LexA Protein Is a Repressor of the Colicin E l Gene*

LexA protein is a represssor of several chromosomal genes involved in the SOS response in Escherichia coli. In previous experiments, we found that LexA protein may also be a repressor of the colicin El gene. We now present evidence that the purified LexA protein strongly repressed the in vitro transcription of the colicin E l gene. As determined in DNase I protection experiments, LexA protein bound with a high affinity to the approximately 40-base pair long sequence between the Pribnow box and the start codon of the colicin E l gene. The sequence of the binding site was composed of two overlapped “SOS boxes” to which the LexA protein bound in a cooperative manner. co-ordinately repressor

phage induction (1,2). The SOS response is controlled coordinately by two regulatory elements, RecA and LexA proteins. LexA protein appears to operate exclusively as a repressor of "SOS genes" (3). RecA protein, which is activated as a protease by signals occurring after SOS-inducing treatments (4), inactivates LexA protein ( 5 ) . Thus, the SOS genes were derepressed to elicit SOS responses.
We reported that the operator-promoter region of the colicin E l gene has DNA sequence homology to the operators of the recA and lexA genes (6), which are commonly repressed by LexA protein (7,8). These findings plus the results of analysis using a double mutant of the recA and lexA genes (9) led to the hypothesis that LexA protein directly represses the colicin E l gene. We now present evidence that LexA protein is indeed a repressor of the colicin E1 gene. The mode of LexA protein binding to the colicin E l operator region is also described.
*This work was supported by Grants-in-Aid for Scientific Research from the Ministry of Education, Science, and Culture of Japan. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisernent" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. $ To whom correspondence should be addressed. §Supported by Grant GM29988 from the National Institutes of Health.

EXPERIMENTAL PROCEDURES
Purification of LexA Protein-LexA protein was purified from RB791 cells harboring pRB451, a plasmid carrying a fusion between the lexA gene and a modified lac promoter that directs the synthesis of large amounts of LexA protein.' Purification procedures were as described by Brent and Ptashne (8), except that Sephadex G-100 was used instead of Bio-Gel P-150 and the final step of the organomercurial-agarose column was omitted. Protein was mesured by the method of Lowry et al. (10) with bovine serum albumin as a standard. Molarities of LexA protein were calculated as concentrations of a monomer, the molecular weight of which is 22,700 (11). The LexA protein thus obtained had a physical purity of at least 95% as judged by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. We confirmed that the protein was cleaved by RecA (ti7) protease (4) and specifically inhibited the synthesis of RecA and LexA mRNAs in vitro (7, 8).
In Vitro Transcription-Transcription was performed as described (6) except that templates were at 33 nM. Incubations were carried out in two stages. (i) LexA protein was added to the reaction mixtures for the transcription, from which RNA polymerase was omitted (ii) after leaving the mixtures for 10 min a t 37 "C to allow LexA protein binding to the operator, RNA polymerase was added to initiate the transcription. After 15 min, the reaction was terminated by shaking with phenol. Transcripts were analyzed on a 6 or 8% polyacrylamide, 8 M urea gel and visualized autoradiographically. For transcription analysis of the colicin E l gene, the 571-bp2 SmaI-HaeIII fragment containing the colicin El regulatory region ( Fig. 1) was used as a template. The colicin E l mRNA transcripts were labeled with [y-"PI GTP as previously described (6). For the analysis of the RecA mRNA synthesis, the 302-bp HaeIII fragment containing the recA regulatory region was used (12). The RecA mRNA and RNA-1 of pA03 (13) were labeled with [y-32P]ATP. DNA was measured by fluorometric assay (14).
DNase I Protection Experiments and Binding Affinity-The experiments of DNase I protection ("footprinting"), to determine the stretch of DNA covered by LexA protein and to estimate the binding affinity of LexA protein to the colicin E l operator, were done by using the procedure of Johnson et al. (15). The reaction mixture (200 SI) containing 10 mM Tris-HCI, pH 7.0, 2.5 mM MgCla, 1 mM CaCl,, 0.1 mM EDTA, 200 mM KCI, 100 Fg/ml of bovine serum albumin, 2.5 rg/ml of salmon sperm DNA, labeled DNA, and various amounts of LexA protein was preincubated for 15 min a t 37 "C. After adding 2 ng/ml of pancreatic DNase I to the reaction, the incubation was continued for 10 min at 37 "C. The reaction was stopped with 50 ~1 of cold 8 M ammonium acetate and tRNA a t 300 pg/ml. The DNA was precipitated with ethanol and analyzed by electrophoresis on an 8% polyacrylamide, 7 M urea gel, followed by autoradiography. operator molecules (8). The intensity of the protected band in footprinting was determined by inspection.
Materials-Restriction endonucleases and T, polynucleotide kinase were purchased from Takara Shuzo (Kyoto, Japan). DNase I was obtained from Boehringer Mannheim. [Y-~'P]ATP (5400 Ci/ mmol) and [y-"P]CTP (36 Ci/mmol) were products of the Radiochemical Centre (Amersham). RNA polymerase was purified by a modification of the method of Chamberlin and Berg (17).

Specific Inhibition of in Vitro Transcription-From in vitro transcription experiments
using the 571-bp SmI-HaeIII DNA fragment containing the promoter-operator region of the colicin E l gene, we previously showed that two species of RNA, 289 nucleotides (mRNA-1) and 205 nucleotides (mRNA-2), were synthesized (6). The mRNA-1 and mRNA-2 started, respectively, 75 bp upstream and 10 bp downstream from the NH2-terminal codon of the colicin E l gene. Although synthesis of mRNA-2 was four times as efficient as that of mRNA-1 in vitro (Fig. 2a), we identified by S1-mapping assay the promoter of mRNA-1 as the authentic promoter of colicin E l mRNA synthesis in vivo (18). The physiological role of the promoter of mRNA-2 is unknown. As shown in Fig. 2a, LexA protein strongly inhibited the synthesis of mRNA-1, but weakly that of mRNA-2. In the control, RNA-1 synthesis of pA03 (13) was not inhibited by the protein (Fig. 2c). These results indicate that in vitro synthesis of colicin E l mRNA is specifically inhibited by LexA protein. Although it was suggested that a peculiar repressor protein of the colicin E l gene might be encoded by the plasmid (19), we obtained evidence that such a repressor protein is not synthesized from the plasmid.3 Therefore, LexA protein is considered to be a proper repressor of the colicin E l gene. The genes, which LexA protein has so far been reported to repress, were the chromosomal genes such as recA (7, 8), l e d (7,8), uvrA (20), uvrB (21), SUM: d i d , and dinB (22). The above findings indicate that the extrachromosomal gene is also subject to repression by LexA protein. It was suggested that LexA protein is also a repressor of the colicin E2 (ll), E3,5 cloacin DF13 (23), and pKMlOl rnuc gene (24).
As shown in Fig. 2a, the in vitro synthesis of colicin E l mRNA was completely inhibited at 100 nM LexA protein. However, at the same concentration of the protein, the in vitro synthesis of RecA mRNA decreased to 60% of the control (Fig. 2b). L e d Protein Binding Site-To assess the mechanism of the strong repression of the colicin E l transcription by LexA protein, we determined the binding site of the protein to the control region of the gene. From the footprinting experiments using both of the strands, we found that LexA protein bound to a segment of DNA downstream from the Pribnow box (Fig.  3, a and b). The segment lies about 10 bp upstream to 30 bp downstream from the start site of the colicin E l mRNA synthesis. It overlies a 37-bp inverted repeat that we previously speculated to be the binding site of LexA protein to the colicin E l gene (6). The protected region of the colicin E l gene seems to be composed of two overlapped "SOS boxes" (7). Strong homology is observed in the sequence of the left binding site, whereas the sequence of the right binding site is less homologous, as shown in Fig. 4.
Affinity of Binding-The equilibrium dissociation constant for LexA protein to the colicin E l control region was measured under the conditions (37 "C, 200 mM KCI) where Kd values of LexA protein for the recA and l e d operators were determined (8). At an operator concentration (50.05 nM), which is far below the dissociation constant for LexA protein to the binding site, the Kd should be approximately equal to the concentration of LexA protein required to occupy one-half of the operator molecules (16). As shown in Fig. 3a, about 70% of the operator molecules were occupied by LexA protein at 0.5 nM, while all of them were bound by the protein at 1.0 nM. Therefore, the apparent Kd of LexA protein for the colicin E l operator was about 0.4 nM. However, the actual Kd may be lower than this value, because some of the -in may be incompetent to bind the operator at these -&Ions.
The apparent Kd value of LexA protein for the recA operator, which was determined under the same conditions, was re- end of Sau3A was used (antisense strand). The chemical sequencing was done by the method previously described (6). b, about 0.5 nM 153-bp ThaI-Sau3A fragment labeled with 32P at the 5' end of ThaI was used (sense strand) to demonstrate the LexA protein binding region. The bases were numbered relative to the starting point of colicin E l mRNA, which was designated + I . Brackets indicate portions of the region that were protected by LexA protein from digestion with DNase I. The lower site of the region protected by LexA protein in b was visualized more clearly by a longer exposure (not shown). ported to be about 2 nM (8). Although a comparison of the values from different experiments may not be valid, the affinity of LexA protein for the colicin E l operator appears higher than that for the recA operator.

~T G T A T A T N N A N N C A~
At 0.25 nM, LexA protein did not occupy the colicin E l operator, but at 0.5 nM the protein simultaneously bound to both of the SOS boxes. Binding of LexA protein to the operator did not show a linear relationship between LexA protein and the operator occupied by the protein. What is observed is consistent with the idea that LexA protein binds to the colicin E l operator in a cooperative manner. Consensus Sequence of L e d Protein Binding Sites-In Fig.  4, the LexA protein binding sequence of the colicin E l gene is compared with operators of l e d (7, 8), recA (7,8), uurA (20), and uurB (21) genes and also with presumed operators of SUM (25) and cloacin DF13 (23) genes. The structure of these binding sites (SOS box) has a consensus sequence, "'CTGTATATNNANNCAGI'. The 3-bp segments at the ends of this 16-bp sequence are related by a 2-fold axis of symmetry. The terminal 3-bp segments are thought to be important to LexA protein binding from the methylation experiments of the recA and l e d genes carried out by Brent and Ptashne (8). Some of the repressor proteins so far characterized exist in a dimeric form which has a %fold symmetry axis coincident with the 2-fold axis in the binding site of the DNA molecule (26,27). Since LexA protein tends to dimerize (8), the protein that undergoes interaction with the above binding site would probably also be in a dimeric form.
The classical B-form DNA has 10.0 bp/turn of the helix. Therefore, the terminal segments in each SOS box separated by 10 bp eventually face to nearly opposite directions on the sides of the DNA molecule. LexA protein would interact with the terminal 3-bp segments which are exposed within the major groove of the DNA as predicted in the case of other repressor proteins (26,27). On the other hand, the D-form DNA which has 8.0 bp/turn reportedly exists in the DNA with a high content of A-T (28,29). In this form of DNA, it is also possible to assume that the protein can interact with the 3-bp segments along one face of the DNA molecule. The sequence of A-T was frequently observed in these SOS boxes.
In the B-form DNA, an axial rise/residue is 0.34 nm, while it is 0.3 nm in the D-form DNA (28,29). Thus, the length of the SOS box should be 5.44 nm in the B-form DNA and 4.8 nm in the D-form DNA. Assuming that a dimer molecule of LexA protein is a sphere, the diameter would be 5.7 nm. Therefore, the LexA dimer could adequately cover the 16-bp sequence of either the B-or the D-form of DNA.
Another feature of the LexA binding site of the colicin E l operator is that there are two overlapping SOS boxes. The structure of the right binding site is less homologous with the typical SOS box. However, the LexA protein binding to both boxes occurred at the same concentration of the protein (Fig.  2a). This hypothetical cooperative binding could occur by at least two mechanisms. (i) Binding of LexA protein to one of the SOS boxes would induce a conformational change in the neighboring box and lead to stimulation of another molecule of LexA protein to bind to the box; ii) LexA protein bound to one of the SOS boxes could interact with another molecule with protein-protein association and stimulate the LexA protein binding to the adjacent box. LexA protein binds with high affinity to the colicin E l operator site which is located in the 3' region of the Pribnow box. This high affinity binding probably accounts for the strong repression of colicin E l mRNA synthesis by LexA protein.