Transcription Activation by the Bacteriophage Mu Mor Protein Requires the C-terminal Regions of Both α and σ70 Subunits of Escherichia coli RNA Polymerase

Middle transcription of bacteriophage Mu requires Escherichia coliRNA polymerase and a Mu-encoded protein, Mor. Consistent with these requirements, the middle promoter, Pm, has a −10 hexamer but lacks a recognizable −35 hexamer. Interactions between Mor and RNA polymerase were studied using in vitro transcription, DNase I footprinting, and the yeast interaction trap system. We observed reduced promoter activity in vitro using reconstituted RNA polymerases with C-terminal deletions in α or σ70. As predicted if α were binding to Pm, we detected a polymerase-dependent footprint in the −60 region. Reconstituted RNA polymerases containing Ala substitutions in the α C-terminal domain were used to assay Mor-dependent transcription from Pm in vitro. The D258A substitution and α deletion gave large reductions in activation, whereas the L262A, R265A, and N268A substitutions caused smaller reductions. The interaction trap assay revealed weak interactions between Mor and both α and σ70; consistent with a key role of α-D258, the D258A substitution abolished interaction, whereas the R265A substitution did not. We propose that: (i) α-D258 is a Mor “contact site”; and (ii) residues Leu-262, Arg-265, and Asn-268 indirectly affect Mor-polymerase interaction by stabilizing the ternary complex via α-DNA contact.

requires activation by Mor and is carried out by the E. coli RNAP holoenzyme containing 70 (4,7). The detailed mechanism by which this activation occurs remains unknown; for example, it might involve protein-protein interactions between Mor and RNA polymerase, conformational changes in the promoter DNA, or a combination of both.
Previous in vivo and in vitro footprinting analysis of P m revealed single-stranded bases resulting from distortion in the Ϫ33 region, close to the predicted interface between Mor and RNAP (8). The distortion was dependent on the presence of both Mor and RNAP in vitro and involved strand separation confined to positions Ϫ35 through Ϫ31, as inferred from sensitivity to KMnO 4 modification and Mung bean or S1 nuclease cleavage following modification with dimethyl sulfate. This unwinding was enhanced or abolished in Up or Down spacerregion mutants, respectively, indicating that it may play a role in the activation of transcription.
The middle promoter possesses characteristic features of a promoter under positive control (9) (Fig. 1); it has a recognizable Ϫ10 hexamer but lacks similarity to the canonical Ϫ35 hexamer (at most, a 2-base pair match to consensus at 16 -18-base pair spacing). Previous analyses demonstrated that Mor forms dimers in solution and recognizes an imperfect dyad-symmetry element centered at Ϫ43.5 (10). The position of the Mor binding site (11), which overlaps the region normally recognized by 70 region 4.2 (12), as well as the absence of the "extended Ϫ10" sequence (13) and Ϫ35 hexamer, lead to the hypothesis that Mor, similar to class II transcriptional activators such as PhoB, cI, and CRP, might use protein-protein interactions with the subunit to activate transcription (14,15). The CRP-dependent galP1 promoter is a particularly well characterized class II promoter (16). The C-terminal part of 70 is required for transcriptional activation of galP1 by CRP (17), with the critical residues located between amino acid positions 529 and 540. The ␣-C-terminal domain (␣CTD) is dispensable for activator function at galP1 (18); however, it apparently interacts with promoter DNA upstream of bound CRP (16) and could be specifically cross-linked to CRP upon the formation of an initiation complex (19). In the absence of CRP, RNAP binds to galP1 and is capable of significant transcription, perhaps due to the presence of the extended Ϫ10 sequence (17,18).
Protein-protein interactions are implicated in both positive and negative control of transcription in E. coli (14 -20). Contact sites for a number of transcriptional activators are located in the C-terminal parts of the ␣ and 70 subunits of polymerase close to or overlapping their DNA-binding regions (17,(21)(22)(23)(24). In this study, we analyzed interactions between Mor and the ␣ and 70 subunits of RNAP by in vitro transcription, DNase I footprinting, and a yeast interaction trap assay system (25). The results demonstrate that the C-terminal regions of both ␣ and 70 subunits play a significant role in P m activation and identify the Asp-258 residue of the ␣ subunit as a primary contact site for Mor.

EXPERIMENTAL PROCEDURES
Media, Strains, Chemicals, and Enzymes-E. coli strain JM109 (mcrA ⌬pro-lac thi gyrA96 endA1 hsdR17 relA1 supE44 recA/FЈ traD36 lacI Q lacZ⌬M15 proAB ϩ ), used as a host in plasmid construction and preparation, was propagated in LB (26) supplemented with 75 g/ml ampicillin (U. S. Biochemical Corp.). Radiolabeled compounds were purchased from DuPont NEN. Acrylamide, bisacrylamide, and TEMED were from Bio-Rad. Calf thymus DNA, tRNA, heparin, dimethyl sulfate, piperidine, and ammonium persulfate were purchased from Sigma; bovine serum albumin (transcription grade) and T4 polynucleotide kinase were from Promega Corp.; deoxynucleotide triphosphates, Sequenase 2.0, and labeling and termination mixes were from U. S. Biochemical Corp. Restriction enzymes, Taq polymerase, and T4 DNA ligase were from Boehringer Mannheim; DNase I was obtained from Worthington, and nucleotide triphosphates were from Pharmacia Biotech Inc. Additional reagents were from Sigma. All enzymes were used according to the manufacturer's instructions.
Proteins-Mor was purified as described previously (10). Wild-type and mutant ␣ subunits were overexpressed and purified as described previously (27,35); the mutant RNAP core enzymes containing mutant ␣ subunits were reconstituted and purified by the standard procedure (27); and the holoenzymes were reconstituted by mixing the core enzymes with a 4-fold molar excess of 70 subunit. The C-terminal truncated 70 subunit was purified as described previously (17). Wild-type RNAP used in DNase I footprinting was a gift from M. T. Record, Jr.
Plasmids and Plasmid Construction-Plasmids pEG202, pJG4-5, and pSH18-34 (28) were used for the interaction trap assay. Plasmids pHT f1␣265A and pHT f1␣258A (29) were used as templates for PCR amplification of the ␣CTD mutants. The "extended Ϫ10" promoter (13) construct pIA51 was made as follows. Oligonucleotides IRI68 (CCGA-AGCTTTCGTTGCGTTTGTTTGCACGAGCTCTATG) and IRI69 (GCC-GGATCCTTAGGAAATTATAACATAGAGCTCGTGCA) were annealed to each other, filled in with Taq polymerase, digested with HindIII and BamHI, and cloned into HindIII and BamHI sites of the pUC19-spf' vector (30). Plasmid pIA54 is a derivative of pKM90 (7) containing a XhoI site inserted immediately downstream of the BamHI site. Plasmid pIA89 contains the NdeI-XhoI mor gene fragment from pIA54, which was cloned between the EcoRI and XhoI sites of pJG4-5, along with an EcoRI-NdeI linker (top strand, AATTGGCTGGTGGTGCTGGAGC; bottom strand, TAGCTCCAGCACCACCAGCC) designed to retain the reading frame of the B42-Mor fusion. Plasmid pIA91 contains sequence encoding the C-terminal domain of the ␣ subunit of RNAP (amino acid residues 236 -329), which was PCR-amplified from genomic DNA of strain MH5385 (11) with oligonucleotides IRI104 (CTAGAATTCGAT-GTACGTCAGCCTGAA) and IRI105 (AGTCTCGAGCGGTTACTCGTC-AGCGAT) in a standard amplification (25 cycles of 40 s at 94°C, 40 s at 55°C, 40 s at 72°C; then followed by 7 min at 72°C); PCR product was purified using a Qia spin PCR purification kit (Qiagen), digested with EcoRI and XhoI, and cloned into similarly digested pEG202. Plasmid pIA93 contains sequence encoding the C-terminal part of the 70 subunit of RNAP (amino acid residues 530 -613), which was amplified as above with oligonucleotides IRI106 (TACGAATTCCTGCCGCTGGATT-CTGCGA) and IRI107 (ATCCTCGAGCGATTAATCGTCCAGGAA) and cloned into pEG202 using the EcoRI and XhoI restriction sites. The mutant ␣CTDs containing Ala substitutions at positions 265 and 258 were PCR-amplified using pHTf1␣265A and pHTf1␣258A plasmids as templates and cloned into pEG202 as described above, resulting in pIA121 and pIA123, respectively. The sequences of amplified fragments were confirmed by dideoxy sequencing analysis of the plasmid clones.
Interaction Trap Assays-Saccharomyces cerevisiae strain EGY48 (28) was transformed by standard methods (31) with plasmids expressing LexA-fusions and B42-fusions, together with the reporter plasmid pSH18 -34; cells were grown on CM (complete minimal) triple dropout plates (Ura Ϫ His Ϫ Trp Ϫ ) supplemented with 2% glucose (31). Liquid cultures for ␤-galactosidase assays were grown in CM triple dropout media supplemented with 2% total sugar (2% glucose or 1.5% raffinose ϩ 0.5% galactose) to A 600 of 0.5-1.0. Samples of 10 ml were added to 200 l of 1% cycloheximide on ice, and cells were collected by centrifugation at 7000 ϫ g for 10 min and resuspended in 1 ml of buffer Z (31). Cells were made permeable by the addition of 50 l each of 0.1% SDS and chloroform and vortexing for 20s; then cells were diluted 10-fold with buffer Z. Samples (1 ml) were preequilibrated at 30°C; assays were initiated by the addition of 200 l of o-nitrophenyl-␤-galactopyranoside (4 mg/ml in H 2 O) and terminated by the addition of 500 l of 1 M Na 2 CO 3 . Activities were determined using the standard procedure (31).
DNase I Footprinting-Linear DNA fragments containing P m sequences from Ϫ115 to ϩ71 were PCR-amplified from pMK100 (11) in a standard reaction (10) using a combination of one unlabeled primer (either top: MLK12, Ϫ115 to Ϫ96; or bottom: MLK16, ϩ71 to ϩ52) and the second primer end-labeled with T4 polynucleotide kinase and [␥-32 P]ATP (3000 Ci/mmol). Fragments containing P m sequences from Ϫ62 to ϩ10 were made analogously from pIA14 (10) using primers IRI21 and IRI22 (10). Complexes were formed for 30 min at 30°C using purified Mor (1 g), RNAP (8 g), and linear DNA fragment (20 ng) in buffer containing 25 mM Tris-HCl, pH 7.9, 50 mM NaCl, 6 mM MgCl 2 , 0.5 mM EDTA, 0.5 mM dithiothreitol, 10% glycerol, 2% polyvinyl alcohol, and 10 ng/l of carrier calf thymus DNA in a 50-l volume. Then 10 ng of DNase I in 50 l of 5 mM CaCl 2 and 10 mM MgCl 2 were added, followed by incubation for 45 s at room temperature. Reactions were terminated by the addition of an equal volume of stop solution (200 mM NaCl, 10 mM EDTA, 1% SDS, and 250 g/ml tRNA), extracted with phenol-chloroform, and precipitated with ethanol. Pellets were washed with 70% ethanol, dried, dissolved in loading buffer, and analyzed on 6% sequencing gels (26). Markers were generated by Maxam-Gilbert sequencing reactions (26) of the same DNA fragments.

RESULTS
In Vitro Transcription with Reconstituted RNAP Holoenzymes Containing C-terminal Deletions of ␣ and 70 -To determine whether the C-terminal regions of the ␣ and 70 subunits of RNAP are involved in middle promoter function, we assayed the ability of reconstituted holoenzymes containing C-terminal deletions in the ␣ or 70 subunits (truncated at amino acids 235 and 529, respectively) to direct transcription from linear templates containing P m or the control promoter P RE# . Because the P RE# promoter contains the "extended Ϫ10" region (13), rendering it active in the absence of the C terminus of 70 , and lacks an UP element, making it insensitive to deletions of the FIG. 1. Middle promoter structure. The critical bases for Mor binding as defined by gel-shift analysis of promoter mutants (10) are located within the imperfect dyad-symmetry element shown by inverted arrows; the Mor DNase I footprint (11) is shown below the sequence. A box indicates the Ϫ10 hexamer; the predicted locations for the Ϫ35 hexamer and extended Ϫ10 region, which are lacking in P m , are underlined. The bent arrow designates the start site for transcription.
␣CTD, this promoter was used to determine the activity of the mutant enzymes and for normalization of activity at P m . The 70 deletion resulted in a dramatic reduction of P m activity relative to P RE# (Fig. 2); in fact, there was no transcript detectable. Deletion of the ␣CTD led to a less dramatic but still substantial (ϳ20-fold) loss of P m transcription. Because the C-terminal regions of ␣ and 70 contain not only activator contact sites but also the DNA-binding regions (17,18,(21)(22)(23)(24), this experiment did not distinguish whether the reduced ability of mutant holoenzymes to support transcriptional activation at P m was due to the loss of DNA binding or contact with Mor.
DNase I Footprinting-To determine the relative positions of Mor and RNAP bound at P m , we used DNase I footprinting with purified wild-type E. coli holoenzyme and purified Mor. Consistent with previous findings (11), Mor protected positions Ϫ56 to Ϫ33 from digestion by DNase I (Fig. 3). The addition of RNAP resulted in extension of the protected region downstream to position ϩ14, a protection pattern characteristic of open and intermediate complexes (33). On the top strand, position Ϫ25 remained sensitive to cleavage; the hypersensitive sites detected previously from Ϫ29 to Ϫ31, using crude extracts of a Mor-overproducing strain (11), were not observed here, raising the possibility that subtle differences in the proteins, the footprinting conditions, or the presence of host factors caused their appearance. The addition of RNAP also resulted in extended protection of positions Ϫ59 to Ϫ62 on the top strand upstream of bound Mor, while the intervening positions Ϫ57 and Ϫ58 remained accessible to cleavage. This pattern of upstream protection remained the same when both short (Fig. 3A) and long (Fig. 3, B and C) promoter fragments were used, suggesting that it does not result from binding of polymerase to the ends of the linear template. On the bottom strand, the effect of RNAP addition on cleavage in the Ϫ59 to Ϫ62 region was more subtle and appeared to be an enhancement rather than reduction in cleavage. This polymerase-dependent upstream protection could be most easily explained by binding of the ␣CTD, which is known to interact with DNA in this region in some promoters, especially those containing UP elements (16,24). Although single-point mutations in this region of P m do not confer a "down" phenotype in vivo (10), the region is AT-rich, as are UP-like elements (24), and might allow specific or nonspecific binding of the ␣CTD.
When binding of RNAP to P m was assayed in the absence of Mor, it did not result in a completely clear footprint, but there was weak protection in the region from Ϫ62 to Ϫ6, and several positions (Ϫ53, Ϫ51, Ϫ15, Ϫ12, ϩ11 on the top strand, and Ϫ47 on the bottom strand) were notably hypersensitive (Fig. 3). We are inclined to believe that this pattern results from specific rather than nonspecific binding of RNAP because we used an excess of competitor DNA and the same amount of RNAP as used for footprinting with Mor, and there was no binding to flanking sequences in the template.
Comparison of the footprints with RNAP alone to those with RNAP and Mor reveals that RNAP interacts with the region to be bound by Mor as well as several bases upstream. The addition of Mor causes dramatic changes in the footprint, clearly altering the association of RNAP with the DNA. Furthermore, the addition of Mor appears to shift polymerase-dependent upstream protection, perhaps by displacing the ␣CTD to a position farther upstream. A similar displacement has been seen at galP1 and shown to require the ␣CTD (16).
Effects of Ala Substitutions in the ␣CTD on P m Activation-The ␣CTD is known to comprise an independently folded protein domain containing two groups of amino acid residues implicated in UP element utilization and, therefore, DNA binding: 262-269 and 296 -299 (34,35). To ascertain whether the role of the ␣CTD at P m involves ␣-DNA interaction, ␣-Mor interaction, or both, we assayed P m transcription using reconstituted holoenzymes containing single Ala substitutions at positions 258 through 275 and positions 297 and 298. Because the activities of the reconstituted holoenzymes could vary, we determined the specific effect of substitutions in the ␣CTD on P m activation by comparison of the P m activity to that observed with the control promoter P RE# . The transcripts produced are shown in Fig. 4A, with the ratio of P m to P RE# promoter activity in Fig. 4B. Among the mutant enzymes tested, there were four that had a significant effect on P m activation, reducing transcription to less than one-half of that observed with wild-type enzyme; they contained substitutions D258A, L262A, R265A, and N268A. The D258A substitution had the greatest effect, resulting in a decrease in promoter activity almost as large as that occurring with the enzyme deleted for the entire ␣CTD. In previous experiments, the three other substitutions (L262A, R265A, and N268A) decreased transcription stimulation by the rrnBP1 UP element, suggesting that they may reduce ␣-DNA binding, but the D258A change did not (34,35). As a control, we tested our reconstituted holoenzymes with D258A, L262A, and R265A substitutions for transcription from an UP element-dependent but activator-independent form of rrnBP1, with results consistent with the previous findings (34,35); both L262A and R265A substitutions resulted in a significant loss of transcription, whereas the activity of the D258A mutant enzyme was not distinguishable from that of the wild-type. 2 Analysis of Mor-RNAP Interactions using the Interaction Trap Assay-To assay directly for Mor-␣ and Mor-interaction in the absence of ␣ and DNA binding, we used the interaction trap assay in yeast (Fig. 5) (25). In this approach, one protein (X) is fused to the DNA-binding domain of LexA protein (pEG202 vector); this LexA-X fusion is expressed constitutively in yeast cells. A second protein (Y) is fused to an acidic activation domain B42; the B42-Y fusion is under the control of the galactose-inducible GAL1 promoter. Expression of both chimeric proteins in a yeast cell containing a lacZ-reporter cloned downstream of one or more LexA binding sites results in the activation of lacZ expression if the chimeric proteins associate. Because the LexA protein fusion might activate transcription by itself, the interaction potential of a chimeric pair is usually estimated from the ratio of galactose-induced levels to noninduced levels of ␤-galactosidase activity. In this study a B42-Mor fusion and several pEG202-derived LexA fusions containing the C-terminal regions of ␣ (either wild-type or with Ala substitutions) or 70 subunits of RNAP (Fig. 5) were expressed and assayed for their ability to activate transcription of the lacZ gene cloned downstream of eight tandem LexA operators (pSH18-34). The ␤-galactosidase values measured for cells grown in galactose ϩ raffinose (conditions inducing B42-Mor expression) or glucose (noninducing) supplemented media, as well as the ratio of those values are presented in Fig. 5. Three conclusions regarding Mor-␣ and Mor-interactions can be drawn from these experiments: (i) transcription increased above the LexA-70 fusion background upon induction of the B42-Mor fusion, indicating weak interaction between Mor and 70 ; (ii) the combination of LexA-␣ and B42-Mor also resulted in a modest but reproducible enhancement of lacZ expression; (iii) the Ala substitution at position 258 of ␣ abolished this effect, whereas the R265A substitution did not. In addition, we found that the LexA-70 fusion activated transcription in the absence of Mor (with and without induction of the pJG4 -5 B42 vector plasmid). Transcriptional activity of the LexA-70 fusion and its apparent interaction with the acidic activation domain of B42 is consistent with the high degree of homology between 70 and eukaryotic general transcription factors (36,37) in regions required for their function. These factors were demonstrated to activate transcription when fused to LexA (38) and interact in vitro with a variety of acidic activation domains (39,40). DISCUSSION In this study, we observed a reduction in P m activity in vitro with reconstituted mutant RNA polymerases containing deletions of the C-terminal regions of either the ␣ or 70 subunits. The C-terminal region of 70 interacts with the Ϫ35 hexamer 2 I. Artsimovitch, unpublished observations.

FIG. 4. In vitro transcription assay with Mor and reconstituted RNAP holoenzymes containing Ala substitutions in the ␣CTD.
A, single-round transcription assays performed in the presence of Mor (5 pmol) and RNAP (1 pmol) using P m and P RE# templates. Reaction products were separated on 6% sequencing gels and visualized by autoradiography. B, a bar graph representing the results of the experiment shown in A expressed as a ratio of the P m to P RE# signal, which was quantitated by densitometry of three different exposures of the same gel using the ScanJet IIcx scanner (Hewlett Packard) and Scan Analysis Software (BioSoft, Cambridge, United Kingdom).

FIG. 5. Interaction trap assay.
This technique relies on the ability of two interacting proteins X and Y fused to separate DNA-binding (LexA) and activation (B42) domains to activate transcription of the reporter gene (lacZ) cloned downstream of the LexA operator sites and TATA box (25). The source and amino acid positions of protein fragments fused to the LexA DNA-binding and B42 activation domains are indicated below the plasmid names. The averaged results of three to six independent assays are presented in table form and expressed as ␤-galactosidase activities measured for cultures grown in inducing (galactose) conditions divided by the activities observed under noninducing (glucose) conditions. For ␤-galactosidase values greater than five units, the variation from the mean for independent assays ranged from 5 to 56% (average, 21%); similarly, the variation from the mean for calculated ratios from independent experiments ranged from 5 to 25% (average, 14%). DNA in typical activator-independent promoters (12); it also contains contact sites used by activators at class II activatordependent promoters to facilitate open complex formation (14,16,17,22,23). Typically, the ␣CTD interacts with the activator at class I promoters, facilitating recruitment of RNAP to the promoter (14,15).
The absence of both the Ϫ35 hexamer and "extended Ϫ10" sequence in P m is consistent with the total dependence of P m promoter activity on the presence of the activator protein Mor. One possible mechanism for Mor activation is that Mor could recruit RNAP to the promoter using protein-protein (Mor-␣ and/or Mor-) interactions (14,15); the results of the interaction trap assay would be consistent with this hypothesis. An alternative possibility is that RNAP can bind to P m to form a closed complex in the absence of Mor; Mor binding might then facilitate isomerization of this complex into a transcriptioncompetent open complex. The altered pattern of DNase I digestion of P m caused by the addition of RNAP alone would lend support to the second hypothesis; the weak protection and strong hypersensitive sites observed would be consistent with the formation of an unstable closed complex, which exists in rapid equilibrium with free RNAP (41). The properties of base substitutions in P m suggest that a flexible spacer is needed to facilitate interactions between RNAP and Mor; they also indicate that the Ϫ35 hexamer is irrelevant to P m promoter function; mutations increasing the fit to the Ϫ35 consensus did not result in increased promoter activity, and several decreased it (10). In contrast, mutations at positions Ϫ29 to Ϫ31 affected the DNA distortion observed at positions Ϫ32 to Ϫ34; mutations that caused Up or Down phenotypes showed increased or decreased distortion, respectively. Nevertheless, these findings do not rule out the possibility of a direct interaction between RNAP and bases in this region; the analysis of the effects of base substitutions on binding of RNAP to P m should be helpful in distinguishing these possibilities.
The results of in vitro transcription assays with reconstituted RNAP containing Ala substitutions in the ␣CTD revealed that four residues, Asp-258, Leu-262, Arg-265, and Asn-268, are critical for Mor-dependent activation. These residues are located relatively close to each other on one side of the ␣CTD (Fig. 6) and could constitute a contact surface for Mor. The residue Asp-258, located in the turn preceding the 260 -263 loop, is also involved in Fis-dependent activation at rrnBP1 (43). The other three, Leu-262, Arg-265, and Asn-268, are believed to be involved in DNA binding because they affect UP element utilization (34); these three residues are also essential for activation by OxyR (44) and CRP (35). Curiously, residue Cys-269, which is needed for UP element utilization (34,35), is not needed for activation of either P m or P lac (35). One possible model, proposed to explain the results from analysis of P lac activation by CRP and mutant RNAP holoenzymes (35), is that the same amino acid residues of the ␣CTD mediate mutually exclusive ␣-CRP and ␣-DNA binding. Although unusual, the existence of domains capable of both DNA binding and proteinprotein interactions is not without precedent. It was recently demonstrated that the zinc-finger domain of the transcription factor GATA-1, in addition to its well documented role in DNA binding, mediates self-association as well as heterotypic interactions with other GATA proteins and Krü ppel-type transcription factors (45,46).
In the case of P m activation, however, we prefer an alternative model in which Asp-258 serves as a specific Mor contact site, and the role of the remaining three residues is to stabilize Mor-␣ interaction by ␣CTD binding to DNA. Several arguments contribute to this preference: (i) the effect of the D258A substitution on activity was almost as large as the effect of deletion of the entire ␣CTD, whereas the other substitutions had lesser effects; (ii) because in galP1 the ␣CTD protects promoter sequences just upstream from CRP, despite the absence of an UP element (16), it appears that RNAP-CRP interactions are sufficient to position the ␣CTD close to the DNA. Thus, an RNAP-activator complex might be mutually stabilized by weak protein contacts and weak DNA binding. This model predicts that ␣ residues involved in DNA binding could affect activation solely due to the loss of favorable DNA interactions and that changes in these residues would cause a less severe reduction in activation than changes in amino acid residues that interact directly with activator; (iii) in the interaction trap assay system, ␣-D258 also played a key role in ␣CTD-Mor interaction; the D258A substitution abolished activation of reporter gene expression, whereas the R265A substitution, which dramatically reduces CRP-dependent activation of lacP1 (35) as well as transcription stimulation by the rrnBP1 UP element (34), had no effect. Nevertheless, since the interaction detected by this assay was weak, it is possible that stable association may require scaffolding by DNA, as suggested previously for CRP (19). Based on the calibration of the interaction trap assay with proteins of known affinities (28), our results suggest that Mor is capable of associating with and ␣ subunits in solution with affinities near the threshold of detection, K d ϳ10 Ϫ6 M, a value similar to that reported for interaction between CRP and RNAP in solution in the absence of their DNA-binding sites (47).
It seems reasonable to think that residues Arg-265 and Asn-268 mediate base-specific or nonspecific interactions with the DNA backbone, because arginine and asparagine are known to participate in such interactions (48). Since Leu-262 may comprise part of the ␣CTD hydrophobic core, the Leu to Ala substitution may lead to an altered conformation in which the presentation of residues directly contacting DNA and/or Mor is affected, resulting in reduced transcription activation. The ab-  (14). Mor is shown as a dimer bound to the DNA between the ␣CTD and 70 subunit of polymerase and making specific interactions (filled ellipse and rectangle) simultaneously with both. The positions of the single-stranded distortion (8) and melted Ϫ10 region are indicated with a diamond and a bubble, respectively. The second ␣CTD is represented by dashed lines to emphasize the uncertainty regarding its location. sence of a role at P m for residue Cys-269 suggests that it is a specific determinant for UP element recognition.
When both Mor and RNAP were used in DNase I footprinting experiments, the presence of RNAP caused an upstream extension of the protected region. The simplest hypothesis, and one consistent with the high AT content of this region, is that this protection is due to binding of the ␣CTD; alternative explanations include: (i) the presence of a MorϩRNAP-induced distortion, rendering DNA resistant to cleavage (compression of the minor groove); and (ii) extension or repositioning of the Mor-DNA contact in response to protein-protein interaction. The DNase I protection experiments with RNAP alone indicate that the enzyme is capable of binding to P m in the region from Ϫ62 to Ϫ6 in the absence of Mor. A similar pattern of protection was observed for me Ada-dependent promoters, aidB and ada, where RNAP apparently recognizes UP element-like sequences in the Ϫ40 to Ϫ60 region, largely overlapping the me Ada-binding site (49). At these promoters, binding of RNAP is not increased by me Ada; instead, the activator seems to function by facilitating contacts of already bound polymerase with core promoter elements at Ϫ35 and Ϫ10. Because P m does not have a Ϫ35 hexamer, it would be tempting to propose that the binding of polymerase to P m upstream of Ϫ10 is mediated by the ␣ subunit rather than the 70 subunit of RNAP.
Our results lead to the following model for interaction of Mor with RNAP during activation of P m transcription (Fig. 7). The central point of this model is that Mor bound as a dimer to a site centered at Ϫ43.5 interacts with both ␣ and 70 subunits of RNAP, making Mor the first reported class IϩII activator. Whether these multiple protein-protein (␣-Mor, -Mor, and Mor-Mor) and protein-DNA (␣-DNA, -DNA, and Mor-DNA) interactions occur simultaneously or sequentially during open complex formation and whether one Mor monomer interacts with ␣ and the other with 70 is not yet known. The torsional stress imposed on the spacer DNA by these contacts could lead to the DNA deformation observed at the predicted interface between the two proteins, with the energy stored in the distortion driving isomerization and strand opening. Whether the dependence of P m activation on the presence of the ␣CTD also reflects increased binding of RNAP to P m remains an open question. Assuming that the protection against DNase I cleavage upstream of Mor is a result of ␣ subunit binding, we propose that only one of the ␣CTDs interacts with P m in this region, because only four additional bases are protected by RNAP. The location of the second ␣CTD (indicated in Fig. 7, dashed line) is uncertain; it might: (i) interact with the opposite face of the DNA helix; (ii) participate in protein-protein contacts with RNAP, Mor, or some other host protein; or (iii) remain free of any interactions.