Active Site of (A)BC Excinuclease I. EVIDENCE FOR 5’ INCISION BY UvrC THROUGH A CATALYTIC SITE INVOLVING Asp”’, Asp4”, Asp4Ffi, AND His5:3R RESIDUES*

(A)BC excinuclease of Escherichia coli removes dam- aged nucleotides from DNA by hydrolyzing the 8th phosphodiester bond 5’ and the 5th phosphodiester bond 3‘ to the modified base. The activity results from the ordered action of UvrA, UvrB, and UvrC proteins. The role of UvrA is to help assemble the UvrB .DNA complex, and it is not involved in the actual incision reactions which are carried out by UvrB and UvrC. To investigate the role of UvrC in the nuclease activity a subset of His, Asp, and Glu residues in the C-terminal half of the protein were mutagenized in vitro. The effect of these mutations on UV resistance in vivo and incision activity in vitro were investigated. Mutations, H538F, D399A, D438A, and D466A conferred ex- treme UV sensitivity. Enzyme reconstituted with these mutant proteins carried out normal 3’ incision but was completely defective in 5’ incision activity. Our data suggest that UvrC makes the 5’ incision by employing a mechanism whereby the three carboxylates acting in concert with H538 and a Mg2+ ion facilitate nucleophilic attack by an active site water molecule. (A)BC excinuclease removes damaged nucleotides from DNA by incising the damaged strand both 5‘ and 3‘ to the lesion (Sancar and Sancar, 1988). The three subunits of the enzyme,

(A)BC excinuclease removes damaged nucleotides from DNA by incising the damaged strand both 5' and 3' to the lesion . The three subunits of the enzyme, UvrA, UvrB, and UvrC function in a partially overlapping manner in the overall repair reaction (Orren andSancar, 1989, 1990;Bertrand-Burggraf et al., 1991): UvrA which has affinity for damaged DNA (Seeberg and Steinum, 1982) associates with UvrB (which has no affinity for DNA) to form an A2B, complex; this complex tracks along DNA (Koo et al., 1991) and delivers UvrB to the damage site and UvrA dissociates from the complex in an ATP-dependent reaction (Orren andSancar, 1989, 1990). The UvrB.DNA complex so-formed is extremely stable ( tIl2 2-3 h) and constitutes a high affinity site for UvrC which upon binding to the UvrB.DNA complex triggers the dual incisions. Thus, the nuclease active site(s) must be located in UvrB, UvrC, or at the UvrB . UvrC interface. In the present study we have used site-specific mutagenesis to find out whether UvrC functions as a nuclease and to identify the active site residues in this subunit.
In our selection of targets for site-directed mutagenesis we were guided by reaction mechanisms of other nucleases. In nearly all nucleases whose reaction mechanisms are known hydrolysis of the phosphodiester bond occurs by general acid-base catalysis (Saenger, 1991). An active site water molecule which is activated either by a metal (Weber et al., 1991;Volbeda et al., 1991;Derbyshire et al., 1991;Beese et al., 1991;Nakamura et al., 1991) or a histidine residue (Suck and Oefner, 1986;Suck et al., 1988) attacks the phosphate, causing cleavage of the phosphodiester bond. Furthermore, in most nucleases characterized to date (excluding site-specific recombinases) His, Asp, and Glu are found in the catalytic site (Saenger, 1991). These residues are either directly involved in activating the water molecule or bind a metal ion which in turn activates the water molecule for nucleophilic attack on the phosphodiester bond. Therefore, we limited our sitespecific mutagenesis to these three types of residues in UvrC. We found that mutations in 4 residues H538, D399, D438, and D466 specifically abolish the 5' incision activity of (A)BC excinuclease without affecting the 3' incision. We conclude that UvrC is the subunit which makes the 5' incision.
UV Suruiual-Cells were grown in Luria broth containing the appropriate antibiotics. The cells were diluted in phosphate-buffered saline, plated on Luria agar plates, and irradiated with 254 nm from a General Electric germicidal lamp at a fluence rate of 5 microwatts/ cm2. The colonies were counted after incubation at 37 "C for 24 h.
Alkaline Sucrose Gradients-The DNA of UV-irradiated cells was analyzed by sedimentation in alkaline sucrose gradients as described by Tang and Ross (1985). Briefly, cells were grown in K medium containing [3H]thymidine (10 wCi/ml) and 200 wg/ml deoxyadenosine to 10' cells per ml, irradiated with 20 Jm-' of 254-nm light. After irradiation the cells were washed with phosphate-buffered saline, resuspended in K medium and incubated for 1 h at 37 "C. Following incubation, 0.2 ml of cells were converted to spheroplasts as described and then layered on top of 3.5 m15 to 20% alkaline sucrose gradient (0.1 N NaOH, 0.1 M NaC1, 0.01 M EDTA). Centrifugation was at 35,000 rpm for 75 min at 18 "C in an SW-60 rotor. Fractions of 0.16 ml were collected, and DNA was precipitated with 10% trichloroacetic acid. One ml of Scintiverse I1 was added directly to the pellet, and the radioactivity in each fraction was quantified by scintillation counting.
Site-specific Mutagenesis-Site-specific mutants of uurC gene were constructed as described by Kunkel et al. (1987) using the Muta-gene 17688 This is an Open Access article under the CC BY license.

Active
Site of (A)BC Excinuclease M13 in uitro mutagenesis kit (Bio-Rad) following the manufacturer's instructions. Briefly, the uurC gene was excised from the overproducing plasmid pDR3274 (Sancar and Rupp, 1983;Sancar et al. 1984) and inserted into M13 mp18 or M13 mp19. Mutations were generated in these constructs by oligonucleotide-directed mutagenesis, mutants were identified by single-stranded DNA sequencing, and the mutated genes were then replaced the wild type uurC in pDR3274. Mutant constructs were confirmed by double-stranded DNA sequencing. Single-and double-stranded DNA sequencing were performed with the Sequenase DNA sequencing kit (United States Biochemical). "Mini-prep" UurC Purification-The following procedure was developed for rapid purification of small quantities of mutant UvrC proteins for preliminary characterization.
Five-ml cultures were grown to AaIo 0.8 a t which time isopropyl-0-D-thiogalactoside was added to 1 mM, and incubation was continued a t 37 "C for 8-12 h. Cells were collected by centrifugation, resuspended in 0.5 ml of buffer A, transferred to 1.5-ml microcentrifuge tubes, and then frozen in a dry ice-ethanol bath. The cells were thawed on ice and sonicated with a Bronson Model W185 sonifier equipped with a micro-tip. Cell debris was removed by centrifugation. The supernatant was transferred into another tube, and 40 pl of single-stranded DNA cellulose suspension in buffer B (-20-p1 packed volume) was added. The tube was inverted gently several times to mix the resin and bind UvrC. The mixture was centrifuged for 5 min, decanted, and the pellet was washed twice with 0.5 ml of buffer B + 0.3 M KCl. Then, UvrC was eluted by resuspending the resin in 200 pl of buffer B + 1.0 M KC1, and the eluted protein was separated from the resin by centrifugation. A 2-pl aliquot of the protein sample was directly used to test for incision activity, and a 50-p1 sample was applied to sodium dodecyl sulfatepolyacrylamide gel electrophoresis to test recovery and purity. Typically 5 pg of UvrC at 40-50% purity was obtained. When necessary the mutant proteins were purified in large scale by the method of Thomas et al. (1985) to >90% purity.
The (A)BC excinuclease reactions were performed in 25-p1 ABC buffer (50 mM Tris-HC1, pH 7.5, 100 mM KC], 10 mM MgCl', 2 mM ATP, 5 mM dithiothreitol, and 50 pg/ml bovine serum albumin) containing 0.1 Kg of undamaged plasmid DNA, -2000 cpm of "'P-labeled 137-mer, 5 nM UvrA, and 80 nM UvrB. The reaction mixture was incubated at 37 "C for 30 min, then 2 ~l of UvrC Mini-prep or 40 nM UvrC was added, and incubation was continued for another 30 min. The reaction was stopped by adding 1 pl of oyster glycogen (10 mg/ml) and 60 p1 of ice-cold ethanol. The DNA was collected by centrifugation, dried, resuspended in a formamide/dye mixture, and analyzed on an 8% polyacrylamide sequencing gel.

Isolation and i n Vivo Characterization of uvrC Mutants-
We have previously shown that the Bacillus subtilis uvrC gene (Chen et al., 1988) complements uvrCmutations in E. coli (Lin and Sancar, 1990) and that the active site of UvrC is within the carboxyl-terminal half of the protein (Lin and Sancar, 1991). Therefore, we restricted our mutagenesis to the Asp, Glu, and His residues within the carboxyl-terminal half of E. coli UvrC that are homologous to the B. subtilis UvrC. The mutant proteins were tested for in vivo complementation and in vitro incision activities. The results are summarized in Tables I and 11. Using a qualitative spot test it was found that mutations in D399, D438, D466, and H538 inactivated UvrC. Furthermore, this loss of activity was not due to unfolding of the protein, because all mutant proteins were overproduced to levels comparable to that of wild type UvrC, were soluble, and behaved identical to wild type protein on several chromatographic resins.
Quantitative UV survival tests were conducted on the UVsensitive mutants to evaluate the level of deficiency in each was used as the host, the UV survival of this strain a t 5 erg/mm' is 3.0 X H538A failed to complement in uiuo; however the mutant protein was only marginally overproduced and therefore was not investigated in uitro.
WT, wild type.  . H538N complemented the mutant to wild type level, while cells carrying UvrC.H538D were slightly more UV-sensitive than wild type. Thus, the H + N and H .--* D changes at position 538 might be considered conservative substitutions.
Incision Activities of Mutant Proteins-When the four UvrC mutants unable to complement uurC34 in vivo were purified and tested in a plasmid-nicking assay (Lin and Sancar, 1989) a paradoxical result was obtained. All mutants, when mixed with UvrA and UvrB, were as active as wild type UvrC in converting UV irradiated superhelical plasmid into open circle form (data not shown). We suspected that this apparent discrepancy between the in vitro and in vivo results might be due to abnormal incision(s) made by enzyme formed Active Site of fA)RC Rxcinuclcasc with mutant UvrC subunit which does not lead to adduct removal. Therefore, we investigated the incision pattern of (A)RC excinuclease reconstituted with mutant UvrC proteins using a uniquely adducted DNA. The results are shown in Fig. 2. All mutant proteins deficient in complementing activity in uiuo were totally lacking the 5' incision activity while retaining normal 3' incision. The simplest interpretation of this data is that UvrC makes the 5' incision and that the 3' incision alone leads to a negligible level of adduct removal by 3' + 5' exonucleases and therefore has only a marginal effect on cell survival. .. D466A. Cultures were grown to stationary phase in Luria hroth. and cells were diluted, plated on h r i a agar plates. and irradiated with 254-nm light from a germicidal lamp. T h e surviving colonies were counted after incuhation a t 37 "C for 24 h.

B
Roles of Active Site I h i d u c s in ('ntn/y.sis-The identification of His and Asp residues in the active site of tivr(' raised the possibility that His:'" and 1 o f t h e :1 Asp residues required for activity along with a Ser residue o r a water molecule in the active site might constitute a catalytic triad (Stryer. 1988). Amino acid arrangements reminiscent o f a catalytic triad (originally described for serine proteases) have recently heen discovered in a number of nucleases, most notably 1)Sase I (Suck and Oefner, 1986; Suck r 7 t a/.. 1988; 1,ahm 6.t a/., 1 9 9 1 ).
In current models for the functioning o f a catalytic triad, His, aided by the electrostatic (hydrogen bondingJ effect o f Asp. facilitates the nucleophilic attack by H-OH o r Ser-OH, first by acting as a general base to take up the liberated proton. and then to facilitate the decomposition of the pentahedral intermediate formed hy donating the proton (general acid).
T o find out if H5.38 had a similar function in tTvrC, we constructed UvrC. H53RN. and I 'vrC. H5.381) mutat ions and tested them in Liuo and in ritro. The former was as active as wild type in uiuo, while the latter had slightly reduced activity (Fig. 1). I n oitro analysis showsed that both mutants were capable of 5' incision ( Fig. 2H, 1nnr.s 7 and 8). A kinetic experiment was carried out to find out if there were any differences hetween the rate enhancements conferred by His,"" and its functionally competent replacements. The results are shown in Fig.   3. The rate o f incision with I'vr('. H538N is approximately 'LO"; and that of 11vrC. HX181) is about 10% of that obtained with the wsild type protein. Since similar mutations in enzymes presumed or known t o act hy a "catal-ytic triad mechanism" reduce k , ,, by 10L10" (see Saenger, 1991 FIG. 3. Kinetics of 3' and 5' incision with wild type and mutant UvrCs. The mixture of equal cpm of 3'-or 5"labeled DNAs was incubated with 5 nM UvrA and 80 nM UvrB for 30 min a t 37 "C in 250 pl of ABC buffer. At "zero time" UvrC was added to 40 nM, and 25-pl samples were taken at the indicated times, the reaction was stopped by adding 60 pl of ethanol and 10 pg of oyster glycogen, the DNA was collected by centrifugation, and the products were analyzed on 8% polyacrylamide sequencing gels. In A lanes 1 and 2 contain 5'-and 3'labeled DNAs, respectively, and lanes 3 and 4 contain 5'-and 3"labeled DNAs treated with (A)BC excinuclease. A, UvrC.WT; B, UvrC. H538N; C, UvrC. H538Y.
The Phenotype of UvrC-Mutants-The biochemical properties of the noncomplementing uvrC mutants provided us the opportunity to explain an unusual feature of uvrC-cells. In contrast to uvrAor uvrB-cells, which fail to nick DNA after UV irradiation, uvrC-mutants accumulate single-strand breaks following exposure to UV (Ogawa et al., 1968;Rupp and Howard-Flanders, 1968;Deutsch et al., 1976;Seeberg et al., 1980;Tang and Ross, 1985).
In light of results presented in this paper it is likely that the uurC-mutants investigated in the previous studies (uvrC34 and uvrC56) are active site mutants which induce the 3' nick only by interacting with the UvrB .DNA complex. This model would predict that AuvrC mutants should not accumulate single-strand nicks but that introduction of a plasmid carrying any of the four mutations D399A, D438A, D466A, and H538F should restore a uvrC34-like behavior in vivo. Fig. 4 shows alkaline sucrose gradient profiles of DNA of E. coli with either UvrB5 or uvrC34 mutation, or AuvrC, and its derivative carrying a plasmid with uvrC-H538F, following UV irradiation. In agreement with the literature, no nicks are formed in E. coli uvrB5 (A) and nicks do accumulate in uvrC34 ( B ) . In contrast, E. coli AuurC behaves nearly identically to E. coli uurB5 (C). As predicted, introduction of a plasmid carrying uvrC-H538F mutation into the AuvrC strain results in a uvrC34-like DNA profile (D). It is reasonable to conclude, then, that uvrC34 and uvrC56 mutants make proteins (Tang et al., 1991) which induce the 3' incision by binding to UvrB.

DISCUSSION
Regarding the reaction mechanism of (A)BC excinuclease two questions are of interest: Which subunit(s) makes the incisions, and which amino acids are involved in catalysis?
The results presented in this study and recapitulated below provide provisional answers to both questions and also provide a possible explanation for a curious phenomenon associated with uvrC-mutants. 1) Mutations in 4 residues of UvrC specifically inhibit the 5' incision without affecting the 3' incision. The mutations apparently do not cause a gross conformational change in UvrC because the mutant proteins have the same solubility and chromatographic properties as the wild type. Furthermore, the mutations must not affect the binding site because the 3' incision which requires specific binding of UvrC to the UvrBSDNA complex occurs at a normal rate. Although, because of its sequential reaction mechanism and lack of catalytic turnover in the absence of Pol1 and helicase 11, it would be difficult to apply standard Michaelis-Menten formalism to (A)BC excinuclease, normal 3' incision might be interpreted to reflect wild type K, for binding to UvrB .DNA by mutant UvrCs, and the four mutations might justifiably be called UvrC V,,, mutations. Thus, we propose that UvrC makes the 5' incision and that these 4 residues are directly involved in catalysis. The possibility that some UvrB residues may also participate in 5' incision cannot be eliminated based on currently available data alone. However, considering that 4 amino acid residues of UvrC are required for the reaction we think the direct participation of additional amino acids in catalysis to be unlikely. Similarly, we consider unlikely a "straw man" model, whereby, the UvrC mutants that abolish 5' nicking catalyzed by UvrB do so by keeping UvrB from attaining proper conformation, because many UvrC mutants with demonstrable deficiencies in their interactions with UvrB interfered with both 3' and 5' nicking but never with 5' nicking alone (Lin and Sancar, 1991).
Our data indirectly implies that UvrB makes the 3' incision for the following reason. First, when incision reactions are carried out with aged UvrC which has greatly diminished 5' incision activity, near-normal 3' incision is observed (Selby and  indicating that oxidation (perhaps of H538) or denaturation that caused UvrC to lose its 5' incision activity did not interfere with its binding to the UvrB .DNA complex and triggering the 3' incision by inducing a conformational change in one or both components of the complex. Second, mutagenesis of all conserved amino acids in UvrC with general acid-base properties (D, E, H) failed to affect the 3' incision site. Finally, the UvrB mutant (E639A) and UvrB' (Arikan et al., 1986) with about 0.1% of wild type 3' and 5' incision activity can be loaded onto 3"incised DNA with UvrA; upon addition of UvrC to these pre(3')-incised DNA. UvrB(UvrB*) complexes normal 5' incision takes place, suggesting that the defect in these mutants is the formation of the 3' incision which is a prerequisite for the conformational change in the UvrB .UvrC.DNA complex that enables UvrC to make the 5' incision (Lin et al., 1992).
2 ) Assuming that UvrC makes the 5' incision it is desirable to know the precise roles of the active site residues. Of the 4 such residues we have identified H538 appears to be the least important as the nonconservative H + D change at this position resulted only in a 10-fold decrease in activity. This result excludes the possibility of a general acid-base catalysis role for this residue. Considering that 2 polar residues at this position were compatible with activity it is possible that H538 contributes to catalysis by hydrogen bonding to substrate or the transition state either directly or through water molecules and that substitutions which do not exclude water from the active site would be compatible with activity.
In contrast, nonconservative changes in any of the three positions D399, D438, and D466 completely abolish activity. Based on crystal structures of nucleases, three not necessarily exclusive roles have been ascribed to acidic residues in nucleases: hydrogen bonding to His in a catalytic triad such as in DNase I , acting as a general acid-base catalyst by directly activating a water molecule such as in RNase H (Nakamura et al., 1991), or liganding metal ions in the active site and thus enabling them to interact with water to generate the attacking hydroxide ion as in 3' + 5' exonuclease of Klenow fragment and P1 nuclease (Beese and Steitz, 1991;Volbeda et al., 1991). The behavior of H538D mutation makes the first mechanism unlikely for UvrC. With regard to the latter two mechanism; it is known that UvrC does not contain zinc (data not shown) but incision by UvrB.UvrC requires ATP (or ATP-yS) and Mg2+ (Orren and Sancar, 1990). Thus, it is conceivable that Mg2+ is involved in catalysis by UvrC as is the case for RNase H. However, our data do not have the resolution to discriminate between mechanisms involving metal-activated or directly carboxylic acid-activated water molecules. Regardless of the details of the actual functions of carboxylate groups in UvrC, it is becoming quite evident that two to three carboxylate groups in active sites of nucleases are a recurring theme (Xu and Schildkraut, 1991;Thielking et al., 1991;Saenger, 1991) and therefore we are quite confident that D399, D438, and D466 are involved in catalysis by UvrC.
3) Finally, the identification of UvrC mutants which lack the 5' incision activity but which enable (or in association with) UvrB to make the 3' incision has helped explain the slow accumulation of nonproductive nicks in uurC34 and uurC56 mutants following UV irradiation without measurable increase in cell survival (Tang and Ross, 1985). Our results suggest that these two mutations inactivate the 5' incision activity of UvrC without interfering with its binding to UvrB . DNA and promoting the 3' incision which does not result in adduct removal. Isolation of the two mutant proteins in sufficient quantities for in vitro assays would be necessary to test this prediction.