAATACGACTCACTATA-30 ) to read through the NdeI site and a second primer designed to (i) truncate at amino acid residues 175, 180, 192 and 208 by inclusion of the UGA termination codon; and (ii) introduce a HindIII site downstream of the termination codon. Following PCR, the product of the reaction was restricted with NdeI and HindIII, gel-puri®ed and cloned into NdeI and HindIII restricted pET28b() (Novagen). Plasmids pMT70/175, pMT70/180, pMT70/192, and pMT70/208 directed the synthesis of amino-terminal 6-His-tagged polypeptides specifying residues 70-175, 70-180, 70-192 and 70-208 of sN. A derivative of rpoN encoding a fragment starting at amino acid 52 was constructed by PCR using plasmid pMB28b:core as template to introduce a NdeI site and an initiation codon at position 52 of sN. The ampli®ed DNA fragment was restricted with NdeI and HindIII and cloned into pET28b() to generate pMT52/324, which directed the synthesis of the amino-terminal 6-Histagged residues 52-324 sN polypeptide. A replacement of the BamHI-HindIII fragment of pMT52/324 by the BamHI-HindIII fragment of pMM70 (Merrick & Gibbins, 1985) produced pMT52/477, encoding the amino-terminal His-tagged 52-477 peptide. Derivatives of rpoN-encoding peptides that start at amino acid residues 86 and 112 were generated by cloning PvuII-HindIII and AccI (end-®lled by Klenow treatment)-HindIII fragments, respectively, from pMB28b:core into pET28b(), restricted with NheI (end®lled by Klenow treatment) and HindIII. Plasmids pMT86/324 and pMT112/324 encode the amino-terminal 6-His tagged polypeptides 86-324 and 112-324. Other deletion derivatives of rpoN were obtained by subloning the PstI-HindIII fragments from pMT70/184, pMT70/192, pMT70/208, pW301 and pMB28b:core (Cannon et al., 1997) into pQE9 (Qiagen). Plasmids pMT120/184, pMT120/192, pMT120/208, pMT120/215 and pW120/324 encode the amino-terminal 6-His-tagged polypeptides specifying residues 120-184, 120-215, 120192, 120-208 and 120-324 of sN. The gene fragment encoding the amino terminal region of sN was generated using PCR to introduce a NdeI site at the initiation codon of sN. The ampli®ed DNA fragment was restricted with NdeI and SalI and cloned into pET29a() to produce pMT-1/71, which directed the synthesis of the carboxy-terminal 6-Histagged residues 1-71 sN polypeptide. Deletion derivatives of pMT-1/71 were obtained by Sau3AI digestion to give pMT1/51 and pMT1/63 or PstI restriction to give pMT-1/71(21-27), encoding the carboxy-terminal 6-Histagged sN peptides 1.51, 1-63 and 1-71(21-27), a fragment with an internal deletion of amino acids 21-27, respectively. The NdeI-SalI fragment from pMT-1/71 was cloned into pET28b() to generate plasmid pMT1/71b, and into pMT70/175 to generate plasmid pMT1/175, which encodes the amino-terminal His-tagged fragment sN comprising amino acid residues 1-175. Other derivatives of rpoN were obtained by subcloning the SalI-HindIII fragment from pMKC8 (Cannon et al., 1997) or pMB28b:core into pMT1/71b to generate pMT1/306 and pMT1/324, respectively, which encode the aminoterminal His-tagged peptides 1-306 and 1-324. A replacement of the BamHI-HindIII fragment of pMT1/306 by the BamHI-HindIII fragment of pMM70 produced
551
N Core Binding Sequences pMTHsN, which encodes the amino-terminal His-tagged 1-477 peptide. The amino terminal His-tagged peptides 1-56 and 1-70 were generated by PCR. rpoN sequences were ampli®ed using a T7 primer to read through the NdeI site and a second primer designed to: (i) truncate at amino acid residues 56 and 70 by inclusion of an UGA termination codon; and (ii) introduce a SalI site downstream of the termination codon. Following PCR, the product of the reaction was restricted with NdeI and SalI, gel-puri®ed and cloned into NdeI and SalI restricted pET28b(). Plasmids pMT1/56 and pMT1/70 directed the synthesis of amino-terminal 6-His-tagged polypeptides specifying residues 1-56 and 1-70, respectively, of sN. Constructs were con®rmed by DNA sequencing. Protein overproduction and purification Puri®ed proteins and fragments are summarised in Figure 6. Plasmids directing overproduction of K. pneumoniae sN peptides were transformed into E. coli B834(DE3) for pET derivatives or JM109 for pQE9 derivatives. The cells were grown in one litre batches at 37 C in 2 YT medium with either 50 mg/ml kanamycin or 100 mg/ml ampicillin from a 20 % inoculum. The cultures were grown to an A600 between 0.6 and 0.8 and then induced with 1 mM IPTG. Cells were harvested after four hours induction at 30 C (25 C for 1-477 and 52-477), resuspended in 25 mM sodium phosphate (pH 7.0), 0.5 M NaCl, 5 % (v/v) glycerol and 1 mM PMSF and broken by treatment with 20 mg/ml of lysozyme and sonication. Following centrifugation at 20,000 g for 30 minutes, the sN peptides were found predominantly (more than 90 %) in the soluble fraction. The sN peptides were puri®ed by nickel af®nity chromatography and eluted with an imidazole gradient. Peak fractions were pooled and dialysed against TGED (50 mM Tris-HCl (pH 8.0), 5 % glycerol, 0.1 mM EDTA, 1 mM DTT) to which 50 % (v/v) glycerol and 50 mM NaCl was added and stored at ÿ70 C (long term storage) or ÿ20 C (shorter term storage). sN and sN-HMK were prepared as described previously (Cannon et al., 1995; Casaz & Buck, 1997). s70, a generous gift from J. Bown, was expressed and puri®ed as described (Igarashi & Ishihama, 1991). Puri®ed E. coli core RNA polymerase was from Epicentre Technologies. A puri®ed C-terminal delete form of activator, PspFHTH (Jovanovic et al., 1996) was used in activation assays. Protein concentrations were determined using the Bio-Rad Protein Assay kit. Core RNA polymerase binding assays Native gel holoenzyme assembly assays These were performed as described (Cannon et al., 1997), but in Tris-NaCl buffer (40 mM Tris-HCl (pH 8.0), 10 % (v/v) glycerol, 0.1 mM EDTA, 1 mM DTT, 100 mM NaCl). Brie¯y, E. coli core RNAP (250 nM) and different amounts of sN peptides were mixed together and incubated at 30 C for ten minutes, followed by the addition of glycerol bromophenol blue loading dye (®nal concentration 10 % glycerol). Samples were loaded onto native 4.5 % polyacrylamide Bio-Rad Mini-Protean II gels and run at 50 V for two hours at room temperature in Trisglycine buffer (25 mM Tris, 200 mM glycine buffer, pH 8.6). Complexes were visualised either by staining with Coomassie blue, phosphoroimaging or immunoblotting.
Nickel spin column affinity method All steps were performed at room temperature in TrisNaCl buffer without DTT. Initially, 400 pmol of 1-56, 2 nmol of 1-71 and 1-71(21-27) and 100 pmol of 120215 His6-sN peptides in 100 ml Tris-NaCl buffer were applied to a 50 ml Ni(NTA) agarose (Qiagen) spin column pre-equilibrated in the same buffer. After ®ve minutes of incubation, the column was centrifuged 720 g for one minute. Core RNAP (10 pmol in 50 ml Tris-NaCl buffer) or sN-holoenzyme (10 pmol core:20 pmol sN in 50 ml Tris-NaCl buffer) was applied to the column and incubated for ®ve minutes. The column was centrifuged (720 g for one minute) and the ¯ow-through saved. After two washes with 100 ml of Tris-NaCl buffer, bound proteins were eluted with 50 ml of Tris-NaCl buffer also containing 200 mM imidazole. Fractions were lyophilised, resuspended in 5 ml of water and analysed on a SDS15 % PAGE gel. Core and sN peptides were detected with a Coomasie blue stain. Sigma labelling with [-32P]ATP A 4 mg sample of sN-HMK (Casaz & Buck, 1997) was labelled suing 50 mCi (5000 Ci/mmole) [g-32P]ATP and one unit of heart muscle kinase (Sigma) for ten minutes at 30 C in 20 mM Tris (pH 7.5), 100 mM NaCl, 12 mM MgCl2, 4 mM DTT in a ®nal volume of 30 ml. Aliquots (25 ng/ml) of labelled protein were stored at ÿ70 C and used only once after thawing. Sigma competition and dissociation assays Competition reactions were carried out in Tris-NaCl buffer containing 200 mg/ml a-lactoalbumin at 30 C. Complexes were formed by mixing 5-50 nM core RNAP, 5-50 nM 32P-labelled sN-HMK and 5-50 nM of s70 (absent in the control reaction) at ratio 1:1:(1) in a total volume of 10 ml for ten minutes. The samples (3 ml) were loaded on a 4.5 % polyacrylamide native gel and electrophoresis was performed in 25 mM Tris, 200 mM glycine buffer (pH 8.6), at 50 V for two hours. For core-sN dissociation assays, complexes were performed in Tris-NaCl buffer containing 200 mg/ml a-lactoalbumin in a total volume of 10 ml. A 50 nM sample of core RNAP was mixed with 100 nM of sN, and after ten minutes of incubation at 30 C, 50 nM 32P-labelled sNHMK was added for 0-50 minutes and samples taken. Core-s70 dissociation assays were performed as above, but with 50 nM core RNAP and 100 nM s70 incubated together for ten minutes at 30 C, followed by the addition of 50 nM 32P-labelled sN-HMK into the s70-core binding reaction. The control reaction was carried out by mixing 50 nM core RNAP with 50 nM 32P-labelled sNHMK for up to 50 minutes. Samples of 3 ml were taken and loaded at the same time on a native gel which was run for two hours at 50 V. Phosphorimager analysis was used to quantify 32P-labelled sN-holoenzyme complexes and the free sN for each sample. Western blotting Native gels were electrotransferred to PVDF membranes (Bio-Rad) and the membranes were blocked for 30 minutes with 10 mM Tris-HCl (pH 8.0), 150 mM NaCl, 0.05 % Tween 20 (TBST) containing 5 % powdered milk. For detection of proteins recognised by an anti-Histag antibody (Pharmacia), monoclonal antibody was
552 diluted 1:30,000 in TBST and incubated with the membrane for one to two hours. After washing in TBST, membranes were incubated with antibody (alkaline phosphatase-conjugated anti-mouse IgG, Sigma) diluted 1:3000 in TBST for one hour. Immunoblots were washed again in TBST and then developed using nitro tetrazolium blue and 5-bromo-4-chloro-3-indolyl phsophate. DNA homoduplex formation Fully complementary 88-mer oligonucleotides (top strand labelled) for preparing short ÿ60 to 28 homoduplexes were annealed together as described (Cannon et al., 1999). Gel shift assays of promoter DNA A gel shift assay (Casaz & Buck, 1997) was employed to detect sN and its holoenzyme bound to a radioactively labelled R. meliloti nifH homoduplex promoter DNA fragment (substituted by unlabelled DNA as appropriate, see the legend to Figure 7). Typical holoenzyme interaction assays included 250 nM core RNA polymerase plus 500 nM sN (substituted by 50 nM 32P-end-labelled sN as appropriate, see the legend to Figure 5), and 16 nM homoduplex (50 nM when using end-labelled proteins) in STA buffer (25 mM Tris-acetate (pH 8.0), 8 mM Mg-acetate, 10 mM KCl, 1 mM DTT, 3.5 % (w/v) PEG 6000). If using labelled sN, 200 mg/ml a-lactoalbumin was added. For activation, 4 mM PspFHTH activator protein and 4 mM GTP (see the legend to Figure 7) were also added. Core RNA polymerase, sigma protein and DNA were pre-incubated at 30 C for ten minutes and then, if necessary, nucleotide and activator was added for ten minutes and, if required, heparin (®nal concentration 100 mg/ml). Samples were then loaded onto native 4.5 % polyacrylamide gels and run at 50 V for 80 minutes at room temperature in Tris-glycine buffer. DNA and proteins from protein-DNA complexes were detected by autoradiography.
Acknowledgements M.T.G. was supported by a Spanish Ministry fellowship and a EC TMR fellowship. Work was supported by a BBSRC grant to M.B. We thank X. Wang for the gift of several puri®ed proteins used in this work, Jon Bown for puri®ed E. coli s70 protein and R. Burgess for communicating data before publication.
References Arthur, T. M. & Burgess, R. R. (1998). Localization of a s70 binding site on the N terminus of the Escherichia coli RNA polymerase b0 subunit. J. Biol. Chem. 273, 31381-31387. Austin, S. & Dixon, R. A. (1992). The prokaryotic enhancer binding protein NTRC has ATPase activity which is phosphorylation and DNA dependent. EMBO J. 11, 2219-2228. Cannon, W., Claverie-Martin, F., Austin, S. & Buck, M. (1994). Identi®cation of a DNA-contacting surface in the transcription factor sigma-54. Mol. Microbiol. 11, 227-236. Cannon, W., Missailidis, S., Smith, C., Cottier, A., Austin, S., Moore, M. & Buck, M. (1995). Core RNA
N Core Binding Sequences polymerase and promoter DNA interactions of puri®ed domains of sN: bipartite functions. J. Mol. Biol. 248, 781-803. Cannon, W. V., Chaney, M. K., Wang, X. & Buck, M. (1997). Two domains within sigma N (sigma 54) cooperate for DNA binding. Proc. Natl Acad. Sci. USA, 94, 5006-5011. Cannon, W. V., Gallegos, M. T., Casaz, P. & Buck, M. (1999). Amino terminal sequences of sigma-N (54) inhibit RNA polymerase isomerisation. Genes Dev., 13, 357-370. Casaz, P. & Buck, M. (1997). Probing the assembly of transcription initiation complexes through changes in sigma N protease sensitivity. Proc. Natl Acad. Sci. USA, 94, 12145-12150. Casaz, P. & Buck, M. (1999). Region I modi®es DNA binding Domain conformation of sigma 54 within the holoenzyme. J. Mol. Biol. 285, 507-514. Chamberlin, M. J. (1976a). Interaction of RNA polymerase with the DNA template. In RNA Polymerase (Losick, R. & Chamberlin, M., eds), pp. 159-191, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Chamberlin, M. J. (1976b). RNA polymerase - an overview. In RNA Polymerase (Losick, R. & Chamberlin, M., eds), pp. 17-67, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Chelm, B. K., Duffy, J. J. & Geiduschek, E. P. (1982). Interaction of Bacillus subtilis RNA polymerase core with two speci®city-determining subunits. Competition between sigma and the SPO1 gene 28 protein. J. Biol. Chem. 257, 6501-6508. Gill, S. C., Weitzel, S. E. & von Hippel, P. H. (1991). Escherichia coli sigma 70 and NusA proteins. I. Binding interactions with core RNA polymerase in solution and within the transcription complex. J. Mol. Biol. 220, 307-324. Gralla, J. D. (1991). Transcriptional control-lessons from an E. coli promoter data base. Cell, 66, 415-418. Greiner, D. P., Hughes, K. A., Gunasekera, A. H. & Meares, C. F. (1996). Binding of the s70 protein to the core subunits of Escherichia coli RNA polymerase, studied by iron-EDTA protein footprinting. Proc. Natl Acad. Sci. USA, 93, 71-75. Gribskov, M. & Burgess, R. R. (1986). Sigma factors from E. coli, B. subtilis, phage SP01, and phage T4 are homologous proteins. Nucl. Acids Res. 14, 67456763. Gross, C. A., Lonetto, M. & Losick, R. (1992). Sigma factors. In Transcriptional Regulation (Yamamoto, K. & McKnight, S., eds), pp. 129-176, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Guo, Y. & Gralla, J. D. (1998). Promoter opening via a DNA fork junction binding activity. Proc. Natl Acad. Sci. USA, 95, 11655-11660. Helmann, J. D. & Chamberlin, M. J. (1988). Structure and function of bacterial sigma factors. Annu. Rev. Biochem. 57, 839-872. Igarashi, K. & Ishihama, A. (1991). Bipartite functional map of the E. coli RNA polymerase a subunit: involvement of the C-terminal region in transcription activation by cAMP-CRP. Cell, 65, 1015-1022. Jishage, M., Iwata, A., Ueda, S. & Ishihama, A. (1996). Regulation of RNA polymerase sigma subunit levels in Escherichia coli: intracellular levels of four species of sigma subunits under various growth conditions. J. Bacteriol. 178, 5447-5451. Joo, D. M., Ng, N. & Calendar, R. (1997). A s32 mutant with a single amino acid change in the highly
N Core Binding Sequences conserved region 2.2 exhibits reduced core RNA polymerase af®nity. Proc. Natl Acad. Sci. USA, 94, 4907-4912. Jovanovic, G., Weiner, L. & Model, P. (1996). Identi®cation, nucleotide sequence, and characterization of PspF, the transcriptional activator of the Escherichia coli stress-induced psp operon. J. Bacteriol. 178, 19361945. Kusano, S., Ding, Q., Fujita, N. & Ishihama, A. (1996). Promoter selectivity of Escherichia coli RNA polymerase Es70 and Es38 holoenzymes. Effect of DNA supercoiling. J. Biol. Chem. 271, 1998-2004. Kustu, S., Santero, E., Keener, J., Popham, D. & Weiss, D. (1989). Expression of sN (ntrA)-dependent genes is probably united by a common mechanism. Microbiol. Rev. 53, 367-376. Lesley, S. C., Brow, M. A. D. & Burgess, R. R. (1991). Use of in vitro protein synthesis from polymerase chain reaction-generated templates to study interaction of Escherichia coli transcription factors with core RNA polymerase and for epitope mapping of monoclonal antibodies. J. Biol. Chem. 266, 2632-2638. Lonetto, M., Gribskov, M. & Gross, C. A. (1992). The sigma 70 family: sequence conservation and evolutionary relationships. J. Bacteriol. 174, 3843-3849. Malhotra, A., Severinova, E. & Darst, S. A. (1996). Crystal structure of a s70 subunit fragment from E. coli RNA polymerase. Cell, 87, 127-136. Merrick, M. J. (1993). In a class of its own-the RNA polymerase sigma factor s54 (sN). Mol. Microbiol. 10, 903-909. Merrick, M. J. & Chambers, S. (1992). The helix-turnhelix motif of sigma 54 is involved in recognition of the ÿ13 promoter region. J. Bacteriol. 174, 7221-7226. Merrick, M. J. & Gibbins, J. R. (1985). The nucleotide sequence of the nitrogen-regulation gene ntrA of Klebsiella pneumoniae and comparison with conserved features in bacterial RNA polymerase sigma factors. Nucl. Acids Res. 13, 7607-7620. Ninfa, A. J., Reitzer, L. J. & Magasanik, B. (1987). Initiation of transcription at the bacterial glnAp2 promoter by puri®ed E. coli components is facilitated by enhancers. Cell, 50, 1039-1046. Oguiza, J. A. & Buck, M. (1997). DNA-binding domain mutants of sigma-N (sN, s54) defective between closed and stable open promoter complex formation. Mol. Microbiol. 26, 655-664. Owens, J. T., Miyake, R., Murakami, K., Chmura, A. J., Fujita, N., Ishihama, A. & Meares, C. F. (1998). Mapping the s70 subunit contact sites on Escherichia coli RNA polymerase with a s70-conjugated chemical protease. Proc. Natl Acad. Sci. USA, 95, 60216026. Polyakov, A., Severinova, E. & Darst, S. A. (1995). Three-dimensional structure of E. coli core RNA polymerase: promoter binding and elongation conformations of the enzyme. Cell, 83, 365-373. Popham, D., Szesto, D., Keener, J. & Kustu, S. (1989). Function of a bacterial activator protein that binds to transcriptional enhancers. Science, 243, 629-635.
553 Popham, D., Keener, J. & Kustu, S. (1991). Puri®cation of the alternative sigma factor, sigma 54, from Salmonella typhimurium and characterization of sigma 54-holoenzyme. J. Biol. Chem. 266, 19510-19518. Rost, B. & Sander, C. (1993). Prediction of protein structure at better than 70 % accuracy. J. Mol. Biol. 232, 584-599. Sasse-Dwight, S. & Gralla, J. D. (1990). Role of eukaryotic-type functional domains found in the prokaryotic enhancer receptor factor s54. Cell, 62, 945-954. Severinova, E., Severinov, K., FenyoÈ, D., Marr, M., Brody, E. N., Roberts, J. W., Chait, B. T. & Darst, S. A. (1996). Domain organization of the Escherichia coli RNA polymerase s70 subunit. J. Mol. Biol. 263, 637-647. Syed, A. & Gralla, J. D. (1998). Identi®cation of an Nterminal region of sigma 54 required for enhancer responsiveness. J. Bacteriol. 180, 5619-5625. Taylor, M., Butler, R., Chambers, S., Casimiro, M., Badii, F. & Merrick, M. J. (1996). The RpoN-box motif of the RNA polymerase sigma factor sN plays a role in promoter recognition. Mol. Microbiol. 22, 10451054. Tintut, Y. & Gralla, J. D. (1995). PCR mutagenesis identi®es a polymerase-binding sequences of sigma 54 that includes a sigma 70 homology region. J. Bacteriol. 177, 5818-5825. Tintut, Y., Wong, C., Jiang, Y., Hsieh, M. & Gralla, J. D. (1994). RNA polymerase binding using a strongly acidic hydrophobic-repeat region of sigma 54. Proc. Natl Acad. Sci. USA, 91, 2120-2124. Tintut, Y., Wang, J. T. & Gralla, J. D. (1995). A novel bacterial transcription cycle involving sigma 54. Genes Dev. 9, 2305-2313. Walter, G., Zillig, W., Palm, P. & Fuchs, E. (1967). Initiation of DNA-dependent RNA synthesis and the effect of heparin on RNA polymerase. Eur. J. Biochem. 3, 194-201. Wang, J. T., Syed, A. & Gralla, J. D. (1997). Multiple pathways to bypass the enhancer requirement of sigma 54 RNA polymerase: roles for DNA and protein determinants. Proc. Natl Acad. Sci. USA, 94, 9538-9543. Weiss, D. S., Batut, J., Klose, K. E., Keener, J. & Kustu, S. (1991). The phosphorylated form of the enhancerbinding protein NtrC has an ATPase activity that is essential for activation of transcription. Cell, 67, 155-167. Williams, K. P., Muller, R., Ruger, W. & Geiduschek, E. P. (1989). Overproduced bacteriophage T4 gene 33 protein binds RNA polymerase. J. Bacteriol. 171, 3579-3582. Wong, C., Tintut, Y. & Gralla, J. D. (1994). The domain structure of sigma 54 as determined by analysis of a set of deletion mutants. J. Mol. Biol. 236, 81-90. Zillig, W., Zechel, K., Rabussay, D., Schachner, M., Sethi, V. S., Palm, P., Heil, A. & Seifert, W. (1971). On the role of different subunits of DNA-dependent RNA polymerase from E. coli in the transcription process. Cold Spring Harbor Symp. Quant. Biol. 35, 47-58.
Edited by J. Karn (Received 28 January 1999; accepted 17 March 1999)