Mutational Analysis of β′260–309, a ς70 Binding Site Located on Escherichia coliCore RNA Polymerase*

In eubacteria, the ς subunit binds to the core RNA polymerase and directs transcription initiation from any of its cognate set of promoters. Previously, our laboratory defined a region of the β′ subunit that interacts with ς70 in vitro. This region of β′ contained heptad repeat motifs indicative of coiled coils. In this work, we used 10 single point mutations of the predicted coiled coils, located within residues 260–309 of β′, to look at disruption of the ς70-core interaction. Several of the mutants were defective for binding ς70 in vitro. Of these mutants, three (R275Q, E295K, and A302D) caused cells to be inviable in an in vivoassay in which the mutant β′ is the sole source of β′ subunit for the cell. All of the mutants were able to assemble into the core enzyme; however, R275Q, E295K, A302D were defective for Eς70 holoenzyme formation. Several of the mutants were also defective for holoenzyme assembly with various minor ς factors. In the recently published crystal structure of Thermus aquaticus core RNA polymerase (Zhang, G., Campbell, E. A., Minakhin, L., Richter, C., Severinov, K., and Darst, S. A. (1999)Cell 98, 811–824), the region homologous to β′260–309 of Escherichia coli forms a coiled coil. Modeling of our mutations onto that coiled coil places the most defective mutations on one face of the coiled coil.

The DNA-dependent RNA polymerase (RNAP) 1 is a multisubunit enzyme that plays a central role in eubacterial gene regulation and expression. This enzyme has two functional forms: core and holoenzyme. Transcription elongation and termination are performed by the core enzyme with ␣:2,␤:1,␤Ј:1 stoichiometry (1). Core RNAP binds one of a variety of subunit species at a given time to form a specific holoenzyme (2). The holoenzyme performs the tasks of promoter recognition and transcription initiation. Each subunit directs its cognate holoenzyme to start transcription from only those promoters containing DNA sequences specifically recognized by that factor (2)(3)(4).
Understanding the molecular basis of binding to core RNAP will aid in the analysis of several questions involving how multiple species come to sequester core and turn on subsequent genes. It has been hypothesized that all of the species bind to the same site/sites on the core enzyme (1,6,7). Studies to identify the core binding site on have resulted in the positive identification of a single binding site located on 70 overlapping conserved region 2.1 (5,8). A single point mutation in the homologous region of Bacillus subtilis E prevented binding to core (9). More recent genetic and biochemical studies suggest that region 2.1 may be only one of multiple contact sites that uses in binding to the core enzyme (6,10,11).
Information about sites on core that bind has come in most part from biochemical assays. Protein footprinting studies of the core enzymes susceptibility to hydroxy radical cleavage upon binding a modified factor containing the cleavage catalyst have revealed that there are three regions on the core, one located on ␤Ј and two on ␤, that are in close proximity to (12). One of the ␤ regions was identified earlier as being near the 70 binding region. It was noticed that this site, in the core enzyme complex, was cleaved by trypsin, whereas formation of E 70 prevented trypsin cleavage at this site (13). Previous in vitro work from our laboratory identified a strong binding site for 70 on the ␤Ј subunit (14). This site was mapped to within residues 260 -309 of ␤Ј.
The predicted secondary structure (15) of ␤Ј 260 -309 has two ␣ helices joined by a random coil. Another structural analysis program indicates that these two helices are amphipathic and have the potential for coiled coil formation (16). The coiled coil motif is based on a heptad repeat of residues designated a-g (17,18) (Fig. 1B). The a and d positions are hydrophobic, whereas the other positions are usually charged or polar. Burial of the a and d hydrophobic residues during coiled coil formation provides a large amount of the binding energy. Specificity in binding comes from the e and g positions, which can form ionic interactions or salt bridges.
We have undertaken a mutational analysis of this region to confirm that our in vitro binding results were relevant to in vivo binding and function. This work presents the analysis of 10 point mutations, most of which are change-of-charge mutations at the e and g residues, in the ␤Ј 260 -309 predicted coiled coil. Three of the mutations (R275Q, E295K, and A302D) were nonfunctional in binding 70 in all of the assays in which they were tested but still able to assemble into the core enzyme. We also report on mutations that were nonfunctional in some of our assays but functional in others, indicating that binding of other sites may compensate for loss of binding at the ␤Ј 260 -309 site. We use this analysis to demonstrate that the binding site identified previously by in vitro methods is important in vivo and that mutations in this region can greatly diminish core binding of 70 and other minor s. In the recently solved crystal structure for the core RNAP of Thermus aquaticus, the region homologous to Escherichia coli ␤Ј 260 -309 was determined to form a "coiled coil-like" structure (19) consistent with our predictions. Modeling of our mutations onto the T. aquaticus struc-ture places all of the nonfunctional mutations on the same face of the ␤Ј coiled coil.

EXPERIMENTAL PROCEDURES
Construction of Plasmids-Plasmid characteristics are described in Table I. Plasmids pTA577 and 600 -620 were made from the base plasmid pRL663 (20). Single HinDIII and BamHI restriction sites, at bases 674 and 952 of rpoC, respectively, were inserted into the rpoC gene of pRL663 via silent mutagenesis to create pTA577. pTA561 was created in the same manner as pTA577 except that pRL308 (22) was the starting plasmid. The HinDIII and BamHI restriction sites were used to insert polymerase chain reaction-generated DNA fragments containing the various mutations to generate pTA600 -609. For pTA620, which contains a truncated rpoC fragment coding for ␤Ј residues 1-319, pRL663 was cut with Xba-HinDIII for insertion of a polymerase chain reaction-generated rpoC truncation. The 70 binding site was previously mapped to residues 260 -309 of ␤Ј; however, we engineered some of the constructs for this work to extend to residue 319. This was done to incorporate the BamHI site mentioned previously. Therefore, the various mutations could be moved into the new plasmid to create pTA610 -619. We have not seen any difference in behavior of the fragments ending at residue 309 as opposed to those ending at residue 319. All sequences generated via polymerase chain reaction were sequenced to ensure that spurious mutations had not been incorporated.
Expression and Purification of 70 -The cells were grown to an A 600 of 0.6 -0.8 in 1-liter cultures at 37°C in LB medium with 100 g/ml ampicillin. Isopropyl-␤-D-thiogalactopyranoside was then added to a concentration of 1 mM. Three hours after induction, the cells were harvested by centrifugation at 8,000 ϫ g for 15 min and frozen at -20°C.
The cell pellet from a 1-liter culture was thawed and resuspended in 10 ml of lysis buffer (40 mM Tris, pH 7.9, 0.3 M KCl, 10 mM EDTA, and 0.1 mM phenylmethylsulfonyl fluoride), and lysozyme was added to 0.1 mg/ml. The cells were incubated on ice for 15 min and then sonicated in three 60-s bursts. The recombinant protein in the form of inclusion bodies was separated from the soluble lysate by centrifugation at 27,000 ϫ g for 15 min. The inclusion body pellet was resuspended by sonication in 10 ml of lysis buffer ϩ 2% (w/v) sodium deoxycholate. The mixture was centrifuged at 27,000 ϫ g for 15 min, and the supernatant was discarded. The deoxycholate-washed inclusion bodies were resuspended in 10 ml of deionized water and centrifuged at 27,000 ϫ g for 15 min. The water wash was repeated, and the inclusion bodies were aliquoted into 1-mg pellets and frozen at Ϫ20°C until use. 70 inclusion bodies (10 mg) were solubilized, refolded, and purified according to a variation of the procedure of Gribskov and Burgess (21). The inclusion bodies were solubilized by resuspension in 10 ml of 6 M guanidine-HCl. The proteins were allowed to refold by diluting the denaturant 64-fold with Buffer A (50 mM Tris, pH 7.9, 0.5 mM EDTA, and 5% (v/v) glycerol) in 2-fold steps over 2 h. One gram of resin (DEAE-cellulose, Whatman) was added and mixed with slow stirring for 24 h at 4°C. The resin was then collected in a 10-ml column and washed, and the protein was eluted with a gradient from 0.1 to 1.0 M NaCl in Buffer A. The 70 fractions were pooled, dialyzed overnight against 1 liter of storage buffer (50 mM Tris, pH 7.9, 0.5 mM EDTA, 0.1 M NaCl, 0.1 mM dithiothreitol, and 50% (v/v) glycerol), and stored at Ϫ20°C.
Quantitative Western Blotting-Protein samples to be quantitated were subjected to SDS-polyacrylamide gel electrophoresis. The proteins were electrophoretically transferred out of the gel onto 0.05-m nitrocellulose. The blot was blocked in Blotto (10 mM Tris, pH 7.4, 150 mM NaCl, 0.1% (v/v) Tween 20, and 3% (w/v) nonfat dry milk) and probed with monoclonal antibodies. The signal was generated using the ECL Plus system (Amersham Pharmacia Biotech) and detected on a Storm PhosphorImager (Molecular Dynamics). The signal was quantitated using ImageQuant software (Molecular Dynamics).
Far Western Blotting-Cells containing truncated ␤Ј expression plasmids pTA610 -620 were grown to A 600 ϭ 0.6 -0.8 and induced with 1 mM isopropyl-␤-D-thiogalactopyranoside. The cells were grown for an additional 30 min. A 200-l sample was removed and sonicated three times for 30 s each. 20 l of glycerol and 20 l of SDS sample buffer were added and heated for 2 min at 95°C, and the sample was then stored at -20°C until use. The lysates were separated by SDS-polyacrylamide gel electrophoresis. The proteins were electrophoretically transferred onto 0.05-m nitrocellulose. The nitrocellulose was blocked by incubating in HYB buffer (20 mM Hepes, pH 7.2, 200 mM KCl, 2 mM MgCl 2 , 0.1 M ZnCl 2 , 1 mM dithiothreitol, 0.5% (v/v) Tween 20, 1% (w/v) nonfat dry milk) for 16 h at 4°C. A chimeric 70 was created by fusing a heart muscle kinase recognition sequence to the N terminus of 70 to facilitate radiolabeling of 70 . Labeling of 70 was done in a 100-l reaction volume. 50 l of 2ϫ kinase buffer (40 mM Tris, pH 7.4, 200 mM NaCl, 24 mM MgCl 2 , 2 mM dithiothreitol) was added to 50 g of heart muscle kinase 70 protein. 240 units of cAMP-dependent kinase catalytic subunit (Promega) was added and the total volume was brought to 99 l with deionized water. One microliter of [␥-32 P]ATP (0.15 mCi/l) was added. The mixture was incubated at room temperature for 30 min. The reaction mixture was then loaded onto a Biospin-P6 column (Bio-Rad) preequilibrated with 1ϫ kinase buffer and centrifuged at 1100 ϫ g for 4 min. The flow-through was collected and stored at Ϫ20°C.
The blocked nitrocellulose was incubated in 10 ml of HYB buffer with 4 ϫ 10 5 cpm/ml 32 P-labeled 70 for 3 h at room temperature. The blot was washed three times with 10 ml of HYB buffer for 3 min each. The blot was dried, and the signal was visualized with a PhosphorImager and quantitated with ImageQuant software (Molecular Dynamics).
Growth Assessment-Plasmids pTA577 and 600 -609 (0.1 g) were transformed into strain RL602 (22,23). After heat shock and incubation on ice, 300 l of LB was added to the 50-l cell mixture. 10 l of the transformation reaction was spotted onto LB plates plus ampicillin (100 g/ml) and incubated at 30°C. Another 10 l was spotted onto identical plates and incubated at 42°C. The plates were incubated for 24 -48 h and assessed for growth.
Purification of Core/Holo Complexes-1-liter flasks containing 200 ml of LB with ampicillin (100 g/ml) and isopropyl-␤-D-thiogalactopyranoside (0.15 mM) were inoculated with 200 l from overnight cultures of cells containing plasmids pTA561, 577, and 600 -609. The cultures were grown at 37°C with shaking until the A 600 reached 0.4 for log phase assays and 2 h longer (A 600 ϳ 2.0) for the early stationary phase assays. The cells were harvested by centrifugation at 6,000 ϫ g for 10 min and stored at -20°C until use. The cell pellets were resuspended in 5 ml of TE (10 mM Tris, pH 7.9, and 0.1 mM EDTA) plus 0.15 M NaCl and lysozyme (0.1 mg/ml) and then incubated on ice for 15 min. The cells were sonicated three times for 30 s each and centrifuged for 25 min at 27,000 ϫ g to remove the insoluble material. The supernatant was loaded onto a 1.5-ml immunoaffinity column containing the polyolresponsive, anti-␤Ј monoclonal antibody NT73 (24). The column was washed with 15 ml of TE plus 0.15 M NaCl followed by a second wash with 10 ml of TE plus 0.5 M NaCl. The protein was eluted from the column with 4 ml of TE plus 0.7 M NaCl and 30% propylene glycol. The eluted sample (4 ml) was diluted with 6 ml of buffer 66 (20 mM Tris, pH 7.9, 500 mM NaCl, 5 mM imidazole, 0.1% (v/v) Tween 20, and 10% (v/v) glycerol) and loaded (2ϫ) onto 500 l of Ni 2ϩ -NTA resin. The resin was washed twice with 5 ml of buffer 66 and eluted with 0.5 ml of buffer 66 plus 0.25 M imidazole. Samples from the elution fractions were assayed by Western blot as described above, using monoclonal antibodies to each subunit or factor. The secondary antibodies were horseradish peroxidase-labeled goat anti-mouse IgG antibodies and the signal was generated using the ECL Plus substrate system (Amersham Pharmacia

RESULTS
Mutational Design-Structural prediction, using the Coils program (16), scored both of the predicted ␣-helices of ␤Ј 260 -309 as having a high probability of forming coiled coils (Fig. 1A). To test this prediction, we constructed two ␤Ј mutants with proline residues inserted into either helix. These ␤Ј mutants were no longer predicted to form helices or coiled coils. When assayed for function in both the far Western and in vivo growth assays, both mutants were found to be nonfunctional (data not shown). We took this to indicate that the helical/coiled coil structure in this region was important for function. The solubility of these mutant proteins was not 100%, so we ceased using them because their loss of function could simply be due to gross folding defects. We decided to concentrate in most part on the e and g positions of the ␣-helices for the next phase of our analysis. The e and g residues of coiled coils often engage in interhelical interactions, such as the formation of ionic interactions or salt bridges (17,18). Such interactions in this case could be intramolecular (between the two helices of ␤Ј 260 -309 ), forming a coiled coil structure necessary for binding by the subunit (Fig. 1B). Alternatively, the e and g residues of ␤Ј 260 -309 could be making intermolecular contacts with upon binding. Our efforts were directed toward making change-of-charge mutations at these residues of ␤Ј (Fig. 1B) and assaying their effects on binding. Two of the mutations described in this work do not involve e or g residues and were chosen for other reasons. Based on the findings that tyrosine and arginine residues are often located in "hot spots" of protein-protein interactions (26), we changed the tyrosine residue at position 269 to an alanine and arginine 297 to a serine. Previous studies in our laboratory found that insertion of a leucine at position 297 generated a ␤Ј subunit that was nonfunctional for binding 70 (unpublished results). Therefore, we were interested to deter-mine whether a less drastic mutation at this position would also affect binding.
Several of the Mutations in the ␤Ј 260 -319 Region Disrupt Interaction with 70 in a Far Western Assay-Previously, we had used far Western blotting to map a 70 binding site to the N-terminal region of the ␤Ј subunit (14). We again applied this procedure as an initial assay for functionality of our ␤Ј mutants. The mutations were cloned into a gene fragment coding for amino acids 1-319 of the ␤Ј subunit. Cells containing these genes were induced for a short period to give moderate levels of the ␤Ј fragment, comparable to other proteins in the extract. Samples were analyzed for binding 70 by far Western analysis as described under "Experimental Procedures." The amount of 70 probe bound by each ␤Ј 1-319 mutant fragment was compared with the amount bound by wt ␤Ј 1-319 fragment. Each signal was normalized to the amount of ␤Ј 1-319 contained in the supernatant as determined by Western blotting.
Five of the mutations (R275Q, R293Q, E295K, R297S, and A302D) were greatly reduced in their ability to bind 70 (Fig. 2). The Q300E and N309D mutations had the opposite effect, binding more 70 than wild type ␤Ј 1-319 . Q300E exhibited an increase in relative binding of greater than 7-fold. There were no effects on binding seen with the N266D, Y269A, or K280E mutations.
Complementation with Mutant ␤Ј Subunits-To assess the importance of the 70 binding site in vivo, we assayed the ability of mutant ␤Ј subunits to function as the sole source of ␤Ј for the cell. Plasmids containing mutant or wild type, fulllength ␤Ј were transformed into strain RL602 (22,23). The chromosomal rpoC gene of RL602 has an amber mutation that prevents functional ␤Ј from being produced in the absence of a suppressor tRNA. RL602 also has a chromosomal, temperature-sensitive amber suppressor. At the permissive temperature (30°C), the amber suppressor is active and allows chromosomal ␤Ј to be produced and the cell can grow. The amber Below the interaction domain are diagrams of the predicted ␣ helices and coiled coils. B, hypothetical helical wheel drawing of predicted ␤Ј 260 -309 coiled coil. The two predicted helices are shown as interacting with one another to form an antiparallel coiled coil. Mutations are shown next to original residues along with the residue number. The N terminus is at amino acid Asn-266 on the right helix. The right helix is drawn as coming out of the page, whereas the left helix is drawn as going into the page and terminates at Asn-309.
suppressor is not active at the nonpermissive temperature (42°C). Therefore, at 42°C, chromosomal ␤Ј is not made, and the cell cannot grow without another source of ␤Ј. If the plasmid-derived ␤Ј can complement the loss of ␤Ј, then the cells will grow and form colonies on plates at the nonpermissive temperature. If the mutant ␤Ј cannot complement, there will be no growth on the plates at this temperature.
Three of the ␤Ј mutants that were defective for binding in the far Western assay (R275Q, E295K, and A302D) could not support growth at the nonpermissive temperature, indicating that these mutations were also caused defects in binding in vivo (Fig. 3). N266D, a mutation that had no detectable effect in the far Western assay, allowed some growth at the nonpermissive temperature but not enough to be considered wild type. In contrast, the R293Q and R297S mutations that did not bind 70 in the far Western assay could support growth in vivo. Mutations Y269A, K280E, Q300E, and N309D had no detectable effects on growth. Expression levels for nonfunctional ␤Ј mutants were determined to be equivalent to that of plasmidderived, wild type ␤Ј when grown at 37°C (data not shown).
Effects of ␤Ј Mutations on Core/Holoenzyme Assembly-An alternate explanation for the inviability caused by some of the ␤Ј mutations would be that they are no longer able to be assembled into the core enzyme. To evaluate the potential assembly defects caused by the various mutations, we expressed His 6 -tagged, mutant ␤Ј subunits in cells that were also expressing wild type, chromosomal ␤Ј proteins. We used a Ni 2ϩ -NTA mediated pull-out assay to purify the mutant ␤ subunits together with associated cell proteins. An immunoaf-finity column was used to clean up the samples in order to reduce any nonspecific binding to the Ni 2ϩ -NTA column.
All of the mutant ␤Ј subunits tested retained the ability to assemble into the core enzyme demonstrated by the association of the ␣ and ␤ subunits throughout the purification (Fig. 4, A  and B). Again, mutations R275Q, E295K, and A302D caused defects in binding 70 in both log and stationary phase samples (Fig. 4C). Also reduced in E 70 formation were N266D in both log and stationary phase samples and R297S in log phase samples. Q300E again showed properties of binding 70 better than wild type. Y269A, K280E, R293Q, and N309D had no detectable effect on E 70 assembly. When a non-His 6 -tagged ␤Ј was expressed from the plasmid, there was no detectable nonspecific binding to the Ni 2ϩ -NTA column.
All of the sample eluates were also assayed for the presence of any minor species. The only minor s of which the concentrations were sufficient for detection were 32 in log phase and 32 and F in stationary phase samples. The results for these s were essentially the same as for 70 with the exception of mutants R297S and Q300E. In stationary phase samples from the Q300E mutant, the 32 and F levels are greatly reduced, whereas the 70 levels are above those in the wt. The log phase samples for this mutant also contained a decreased amount of 32 indicating a defect in E 32 formation but not as severe as in stationary phase.  sequence conservation with its T. aquaticus homolog (Fig. 5A). 2 This region of the T. aquaticus ␤Ј subunit forms a coiled coillike structure. When the mutations studied here are modeled onto the T. aquaticus structure using the Rasmol program (25), those that are most defective in binding are grouped on one face of the coiled coil. Those that had defective phenotypes in some assays but not others are on the outer edges of this face. Mutations that had no detectable effects are clustered on the opposite face of the coiled coil, with the exception of N309D, which is located at the very C terminus of the coiled coil immediately next to the "rudder" (19) (Fig. 5, B and C). DISCUSSION Binding of various factors to the core polymerase is a major step in the process of global gene expression and regulation. It is not known whether this step is part of the regulation, via a competition for binding to a limited core population, or merely a straight-forward binding of free s to an excess of core (27)(28)(29)(30). If there is competition among populations of species for core binding, that competition may be influenced by binding specificity of the s. In light of the high sequence conservation of most species, it has been thought that all factors bind to the same location(s) on the core enzyme (1). We had previously identified a binding site for 70 in vitro (14). In this report, we verify the importance of this site for in vivo binding and function, identify important residues for binding, define a potential binding interface for the -core interaction, and show that this binding site is involved in binding at least some of the minor factors.
Our mutational analysis was designed to look for loss of 70 binding by targeting residues of the 260 -309 region of the ␤Ј subunit that were identified as occupying e or g positions in the FIG. 4. Assembly of core and/or holoenzyme. Cells grown with wt or mutant ␤Ј expression plasmids were harvested and subjected to purification to isolate the plasmid-derived, His 6 -tagged ␤Ј and any of its assembled complexes. Proteins from Ni 2ϩ -NTA-purified samples were separated via SDS-polyacrylamide gel electrophoresis and blotted to nitrocellulose. The blots were then probed with monoclonal antibodies against the indicated subunits. A, log phase samples. B, stationary phase samples. No His 6 , strain expressing plasmid-derived, wt ␤Ј without a hexahistidine tag. C, quantitation of relative 70 binding for the mutants versus wild type ␤Ј, normalized to the amount of the ␣ subunit retained (wt ϭ 1.0). Results are the average of three different experiments. Error bars represent standard deviation. D and E, log and stationary samples, respectively, probed for minor factors.
predicted coiled coil structure. Based on our results, we have divided the mutations into three groups: nonfunctional for binding in all assays tested, nonfunctional in some assays but not others, and functional in all assays tested. The first group contains the mutations R275Q, E295K, and A302D. These three mutations were nonfunctional for 70 binding in vitro and in vivo, indicating that they play a very important role in binding 70 . Arginine 275 is located near the C terminus of the first of the two putative helices, whereas glutamate 295 and alanine 302 are in the middle and near the C terminus of the second helix, respectively. This confirms what we had found with our earlier mapping work that both predicted helices of ␤Ј 260 -309 were involved in binding 70 . The mutations made at these residues were the only ones tested that could not support any detectable growth when the expression of the chromosomal ␤Ј subunit was turned off. This is significant in light of the fact that these mutant ␤Ј subunits, along with all those tested in this study, had no detectable defect in interactions with the ␣ and ␤ subunits necessary to form the core enzyme. We conclude from these results that there are no gross folding defects that are responsible for the lack of 70 binding.
It is possible that the local structure of these mutant proteins is disturbed. Sequence analysis of all 10 mutant subunits predicted no change in secondary structures as compared with the wild type protein (15,31). The A302D change would be the most likely, though, of the three group 1 mutations to be disturbing the local structure. This introduces a bulky charged side chain in place of a single methyl group. Also, based on the crystal structure of the T. aquaticus core RNAP (19), the Ala-302 ␣ carbon is directed more toward the opposite helix of the coiled coil than are the side chains of Arg-275 or Glu-295, which are solvent-exposed. If R275Q and E295K are not affecting the local ␤Ј structure, then most likely, the negative 70 binding properties are coming from steric hindrance, charge repulsion, or loss of a specific interaction with the subunit.
The group 2 mutants, N266D, R293Q, and R297S, are of particular interest because they seem to have some function depending on the assay in which they are analyzed. R293Q and R297S were not functional for in vitro 70 binding in far Western assays but could support growth and were able to form core enzymes that were capable of binding 70 , although R297S does cause a decrease in the binding efficiency of the mutant core enzyme in log phase. The differences in the in vivo and in vitro assay results for these mutants can be explained in multiple ways. First, a positive result from the far Western assay requires that a ␤Ј fragment (amino acids 1-319) refold the secondary structure needed to bind 70 , whereas part of the protein is immobilized on a membrane. Therefore, mutations found to cause defects may be introducing in vitro folding deficiencies. Secondly, the in vivo assays are analyzing binding to the multisubunit core enzyme and not just an individual subunit or fragment. A great deal of evidence has been reported suggesting multiple binding sites on core RNAP for the factor (6, 10 -12). Thus, loss of one of those sites may be compensated for by the remaining binding interactions. We believe that whereas R293Q and R297S mutations are disrupting binding to ␤Ј 260 -309 , they are not obstructing 70 from making its other contacts on core polymerase.
In contrast to the previous group 2 mutations, N266D had no effect on 70 binding to ␤Ј but caused reductions in E 70 formation comparable to group 1 mutations and had a weak growth deficiency. Asn-266 is located at the base of the coiled coil and, when mutated, could change the local structure. This change may be causing a shift in the orientation of the coiled coil with respect to the rest of the core enzyme. This would not affect the binding of 70 to the coiled coil but may disrupt other contacts normally made by 70 with core.
The group 3 mutants, Y269A, K280E, Q300E, and N309D,  (25). B, oriented looking down the center of the coiled coil, toward the polymerase. C, side view of the coiled coil. The mutations that were defective in all assays tested are colored green. Mutations that were defective in some assays, but not all, are colored cyan. Mutations that were always functional are colored purple. Rudder, colored maroon, is added to orient the structure (19).
were all fully functional, indicating that these residues are not making critical contacts with 70 . The Q300E change was rather interesting. This mutation seems to cause an increase in binding of 70 to ␤Ј. The large increase in relative binding seen in the far Western may not have been derived strictly from an increase in affinity of the mutant ␤Ј fragment for 70 . The ␤Ј fragment containing this mutation could be better suited than the wild type fragment to refold its native conformation while attached to the nitrocellulose. Therefore, a larger population of properly folded protein may exist to bind 70 . The increase in relative 70 binding by the Q300E ␤Ј mutant in the far Western was not as dramatic in vivo possibly due to the 70 -core interaction having a larger K eq than the 70 -␤Ј interaction. However, the assembly of E 70 for this mutant was still almost twice that of wild type. Inhibitors based on coiled coil interactions have proven to be useful in disrupting such processes as viral entry into cells and topoisomerase activity (32)(33)(34). Our laboratory has begun work to design inhibitors of the -core interaction for potential use as antibacterial therapeutics. The Q300E mutation may provide useful information on increasing the binding constant of such an inhibitor.
The alternative factors have been thought to bind to the same sites on core RNAP as 70 . Mutating conserved residues of different species will disrupt core binding (6). Traviglia et al. (7) used tethered Fe-EDTA cleavage to determine that several of the minor species of E. coli are in close proximity to the same regions of core RNAP as 70 within the E complex. We found that, at least for 32 and F , minor factors do bind one of the same sites on core as 70 . Although they are binding to the same site, there is some difference in the manner of binding. The Q300E mutation that increased binding of 70 had the opposite effect, especially in stationary phase, on 32 and F . R297S also had different binding properties for the factors. This mutation caused an increased binding of the minor s and reduced binding of 70 . It is interesting that these mutations both had opposite effects on 70 and the minor s, although only two minor s were at detectable levels. This suggests that changes in the local environment could favor or hinder minor binding as a whole as compared with 70 . The patterns of binding to the various ␤Ј mutants need to be determined for more of the minor s in order to substantiate this hypothesis.
Finally, the crystal structure of T. aquaticus core RNAP has been of great utility in trying to understand the results of the mutations. Based on the computer predictions and our mutational results, we could not have concluded that ␤Ј 260 -309 formed a coiled coil structure. However, combining this information with the T. aquaticus versus E. coli ␤Ј sequence alignment and the T. aquaticus crystal structure, it is clear that ␤Ј 260 -309 adopts a coiled coil conformation. Upon binding, though, it is not clear what structure this region takes on. Conserved region 2.1 of 70 , implicated in core binding (8), forms a coiled coil with region 1.2 in the crystal structure of the 70 protease-resistant domain (35). Also, a predicted coiled coil in 54 of E. coli has been found to be important in the -core interaction (36). These factor structures may be interacting with ␤Ј 260 -309 to form a four-helix coiled coil. It is also known that the factor undergoes a conformational change upon binding core (11,37,38). This may be caused by a rearrangement of the coiled coils to form new contacts (39,40).
From the clustering of the group 1 mutations on the same face of the coiled coil structure and the positioning of the group 2 mutations on the edges of the cluster and the group 3 members on the opposite side of the coiled coil, we conclude that we have defined the binding interface for 70 on ␤Ј. Recent work in our laboratory has localized the region of 70 that is interacting with ␤Ј 260 -309 to a peptide containing a portion of the noncon-served region and region 2.1 of the 70 subunit (41). ␤Ј region 198 -237 was identified by Brodolin et al. (42) as interacting with the nontemplate strand of the lacUV5 promoter, which also is known to be contacted by region 2.4 of 70 (43,44). We are now in a position to model the -core interaction using these results as landmarks.