Identification of RNA polymerase beta' subunit segment contacting the melted region of the lacUV5 promoter.

Identification of the RNA polymerase functional regions involved in interactions with promoter is a basis for understanding the mechanism of transcription initiation. We have used formaldehyde cross-linking to identify a region of Escherichia coli RNA polymerase beta' subunit contacting lacUV5 promoter in open complex. Treatment of open complex with formaldehyde results in cross-linking of beta' and sigma(70) subunits at positions -5 and -3 on the nontemplate strand of the promoter DNA. These cross-links reflect specific interactions between RNA polymerase and promoter established in open complex. The positions of formaldehyde cross-links in the beta' subunit were mapped to the N-terminal segment (Cys(198)-Met(237)), which is contiguous to the evolutionary conserved region B. The proximity of the beta' and sigma cross-links suggest that the N-terminal region of the beta' subunit, interacting with single-stranded promoter DNA, can cooperate with the sigma subunit in the process of open complex formation.

Escherichia coli RNA polymerase (RNAP) 1 is a multisubunit enzyme consisting of the catalytic core (␣ 2 ␤␤Ј) and subunit. The ␤ and ␤Ј subunits contain several regions (from A to H in ␤Ј and from A to I in ␤ subunit) conserved between prokaryotic and eukaryotic RNAPs (1,2). This conservation permits a universal role for the homologous regions of the subunits in transcription.
Initiation of transcription by RNAP is a complex process including several steps (3,4). A critical step is formation of an "open" binary complex between RNAP and promoter accompanied by "melting" of promoter DNA around the transcription start site (3,5). The open complex is competent to initiate transcription upon the addition of nucleotide substrates. The driving forces and mechanism of the promoter melting are poorly understood. Genetic and biochemical studies have demonstrated that regions 2.3 and 2.4 of the subunit participate in recognition of promoter consensus sequence 5Ј-TATAAT-3Ј (Ϫ10 consensus) and promoter melting (6 -9). Recently, the locations of the subunit conservative regions over the promoter DNA were mapped using cystein-tethered chemical nu-clease (p-bromoacetamidobenzyl)-EDTA⅐Fe (10). It is evident that the ␤ and ␤Ј subunits cooperate with in the process of open complex formation (11)(12)(13)(14); however, which regions of ␤ and ␤Ј subunits are directly involved in this process is not shown. The available data indicate that the ␤Ј subunit is involved in interactions of RNAP with DNA. Using UV-induced cross-links, it has been shown that the evolutionarily conservative region A of ␤Ј containing the zinc finger participates in DNA binding and stabilization of the RNAP ternary complexes (15). Region C of the ␤Ј subunit shows some homology with the DNA-binding domain of DNA polymerase I (1) and exhibits nonspecific DNA binding capacity (16).
Cross-linking techniques are very effective tools for the identification of RNAP domains interacting with or positioned close to DNA in transcription complexes. Thus, the 70 subunit interactions with the bases of the promoter DNA in open promoter complex were determined using photochemical crosslinking techniques (17,18). Formaldehyde cross-linking links atoms at the 2 Å distance ("one atom" cross-linking) and permits analysis of a different set of DNA-protein contacts than "zero-length" UV cross-linking (19). Formaldehyde is widely used now for the in vivo studies of protein-DNA interactions in chromatin (20). Here, we demonstrated the possibility of using this technique for the mapping of the protein and DNA entities located in close contact in the RNAP-promoter complex. It was shown previously that two subunits of RNA polymerase could be cross-linked by formaldehyde to the nontemplate DNA strand of the lacUV5 promoter melted region in the open complex (21). We have identified these subunits as ␤Ј and . The subunit regions involved in the interactions with promoter DNA were extensively studied, whereas the corresponding regions of the catalytic RNAP subunits are not defined. Therefore, we have attempted to map the region of the ␤Ј subunit cross-linked to promoter DNA and have developed a new approach for mapping the DNA bases involved in formaldehyde cross-links using hydroxyl radicals. The methodology that we developed can be applied in studies of protein regions that interact with single-stranded DNA in different protein-DNA complexes.

EXPERIMENTAL PROCEDURES
DNA and Proteins-E. coli RNAP containing a His 6 tag in the ␤Ј subunit was isolated as described in Refs. 15 and 22. RNAPs containing His 6 tags in either the ␤Ј, ␤, or ␣ subunit were assembled from individual subunits (23). Recombinant RNAP containing a split ␤Ј subunit was prepared as described previously (24). The expression plasmid pET15, which expresses 70 bearing an N-terminal His 6 tag, was constructed by first destroying (by site-directed mutagenesis) the natural NdeI and XhoI sites in the cloned rpoD gene. The mutations did not change the protein sequence due to degeneracy of the genetic code. The rpoD gene was then amplified with primers containing artificial NdeI (5Ј primer) and XhoI (3Ј primer) sites and cloned in the appropriately treated pET15b (Novagene) plasmid. His 6 -tagged 70 was expressed in BL21 (DE3) cells and purified from inclusion bodies by renaturation from 6 M guanidine-HCl, followed by immobilized metal affinity chromatography and MonoQ chromatography. The specific activity (measured as the ability to support promoter-dependent transcription) of the tagged protein did not differ from that of control, untagged protein. The 123-and 167-base pair lacUV5 promoter fragments were isolated from pGEM1 plasmid as described in Ref. 21. DNA was 3Ј end labeled by Klenow fragment of DNA polymerase I, digested with HaeIII, and purified by gel electrophoresis.
Formaldehyde Cross-linking and Purification of Cross-linked Complexes-RNA polymerase (final concentration, 200 nM) and the lacUV5 promoter DNA fragment (final concentration, 25 nM) were mixed in 20 l of binding buffer (50 mM HEPES (pH 8.0), 50 mM NaCl, 5 mM MgCl 2 , 5% glycerol (XLB)) and incubated at 37°C for 5 min. Formaldehyde (prepared from 37% stock (Aldrich)) was added to a concentration of 20 mM. The cross-linking reaction was carried out for 30 s and stopped by the addition of equal volume of the stop buffer (2% SDS, 10 mM dithiothreitol, 10% glycerol, 125 mM Tris-HCl, pH 6.8). If the cross-linking was followed by fractionation on Ni 2ϩ -NTA agarose (Qiagen), the reaction was stopped by addition of 2 volumes of 10 M urea. 10 l of Ni 2ϩ -NTA agarose was added and incubated for 40 min at 25°C with gentle agitation. The resin was washed three times with 150 l of XLB containing 7 M urea. Bound cross-linking complexes were eluted in stop buffer and loaded on SDS-PAGE gels. Exonuclease mapping of protein-DNA cross-links was done as described (21) except that digestion was performed with 0.05-5 units of phage exonuclease (Life Technologies, Inc.) for 15 min at 30°C. DNA was eluted and loaded onto 6% sequencing gels.
Limited Chemical Cleavage of Cross-linked Complexes-Formaldehyde cross-linked complexes of ␤Ј or subunits and 3Ј end labeled lacUV5 promoter DNA were obtained after the cross-linking of 6 g of RNAP. For the cleavage at Cys residues the His-tagged ␤Ј-DNA crosslinked complexes were purified on Ni 2ϩ -NTA agarose and then treated with NTCBA (25) directly on the resin. Cleavage at Met residues with CNBr was performed as described (26), with the following modifications. The cross-linked complexes were separated by 5% SDS-PAGE, and gel slices containing radioactive bands were cut out and placed in Eppendorf tubes. The gel slices were washed with 1 ml of distilled water three times, and the cross-linked complexes were eluted from the gel for 1.5 h in 0.5% SDS and precipitated with 4 volumes of cold acetone. The pellet was then dissolved in 1% SDS, 50 mM HCOOH-NaOH buffer, pH 4.0. Then, CNBr was added to 50 mM and incubated for 5-10 min at 23°C. The products of all cleavage reactions were analyzed by SDS-PAGE.
Hydroxyl Radical Footprinting of Cross-linked Complexes-6 g of RNAP was cross-linked with lacUV5 promoter in 40 l of XLB, and then SDS was added to 0.5% to terminate the reaction. The samples were passed through a Chromo-Spin 30 column (CLONTECH), and the reaction volume was adjusted to 50 l with 0.5% SDS. Then, 6 l each of 1 mM Fe-EDTA, 20 mM ascorbate, and 0.6% H 2 O 2 (27) were added and incubated 4 min at room temperature. The reaction was quenched by the addition of glycerol to 10% and stop buffer, and the solution was immediately loaded on SDS-PAGE gels. The gels were autoradiographed, and the cross-linked complexes were cut off the gel. DNA was eluted and analyzed on an 8% sequencing gel.
KMnO 4 Probing of RNAP-Promoter Complexes-RNAP promoter complexes formed in 10 l of XLB were incubated at the desired temperature and treated with 5 mM KMnO 4 for 30 s. The reaction was stopped by the addition of 1.5 M potassium acetate and 1 M ␤-mercaptoethanol. DNA was ethanol-precipitated, treated with 0.5 M piperidine, and analyzed on 6% sequencing gel. 70 and ␤Ј Subunit to the lacUV5 Promoter-Treatment of RNAP-lacUV5 promoter open complex at 37°C with formaldehyde results in appearance of several protein-DNA cross-linked complexes having different electrophoretic mobilities. In a previous work (21), cross-linking was performed for a long time (10 min at 37°C) at 33 mM formaldehyde. This resulted in "multiple-hit" conditions and, as a consequence, formation of several cross-linking sites per molecule and multiprotein cross-linked complexes. In order to avoid these events in the current work, we reduced the time of crosslinking to 30 s at a 20 mM concentration of formaldehyde. Under these conditions, only two protein-DNA complexes, containing one cross-linked subunit per DNA fragment, were formed ( Fig. 1). Identification of the cross-linked species was performed by using Ni 2ϩ -NTA agarose purification. The complexes obtained after cross-linking of RNAP containing His 6 tag in the ␤, ␤Ј, or ␣ subunit were denatured and fractionated on the Ni 2ϩ -NTA agarose in the presence of 7 M urea ( Fig. 1). Only cross-linked protein-DNA complexes with slow mobility formed by RNAP containing the His 6 tag in ␤Ј subunit were effectively retained on the resin. Small amounts of the crosslinked DNA were retained on the resin when His 6 tag was located in other subunits of RNAP core. This can be attributed to nonspecific binding of denatured proteins to the resin. The result obtained shows directly that the slow mobility complex contains only the ␤Ј subunit cross-linked to DNA, whereas the ␤ and ␣ subunits were not cross-linked. Identification of the subunit was performed by using holoenzyme reconstituted from the His-tagged core (His 6  , and DNA was analyzed on a 6% sequencing gel. As a control, non-cross-linked DNA was treated under the same conditions. Label was at the 3Ј end of nontemplate (a and b) and template (c) DNA strands. A Maxam-Gilbert AϩG sequencing marker (AϩG) is shown.

Cross-linking of
subunit containing an N-terminal His 6 tag (Fig. 1). As a control, we used the His-tagged core reconstituted with the 70 subunit without His 6 tag. Fractionation of the complexes obtained after the cross-linking of the reconstituted holoenzyme containing His-tagged has shown that both ␤Ј and crosslinked complexes were retained on the resin, whereas only ␤Ј complex was retained on the resin if the subunit without His 6 tag was used for holoenzyme reconstitution. This gave direct proof that the faster mobility complex is formed by the 70 subunit cross-linked to DNA.
Formaldehyde-induced Cross-links Are Localized Only in the Melted Promoter Region-To map the linear arrangement of cross-links on the DNA, we used phage exonuclease, which degrades DNA in the 5Ј 3 3Ј direction and ceases the hydrolysis when it encounters a cross-linked protein (21). Crosslinked complexes of the ␤Ј or subunit and end-labeled promoter DNA were separated by SDS-PAGE, excised from the gel, and treated with exonuclease. Subsequently, cross-links were reversed by heating, and the DNA fragments were analyzed on a sequencing gel (Fig. 2). In the case of the ␤Ј subunit, the major exonuclease stops appeared at positions Ϫ8, Ϫ7, Ϫ6, and Ϫ5 of the nontemplate DNA strand, as was reported previously (21). In the case of the subunit, exonuclease terminates at positions Ϫ9, Ϫ8, and Ϫ7, which leads to a prediction that cross-linking sites may be located between positions Ϫ7 and Ϫ3. Moreover, we have observed exonuclease stops between positions Ϫ7 and ϩ15 on the template DNA strand for both subunits (Fig. 2c). Because these stops were dispersed over a wide region and were not observed previously, we doubt that they reflect cross-links with the template DNA strand. To test this, we performed titration of cross-linked complexes with different amounts of exonuclease (Fig. 2, b and c). The exonuclease stops at positions Ϫ8 to Ϫ5 of the nontemplate strand accumulated and finally were the only ones observed. Some minor exonuclease stops (for example, those observed at positions ϩ6 or Ϫ50) have disappeared. In contrast, the exonuclease stops on the template DNA strand progressively disappeared when the amount of exonuclease was increased. Therefore, we concluded that cross-links are localized only on nontemplate DNA strand, and the stopping of exonuclease on template strand was also due to these cross-links. Such behav-ior can be explained by structure of exonuclease, which holds both DNA strands during reaction (28). At a high concentration of exonuclease, the partially digested DNA duplexes can be destabilized, and the subsequently released template DNA strand can be completely digested with exonuclease, which can degrade single-stranded DNA (29). Digestion of the nontemplate DNA strand is blocked by cross-linked protein. Therefore, if there are any cross-links with template DNA strand, these cross-links represent a minor fraction. There are also exonuclease stops located close to the DNA fragment ends observed on non-cross-linked DNA, which could be a result of stalling of the exonuclease at the DNA fragment ends.
For precise mapping of cross-links, we performed hydroxyl radical footprinting. The cross-linked complexes were treated with Fe-EDTA in the presence of SDS and separated by SDS-PAGE afterward (Fig. 3). We expected that DNA retarded in cross-linked complexes would not be cleaved at the site of cross-linking and that these sites could be identified with single base precision. Surprisingly, we observed the extended protection over both DNA strands at the place of the cross-links. The cross-linked ␤Ј subunit protects DNA backbone from position Ϫ5 to ϩ2, with maximum protection at positions Ϫ3 and ϩ2. The cross-linked subunit has a more extended protected region, from Ϫ6 to ϩ2, with maximum protection at Ϫ5 and Ϫ3. Thus, the cross-linked protein is closely positioned to the DNA backbone of both strands near the cross-linked site. We propose that the bases with maximum protection are likely to represent cross-linking sites. Thus, the major site of cross-linking of ␤Ј is Ϫ3 thymine, and the second site is probably located at position ϩ2. For the subunit, probable sites are Ϫ5 and Ϫ3 thymines.
The ␤Ј and Cross-links Are Specific for Open Promoter Complex-We have investigated the functional relation of the cross-links formed by formaldehyde. Cross-linking of the RNAP-promoter complexes formed at 0, 14, and 37°C has shown that cross-links correlate with formation of open complex, as was monitored by potassium permanganate footprinting. No cross-links were observed in closed complex (Fig. 4). Therefore, a complete melting of the Ϫ10 region is required for establishing the specific open complex conformation competent for cross-linking. It is known that core RNAP interacts with promoter fragments nonspecifically (30). We have not detected any cross-links between core RNAP and promoter at 37°C (data not shown). These results confirm our conclusion that formaldehyde cross-links reflect specific interactions of RNAP with promoter in open complex.
Mapping of the Cross-linked Region within the ␤Ј Subunit-It was shown that recombinant RNAP containing the ␤Ј subunit reconstituted from the N-terminal (residues 1-875) and C-terminal fragments (residues 850 -1407) is functional (24). We used this RNAP to identify which part of the ␤Ј subunit was cross-linked to the promoter. The electrophoretic mobility of the ␤Ј-subunit N-terminal fragment in SDS-PAGE is slower than that of the subunit, whereas the C-terminal fragment runs faster than the subunit. Cross-linking of this recombinant RNAP with promoter results in the appearance of a cross-linked complex (containing the ␤Ј subunit fragment) with mobility faster than that of the full-length ␤Ј-DNA complex but slower than that of the -DNA complex (Fig. 5a). Therefore, we have concluded that cross-links are located between the N terminus and residue 850 of the ␤Ј subunit.
To map the region of the ␤Ј subunit cross-linked to DNA precisely, we employed approaches used previously for the mapping of protein-RNA contacts (25,26). Thermolability of formaldehyde cross-linking puts some limitations upon the mapping method. The cross-linked complexes cannot be heated; therefore, denaturation of some protein domains may not be complete, hereby reducing the cleavage efficiency at some sites. Moreover, a 123-base pair DNA fragment cross-linked to a protein adds ϳ80 kDa to the mobility of the peptide; therefore, the standard molecular weight markers previously used for the determination of the molecular weight of the cleavage products are not directly applicable. However, the Cys-specific cleavage patterns of the ␤Ј subunit labeled at either the C or the N terminus are well distinguished because of the asymmetric distribution of the Cys residues over the ␤Ј polypeptide (Fig.  5c). The ␤Ј-DNA complexes obtained after cross-linking of the RNAP containing His 6 tag in ␤Ј subunit and radiolabeled lacUV5 promoter fragment were purified on Ni 2ϩ -NTA agarose and treated with NTCBA under single-hit conditions to obtain the Cys-specific cleavage. As a marker, we used ␤Ј subunit labeled with rifampicin-GpCpT* (31) at the C terminus (Fig.  5b) and treated with NTCBA under the same conditions as cross-linked ␤Ј. Identification of the peptides rests on the assumption that molecular mass of cross-linked DNA fragment makes the same contribution to the electrophoretic mobility of all peptides. Therefore, the cleavage pattern of the cross-linked subunit should be the same as the cleavage pattern of the non-cross-linked subunit but shifted upward to a larger molec-  FIG. 6. Mapping of formaldehyde cross-linking site in the ␤ subunit using Met-specific cleavage. Cross-linked complexes of the ␤Ј subunit and end-labeled lacUV5 promoter fragment were treated with NTCBA for 20 min or CNBr for 5 or 10 min and separated by 6% SDS-PAGE. The theoretical Met cleavage pattern, predicted on the basis that the label is located at the N terminus, is shown on the right. ular weight. Indeed, in the case of cross-linked ␤Ј, we observed the same ladder of four bands seen in the control lane, which can be derived only from the cleavages at positions 366, 454, 517, and 608 (Fig. 5b). However, this pattern could be generated if the cross-link is localized at the C terminus or at the N terminus. We did not observe the C-terminal products of cleavages at Cys 814 and Cys 869 -Cys 898 observed in the marker lane, whereas the corresponding N-terminal products migrating close to full-length ␤Ј subunit were visible. Consequently, the peptide with the faster mobility corresponds to the C-terminal product of cleavage at Cys 608 or N-terminal product of cleavage at Cys 366 . Because we had already localized the cross-link at the N-terminal side from position 850, it must be positioned either at the middle of the ␤Ј polypeptide (between Cys 608 and Cys 814 ) or between the N terminus and Cys 366 . The localization of the cross-link in the middle of ␤Ј is unfeasible because this would result in an overlapping of N-and C-terminal patterns, which was not observed in our case. These observations allow to conclude that the cross-link is positioned between the Cys 366 and N terminus. In order to strengthen our conclusion, we estimated the molecular mass of the cross-linked complexes (Table I) and of ␤Ј subunit complex (ϳ240 kDa) on the basis of C-terminal marker ladder (Fig. 5b, lane M). Consequently, we estimate the contribution of DNA fragment as 85 kDa, which is close to expected molecular mass for the 123-base pair double-stranded DNA fragment (77.5 kDa) (Fig. 5b). Subtraction of the DNA molecular mass from the observed molecular mass of the cross-linked protein-DNA complexes gives the expected molecular mass of cross-linked peptides ( Table I). The C-terminal product of the cleavage at Cys 608 has a predicted molecular mass of 86 kDa, which is much higher than estimated molecular mass of the shortest cross-linked peptide (ϳ33 kDa); therefore, this peptide can correspond only to N-terminal product of Cys 366 cleavage (42 kDa). This gives the localization of crosslink between Cys 366 and the next Cys residue, at position 198.
The location of the cross-links was further narrowed down by single-hit cleavage at Met residues with CNBr. The cleavage was performed under mild conditions at pH 4.0, because the formaldehyde cross-links were not stable at pH 2.0, used for standard cleavage with CNBr. This change in pH does not affect the cleavage efficiency at Met residues in the protein. 2 The ␤Ј-DNA cross-linked complexes were treated with CNBr, and the products of cleavage were analyzed by SDS-PAGE (Fig.  6). To estimate the molecular mass of the produced peptides, the cross-linked ␤Ј-DNA complexes treated with NTCBA were used as molecular mass markers. This method of calculation provides a good approximation of the observed masses. The peptide with the highest mobility has a molecular mass of around 25.7 Ϯ 1.7 kDa and can correspond to the N-terminal peptide produced after cleavage at Met 237 (ϳ27.5 kDa). The next cleavage site, Met 192 , is located outside the cross-linked region defined by Cys-specific mapping. Therefore, we have concluded that the cross-link is localized between Met 237 and Cys 198 .

DISCUSSION
The ␤Ј Subunit Segment Contacting Single-stranded DNA Near the Ϫ10 Consensus of the Promoter-Available data support the assumption that RNAP is involved in base-specific interactions with the single-stranded DNA of the promoter melted region (7,18,32,33). Formaldehyde cross-linking is an effective method for the study of protein regions involved in these interactions. Formaldehyde reacts with the amino and imino groups of the DNA bases, which can be involved in hydrogen bond contacts with protein (19,34,35). The chemistry of formaldehyde predicts that thymine bases that are primary sites of formaldehyde reaction on single-stranded DNA are likely candidates for cross-linking. In the open complex the lacUV5 promoter is melted at positions Ϫ11 to ϩ4 (36). In this study, we showed that ␤Ј and subunits cross-link within positions Ϫ5 to ϩ2 of nontemplate DNA strand in open complex, whereas no cross-links appeared in closed complex. There are only two thymines in position Ϫ5 and Ϫ3, which can be involved in cross-linking (Fig. 7). The role of this promoter region, located between the Ϫ10 consensus and transcription start, is not defined. It is known that base-specific interactions of RNAP with the nontemplate DNA strand of the Ϫ10 consensus are important for open complex formation and stability (33). It was proposed that conservative regions 2.   366  112  42  118  33  454  103  52  140  55  517  96  59  145  60  608  86  69  168  83 the subunit are responsible for these interactions (7,8,9). Cross-linking and chemical footprinting studies support this assumption (10,18,37). However, some evidence indicates that there are additional components of holoenzyme and promoter that are responsible for the formation of the heparin resistant open complex. In particular, mutations in region 2.3, which affect the formation of open complex, did not impair the ability of RNAP to form heparin-resistant complex on bubble template (8). Laser UV cross-linking detects close contacts between subunit and positions Ϫ5 and Ϫ3 of lacUV5 promoter nontemplate strand (18). Our data directly show that not only the but also the ␤Ј subunit may be involved in interactions with nontemplate DNA strand of this promoter region. Recently, it was shown that interaction of the core component of holoenzyme with the bases from Ϫ7 to ϩ1 of the "fork junction" template DNA strand may be responsible for heparin resistant complex formation (38). We have not detect any cross-links with the template DNA strand. This can be explained by the difference in structure of the "native" RNAP complex containing complete promoter and of the complex with synthetic fork junction construction. An alternative explanation is the absence of crosslinkable groups in contacting regions of RNAP and DNA. Interestingly, the depurination or prenicking at the positions from Ϫ8 to Ϫ4 of template DNA strand of prmup-1⌬265 promoter (which can facilitate formation of contacts with the nontemplate strand) strongly stimulates open complex formation (39). Along with these results, our data suggest that specificity of interactions of RNAP with the promoter melted region can be a result of cooperative action of and ␤Ј subunits. We have mapped the region of the ␤Ј subunit involved in these interactions to the N-terminal segment (Cys 198 -Met 237 ), which overlaps with the left edge of the evolutionary conserved region B (residues 233-254) (Fig. 7). This region has not been previously implicated in any particular biochemical functions. Analysis of the ␤Ј subunit sequences from different species has shown that this region contains two conserved aromatic residues (Phe 227 and Trp 236 ) that may be involved in intercalation or stacking with single-stranded DNA (9,40,41). Moreover, the N-terminal part of the cross-linked region adjacent to region B is enriched in basic amino acids that form DNA binding surface in single-stranded DNA-binding proteins (40,42).
If the cross-linked region is involved in protein-DNA interactions, we would expect that the mutations in this region affect the process of open complex formation or its stability. Consistent with this notion is the finding that a deletion of basic residues 215-220 adjacent to the B region affects isomerization of closed to open complex on rrnB1 promoter (13).
Interface between ␤Ј and Subunits-Previously, using formaldehyde cross-linking, preferential formation of the ␤Јcross-links in the RNAP holoenzyme had been shown (21). However, the cross-linked regions were not identified. The crystal structure of 70 proteolytic fragment (43) have shown that the melting domain (region 2.3), contacting the Ϫ10 consensus, and "core binding" domain (region 2.1 (44)) are located closely in space facing the opposite sides of the structure. Our data have shown that subunit is cross-linked near the Ϫ10 consensus of the promoter, adjacent to the cross-linking site of ␤Ј subunit; therefore, we predict that the core binding domain of is involved in protein-protein interaction with the region adjacent to the conservative region B of ␤Ј subunit. This suggestion is supported by the finding that segment 260 -309, adjacent to the cross-linked region of the ␤Ј subunit (residues 198 -237), represents the binding site for the subunit (45) (Fig. 7). Using the chemical nuclease (p-bromoacetamidobenzyl)-EDTA⅐Fe, the subunit region 2 was shown to be proximal to the ␤Ј subunit segment localized between regions B and C (46), and the core binding region 2.1 was located in proximity to positions Ϫ4 to ϩ3 of the lacUV5 promoter open complex (10). Together with these results, the results of our study predict that the subunit can direct the correct positioning of the RNAP catalytic center during open complex formation via the interactions with the ␤Ј subunit.