b*, a UV-inducible Smaller Form of the b Subunit Sliding Clamp of DNA Polymerase III of Escherichia coli

The 40.6-kDa β subunit of DNA polymerase III of Escherichia coli is a sliding DNA clamp responsible for tethering the polymerase to DNA and endowing it with high processivity (Stukenberg, P. T., Studwell-Vaughan, P. S., and O'Donnell, M.(1991) J. Biol. Chem. 266, 11328-11334). UV irradiation of E. coli induces a smaller 26-kDa form of the β subunit, termed β*, that, when overproduced from a plasmid, increases UV resistance of E. coli (Skaliter, R., Paz-Elizur, T., and Livneh, Z.(1996) J. Biol. Chem. 271, 2478-2481). Here we show that this protein is synthesized from a UV-inducible internal gene, termed dnaN*, that is located in-frame inside the coding region of dnaN, encoding the β subunit. The initiation codon and the Shine-Dalgarno sequence of dnaN* were identified by site-directed mutagenesis. The dnaN* transcript was shown to be induced upon treatment with nalidixic acid, and transcriptional dnaN*-cat gene fusions were UV inducible, suggesting induction of dnaN* at the transcriptional level. Analysis of translational dnaN*-lacZ gene fusions revealed that UV induction was abolished in strains carrying the recA56, lexA3, or ΔrpoH mutations, indicating involvement of both SOS and heat shock stress responses in the induction process. Expression of dnaN* represents a strategy of producing several proteins with related functional domains from a single gene.

UV irradiation of Escherichia coli cells leads to the formation of both mutagenic and inactivating DNA lesions (1). The cells respond by an immediate arrest of DNA replication, followed by a period of extensive DNA repair, that operates to eliminate DNA damage in order to prevent replication obstacles (2). These processes are controlled primarily by the SOS stress regulon, which involves more than 20 genes that are commonly regulated by the LexA repressor and the RecA activator (3,4). However, UV irradiation induces change also in heat shock genes (5) and other genes (6) which affect the post-UV physiology of the cell. We have previously found that the ␤ subunit of DNA polymerase III holoenzyme, the major replicase of the E. coli chromosome (7), limits the ability of the purified polymerase to replicate UV-irradiated single-stranded DNA (8). Consistent with this result, overproduction of the ␤ subunit from a plasmid caused a reduction in UV resistance and in UV mutagenesis of E. coli cells (9).
This involvement of the ␤ subunit in UV irradiation effects prompted us to examine whether it may be present in a different form in UV-irradiated cells. We found that upon UV irradiation a smaller form of the ␤ subunit, termed ␤*, was induced. When overproduced from a plasmid under the inducible lac promoter, ␤* caused up to a 6-fold increase in UV resistance of E. coli cells, suggesting a role in recovery from UV damage, e.g. by involvement in DNA repair or reactivation of DNA replication (48).
Smaller derivatives of proteins that are found in cells are frequently generated by proteolysis, as in the case of the mutagenesis protein UmuDЈ that is formed from UmuD by specific cleavage promoted by the RecA protein (10). Alternatively, the protein can be translated from the overlapping mRNA by a de novo internal translational start, or it can be expressed from an internal in-frame gene. The present study shows that ␤* is synthesized from a novel UV-inducible gene which is located in-frame inside the coding region of the dnaN gene, and it is subjected to indirect regulation by both the SOS and heat shock stress responses.
Proteins-LexA repressor and the ␤ subunit of DNA polymerase III were purified as described by Little (12) and Johanson et al. (13), respectively. ␤* was purified from an overproducer strain that was constructed in our laboratory (49). Anti-␤ antibodies were affinitypurified on ␤* immobilized on nitrocellulose as previously described (48). 32 RNA polymerase was a gift from R. Burgess (University of Wisconsin). DNA polymerase I, T4 DNA ligase, alkaline phosphatase, T7 RNA polymerase, E. coli RNA polymerase, RNase T 1 and RNase A, and bovine serum albumin were purchased from Boehringer Mannheim. Restriction endonucleases were purchased from New England Biolabs. Polynucleotide kinase was from U. S. Biochemical Corp. Proteinase K, chicken egg white lysozyme, and anti-␤-galactosidase antibody were obtained from Sigma.
Bacterial Strains and Bacteriophages-The bacterial strains used in this study are listed in Table I. E. coli R40NL8 was obtained by removing the imm21 prophage from E. coli R40 (14) by superinfection with a heteroimmune derivative, b2 ( immunity). Phage b2 has a deletion at the att site so it cannot integrate into the chromosome; rather it enters the lytic pathway, and can supply the proteins needed for excision of the prophage from the chromosome. The reduced level of heat shock proteins (that are needed for the life cycle of ) slows down the lytic infection by b2. This raises the probability of obtaining colonies of -free E. coli R40 cells as a result of asymmetric segregation during cell division that has occurred after excision. E. coli R40NL8 was sensitive to all types of , as expected from a non-lysogen, and remained temperature-sensitive like the parental R40. Transformation of R40NL8 with plasmid pFN97, carrying the rpoH gene, yielded temperature-resistant colonies suggesting that the temperature sensitivity of the cells was indeed due to the ⌬rpoH mutation.
Construction of Plasmids-The plasmids used in this study are presented in Table II. Plasmid pRPHF11 that served as a template for the synthesis of the dnaN* riboprobe is a pBluescript SK ϩ derivative in which the HpaII-(1942)-FspI(2142) dnaN DNA fragment was cloned in an orientation opposite to the T7 RNA polymerase promoter. Plasmid pCAT was derived from plasmid pCM4, a pBR327 derivative containing the coding sequence of the cat gene downstream to the tet promoter (Fig.  1). The BamHI site located at the 3Ј end of the cat gene was eliminated by a partial digestion of plasmid pCM4 to full-length linear DNA, followed by filling-in of the termini and self-ligation to yield plasmid pCMB. The T 1 rrnB terminator of E. coli was isolated from plasmid pPS1 (15) by digesting with restriction endonucleases HindIII and MboII, to produce a 326-bp fragment, followed by further digestion with restriction endonuclease AhaII to generate the desired 247-bp HindIII-AhaII fragment. It was cloned into the AatII site of pCMB, to yield plasmid pCAT. Plasmid pCAT was the parent for the various cat transcriptional gene fusions. They were constructed by eliminating the tet promoter from pCAT by cleavage with restriction nucleases EcoRV and ClaI ( Fig. 1) and ligating the cleaved vector to the appropriate dnaN* promoter fragment as follows. Plasmid pNCB17, BstUI-(1896 -2228); pNCH6, BstUI-(1896)-HgaI-(2037); pNCS14, SfaNI-(1918 -2090); pPC1, BstUI-(2228)-SfaNI-(2090), P x promoter; pRC5, NciI-(62-219), recA promoter; pNCH20, HgaI-(2037)-BstUI-(1896). Plasmids pTEN5, pNB3, and pSB6 are dnaN*-lacZ translational gene fusions. They were constructed by cloning dnaN* fragments into the SmaI site in plasmid pMC1403 (16) leading to the fusion of parts of ␤* to the 8th amino acid of ␤-galactosidase. The following dnaN* gene fragments were used. Plasmid pTEN5, SfaNI-(1918 -2090); Plasmid pNB3, BstUI-(1896 -2228); plasmid pSB6, AccI-(1965)-BstUI-(2228). The coordinates of the dnaN* gene are according to Ohmori et al. (17). Control cI-lacZ and dnaA-lacZ translational gene fusions were constructed as follows. Plasmid pHSC6 (cI-lacZ fusion) was constructed by isolating from plasmid pUN121 (18) the 550-bp HincII DNA fragment containing the control region and the coding sequence for the first 168 amino acids of cI repressor gene of phage and cloning it into the SmaI site in plasmid pMC1403. Plasmid pHSA2 (dnaA-lacZ fusion) was constructed by isolating from plasmid pHB21 (17) the 450-bp EcoRI fragment carrying the control region and the coding sequence for the first 20 amino acids of the dnaA gene and cloning it into the EcoRI site in plasmid pMC1403 after filling-in the termini of both vector and insert in order to synchronize the reading frames of dnaA and lacZ. The in-frame fusion at the cloning junctions was verified by DNA sequence analysis. Plasmid pCM1 is a 6-kb pBR322-derivative that carries the rpoH gene and the cat gene, conferring resistance to chloramphenicol. It was constructed from plasmid pFN97 (19) that carries the rpoH and bla genes, by inserting the 1.6-kb NheI-HincII fragment carrying the cat gene from plasmid pACYC184 (20) into the ScaI site inside the bla gene.
Kinetics of ␤* Synthesis from the Phage T7 Promoter-E. coli BL21(DE3) cells harboring plasmid pNLW1 (wild-type), pNLM11 ( 2043 ATG 3 2043 CTG mutation, Met1) and pNLM21 ( 2076 ATG 3 2076 CTG mutation, Met2) were grown to OD 595 ϭ 0.4 in minimal medium supplemented with ampicillin (20 g/ml), MgSO 4 (1 mM) and glucose (0.4%). Expression of dnaN* was triggered by the addition of IPTG (0.5 mM). After 40 min of induction, rifampicin (200 g/ml) was added, to inhibit transcription by the host RNA polymerase, and incubation was continued for 30 min. At this point 300 Ci (20 l) of [ 35 S]methionine were added to a cell culture of 4 ml. At various time points 0.5-ml samples were withdrawn, and added to 0.1 ml of a solution of 0.3% sodium azide containing 120 g of unlabeled L-methionine to chase the metabolic labeling. The culture was cooled for 10 s in a dry ice-ethanol bath, and then kept on ice until assayed. The cell extracts were fractionated by 15% SDS-PAGE after which the gel was dried and fluorographed using a Kodak XAR-5 x-ray film. The intensities of the bands were quantified by scanning with a Molecular Dynamics 300 A computing densitometer.
Kinetics of Degradation of ␤*-Four ml of E. coli BL21(DE3) cells harboring plasmid pNLW1 (wild-type) were pulse-labeled for 5 min with 300 Ci of [ 35 S]methionine, and then chased with unlabeled 960 g of L-methionine. At various time points after the addition of the unlabeled methionine, 0.5-ml samples were withdrawn and added to 0.1 ml of 0.3% sodium azide. Samples were heated at 100°C for 5 min and analyzed by 15% SDS-PAGE. The gel was dried, fluorographed, and the bands were quantified by scanning with a Molecular Dynamics 300 A computing densitometer.
Immunoblot Analysis of ␤*-E. coli AB1157XL harboring plasmids that carried mutated dnaN* gene fragments expressed from the lac promoter were grown in LB medium supplemented with ampicillin (100 g/ml) and glucose (0.2%) at 30°C to OD 595 ϭ 0.5. The cells were induced with IPTG (0.5 mM) for 1.5 h at 30°C, after which they were precipitated, resuspended to OD 595 ϭ 20 in 10 mM Tris⅐HCl, pH 7.5, 0.15 M NaCl, and frozen in liquid nitrogen. Aliquots of 500 l of the frozen cells were thawed, disrupted by sonication, and analyzed by Western blot analysis. Protein concentration was determined in 0.1 M NaOH according to Bradford (21). Aliquots containing 5 g of protein were separated by 10% SDS-PAGE, blotted to a nitrocellulose membrane, and probed with affinity-purified anti-␤ antibodies as described by Skaliter et al. (48) using enhanced chemiluminescence (ECL, Amer-sham Corp.) for detection.
Nalidixic Acid Induction of the dnaN* Transcript-AB1157 cells were grown to OD 595 ϭ 0.4 -0.5, at which nalidixic acid (40 g/ml) was added. At the desired time interval, samples were withdrawn, and cells were precipitated and resuspended in a solution of 10 mM Tris⅐Cl, pH 7.5, 1 mM EDTA, and 150 mM NaCl. They were frozen and kept in liquid nitrogen. Total RNA was isolated as described elsewhere (22) and further purified by isopicnic centrifugation in CsCl (1.799 g/ml) containing ethidium bromide, followed by isopropanol extraction. Riboprobes were prepared from plasmid pRPHF11, in which the control region of dnaN* was cloned in an orientation opposite to the T7 RNA promoter. The riboprobe was synthesized with T7 RNA polymerase in the presence of [␣-32 P]UTP, after which it was purified on a 6% denaturing urea-polyacrylamide gel. Analysis of transcription initiation sites was performed by the RNase protection assay as described (22). A hundred and fifty g of RNA were hybridize with the radiolabeled riboprobe (2 ϫ 10 6 cpm/reaction) in a final volume of 30 l of a buffer containing 40 mM PIPES, pH 6.4, 1 mM EDTA, 0.4 M NaCl, and 80% formamide. The mixture was heated at 85°C for 5 min, and then it was incubated at 45°C for 12 h to allow annealing of the riboprobe to the specific RNA. The mixture was then cooled to room temperature, and 300 l containing 0.3 M NaCl, 10 mM Tris⅐Cl pH 7.4, 5 mM EDTA, 2 g/ml RNase T 1 , and 40 g/ml RNase A were added. The mixture was incubated for 1 h at 30°C, after which the RNases were digested with proteinase K (0.3 mg/ml) in the presence of SDS (0.6%) at 37°C for 30 min. The RNA hybrid was extracted with phenol/chloroform and precipitated with ice-cold ethanol in the presence of 20 g of tRNA as a carrier. The hybrid was fractionated on a 6% polyacrylamide gel containing urea.
Kinetics of UV Induction of the cat Gene Fusions-E. coli AB1157 cells harboring cat gene fusion plasmids, were grown in M9 medium supplemented with amino acids and glucose at 37°C to OD 595 ϭ 0.1. The cultures were then divided into two 20-ml portions, one of which  (18) FIG. 1. Vectors used to construct cat gene fusions. Plasmid pCAT was derived from pCM4 by eliminating the BamHI site at the 3Ј side of cat, and by inserting the T 1 rrnB transcription terminator into the AatII site upstream to cat. Transcriptional gene fusions to cat were constructed by replacing the tet promoter (P) located on the EcoRV-ClaI fragment with the promoter to be studied. See "Experimental Procedures" for details.
was UV-irradiated and returned for further growth at 37°C. Usually a UV dose of 30 J m Ϫ2 was used. Samples of 1.5-6 ml were withdrawn at various time points, sedimented, and resuspended in an extraction buffer (50 mM Tris⅐HCl, pH 7.5, 30 mM dithiothreitol). The cell suspension was then frozen in liquid nitrogen, thawed, and sonicated for 30 s. The mixture was centrifuged and the concentration of the soluble protein fraction was measured according Bradford (21). CAT activity was assayed essentially as previously described (23), based on the ability of CAT to transfer a butyryl residue from butyryl-CoA to one or two hydroxyl residues on the chloramphenicol molecule, thus causing its inactivation. Butyrylation of [ 3 H]chloramphenicol was followed by extracting the butyrylated chloramphenicol from the reaction mixture with an organic solvent. The reaction mixture (100 l) contained 0.2 Ci of [ 3 H]chloramphenicol (preextracted twice with xylene), 25 g of butyryl CoA, 20 mM Tris⅐HCl, pH 8, and the protein extract. The reaction mixtures were incubated for 1 h at 37°C, after which they were extracted with 200 l of 2:1 tetramethylpentadecane:xylene, and the amount of radioactivity extracted into the organic phase was determined by liquid scintillation counting.

Identification of the ATG Initiation Codon and the Shine-Dalgarno
Sequence of dnaN*-The DNA sequence of the beginning of the dnaN* gene is shown in Fig. 2, including the putative control elements. Two ATG sequences were candidates for the initiation codon of dnaN*: 2043 ATG and 2076 ATG (Fig. 2). The sequence GCAGG, homologous to the Shine-Dalgarno sequence, was located 5 nucleotides upstream to 2043 ATG, whereas no such sequence was found at the appropriate distance upstream to 2076 ATG (Fig. 2), suggesting that 2043 ATG serves as the initiation codon of dnaN*. In order to examine this possibility we have mutated each of the ATG codons into a CTG codon by site-directed mutagenesis.
The mutated dnaN* genes were cloned under the strong phage T7 promoter in plasmid pBluescript SK ϩ . E. coli BL21(DE3) cells harboring these plasmids grew poorly, and they had variations in plasmid copy number. This was caused most likely by the induction of the T7 RNA polymerase due to titration of LacI by the lacP promoter present on the high copy number plasmid. Indeed, deletion of the lacP segment from the plasmids resolved the problem, and the synthesis of ␤* could then be quantitatively monitored by metabolic labeling with [ 35 S]methionine, followed by SDS-PAGE and fluorography. To facilitate detection of ␤*, transcription by the cellular RNA polymerase was inhibited by the addition of rifampicin, such that transcription was selectively initiated from the T7 promoter by the T7 RNA polymerase expressed from a prophage in the host cell. Indeed, as can be seen in Fig. 3, in the presence of rifampicin, the synthesis of ␤* could be easily detected.
The validity of measuring rates of protein synthesis by this procedure depends on the turnover of ␤* in the cell. The halflife of ␤* was determined by pulse-labeling of the protein with [ 35 S]methionine, followed by a chase with unlabeled methio- nine. From the decay in the amount of radiolabeled ␤*, its half-life is approximately 40 min (Fig. 4), much higher than the time scale used to estimate metabolic rates of synthesis. Thus, ␤* is relatively a stable protein, and its degradation is not expected to affect significantly the measurements of its synthesis.
Mutating 2076 ATG (Met2) had essentially no effect on the rate of synthesis of ␤* (Fig. 3). In contrast, mutating 2043 ATG (Met1) caused a 3-fold reduction in the rate of synthesis of ␤* (Fig. 3), suggesting that 2043 ATG is the initiation codon of dnaN*.
To further support this conclusion, we have constructed another set of plasmids, in which the dnaN* gene was cloned under the lac promoter in plasmid pUC18. In this case we detected ␤* by Western blot analysis of cell extracts, using affinity-purified anti-␤ antibodies. As can be seen in Fig. 5  (lanes 7-12), when the lac promoter was repressed by glucose, ␤* was not produced. Upon induction by IPTG, the dnaN* plasmid yielded two products, a major product that comigrated with a sample of ␤* purified from an overproducing cell and a minor product that migrated slightly faster (Fig. 5, lane 2). The major product comigrated with ␤* synthesized in vivo from the chromosome (48). As can be seen in Fig. 5 (lane 3), mutating the 2043 ATG codon resulted in disappearance of the major ␤* band, whereas the minor ␤*-related band remained unchanged. On the other hand, mutating 2076 ATG (Fig. 5, lane 4) eliminated the minor band, and caused also a reduction in ␤*. These results indicate that 2043 ATG is the initiation codon of dnaN*. The minor band seems to be the result of an alternative initiation from 2076 ATG when ␤* was present on a plasmid.
The assignment of 2043 ATG as the initiation codon of dnaN* pointed to the GCAGG sequence as a likely Shine-Dalgarno sequence involved in ribosome binding. In order to examine this possibility we prepared two Shine-Dalgarno double mutants: GCA GG 3 ACA AG (SD-S) and GCA GG 3 GCA AA (SD-M). As can be seen in Fig. 5 (lanes 5 and 6), both mutants exhibited reduced expression of dnaN*, consistent with the suggested role of the GCAGG sequence in ribosome binding.
The dnaN* Transcript Is Induced by Nalidixic Acid-Total RNA was isolated from E. coli cells, and the 5Ј termini of mRNAs initiating at the promoter region of dnaN* were analyzed using the RNase protection techniques. Several transcription initiation sites could be detected in the region analyzed, including the major recF transcript initiating at promoter P1, and a fully protected RNA probe, which represents the overlapping dnaN mRNA (Fig. 6, P N ). In the dnaN* promoter region, a band of approximately 130 bases was detected, suggesting that transcription of dnaN* starts near po-sition 2013 (Fig. 2).
The promoter region of the dnaN* gene contains the sequence 5Ј-CGCTGTCTACCCTGCCAGCG-3Ј (positions 1960 -1979; Fig. 2), resembling the consensus sequence of the binding site of the LexA repressor, 5Ј-NNCTGTNTatNcaNNCAGNN-3Ј (3). The most conserved 8 nucleotides are present in the dnaN* SOS box-like sequence, including the inverted repeat CTGN 10 CAG, which in our case is part of a pentanucleotide inverted repeat, CGCTGN 10 CAGCG. If indeed this sequence binds LexA, it is expected that the gene will be inducible by agents that induce the SOS regulon. In agreement with such a prediction, UV irradiation of E. coli cells was found to cause induction of ␤*, the dnaN* gene product (48). As can be seen in Fig. 6, treatment of cells with nalidixic acid, a potent inducer of the SOS and the heat shock responses, caused a 4 -5-fold induction in the dnaN* transcript. Thus, induction of dnaN* expression is regulated, at least in part, at the transcriptional level.
In order to examine whether dnaN* is controlled directly by LexA, the global SOS repressor, we studied the binding of purified LexA repressor to the promoter region of dnaN*, using the gel mobility shift assay (22). Binding of LexA to the promoter region of recA has been demonstrated by this technique (24). Indeed, the LexA protein caused specific retardation of a 148-bp MspI restriction DNA fragment carrying the recA promoter which served as a positive control (data not shown). However, we could not to detect any specific binding of LexA to the dnaN* promoter region under a variety of condition (data not shown). This suggests that the inducibility of the dnaN* gene is not regulated directly by LexA. The region of the recF promoters located inside the dnaN* gene contains an antisense promoter, termed P x (25) (Fig. 2). The transcript directed by this promoter is complementary the first 86 nucleotides of the dnaN* transcript. We have confirmed the activity of this promoter in our cells and found that its transcript was unaffected by treatment with nalidixic acid (data not shown). The role of this transcript is not clear, although it may function to regulate dnaN* and/or dnaN expression.
Transcriptional dnaN*-cat Gene Fusions Are UV-inducible-In order to examine by an independent method the induction of dnaN* at the transcriptional level we have utilized dnaN*-cat gene fusions. Since dnaN* has a weak promoter, it was necessary to use a vector with a low background of cat activity. Our vector, termed pCAT, carries the coding sequence of the cat gene under the control of the tet promoter (Fig. 1) and the independent T 1 rrnB terminator located upstream to cat. In order to examine the background CAT activity of a promoterless cat gene in plasmid pCAT we have deleted the 162-bp ClaI-EcoRV fragment (23-185), containing the tet promoter and inserted a spacer sequence between the terminator and the cat open reading frame. The spacer was the 141-bp HgaI-(2037)-BstUI-(1896) fragment from the dnaN gene, lacking any known promoter sequence (in the orientation opposite to dnaN*). Protein extracts prepared from cells harboring the resultant plasmid, termed pNCH20, had a background of 0.025 unit. UV irradiation of these cells did not affect CAT activity, as expected.
We have used the recA promoter, a classical SOS promoter, to serve as a positive control for a UV-inducible gene (Fig. 7). The CAT activity of cells harboring plasmid pRC5, containing the recA-cat fusion, was 4 units/mg of protein. UV irradiation at various doses led to the induction of CAT activity peaking at 30 min after irradiation. The extent of induction increased with increasing UV dose, up to an effect of 8-fold at 30 J m Ϫ2 (Fig. 7). The increase in recA transcription as assayed by this recA-cat gene fusion is similar to the results obtained by assaying directly the level of recA mRNA, where a maximal 8 -9-fold induction was found after 20 min (26). Based on these results we used an inducing dose of 30 J m Ϫ2 for examining the UV inducibility of dnaN*.
We have constructed three dnaN*-cat fusions plasmids, containing various portions of the dnaN* gene (Fig. 8). Plasmid pNCB17 contains the 332-bp BstUI (1896)-BstUI (2228) fragment of dnaN containing the two recF promoters, and 147 nucleotides upstream to the initiation codon of dnaN* including the SOS box-like sequence, the promoter, and the Shine-Dalgarno sequence. Plasmid pNCS14 contains the 172-bp SfaNI-(1918 -2090) DNA fragment of dnaN, including the control region of dnaN*, but lacking the recF promoters. Plasmid pNCH6 contains the 141-bp BstUI-(1896)-HgaI-(2037) fragment, containing the SOS box-like sequence and the promoters of dnaN*, but no coding sequences. All three dnaN* gene fragments exhibited weak promoter activities as judged by the level of CAT activity (Fig. 9). The activity varied from 0.04 to 0.1 unit, which is 2-4-fold higher than the background activity of the control plasmid without a promoter (pNCH20). UV irradiation of cells harboring the dnaN*-cat gene fusion plasmids caused a 3-fold induction of CAT activity (Fig. 9), consistent with the dnaN* transcript analysis (Fig. 6). This included plasmid pNCS14 which did not contain the recF promoters. In order to analyze the P x promoter we have constructed a P x -cat fusion using the 138-bp BstUI-(2228)-SfaNI-(2090) DNA frag-FIG. 6. Mapping of transcription initiation sites in the dnaN* control region by the RNase protection technique. Upper panel, cellular RNA was extracted from MC4100 wild-type cells at the indicated time points after induction of the SOS response by nalidixic acid (40 g/ml). The RNA was purified and then hybridized to a uniformly labeled RNA probe transcribed from plasmid pRPHF11. The hybrids were digested with RNase A and RNase T 1 , then treated with proteinase K, extracted with phenol, and separated on a denaturing 6% ureapolyacrylamide gel, after which the gel was dried and autoradiographed. P, P N* and P F1 represent transcription initiation at the promoters of dnaN, dnaN*, and the major promoter of the recF gene, respectively. The weak second promoter of recF, P F2 , is hardly seen under our conditions. The details are presented under "Experimental Procedures." Lower panel, the riboprobe used in the assay shown in the upper panel and the predicted sizes of its protected regions that hybridize to mRNAs initiating inside the dnaN* gene. ment (Fig. 8), containing the P x promoter. The resultant plasmid, termed pPC1, had a basal activity of 0.35 unit, which was 3-4-fold higher than the dnaN* gene fusion (Fig. 9). UV irradiation of cells harboring plasmid pPC1 did not affect CAT activity (Fig. 9). Thus, in contrast to the dnaN*-cat fusions, the antisense P x -cat fusion was not inducible by UV light, in agreement with the P x transcript analysis.
Plasmids pTEN5 and pNB3 gave rise to similar constitutive levels of ␤-galactosidase activity, indicating that the recF promoters did not contribute to the expression of the gene fusion (Fig. 10, A and C). Upon UV irradiation of cells harboring these gene fusions, an increase of ␤-galactosidase activity was observed. The level of induction was 5-10-fold, and was the same also for plasmid pSB6, in which the 5Ј-half of the SOS box-like sequence was deleted (Fig. 10D). The UV induction was completely abolished in isogenic mutant cells with either a lexA3 or recA56 mutation (Fig. 10). The lexA3 mutation renders the LexA repressor non-cleavable by activated RecA protein, whereas the recA56 mutation inactivates the RecA protein, thus the SOS response cannot be induced in cells carrying either of these mutations. The noninducibility of dnaN* in these strains suggested that its expression is under the control of the SOS stress regulon. The uvrA6 mutation, which inactivates nucleotide excision repair, and the umuC36 mutation, that inactivates UV mutagenesis, did not affect the UV inducibility of the dnaN*-lacZ fusions, indicating that the induction was not dependent on excision repair or UV mutagenesis (data not shown).
The control fusion of the dnaA gene did show UV induction consistent with the report on the inducibility of dnaA by mitomycin C (27). However, as can be seen in Fig. 10B, the UV induction was not dependent on the recA gene product, implying that the SOS response was not involved. The negative control for induction was the cI-lacZ fusion that was noninducible by UV irradiation (Fig. 10B).
The kinetics of induction of the dnaN*-lacZ fusions showed peak levels of ␤-galactosidase activities at 3-5 h after irradiation (Fig. 10), whereas many SOS functions (26), as well as the UV induction of ␤* (48) peak an hour or less after irradiation. A similarly slow induction of SOS-inducible genes fused to lacZ was observed before (28). This may be the result of the fact that the active structure of ␤-galactosidase is a tetramer (29), and that oligomerization of the fused ␤-galactosidase molecules might be slow, particularly when their concentration is low. Indeed, when the induction of the ␤*-␤-galactosidase protein was examined at the protein level, by Western blot analysis using polyclonal antibodies against ␤-galactosidase, maximal induction of ␤-galactosidase occurred approximately 60 -90 min after UV irradiation (Fig. 11). This time period is close to the time of induction of the ␤* protein (48). This result shows that the UV-induced increase in the activity of ␤-galactosidase was indeed due to an increase in the synthesis of the enzyme, and it is consistent with the suggestion that the slower kinetics of induction of the activity of the fused enzyme was due to the slow rate of assembly of the active tetrameric structure. dnaN* Is under the Control of the Heat Shock Response-UV irradiation and nalidixic acid induce in E. coli both the SOS and heat shock responses (5). In order to test the involvement of the heat shock response in UV induction of dnaN we assayed a dnaN*-lacZ fused gene in cells which lack the heat shock 32 subunit. E. coli R40NL8 is a derivative of E. coli MC4100 that has a deletion in rpoH, the gene encoding 32 . Because 32 is essential for growth at temperatures above 20°C, R40NL8 (like its parent R40) (14) contains a suppressor mutation that causes overproduction of groE, and thus enables it to grow at 37°C (but not at 42.5°C). The kinetics of ␤-galactosidase induction from the dnaN*-lacZ gene fusion was determined after UV-irradiation of the ⌬rpoH30 mutant and its isogenic wildtype parent. As seen in Fig. 12B, there was no UV induction of ␤-galactosidase activity from pTEN5 in cells which lack the heat shock 32 subunit, whereas a 10-fold increase in activity was found in the isogenic wild-type cells (Fig. 12A). Interestingly, the basal level of ␤-galactosidase from plasmid pTEN5 was the same in the wild-type and ⌬rpoH30 strains. In order to establish that the loss of UV inducibility in strain R40NL8 was indeed due to the absence of the 32 subunit, we supplied it in trans from plasmid pCM1. This plasmid did not affect the UV inducibility of the dnaN*-lacZ fusion from pTEN5 in the wildtype parent MC4100 (Fig. 12A). However, introduction of pCM1 into R40NL8 harboring pTEN5, rendered the dnaN*-lacZ fusion UV-inducible again (Fig. 12B). The extent of UVinduction of the dnaN*-lacZ fusion was 3-fold lower than in the wild-type strain. This may be due to the fact that the strains are not fully isogenic (i.e. R40NL8 but not MC4100 overpro-duces groE). In any case, it is clear that introducing the plasmid that expressed 32 made the dnaN*-lacZ gene fusion UVinducible again, suggesting that dnaN* is controlled by the heat shock activator, the RNA polymerase 32 subunit.
We attempted transcribe dnaN* in vitro using purified RNA polymerase. We were unable to detect any in vitro initiation of transcription from the dnaN* promoter using either the regular 70 RNA polymerase, or the heat shock-specific 32 RNA polymerase, although the recF transcripts were observed (data not shown). Thus, it seems that transcription of dnaN* requires additional factors, or possibly another subunit. Possible candidates are 24 , which is specific for some heat-induced genes (30,31), and s , which transcribes stationary phase genes (32,33). Consistent with such a possibility we found a higher amount of ␤* in stationary phase cells (48). DISCUSSION We have previously shown that a smaller form of the ␤ subunit of DNA polymerase III holoenzyme is induced in E. coli by UV irradiation (48). Such a protein can be generated by proteolytic processing, like the mutagenesis protein UmuDЈ, that is formed from UmuD by specific proteolysis promoted by the RecA protein (10). Alternatively, the protein can be translated from the dnaN mRNA by a de novo translational start, or it can be expressed from an internal in-frame gene.
The data presented here suggests that ␤* is expressed from an internal in-frame gene termed dnaN*. This is based on the following observations. 1) The ATG initiation codon of dnaN* and its Shine-Dalgarno sequence were identified by site-directed mutagenesis. 2) A transcription initiation site was mapped inside dnaN, upstream to a Shine-Dalgarno sequence.
3) Plasmids carrying the dnaN* gene expressed ␤*. 4) When cloned into a plasmid, the promoter region of dnaN* directed the expression of a promoter-less cat gene. 5) When the control region of dnaN*, including the beginning of its coding region, was fused in-frame to a portion of the lacZ gene lacking all transcriptional and translational control elements as well as its first 8 codons, it directed the synthesis of a fused ␤*-␤-galactosidase protein.
The expression of dnaN* is complex and is likely to be regulated via several mechanisms. Transcription of dnaN* was not observed in vitro using either 70 or 32 RNA polymerase, suggesting that another transcription factor is required. Internal initiation of translation at the dnaN* ATG initiation codon on the intact dnaN mRNA seems to be very inefficient. This is FIG. 11. Kinetics of UV-induction of a ␤*-␤-galactosidase fused protein assayed by immunoblot analysis. E. coli MC4100 cells harboring plasmid pTEN5 or the control plasmid pMC1403 were UVirradiated at 50 J m Ϫ2 and assayed for the induction of the ␤*-␤galactosidase protein by immunoblot analysis using anti ␤-galactosidase antibodies and the enhanced chemiluminescence method for detection. Lane M contains a marker of ␤-galactosidase. indicated by the fact that overexpressing dnaN mRNA from the lac promoter on a plasmid did not yield any detectable ␤*. Only after introducing a frameshift mutation into dnaN, upstream to dnaN*, that eliminated overproduction of the ␤ subunit, expression of ␤* was observed from dnaN mRNA (48). Thus, it seems that, under normal conditions, synthesis of ␤* from dnaN mRNA is strongly inhibited, e.g. due to its engagement in translation of the ␤ subunit or due to direct inhibition by the ␤ subunit. The antisense transcript originating from P x , may also be involved in the down-regulation of the expression of dnaN*.
UV induction of dnaN* is regulated at the transcriptional level, and subjected to control by both the SOS and heat shock responses, as indicated by the dependence of UV induction of dnaN*-lacZ gene fusions on recA, lexA, and rpoH. However, this dual regulation is indirect, since dnaN* did not bind LexA, and it was not transcribed by 32 RNA polymerase. Thus, another factor(s) that is controlled by these major stress responses, is responsible for the UV induction of dnaN*. The role of the SOS box-like sequence in the promoter region of dnaN* is puzzling. It may represent a degenerated LexA binding site, or it may be a coincidental homology of no functional role, especially since its 5Ј-half was found to be dispensable for UV induction of dnaN*-lacZ gene fusions. It should be noted that if the sequence 5Ј-TACTGTATATATATACAGTA-3Ј is taken as the consensus LexA binding site, then based of the differences between it and the dnaN* SOS box-like sequence (34), the latter is predicted to have no specific binding to LexA. Similar SOS box-like sequences, that did not bind LexA, were found in the phr gene, encoding DNA photolyase (35), and in the uvrC gene, encoding a subunit of the UvrABC repair excinuclease (36); however, their significance remains unclear. In addition to dnaN* at least three other genes are inducible by DNA-damaging agents in a recA-and lexA-dependent pathway, but are not directly regulated by LexA: The phr gene mentioned above (35), the dnaQ gene encoding the proofreading ⑀ subunit of DNA polymerase III (37), and the dnaN gene (37,38). The mechanism of this regulation is unknown yet, representing another layer of complexity of the SOS regulatory network. It may be performed by a factor which is by itself repressed directly by LexA.
Genes whose coding sequences overlap are not rare; however, extensively overlapping genes, or genes nested within other genes, are not common in the chromosome (39). A well documented case is the phage T7 gene gp4, encoding a helicaseprimase. The gene encodes two proteins of 63 and 56 kDa, the latter generated by an internal in-frame start site (40). The dnaX gene encodes two subunits of DNA polymerase III holoenzyme: and ␥. They both start at the same site, but ␥ is terminated before by a mechanism of ribosomal frameshifting, leading to the production of proteins of 47.5 and 71 kDa (7).
The expression of the internal dnaN* gene produces a protein that lacks precisely one of the three repeating domains of the ␤ subunit. In this respect it belongs to a family of mechanisms such as alternative splicing, that produce from a single gene more than one protein, differing by a one or more defined functional domains. Such mechanisms generate a protein (or more) with a subset of the properties of the parental intact protein. They might be required to fulfill biochemically similar reactions under different conditions, or with conjunction with different counterpart proteins. Such are the cases of the dnaX gene and the T7 gp4 genes. The intact ␤ subunit forms a ␤ 2 ring-shaped sliding DNA clamp, that confers high processivity on DNA polymerase III holoenzyme by tethering it to the DNA (41,42). As shown in a companion study (49), ␤* forms an alternative DNA clamp for DNA polymerase III that may have a specialized function connected to DNA synthesis in the UVirradiated cell. The increase in UV resistance caused by overproducing ␤* is consistent with such a model (48).