Identification of the DNA-binding Domain of the OmpR Protein Required for Transcriptional Activation of the ompF and ompC Genes

Expression of the ompF and ompC genes of Escherichia coli requires the OmpR protein for transcriptional activation. In vivo binding of the OmpR protein to the ompF and ompC promoter regions was observed using an in vivo dimethyl sulfate DNA footprinting technique. Two different sequence motifs were found to be protected by OmpR in both the ompF and ompC promoter regions. This technique was further used to localize the DNA-binding domain of OmpR to be within the C-terminal117 amino acid residues. Binding of the C-terminal portion OmpR to the ompF and ompC promoter regions, however, did not result in activation of transcription. Our results, together with sequence homologies between OmpR and other regulatory proteins, suggests that OmpR has separable domain structures: the C-terminal portion for binding-specific DNA sequences and the N-terminal portion for interacting with RNA polymerase and/or other transcription factors.

The OmpR protein, encoded by the ompR gene of the ompB locus of Escherichia coli, is known to be a transcriptional activator which is involved in the expression and regulation of the ompF and ompC genes (Hall and Silhavy, 1981a; for review, see Forst and Inouye, 1988). Mutations in the ompR gene have been shown to result in abolished or altered expression of OmpF and OmpC in the outer membrane at the transcriptional level (Hall and Silhavy, 1981b; also see a review by Forst and Inouye, 1988). Purified OmpR has been shown to bind to the -60 to -100 regions of the ompF and ompC promoters and, in the presence of RNA polymerase, transcription from the ompF promoter was activated by OmpR in vitro (Norioka et al., 1986). The OmpR protein is considered to consist of at least two domains, a modulator domain and a DNA-binding domain. The modulator domain may interact with the EnvZ protein, a component of the inner membrane, considered to function as an osmosensor or a signal transducing protein (Forst and Inouye, 1987). This presumed interaction of OmpR and EnvZ may govern the binding of OmpR to the ompF and ompC promoters by the second domain of OmpR, the DNA-binding domain. Upon binding of OmpR to the promoters, the modulator domain may activate transcription by stimulating the association of RNA polymerase to the promoters. The DNA-binding domain * This work was supported by Grant GM19043-16 from the National Institutes of Health (to M. I.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ To whom correspondence should be addressed.
of OmpR recognizes the specific DNA sequences upstream of the ompF and ompC promoters so that the modulator domain is then able to activate transcription of the gene. Comparison of sequence similarity between OmpR, PhoB, CheB, and NtrC, as well as other regulatory proteins has suggested a common domain at the N-terminal portions of these proteins (Winans et al., 1986). Various techniques have been developed to study protein-DNA interactions in vivo (Church and Gilbert, 1984;Martin et d., 1986;Gilmour and Lis, 1984;Solomon et al., 1988). One of these methods is in vivo chemical (dimethyl sulfate) DNA footprinting either on genomic DNA (Church and Gilbert, 1984) or on plasmid DNA (Martin et al., 1986). Due to technical difficulties, however, in vivo chemical (dimethyl sulfate) DNA footprinting is not widely applied compared to the conventional in vitro chemical (dimethyl sulfate) DNA footprinting method. In this paper, using a modified method for in vivo chemical (dimethyl sulfate) DNA footprinting, we investigate OmpR-DNA interactions in uiuo.
It was found that two different sequence motifs were protected by OmpR in vivo in both the ompF and ompC promoter region.
In addition, both motifs were recognized by the C-terminal domain of OmpR consisting of 117 amino acid residues.
Binding of the C-terminal half of OmpR, however, could not activate transcription of either the ompF or ompC genes.
These results indicate that the DNA-binding domain of OmpR resides within the C-terminal portion of the protein and that the activation function resides in the N-terminal portion which is considered to interact with both EnvZ and RNA polymerase.

EXPERIMENTAL PROCEDURES
Plasmids-The plasmid pKI0033 containing the ompF promoter sequences has been described previously (Norioka et al., 1986). This plasmid was used for in vivo footprinting. The ompF-lacZ fusion plasmid (pKL0533) was constructed by inserting the ompF promoter XbaI-BamHI DNA fragment of pKI0033 into the corresponding sites of pKM005, a promoter proving vector (Masui et al., 1983). The ompF-lac2 fusion plasmid was used for the P-galactosidase activity assay. The ompC promoter containing plasmid used in the in vivo footprinting experiment is pKI0041-C (BH). This plasmid contains both the ompC and micF sequences (Mizuno et al., 1983). The reading frame of the ompC gene was disrupted at two positions. One is at the 4th codon by inserting a BamHI linker sequence and the other one is at the 38th codon by filling in the natural BglII site and religation. The ompC-lucZ fusion plasmid contains both the ompC promoter and the upstream micF gene and has been described previously (Norioka et al., 1986).
In Vivo DNA Footprinting-An outline for the in uiuo chemical (dimethyl sulfate) footprinting method is presented in Fig. 1. A midlog culture (50 ml) grown in LB medium at 37 "C was transferred to a 500-ml beaker before dimethyl sulfate treatment. The beaker was incubated in a 37 "C water bath. Dimethyl sulfate (100%) was then added to the culture with vigorous shaking to a final concentration of 0.08%. After 20 s, EDTA (0.5 M, adjusted to pH 8.0) was added to the culture to a final concentration of 45 mM and the incubation was continued for another 10 s. Immediately after adding 30 g of ice the beaker was transferred to an ice-water bath. The cells were then collected by centrifugation as soon as possible. The cell pellet was washed once with 10 ml of M9 salt (Miller, 1972). The washed cell pellet was suspended in 0.5 ml of 25 mM Tris-HC1 (pH 8.0) containing 10 mM EDTA (pH 8.0), 25% glucose, and 50 mg/ml lysozyme, and the mixture was incubated at room temperature for 5 min. After adding 1 ml of 0.2 N NaOH containing 1% sodium dodecyl sulfate the cell lysate was incubated on ice for 2 min followed by addition of 0.75 ml of 3 M sodium acetate (pH 4.8). The incubation was continued on ice for another 5 min. The resulting turbid suspension was centrifuged at 20,000 X g for 15 min at 4 "C. The clear supernatant was transferred to another clean centrifuge tube and DNA was precipitated by adding 5 ml of 95% cold (-20 'C) ethanol. The DNA precipitate was collected by centrifugation. The pellet was resuspended in 0.4 ml of HzO and the DNA was precipitated again in a 1.5-ml Eppendorf tube with 0.9 ml of 95% cold ethanol. Plasmid DNA isolated through the above steps is usually heavily contaminated with RNA and randomly sized chromosomal DNA due to the dimethyl sulfate treatment. Therefore it is important to separate these contaminants as much as possible before 32P labeling of the target fragment. For these reasons, the first enzyme digestion (BamHI) was carried out in the presence of ribonuclease A. The linear plasmid DNA was then purified by electrophoresis on a preparative agarose gel (0.7%, E buffer). In the present in vivo DNA footprinting method, it is most critical how well the dimethyl sulfate-treated plasmid can be separated from other contaminants on the agarose gel. Therefore, special attention should be paid to: 1) insoluble materials should be removed before loading samples on the agarose gel by centrifuging samples after solubilizing the DNA pellet; 2) plasmid preparation should be performed quickly and as soon as possible. When the gel was stained with ethidium bromide (0.5 pg/ml) a very heavy background smear of DNA usually appeared. Linear or supercoiled plasmid DNA bands appearing on the background of smeared DNA were cut out of the gel and the DNA was electroeluted. If the plasmid DNA was not digested before agarose gel electrophoresis, the purified plasmid DNA was subsequently digested with an appropriate restriction enzyme. The linear DNA fragment was then 32P-labeled at the 5' end with T4 kinase or at the 3' end with Klenow enzyme. After digestion with a second restriction enzyme, the 3ZP-labeled fragment was purified by gel electrophoresis. The DNA fragment was then cleaved with 1 M piperidine at 90 "C for 30 min and analyzed on a 6% sequencing gel.
Chimeric OmpRs and Their Target Genes-Chimeric OmpRs were constructed using pITlacI (1ppp"-lacZ), a derivative of pBR322. The target ompF and ompC promoter regions were cloned in pACYC184 (Chang and Cohen, 1978). This target plasmid is compatible with pBR322. E. coli strain SG48A76 (ompB deletion) was transformed sequentially with a plasmid carrying a gene for a chimeric OmpR and a target plasmid. Double transformants were selected for both ampicillin (for factor plasmids) and chloramphenicol (for target plasmids) resistance. Cells containing both plasmids were used for in vivo DNA footprinting experiments as described in the previous section. In order to construct genes for the chimeric OmpRs, the ompR gene in pAT428 was disrupted by an XbaI linker (CTCTAGAG) insertion at the 54th codon. pAT428 is a clone of the ompB locus consisting of ompR and envZ (Ikenaka et al., 1988). This plasmid was then digested with XbaI and the resultant linear DNA fragment was treated with Bal31 nuclease for various periods of time to progressively remove the N-terminal coding region of ompR. The nuclease-treated fragment was ligated with the EcoRI linker (CGGAATTCCG), and after screening a group of fragments for proper lengths of the truncated Cterminal coding region, they were inserted into the downstream of the EcoRI site of the lacZ gene. As a result, a long N-terminal fragment of 8-galactosidase consisting of 1006 amino acid residues was fused to a C-terminal fragment of OmpR of various sizes. Fusion of the full-sized OmpR to 8-galactosidase was carried out with use of an EcoRI fragment isolated from plasmid pSN142 (Norioka et al., 1986). The resultant plasmid produces a hybrid protein consisting of the N-terminal @-galactosidase fragment (1006 amino acids residues) and full-sized OmpR (238 amino acid residues). We also constructed a hybrid protein consisting of the N-terminal OmpR fragment (192amino acid residues) and the C-terminal portion of 8-galactosidase as follows. The Ssp1 site at the 192nd codon of ompR was converted to BglII site. The DNA fragment containing the ompB promoter and the OmpR sequence up to the BglII site was then inserted to the upstream of the 8th codon of lac2 at the BamHI site in plasmid pKM005 (Masui et al., 1983).
Other Method-Assay for 8-galactosidase activity was performed as described previously (Miller, 1972).

OmpR-binding Motifs in the ompF and ompC Promoter
Regions in Vivo-Mutations or deletions in ompR are known to cause altered or abolished expression of the ompF and/or the ompC genes (Forst and Inouye, 1988). The ompB deletion (deletion of both ompR and envZ genes) strain, SG480A76, does not produce either OmpF or OmpC (Garrett et al., 1985). Other strains, including the OmpR wild type and various OmpR mutant strains which were used in the footprinting experiments are listed in Table I. The effect of various ompR strains on transcription of cloned ompF and ompC was measured by using ompF-lac2 and ompC-lac2 fusion plasmids. 8-Galactosidase activity was found to be dependent upon ompR which is in agreement with the phenotypes of the host strains ( Table I). The general method for in vivo DNA footprinting is outlined in Fig. 1. In both the case of ompF and ompC, footprinting of the respective promoter region was carried out in a wild type (MC4100) (Casadaban, 1976) and an isogenic

DNA-binding Domain
of OmpR ompB deletion strain (SG480A76) (Garrett et al., 1985). In addition, OmpR protection patterns were examined in two OmpR mutant strains. The ompRl (MH1160) mutant (Hall and Silhavy, 1979) is phenotypically OmpF-/OmpC-while the ompR2 (MH760) mutant (Hall and Silhavy, 1981a) is OmpF+/OmpC-. Results from these experiments are presented in Figs. 2 and 3. The methylation protection pattern of the ompF promoter minus (-) strand (bottom strand in Fig. 4A) is presented in Fig. 2 A . The guanine residues at base position -96, -86, -76, and -46 (+1 as the mRNA start site) were protected from methylation in wild type and ompR2 cells when compared with the methylation pattern from the ompB deletion and ompRl cells. Also, the guanine residue at -65 position was hypermethylated in wild type and ompR2 cells.
On the plus (+) strand (top strand in Fig. 4A), shown in Fig.  2B, guanine residues at base position -70 and -50 were protected from methylation and the residues at -90 and -91 were hypermethylated in wild type and ompR2 cells compared with patterns from ompB deletion and ompRl cells. The methylation protection pattern of the ompC promoter is shown in Fig. 3A for the minus (-) strand and in Fig Table I   promoter in ompRl cells and no OmpR mediated protection was seen (data not shown).
A summary diagram of the OmpR-DNA sequence-specific interaction is shown in Fig. 4A. From the DNA sequence of the protected region, two consensus sequence motifs were deduced, the F box and the C box (Fig. 4B). Wild type as well as one mutant OmpR, OmpR2, were found to interact with the 3 juxtaposed F boxes (Fa, Fb, Fc) of the ompF promoter. The C type box (Cd) located downstream of the F boxes in the ompF promoter was also protected in wild type cells. However, in the ompR2 mutant cells the protection at the Cd box appears to be weaker than in the wild type cells (Fig. 2 A , -46 position and Fig. 2B, -50 position). In the ompC promoter region, the three C boxes (Ca, Cb, Cc), each separated by 10-11 base pairs and one F box (Fd) overlapping with the most upstream C box (Ca), were protected only by wild type OmpR in the following order: Fd, Ca > Cb > Cc (Fig. 3). Two striking features of the OmpR binding pattern should be noted. First the DNA sequence regions interacting with OmpR in both the ompF and ompC promoters are in similar locations (-40 to -100) relative to the start site of transcription (+l). Second, all C and F boxes are in-phase in terms of DNA helix structure. That is, OmpR, when bound to these boxes, is expected to be found on the same side of the DNA helix.
The DNA-binding Domain of OmpR-To determine which portion(s) of OmpR is required for binding to the ompF and/ or ompC promoter regions, protein fusions consisting of either the C-terminal portion of OmpR or the N-terminal portion of OmpR fused to @-galactosidase were constructed as shown in Fig. 5. A plasmid mediated factor-target system (Fig. 6) in combination with in vivo chemical DNA footprinting was used to examine the DNA-binding properties of the chimeric OmpR proteins. In these experiments the host cell for the two compatible plasmids was the ompB deletion strain (SG480A76) which displayed no protection of guanine residues in the F and C boxes (Figs. 2 and 3). The OmpR (192 amino acids)-@-galactosidase fusion protein, which was expressed in vivo close to 1% of total cellular proteins, did not bind to the sequence motifs in the ompF and ompC promoter regions (data not shown). On the other hand, chimeric proteins composed of the C-terminal portions of OmpR fused to the N terminus of @-galactosidase were able to bind both the F and C sequence motifs in vivo (Fig. 7, B and C). The expression of these OmpR fusions was under the control of 1ppP-lacP/" which is inducible with IPTG' (Masui et al., 1983).
Fusion proteins with the C-terminal 238, 156, and 117 amino acid residues of the 239 amino acid residues comprising fulllength OmpR were produced in the host cell in response to the presence of IPTG (Fig. 7A). The corresponding methylation protection pattern for the ompC promoter region is shown in Fig. 7B. In Fig. 7C the methylation protection pattern for the ompF promoter region is presented. In this experiment IPTG was used in all the cell cultures and one control plasmid was included which contained the C-terminal238 amino acid coding sequence of the ompR gene fused out of reading frame to the lac2 gene (Fig. 7C, lane 0 a.a.). The C-terminal region of OmpR was able to bind to both the F and C boxes as in the case of native OmpR as observed from the protection pattern of both the ompF and ompC promoters. As the Cterminal portion of OmpR was shortened, however, protection indicating binding decreased (e.g. the -43 position in Fig.   7B). The shortest C-terminal OmpR (117 amino acid residues) displayed much more reduced binding to the ompF than the ompC promoter regions. This short fusion protein apparently is able to bind to the ompF promoter, however, since one can observe the protection at base position -96, -86 and hypermethylation at base position -65 (Fig. 7C, lane 117 a.a.). Additionally, all of the C-terminal OmpR fusion proteins shown here displayed an inhibitory effect on the production of the OmpF protein when introduced into wild type cells (data not shown). Further shortening of the C-terminal portion of OmpR (e.g. 82 amino acid residues) did not show any inhibitory effect in wild type cells and did not bind to the ompF and ompC promoter regions in vivo (data not shown).
The C-terminal OmpR fusion proteins including the near fulllength OmpR fusion (238 a.a), although capable of binding to specific DNA sequences to the same extent as native OmpR, did not activate transcription from either the ompF or ompC promoters in vivo when introduced into ompB deletion and ompRl cells.

DISCUSSION
The DNA sequence motifs in the promoter region of the ompF and ompC genes recognized by the OmpR protein in vivo have been identified by using an in vivo (dimethyl sulfate) chemical DNA footprinting technique. The location of these sequence motifs found in the present report is in agreement with the findings reported by another group using the in vitro DNase I footprinting technique  and is extended further downstream from the regions reported by our group in an in vitro study (Norioka et al., 1986). The possible explanation for the failure of identifying broader OmpR-binding regions in our previous study is that the amount of purified OmpR used may not have been high enough to bind to low affinity sites (Cc, Cd, for example). In our previous report (Norioka et al., 1986) a homologous region found in both the ompF (-74 to -66) and ompC (-92 to -84) promoter regions, protected in vitro by OmpR protein from DNase I digestion, was named sequence 11. In the present study, no identical protection/enhancement was observed for these regions in the ompF and ompC promoters (see Fig. 4A). When considering the protection patterns obtained from the in vivo footprinting we could, instead, find two types of OmpR binding motifs with different consensus sequences (Fig. 4B). Methylation of two base positions (-65 and -70) near the Fc box in the ompF promoter were also found to be affected by OmpR binding in vivo. These  ruled out that other factors may bind to this region and play some regulatory functions. The lack of apparent symmetry of the arrangement as well as the DNA sequences of the two OmpR recognition motifs seem to support the idea that OmpR binds to DNA as a monomer. Three juxtaposed F boxes (Fa, Fb, and Fc) were assigned to the ompF promoter region. This assignment was made since the DNA sequence from -70 to -100 consists of three homologous "boxes" and the removal of box Fa and Fb in stepwise deletion still gave weak but detectable OmpR-dependent methylation protection in the remaining box(s).' Although the consensus DNA sequences of the F box (TTTACA(T)TTTT) and the C box (TGA(T)ANCATNT) share over 50% homology (Fig. 4B), they represent two independent sequences recognized by OmpR. Evidence for this assumption is from the observation that OmpR2 failed to bind to the Ca, Cb, and Cc boxes and bound the Cd box with less degree. The reason that this mutant OmpR also failed to bind to one of the F boxes (Fd) cannot be explained at the present time. The Fd sequence, however, was found to be protected in wild type cells in vivo (present work) as well as in vitro (Norioka et aL, 1986;Mizuno et al., 1988) and it shares high homology with the Fa, Fb, and Fc boxes. No clear function can be assigned to this sequence box at present since point mutations' and partial deletion of Fd (Mizuno and Mizushima, 1986) did not affect transcriptional activation. On the other hand, the Fa and Ca boxes have been shown to be essential for transcriptional activation. Systematic deletion from upstream ending downstream of these two boxes resulted in complete loss of promoter activity (Mizuno and Mizushima, 1986;Ramakrishnan et al., 1986).
It is interesting to note that the locations of the Cd and Cc boxes are near the -35 regions of ompF and ompC promoters where only medium homologous to consensus sequence elements can be found. The close locations of OmpR binding site to that of the putative RNA polymerase binding site suggests a possible protein-protein interaction between OmpR and RNA polymerase. The DNA sequence spanning the Cc box has been replaced in another report (Maeda et al., 1988) and 90% of the promoter activity was lost. We have made point mutations in the Cc box and a similar loss of promoter activity was observed.2 These results indicate that the Cc box is required for activation. The Cd box, however, may have a different role from the Cc box. First, we have made point mutations in this box and no significant loss of promoter activity was observed.' Second, the OmpR2 mutant protein was found to bind to this box less tightly compared with wild type OmpR (Fig. 2) and the ompF promoter activity in ompR2 cells was found almost 2-fold higher than in isogenic wild type cells (Table I). Therefore, the Cd box may function as a repressor site for the ompF gene.
It is important to point out that the Ca, Cb, and Cc boxes on the ompC promoter are separated from each other by 10 or 11 base pairs, indicating that OmpR binds to the same side of the DNA helix. Such binding may be important for cooperative binding of a transcriptional activator to a gene. In contrast to the C box, the Fa, Fb, and Fc boxes have no space between them. It is not clear at present if three OmpR molecules are able to bind to the ompF promoter at the same time.
C-terminal OmpR fusion proteins were shown to be able to bind both the F and C boxes, and the present data demonstrates that the OmpR DNA-binding domain lies within the C-terminal domain consisting of 117 amino acid residues. The function of the N-terminal portion of OmpR has not been demonstrated directly. This portion shares very high homology with several bacterial regulatory proteins including Dye, PhoB, and NtrC all of which are known to have activation functions (Winans et aL, 1986). It would be reasonable to assume that this family of transcriptional activators uses a common mechanism for activating transcription. Binding of the C-terminal OmpR to the specific sequences in ompF and ompC promoter regions did not give transcriptional activation. The fact that even with almost full-length OmpR (238 out of 239 amino acids) fused to &galactosidase, no transcriptional activation was achieved suggests that the N-terminal portion of OmpR plays an important role in interaction with other protein(s) such as RNA polymerase and EnvZ. Fusion of /3galactosidase at the N-terminal region probably interferred with the interaction of these proteins.
The expression of the OmpF and OmpC proteins in E. coli is regulated according to the osmolarity of growth media, and the regulation has been identified at least at the level of transcription (Hall and Silhavy, 1981a). It is interesting to apply the in vivo DNA footprinting technique in the study of OmpR binding under different growth conditions. This work is currently in progress in our laboratory.