Characterization of the DNA Damage-inducible Helicase DinG from Escherichia coli *

The dinG promoter was first isolated in a genetic screen scoring for damage-inducible loci in Escherichia coli (Lewis, L. K., Jenkins, M. E., and Mount, D. W. (1992) J. Bacteriol. 174, 3377–3385). Sequence analysis suggests that the dinG gene encodes a putative helicase related to a group of eukaryotic helicases that includes mamma-lian XPD (Koonin, E. V. (1993) Nucleic Acids Res. 21, 1497), an enzyme involved in transcription-coupled nucleotide excision repair and basal transcription. We have characterized the dinG gene product from E. coli using genetic and biochemical approaches. Deletion of dinG has no severe phenotype, indicating that it is non-essential for cell viability. Both dinG deletion and overexpression of the DinG protein from a multicopy plasmid result in a slight reduction of UV resistance. DinG, purified as a fusion protein from E. coli cells, behaves as a monomer in solution, as judged from gel filtration experiments. DinG is an ATP-hydrolyzing enzyme; sin-gle-stranded (ss) DNA stimulates the ATPase activity 15-fold. Kinetic data yield a Hill coefficient of 1, consistent with one ATP-hydrolyzing site per DinG molecule. DinG possesses a DNA helicase activity; it translocates along ssDNA in a 5 (cid:1) 3 3 (cid:1) direction, as revealed in experiments with substrates containing non-natural 5 (cid:1) –5 (cid:1) and

dinG was identified as a DNA damage-inducible gene in a genetic screen scoring for induction of the transcription of galactokinase gene fusions after treatment of Escherichia coli cells with mitomycin C. Transcription of the dinG::galK fusion was suppressed by overexpression of the LexA protein, suggesting the SOS nature of the induction (1). Indeed, sequencing of the dinG promoter revealed an asymmetric nucleotide sequence TTG(N 10 )CAG that was similar but not identical to the canonical, fully symmetrical CTG(N 10 )CAG SOS box. Despite the deviation of the SOS box found in the dinG regulatory region from the consensus sequence of the LexA-binding box, the double-stranded (ds) 1 oligonucleotide TTGG(N 8 )ACAG bound the LexA repressor with high affinity in an electrophoretic mobility shift assay (2). The dinG promoter was also up-regulated upon DNA damage by nalidixic acid (3).
dinG, along with lexA and dinI, was isolated in another genetic screen aimed at isolating multicopy suppressors of the cold-sensitive phenotype of the DinD68 mutation. This particular mutation in the DNA damage-inducible dinD gene, which is also regulated by the LexA-RecA system, results in the constitutive expression of the SOS response at lowered temperature (Ͻ20°C) (4). Because both dinG and dinI are part of the SOS response (1,2,5) and they suppress an SOS phenotype of the dinD68 mutation (6), dinG could also be a negative regulator of the SOS response in a manner similar to dinI (7,8).
Analysis of the protein sequence of E. coli dinG reveals that it encodes a putative DNA helicase related to yeast DNA helicases Chl1 and Rad3 from Saccharomyces cerevisiae, Rad15 from Schizosaccharomyces pombe, and the human helicases XPD and BACH1 (9,10). The mutant forms of the last two proteins result in well described human diseases, three human recessive photosensitive syndromes for XPD, and early onset breast cancer for BACH1 (10,11). DinG and its eukaryotic counterparts have been classified as superfamily II helicases on the basis of the presence of seven canonical helicase motifs in their sequences (9,12,13). Still, the presence of the helicasespecific motifs in the protein amino acid sequence per se does not necessarily imply that it is a bona fide helicase. Proteins having helicase motifs but lacking a helicase activity are well known. Among them are the endonuclease (R) subunits of type I and type III restriction-modification enzymes (14), both bacterial and human transcription-repair coupling factors Mfd (15), and CSB/ERCC6 (16), members of SWI2/SNF2 family chromatin remodeling factors (17) and the RAD54 recombinational DNA repair protein (18).
To prove that DinG is a true helicase, we carried out the purification and biochemical characterization of the E. coli DinG protein. In agreement with the prediction (9), DinG possesses DNA-dependent ATPase and helicase activities. We discuss the possible biological role that DinG helicase might play.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Plasmids-Gene deletions were created using a combination of described procedures (19,20). To construct the metB gene deletion, plasmid pKD3 (19) was used to amplify a deletion cassette containing the cat gene flanked by FLP recognition target sites and 36-nucleotide extensions homologous to regions of DNA directly * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  surrounding the metB gene. When the deletion cassette was transformed into strain DY330 (W3110 ⌬lacU169 gal490 ⌬cl857 ⌬(cro-bioA)) (20), expressing recombination functions, the cat gene replaced the chromosomal copy of metB. ⌬metB mutants were selected by screening for chloramphenicol resistance and verified by PCR. The dinG gene deletion was created with a similar approach, using plasmid pKD4 (19) to amplify a deletion cassette containing the kan gene and screening for kanamycin resistance. Strain FV002 was constructed by using P1 phage transduction to transfer the metB gene deletion into strain MG1655 (F Ϫ Ϫ ilvG Ϫ rfb-50 rph-1). Strain FV004 was constructed by transferring the metB and dinG gene deletions into strain MG1655. The cat and kan genes were removed from the strains by expressing FLP from plasmid pCP20 (19), resulting in 85-nucleotide scars in place of the inserted resistance genes.
The dinG gene was amplified from E. coli strain W3110 genomic DNA (Sigma) with either DinG-5Ј-pET30 and DinG-3Ј-pET30 or DinG-5Ј-Xba and DinG-3Ј-Xho pairs of primers. The former PCR fragment was inserted into the pET-30 Xa/LIC expressing vector (Novagen) with a ligation-independent cloning procedure, resulting in plasmid pET30DinG. The latter PCR fragment was cloned between XbaI and XhoI sites of pBluescript II SKϩ (Stratagene) yielding plasmid pDinGPro. pET30DinG drives the T7 polymerase-mediated expression of the DinG protein in the form of an 86.5-kDa fusion with a His 6 tag and an S tag at the N terminus. The fusion part of the recombinant protein can be cleaved off precisely by treatment with factor Xa, producing the intact 81.5-kDa DinG protein. Constitutive expression of the DinG protein from pDinGPro plasmid was confirmed by Western blot analysis (data not shown). Nucleotide sequences of the inserts in the created plasmids were verified by DNA sequencing.
DNAs-All natural and modified oligodeoxynucleotides with switched polarity (21) were synthesized on an Applied Biosystems 380A synthesizer and purified by denaturing PAGE. Sequences of the synthetic oligodeoxynucleotides are shown in Table I. M13mp18 ssDNA was from U. S. Biochemical Corp. The helicase substrate M1, used in most experiments, was prepared by annealing the 32 P-labeled oligonucleotide 55B to M13mp18 ssDNA. Bifurcated substrates, termed D1, P1, P2, and P3, were made by hybridization of oligonucleotides 32 P-T8 and B9, 32 P-T8 -3Ј-3Ј and B9, 32 P-T8 and B9 -5Ј-5Ј, and 32 P-T8 -3Ј-3Ј and B9 -5Ј-5Ј, respectively. DNAs were hybridized by slow cooling down from 85°C to room temperature in annealing buffer (50 mM Tris⅐HCl at pH 7.4, 50 mM NaCl, 10 mM MgCl 2 ). An excess of free 55B oligonucleotides was separated from annealed M1 substrate by gel filtration on a Chromaspin 400 column (Clontech) equilibrated with annealing buffer. Bifurcated substrates were purified after electrophoresis in 10% polyacrylamide gel and were recovered in annealing buffer. Duplex BS (22), composed of two fully complementary 53-mer BS-1 and BS-2 oligonucleotides, was gel-purified and stored in TE buffer (10 mM Tris⅐HCl at pH 7.5, 1 mM EDTA). Unless specified otherwise, the DNA concentration is expressed in terms of molecules.
UV Sensitivity Assay-We studied UV sensitivity of three strains: FV002 transformed with pBluescript II SKϩ, FV004 (⌬dinG) transformed with pBluescript II SKϩ, and FV002 overexpressing DinG from the multicopy plasmid pDinGPro. Overnight cultures were prepared in LB media supplemented with 100 g/ml ampicillin. One hundred microliters of serial dilutions of the cells were plated on LB plates with 100 g/ml ampicillin and irradiated with different doses of shortwave UV light from a Stratalinker 1800 (Stratagene). The plates were incubated in the dark overnight at 37°C, and the colonies derived from surviving cells were counted.
Purification of the DinG Protein-Overexpression of the DinG protein was performed in BLR(DE3)pLysS strain (F Ϫ ompT hsdS B (r B Ϫ m B Ϫ ) gal dcm ⌬(srl-recA)306::Tn10(Tc R ) (DE3) pLysS (Cm R )) trans-formed with pET30DinG upon induction with IPTG. Four flasks with 1 liter of a bacterial culture in LB medium supplemented with 30 g/ml kanamycin and 25 g/ml chloramphenicol were grown at 37°C to A 600 ϭ 0.6, and the protein expression was induced with 1 mM IPTG over a 1.5-h period. The cells were collected by centrifugation at 7,000 rpm for 10 min in a Sorvall SLC-400 rotor, washed with 80 ml of Buffer I (50 mM sodium phosphate at pH 8.0, 300 mM NaCl, 10 mM imidazole, 10% glycerol, and 0.1% Triton X-100), and frozen at Ϫ70°C. All purification procedures were performed at 4°C. The cell paste was resuspended in 50 ml of Buffer II (50 mM sodium phosphate at pH 8.0, 1 M NaCl, 10 mM imidazole, 10 mM ␤-mercaptoethanol, 0.1% Triton X-100, and 30% glycerol) and sonicated. The lysate was cleared by centrifugation at 18,000 rpm for 1 h in an SS-34 Sorvall rotor. The supernatant was added to 20 ml of drained and equilibrated with Buffer II Ni 2ϩ -NTA-agarose (Qiagen). Binding of the His 6 -tagged protein to the resin was done in batch on a rocker overnight. The slurry was transferred to a disposable column and washed 6 times with 12 ml of Buffer II. The protein was eluted with 72 ml of Buffer III (50 mM sodium phosphate at pH 8.0, 1 M NaCl, 250 mM imidazole, 10 mM ␤-mercaptoethanol, 0.1% Triton X-100, and 30% glycerol). Fractions of 6 ml were collected and analyzed by SDS-PAGE. Three DinG-containing fractions were combined and concentrated to a volume of 10 ml with a Vivaspin 20 concentrator (Sartorius, 50,000 molecular weight cut-off). The buffer was exchanged to Buffer IV (50 mM sodium phosphate at pH 6.8, 100 mM NaCl, 10 mM ␤-mercaptoethanol, 0.1% Triton X-100, and 30% glycerol) on a HiPrep 26/10 desalting column (Amersham Biosciences), and the sample was applied to a tandem of MonoQ HR 5/5/MonoS HR 5/5 columns (Amersham Biosciences). After the MonoQ column was disconnected, a MonoS column was developed with 80 ml of linear gradient Buffer IV to Buffer V (50 mM sodium phosphate at pH 6.8, 600 mM NaCl, 10 mM ␤-mercaptoethanol, 0.1% Triton X-100, and 30% glycerol). Fractions containing DinG (eluted at ϳ150 mM NaCl) were pooled and concentrated on Vivaspin 20, and the protein was transferred to Buffer VI (50 mM Tris⅐HCl at pH 8.0, 100 mM NaCl, 10 mM ␤-mercaptoethanol, and 30% glycerol) using a desalting column. The protein was applied to a HiTrap Blue column (Amersham Biosciences) and was eluted with 30 ml of a linear gradient Buffer VI to Buffer VII (50 mM Tris⅐HCl at pH 8.0, 2 M NaCl, 10 mM ␤-mercaptoethanol, and 30% glycerol). The DinG protein eluted at ϳ1 M NaCl was concentrated, dialyzed against Buffer VIII (25 mM Tris⅐HCl at pH 8.0, 1 M NaCl, 10 mM ␤-mercaptoethanol, and 50% glycerol), and stored in aliquots at Ϫ80°C. The concentration of the fusion protein was measured using the extinction coefficient of 80,435 M Ϫ1 cm Ϫ1 at 278 nm, calculated with a ProtParam tool (www.expasy.ch/tools/protparam.html). In all experiments, described in this paper, we use fusion His 6 tag-S tag-DinG and refer to it as DinG.
Analytical Sizing Chromatography-The molecular mass of DinG in different conditions was determined by gel filtration on a Superdex 200 PC 3.2/30 column (Amersham Biosciences) equilibrated with the running buffer (25 mM Tris⅐HCl at pH 8.0, 1 M NaCl, 10 mM ␤-mercaptoethanol, and 100 M EDTA). The separation was conducted on a SMART System (Amersham Biosciences) at 40 l/min. DinG resuspended in the running buffer was injected at concentrations of 0.86 and 20.9 M. The molecular mass of DinG was estimated as described in the manual of the gel filtration calibration kit (Amersham Biosciences) using a calibration curve obtained with ribonuclease A (13.7 kDa), bovine serum albumin (67 kDa), ferritin A (440 kDa), thyroglobulin (669 kDa), and blue dextran 2000 (Ϸ2000 kDa). Molecular weight markers were run under conditions identical to those used for DinG.
ATPase Assay-The steady-state ATPase activity of DinG was measured using a coupled enzymatic system (23) as described (24). Reaction mixtures contained 50 nM DinG protein, 7.8 -1000 M ATP, 42.5 mM Tris⅐HCl at pH 7.5, 100 mM NaCl, 5 mM MgCl 2 , 1 mM ␤-mercaptoethanol, 0.1 mM EDTA, 5% glycerol, 50 g/ml bovine serum albumin, 2 mM phosphoenolpyruvate, 150 M ␤-NADH, 200 units/ml pyruvate kinase, and 100 units/ml lactate dehydrogenase. If DNA was included in the reaction, its concentration was 6 M, expressed as the concentration of bases for ssDNA or of base pairs for dsDNA. Hydrolysis reactions were performed in a total volume of 100 l at 37°C. The time course of the reaction was monitored using the kinetics mode of a Hewlett-Packard 8453 spectrophotometer equipped with an 89090A temperature controller and Peltier thermostated cell holder. V max and K m values were calculated from the initial rates of ATP hydrolysis using the EnzymeKinetics version 1.11 program (Trinity Software).
Helicase Assays-In a typical experiment, 2 nM helicase substrate M1 or 0.5 nM bifurcated substrate was incubated with DinG in 15 l of the reaction mixture containing 42.5 mM Tris⅐HCl at pH 7.5, 100 mM NaCl, 5 mM MgCl 2 , 1 mM ␤-mercaptoethanol, 0.1 mM EDTA, 5% glycerol, 50 g/ml bovine serum albumin, and 2 mM ATP. Unless otherwise specified, the concentration of DinG was 100 nM, and incubations were conducted for 20 min. at 30°C. Every experiment contained a "positive control" in which the helicase substrate was denatured by incubation at 95°C for 5 min without DinG. We could detect no more than 5% of reannealing of the denatured 2 nM M1 substrate after incubation for 20 min at 30°C in the standard buffer (data not shown). In some cases, Tris⅐HCl at pH 8.0 or sodium cacodylate at 6.0 was used instead of Tris⅐HCl at pH 7.5; ADP or ATP␥S was used in place of 2 mM ATP; 5 mM MgCl 2 was substituted with 20 mM EDTA, 5 mM ZnCl 2 , 5 mM MnCl 2 , or 5 mM CaCl 2 . After incubation, the reactions were placed on ice and quenched by addition of SDS and EDTA to 1% and 10 mM, respectively. Deproteinized products were separated in 1% TAE-agarose gels for 1.5 h at 5 V/cm (M1 substrates) or in 10% (19:1) TTE-polyacrylamide gels for 1.5 h at 0.25 V/cm (substrates D1, P1, P2, and P3). Gels were dried and quantitated using a BAS 2500 Bio-Imaging Analysis System (Fuji Medical Systems). Percent of the substrate unwound by helicase was calculated using Equation 1: where C Prod,i and C Substr,i are counts of the displaced strand and the remaining substrate, respectively, and C Total,i is their sum. Equation 1 contains two factors ␣ and ␤, which correct Equation 1 for helicaseindependent melting of the substrate and reannealing of the unwound substrates after termination of the reaction. For each set of experiments, the correction factors ␣ and ␤ were calculated using Equation 2 and Equation 3, respectively.
where C Prod,0 are the counts of the single-stranded oligonucleotide generated during storage of the substrate, and the protein-free incubation in the mock experiment and C Substr,0 are the counts of the intact substrate remaining under the same circumstances.
where C Total,H values are the combined counts of the initial substrate C Substr,H and the product C Prod,H in the control experiment where the substrate was heat-denatured.

RESULTS
Overexpression and Deletion of dinG Do Not Significantly Impair UV Sensitivity of E. coli Cells-We tested UV sensitivity of the strains lacking and overexpressing dinG. UV resistance of the bacterial cells is one of the manifestations of the SOS response. Fig. 1 shows that both the null mutation of dinG and overexpression of the protein from the multicopy pBluescript II SKϩ-derived plasmid confer only a mild UV sensitivity to the FV002 strain, which is wild type with respect to recombination, DNA repair, and replication functions. The lack of severe UV sensitivity in a ⌬dinG strain is in agreement with the observation of Yasuda et al. (6), who reported that inactivation of dinG by mini-Tn10 did not have a considerable effect on the survival of cells with a dnaA5(Ts) rnhA genetic background after UV irradiation. At the same time, these results are different from those for the dinI gene, where deletion of dinI had virtually no effect (7), and its overexpression led to a significant UV sensitivity (7,8).
Purification of DinG-The fusion DinG protein was expressed in BLR(DE3)pLysS cells carrying the pET30DinG plasmid. Induction with IPTG resulted in appearance of an 86-kDa polypeptide in cell lysates (Fig. 2, compare lanes 2 and 3). Chromatography on Ni 2ϩ -NTA was the initial step in the purification procedure (Fig. 2, lanes 5 and 6). Next we used a combination of two ion exchange columns, MonoQ and MonoS, to further purify DinG. Because the theoretical pI of the fusion protein, calculated with a ProtParam tool, is 7.66, it should bind to a cation exchange, but not to an anion exchange, column at pH 6.8. Accordingly, we passed the imidazole elution fractions through a MonoQ-MonoS tandem column at pH 6.8. Impurities with pI values below pH 6.8 bound to the MonoQ column, whereas the fusion protein passed through MonoQ was retained on the MonoS column. After the MonoS column was disconnected, DinG was eluted from MonoS with a salt gradient (Fig. 2, lane 7). Further purification was achieved after chromatography on a HiTrap Blue column (Fig. 2, lane 8). We stored purified DinG at pH 8.0 in a buffer containing 1 M NaCl at Ϫ80°C. Prolonged storage of the protein at pH 7.5 in a low salt buffer resulted in its precipitation.
DinG Is a DNA-dependent ATPase-Helicases are molecular motors coupling the ATP hydrolysis to unwinding of the complementary strands of DNA or RNA. NTP binding and hydrolysis is a prerequisite for the strand-separating activity for any of the known helicases. We began the biochemical characterization of DinG with the evaluation of its ATPase activity. As evident from Fig. 3A, DinG possesses an ATPase activity that can be stimulated by DNA. ssDNA and dsDNA differ in their ability to activate the ATPase of the DinG protein. Whereas addition of either a short 53-bp (Fig. 3A) or 3-kb DNA duplex (Table II) resulted in only a slight increase in the rate of ATP hydrolysis, single-stranded DNA, either a 55-mer oligonucleotide or long circular M13mp18 DNA, stimulated ATP hydrolysis about 15-fold ( Fig. 3A and Table II). ssDNA also reduced the K m for ATP to a greater extent than dsDNA. Kinetic parameters of the ATP-hydrolyzing activity of DinG are summarized in Table II. The ssDNA-stimulated k cat value of 24 s Ϫ1 places DinG among the most efficient ATPases in the helicase realm. Typical values of stimulated k cat for helicases range from 0.6 to 30 s Ϫ1 (25).
In order to estimate the minimum number of nucleotidebinding sites involved in the ATP hydrolysis reaction, we fitted our kinetic data to the Hill equation. A Hill coefficient of 1, which is the slope of the line in Fig. 3B, suggests only one ATP-binding site per DinG molecule. This is consistent with, although it does not prove, the idea that DinG functions as a monomer during ATP hydrolysis.
DinG Exists as a Monomer in Solution-We used gel filtration to estimate the assembly state of purified DinG in solution. Irrespective of the protein concentration (0.86 or 20.9 M), DinG eluted as a globular protein with a molecular mass of 74 -86 kDa (Fig. 4). This indicates that DinG is a monomer in solution, in accord with the value of the Hill coefficient of 1. In order to minimize nonspecific protein associations and prevent protein precipitation, we performed all the gel filtration experiments in 1 M NaCl, conditions quite different from the intracellular environment. Furthermore, the fact that the helicase is present as a monomer in solution does not necessarily imply that the functional form of the enzyme is monomeric. Helicases can oligomerize when bound to nucleotide cofactors and DNA substrates. Models for DNA unwinding that utilize monomeric, dimeric, and hexameric helicases have been proposed (25)(26)(27)(28).
DinG Is a Helicase-An examination of the amino acid sequence of DinG (9) revealed seven signature helicase motifs placing the DinG protein in the superfamily II of helicases (12). We carried out experiments aimed at demonstrating that the DinG protein is a bona fide helicase. The helicase activity was assayed using substrate M1 that was composed of a 32 P-labeled 55-mer oligonucleotide annealed to M13mp18 ssDNA. Indeed, the DinG protein is a DNA helicase, as evident from Fig. 5. Incubation of the substrate with increasing amounts of the protein leads to the DinG concentration-dependent release of the oligonucleotide 55B from the M1 complex (Fig. 5, A and B). DinG at concentrations below 50 nM resulted in 50% of product formation, and the amount of the displaced strand reached the plateau at a concentration of the protein between 100 and 200 nM. In the experiments shown in Fig. 5A, the reaction was allowed to proceed for 20 min at 30°C. To get a better estimation of the rate of DNA unwinding, we incubated 2 nM M1 substrate with 100 nM DinG at 30°C for different times. As one can see in Fig. 5, C and D, a 55-mer oligonucleotide is displaced nearly quantitatively in 2.5 min. Nevertheless, we would like to point out that the actual rate of unwinding can be much higher because ssDNA displacement is achieved via two consecutive steps: a possible relatively slow binding of the DinG to the helicase substrate and the unwinding step itself.
Reaction Requirements for the DinG Helicase Activity-Next we characterized the cofactor requirements for the DinG helicase activity. Whereas DinG unwound the M1 substrate at mild alkaline conditions, no helicase was observed at slightly acidic pH (Fig. 6A, lane 3). The lack of helicase activity at a pH below neutral may be attributable to the rapid denaturation or precipitation of DinG. This suggestion is in agreement with the observation that DinG precipitates upon prolonged storage at pH 7.5.
As expected, the ability of DinG to unwind a DNA duplex completely depends on the presence of a nucleotide triphosphate cofactor. Whereas DinG acts as a very efficient helicase when ATP is added to the reaction (Fig. 6B, lane 3), we could not detect any unwinding in the presence of nonhydrolyzable ATP analog ATP␥S (Fig. 6B, lane 9) or in the absence of a nucleotide (Fig. 6B, lane 6). Because ATP␥S is a competitive  inhibitor of ATP hydrolysis by DinG (data not shown), these findings suggest that not only binding of ATP but also its hydrolysis is essential for unwinding. ADP did not support unwinding either (Fig. 6B, lane 12). Apparently, the DinG helicase does not discriminate between ribonucleotide and deoxyribonucleotide triphosphates, because dATP is a competent ATP substitute in the helicase reaction. We could see no difference in the DinG-mediated ssDNA displacement when dATP substituted for ATP (Fig. 6D). All enzymatic reactions involving nucleotide triphosphates require divalent cations. DinG is no exception to this rule. The omission of bivalent metals from the reaction resulted in com- plete inhibition of the helicase activity (Fig. 6C, lane 6). As expected, the DinG helicase was fully active in the presence of Mg 2ϩ ions; Mn 2ϩ and Ca 2ϩ were valid substitutes for Mg 2ϩ (Fig. 6C, lanes 12 and 15), but no helicase activity was detected in the presence of Zn 2ϩ (Fig. 6C, lane 9). There are several possible explanations for the absence of helicase activity in the presence of zinc ions. The first consists of the inability of DinG to hydrolyze ATP chelated to zinc, hence the absence of the energy needed for DNA unwinding. A greater stability of DNA or much faster reannealing of unwound ssDNAs in the presence of Zn 2ϩ is a second possibility. This suggestion is supported by the fact that heat denaturation of the M1 substrate in the presence of Zn 2ϩ produced little of the product ssDNA (Fig.  6C, lane 7). Inactivation or denaturation of the enzyme by zinc ions might be another reason for the absence of helicase activity. Finally, it could be a combination of any of these possibilities.
Polarity of DNA Unwinding by DinG-All known helicases translocate along one strand of DNA duplex unidirectionally. The directionality, or polarity, of DNA unwinding is an intrinsic feature of the helicase. It is linked to the biological process in which the protein is mechanistically involved. Hence knowledge of the polarity of unwinding could provide clues about the molecular transaction in which the helicase might be engaged.
To determine the polarity of DNA unwinding by DinG, we used bifurcated substrates D1, P1, P2, and P3 composed of natural and modified oligonucleotides with switched polarity. P-series substrates are derivatives of the D1 substrate, depicted in Fig. 7A. All of the forked substrates share a common feature; they are 34-mer duplexes with unpaired 31-nucleotide extensions at the 3Ј-end of the top strand and at the 5Ј-end of the bottom strand. Although D1 is an "all natural" substrate, P1 has a 3Ј-3Ј linkage in the top strand, P2 contains a 5Ј-5Ј linkage in the bottom strand, and P3 combines both modifications. Non-natural linkages are located exactly at the border of the duplex and the single-stranded region.
The prediction is that, depending on their polarity, 3Ј 3 5Јand 5Ј 3 3Ј-helicases should unwind different subsets of the bifurcated substrates. Both 3Ј 3 5Ј and 5Ј 3 3Ј enzymes should work on D1 substrate, and substrate P3 should be immune to unwinding by either type of the helicase. The polarity of the helicase can be elucidated from its choice between P1 and P2 substrates. A 5Ј 3 3Ј-helicase would unwind P1 but not P2, and a 3Ј 3 5Ј-helicase would be active on P2 but not P1. The above reasoning assumes that DinG is incapable of unwinding DNA from blunt ends; this is the case for DinG. 2 The results of the helicase activity of DinG on bifurcated substrates are presented in Fig. 7, B-F. The protein was active on substrates D1 and P1 only, as expected for a 5Ј 3 3Ј-helicase (Fig. 7, B and C). This is in agreement with the directionality of DNA unwinding by the eukaryotic counterparts of DinG (29,30). Surprisingly, DinG had a much greater activity on the modified P1 substrate compared with the wild type D1 substrate (Fig. 7F). The formation of secondary structures in the 5Ј extension of the bottom strand or interaction between noncomplementary single-stranded tails could account for the re- duced helicase activity observed with D1. Consistent with either of these two possibilities is the fact that the band corresponding to D1 smears during gel electrophoresis (Fig. 7B). Such a behavior is usually indicative of the formation of secondary structures.
To validate use of the substrates with unnatural linkages in assessing the polarity of the DinG-catalyzed DNA unwinding, we also performed control experiments with a more conventional substrate, a blunt-ended duplex with a single-stranded gap. After binding to the single-stranded region, the unidirectionally moving helicase can unwind only one of the two avail-able duplexes. Results obtained with this kind of substrates confirmed that DinG is a 5Ј 3 3Ј-helicase (data not shown). DISCUSSION We characterized biochemical properties of the DNA damage-inducible protein DinG from E. coli (1)(2)(3). DinG is a protein of great interest because it is a bacterial homolog of two human helicases, XPD and BACH1 (Fig. 8), implicated in human disorders. The human XPD protein is a subunit of the large protein complex TFIIH that has a dual role: in basal transcription and in nucleotide excision repair (11,31). Three human recessive photosensitive syndromes, xeroderma pigmentosum (XP), xeroderma pigmentosum/Cockayne syndrome, and trichothiodystrophy, are associated with different mutations in the XPD gene (32). Totally, 14 mutations in XPD have been described in humans: 7 mutations are associated with XP, 6 mutations associated with trichothiodystrophy, and 1 mutation associated with XP/Cockayne syndrome (33). Of them, 7 mutations were found in, or 10 amino acid residues from, helicase motifs. BACH1 is a recently described putative helicase that shares a significant homology with the Rad3/XPD family of eukaryotic helicases (10). BACH1 physically interacts with the C terminus of the BRCA1 protein and contributes to its DNA repair function. That the overexpression of a BACH1 gene carrying Lys-52 3 Arg mutation in the conservative helicase motif I greatly decreased the ability of the cell to repair doublestranded breaks suggests that the helicase activity of BACH1 is required for DNA repair (10). However, the DNA helicase activity of BACH1 has not yet been demonstrated. The potential involvement of BRCA1 in transcription, replication, and DNA repair opens the possibility that BACH1 could participate in any of these cellular processes as well.
Similarly, the DinG helicase, a prokaryotic counterpart of Rad3/XPD and BACH1, could be involved in different aspects of nucleic acid metabolism. As there are no molecular structures for any of the members of this helicase family, the bacterial DinG protein could pave the way for the elucidation of such a structure. Thus, genetic, mechanistic, and structural studies of this very tractable protein could shed some light on the details of how XPD and BACH1 work.
DinG is one of numerous E. coli helicases. Analysis of all the open reading frames found in the genome sequence of E. coli K-12 (34) predicted that the genome encodes nearly two dozen potential nucleic acid unwinding enzymes, DNA and RNA helicases. Twelve E. coli proteins (11 chromosomal and 1 F factorencoded) with biochemically documented DNA-DNA or DNA-RNA strand-separating activity are presented in Table III. Not listed in Table III is a helicase-like enzyme UvrB capable of limited strand separation in a complex with the UvrA protein during nucleotide excision repair (35). Two E. coli helicases, DbpA (36,37), involved in ribosome maturation, and RhlB (38), a subunit of the E. coli RNA degradosome, have been shown to possess RNA-RNA unwinding activity. The genes hepA (39), hrpA and rhpB (40), rhlE (41), phoH and yjhR (42), lhr (43), yejH (34), srmB (44), b1808 (34) and deaD (45) have been predicted to encode DNA and RNA helicases, but the corresponding proteins have never been purified, and their helicase activities have yet to be demonstrated. DinG and UvrD are the only E. coli helicases induced by DNA damage. Currently we can only speculate about the biological function of the DinG protein.
That dinG deletion does not significantly compromise UV resistance of FV004 cells indicates that it is dispensable for induction and maintenance of the SOS response and the repair of photolesions in DNA. The absence of a profound UV-sensitive phenotype in FV002 cells carrying the plasmid pDinGPro implies that DinG probably is not involved in the down-regulation of SOS functions. Nevertheless, the slight but reproducible UV sensitivity of cells lacking or overexpressing dinG suggests its possible involvement in the control of the SOS response and/or DNA repair processes; the loss of a DinG activity might be backed up by a redundant compensatory pathway(s).
The SOS response is not the only cellular process triggered by DNA damage. The alternative pathway of DNA replication, inducible stable DNA replication (iSDR), can be initiated at chromosome sites different from oriC. iSDR was discovered as a DNA damage-resistant replication induced in the absence of protein synthesis (46). In fact, there is a striking similarity between the stimuli that induce SOS functions and those that result in stable DNA replication: thymine starvation, UV irradiation, incubation of dnaB(Tc) mutants at the restrictive temperature, and exposure to DNA-damaging agents (47). Moreover, the genetic requirements for the induction of the SOS and iSDR are the same; both processes are inhibited by a lexA(Ind Ϫ ) mutation and a recA mutation allowing constitutive RecA function. The complete dependence of iSDR on the recombination activity of RecA and the helicase activity of RecBC (48,49) suggests a D-loop model for the initiation of iSDR.  (12) are marked with roman numerals. Amino acid residues identical in at least two proteins are shown in red, and similar residues are shown in blue. The distances between motifs and from the protein termini are indicated with arabic numerals. B, the identity (red) and similarity (blue) between DinG, XPD, and BACH1 in the regions of helicase motifs.
The D-loop model presumes generation of double-strand breaks (DSBs) at the oriM, the origin for iSDR. The DSB is introduced by a hypothetical oriM-specific endonuclease induced during the SOS response (47). At the same time, DSBs generated at sites of DNA lesions (50) can be converted into substrates for the homologous recombination machinery (tailed duplexes and relatively short fragment of ssDNA) after processing by the helicase and endonuclease activities of the RecBCD enzyme. This would result in the uncontrolled formation of D-loops in numerous illegitimate chromosomal locations. Such "junk" D-loops will interfere with the normal progression of the replication fork and must be removed to ensure efficient replication. A structure-specific helicase recognizing and specifically destroying D-loops could be responsible for clearing the DNA duplex. DinG could play the role of such a "janitor" sweeping the road ahead of a replication fork. Our observations that the helicase activity of DinG could be modulated by the secondary structure of the substrate (Fig. 7, B and F, substrate D1) and by the nature of the substrate ends (compare substrates M1 and P1 in Fig. 5B and Fig. 7F) points to the possibility that the DinG protein is a structure-specific helicase that unwinds, among other structures, D-loops. We are currently testing the structure specificity of the DinG-mediated DNA unwinding. Our preliminary observations suggest that DinG can unwind the heteroduplex part of the model D-loop structures (data not shown).
Constitutive SDR is another type of oriC-independent replication. It is primed at oriK sites from R-loops that are generated by annealing an RNA transcript to one of the strands in a DNA duplex accompanied by the displacement of the complementary strand. Normally, constitutive SDR is suppressed by the presence of RNase H, the enzyme that degrades RNA in DNA-RNA heteroduplexes. However, mutations in the rnh gene allow this alternative pathway of DNA replication under conditions where physiological replication from oriC is impossible, e.g. in dnaA(Ts) strains under a non-permissive temperature (51). Ohmori and co-workers (6) have shown that dinG mutants created in dnaA5(Ts) rnhA strain grow normally in minimal media but are temperature-sensitive when placed in rich media. They proposed that the absence of a functional DinG did not permit the removal of RNA transcripts hybridized to DNA under conditions of active transcription (6). This speculation is consistent with a role for DinG in the removal of R-loops. The S. cerevisiae RAD3 has been shown to possess DNA⅐RNA helicase activity (52). In this regard, it would be of interest to test the ability of DinG to unwind DNA-RNA hybrids and R-loop structures.