Structure-guided Mutational Analysis of the Nucleotidyltransferase Domain of Escherichia coli DNA Ligase (LigA)*

NAD+-dependent DNA ligases (LigA) are ubiquitous in bacteria, where they are essential for growth and present attractive targets for antimicrobial drug discovery. LigA has a distinctive modular structure in which a nucleotidyltransferase catalytic domain is flanked by an upstream NMN-binding module and by downstream OB-fold, zinc finger, helix-hairpin-helix, and BRCT domains. Here we conducted a structure-function analysis of the nucleotidyltransferase domain of Escherichia coli LigA, guided by the crystal structure of the LigA-DNA-adenylate intermediate. We tested the effects of 29 alanine and conservative mutations at 15 amino acids on ligase activity in vitro and in vivo. We thereby identified essential functional groups that coordinate the reactive phosphates (Arg136), contact the AMP adenine (Lys290), engage the phosphodiester backbone flanking the nick (Arg218, Arg308, Arg97 plus Arg101), or stabilize the active domain fold (Arg171). Finer analysis of the mutational effects revealed step-specific functions for Arg136, which is essential for the reaction of LigA with NAD+ to form the covalent ligase-AMP intermediate (step 1) and for the transfer of AMP to the nick 5′-PO4 to form the DNA-adenylate intermediate (step 2) but is dispensable for phosphodiester formation at a preadenylylated nick (step 3).

Escherichia coli NAD ϩ -dependent DNA ligase (LigA) is the founding member of a family of DNA replication/repair enzymes present in all known bacteria as well as certain halophilic archaea (1). LigA seals 3Ј-OH/5Ј-PO 4 DNA nicks via three nucleotidyl transfer reactions: (i) LigA reacts with NAD ϩ to form a covalent ligase-(lysyl-N)-AMP intermediate and release NMN; (ii) LigA-AMP binds to nicked DNA and transfers the adenylate to the 5Ј-PO 4 strand to form an adenylylated nicked intermediate, AppDNA; and (iii) LigA remains bound to the adenylylated nick and directs the attack of the 3Ј-OH on the 5Ј phosphoanhydride linkage, resulting in a repaired phosphodiester and release of AMP (2)(3)(4).
Domain Ia is unique to NAD ϩ -dependent ligases and is the determinant of NAD ϩ specificity (7, 10 -12). During the ligase adenylylation reaction (step 1), the Ia domain binds the NMN moiety of NAD ϩ via contacts to essential conserved amino acids (Tyr 22 , His 23 , Asp 32 , Tyr 35 , and Asp 36 in E. coli LigA) and thereby orients the NMN leaving group apical to the attacking lysine nucleophile (Lys 115 in E. coli LigA) (11). Deleting the Ia domain abolishes ligase adenylylation without affecting phosphodiester formation at a preadenylylated nick (step 3). A fragment of LigA comprising only the Ia and NTase domains suffices for ligase adenylylation (13)(14)(15)(16) but cannot perform the second and third steps of the pathway. The several C-terminal domains of NAD ϩ -dependent ligases are thereby implicated in recognition of the nicked DNA substrate.
The crystal structure of E. coli LigA bound to the nicked DNA-adenylate intermediate consolidated this point by showing that LigA encircles the DNA helix as a C-shaped protein clamp (9). The protein-DNA interface entails extensive DNA contacts by the NTase, OB, and helix-hairpin-helix domains over a 19-bp segment of duplex DNA centered about the nick (see Fig. 1). A structure-guided mutational analysis (17) pinpointed essential functional groups in the OB domain that engage the DNA phosphodiester backbone flanking the nick (Arg 333 ), penetrate the minor grove and distort the nick (Val 383 and Ile 384 ), or stabilize the OB fold (Arg 379 ). The essential constituents of the helix-hairpin-helix domain include four glycines (Gly 455 , Gly 489 , Gly 521 , and Gly 553 ) that bind the phosphate backbone across the minor groove at the outer margins of the LigA-DNA interface; Arg 487 , which penetrates the minor groove at the outer margin on the 3Ј-OH side of the nick; and Arg 446 , which promotes protein clamp formation via contacts to the nucleotidyltransferase domain (17). The zinc finger module appears to play a purely structural role in bridging the OB and helix-hairpin-helix domains (9). Zn 2ϩ coordination is critical for E. coli LigA activity, insofar as single alanine substitutions at three of the four cysteines (Cys 408 , Cys 411 , and Cys 432 ) suppressed nick sealing by 3 orders of magnitude (18). Deletion analysis revealed that the BRCT domain of E. coli LigA is required in its entirety for effective nick sealing in vitro (17). However, a structural interpretation of this result is elusive because no electron density was observed for this segment of DNA-bound E. coli LigA (9).
Whereas NAD ϩ and DNA recognition are aided by the flanking domains discussed above, the chemical steps of nick sealing are performed by the NTase domain. Initial mutational analyses of the NTase domain of E. coli LigA (18,19) focused on: (i) residues within a set of conserved peptide motifs that define the "covalent nucleotidyl-transferase superfamily," which includes ATP-dependent DNA ligases, ATP-dependent RNA ligases, and GTP-dependent mRNA capping enzymes (20); and (ii) amino acids unique to and conserved among members of the NAD ϩ -dependent DNA ligases clade. Ten individual residues of the NTase domain were identified as critical for E. coli LigA function: Lys 115 (the site of covalent AMP attachment), Asp 117 , Gly 118 , Glu 173 , Arg 200 , Arg 208 , Arg 277 , Asp 285 , Lys 290 , and Lys 314 (19). Reference to a crystal structure of LigA bound to NAD ϩ (7) revealed atomic contacts with the AMP moiety that could plausibly account for the essentiality of Glu 173 (which coordinates the ribose O2Ј), Lys 290 (which donates a hydrogen bond to adenine-N1), and Lys 314 (which coordinates the AMP phosphate (see Fig. 2C).
The roles of Arg 200 and Arg 208 were illuminated by the crystal structure of the LigA-DNA-adenylate intermediate. Arg 200 and Arg 208 reside within a conserved ␣-helical peptide ( 197 ANPRNAAAGSLRQ 209 ) that engages the DNA minor groove on the 3Ј side of the nick. Arg 200 is located at the nick, where it: (i) stacks over and makes van der Waals' contacts with the terminal nucleoside sugar of the 3Ј-OH strand, (ii) contacts via water the 3Ј-OH terminal base, and (iii) forms a salt bridge to the essential Asp 117 side chain (see Fig. 2B). Arg 208 penetrates deeply into the minor groove, where it contacts several nucleoside sugars and bases (see Fig. 2B).
Here, we used the E. coli LigA-DNA crystal structure to guide a new round of mutational analysis of the NTase domain, focusing on side chains that make atomic contacts to the DNA duplex and the 5Ј-adenylate. We thereby identified several conserved amino acid functional groups essential for nick sealing in vitro and in vivo. New mechanistic insights were gleaned from an analysis of mutational effects on individual steps of the ligation pathway.

EXPERIMENTAL PROCEDURES
Ligase Mutants-Missense and nonsense mutations were introduced by PCR into the pET-EcoLigA expression plasmid as described previously (12). The entire ligA gene was sequenced in every case to confirm the desired mutation and exclude the acquisition of unwanted changes during PCR amplification and cloning. The expression plasmids were transformed into E. coli BL21(DE3). Mutant and wild type ligases were purified from the soluble lysates of isopropyl ␤-D-thiogalactopyranoside-induced BL21(DE3) cells by nickel-agarose chromatography as described (12). The protein concentrations were determined using the Bio-Rad dye reagent with bovine serum albumin as a standard.
Nick Ligation-Reaction mixtures (20 l) containing 50 mM Tris-HCl (pH 7.5), 10 mM (NH 4 ) 2 SO 4 , 5 mM dithiothreitol, 5 mM MgCl 2 , 20 M NAD ϩ , 1 pmol of 5Ј 32 P-labeled nicked duplex DNA substrate, and aliquots of serial 2-fold dilutions of wild type or mutant ligases were incubated at 22°C for 20 min. The products were resolved by electrophoresis through a 15-cm 18% polyacrylamide gel containing 7 M urea in 0.5ϫ TBE (45 mM Tris borate, 1.25 mM EDTA). The extents of ligation were determined by scanning the gel with a Fujix BAS2500 imager. The specific activities of wild type and mutant ligases were determined from the slopes of the titration curves in the linear range of enzyme dependence. The activities of the mutant ligases were normalized to the specific activity of wild type LigA protein purified in parallel with that mutant and assayed in a parallel with the same preparation of radiolabeled DNA substrate. On average, the wild type LigA sealed 63 Ϯ 21 fmol of nicks/fmol of input enzyme (average value for 20 different titration experiments with three different preparations of LigA and multiple different preparations of 32 P-labeled singly nicked DNA substrate).
Ligase Adenylylation-Reaction mixtures (20 l) containing 50 mM Tris-HCl (pH 7.5), 5 mM dithiothreitol, 5 mM MgCl 2 , 1 M [ 32 P-adenylate]NAD (purchased from PerkinElmer Life Sciences), and 8 pmol of wild type or mutant LigA were incubated at 22°C for 10 min. The reactions were quenched with SDS, and the products were analyzed by SDS-PAGE. The ligase-[ 32 P]AMP adduct was visualized by autoradiography and quantified by scanning the gel with a Fujix imager.
Ligation at a Preadenylated Nick-The nicked DNA-adenylate substrate was prepared as described previously (19). In brief, the 5Ј adenylated 32 P-labeled 18-mer strand (AppDNA) was prepared by in vitro reaction of Mycobacterium tuberculosis LigC with a singly nicked three-piece duplex DNA containing a 5Ј 32 P-labeled 18-mer strand at the nick. The AppDNA strand was then purified by PAGE, and the nicked DNA-adenylate substrate was formed by annealing the purified AppDNA strand to the complementary 36-mer and a 3Ј-OH acceptor strand at a molar ratio of 1:4:4. AppDNA ligation reaction mixtures (100 l) containing 50 mM Tris-HCl (pH 7.5), 5 mM dithiothreitol, 5 mM MgCl 2 , 600 fmol of nicked DNA-adenylate substrate, and 6 pmol of wild type or mutant LigA proteins were incubated at 22°C. The sealing reactions were initiated by adding ligase. Aliquots (10 l) were withdrawn at 0.25, 0.5, 1, 2, and 5 min and quenched immediately with EDTA and formamide. The products were resolved by denaturing PAGE, and the extents of ligation were determined by scanning the gel with a Fujix imager. At maximum, ϳ80% of the input AppDNA strand was ligated at the reaction end point. (This reflects incomplete annealing of the labeled AppDNA to the template and 3Ј-OH strands.) A plot of the kinetic profile of AppDNA sealing by wild type LigA (with each datum being the average of 11 separate experiments) fit to a single exponential with a rate constant of 0.64 Ϯ 0.019 min Ϫ1 , as calculated in Prism by nonlinear regression curve fitting.
Assay of LigA Activity in Vivo by Complementation of cdc9⌬-NdeI-BamHI restriction fragments containing LigA mutant alleles were excised from their respective pET-based LigA plasmids and inserted into pYX1-His (CEN TRP1). In this vector, expression of ligA is under the control of the yeast TPI1 promoter. The pYX1-LIGA plasmids were transformed into a yeast cdc9⌬ strain bearing a CEN URA3 CDC9 plasmid. Trp ϩ isolates were streaked on agar plates containing 0.75 mg/ml of 5-fluo-roorotic acid (FOA). Lethal mutations (Ϫ growth) were those that formed no FOA-resistant colonies at 25°or 30°C. The viable cdc9⌬ ligA yeast strains were tested for growth on YPD agar at 25 and 30°C. ϩϩϩ indicates colony size indistinguishable from a cdc9⌬ strain expressing wild type LigA; ϩϩ indicates smaller colony size; and ϩ indicates pinpoint colonies.

New Round of Alanine Scanning of the LigA NTase Domain-
The NTase domain of E. coli LigA contains an adenylate-binding pocket composed of a cage of ␤ strands and interstrand loops that includes the six defining motifs (I, Ia, III, IIIa, IV, and V) of the covalent nucleotidyl-transferase superfamily (20). The motif I ( 115 KLDG) lysine nucleophile is located in a loop between the two antiparallel ␤ sheets that form the binding site for the reactive AppN dinucleotide of the nicked DNA-adenylate intermediate. Lys 115 contacts the 5Ј-phosphate of the App-DNA strand (see Fig. 2B). Motif Ia ( 135 TRG 137 ) is located within an interstrand loop that contacts both sides of the nick. The pyrophosphate bridge of AppN is coordinated by motif Ia residue Arg 136 and by Ser 81 (see Fig. 2A). The two phosphodiesters flanking the nick 5Ј terminus (AppNpNpN) are engaged by Asn 84 , directly and via water (see Fig. 2B). The three phosphodiesters flanking the 3Ј-OH side of the nick are coordinated by: (i) the motif Ia Thr 135 side chain to NpNpNpN OH ; (ii) the motif Ia Arg 136 side chain to NpNpNpN OH ; and (iii) Gln 72 via water to NpNpNpN OH (see Fig. 2A). Asn 201 engages in a network of water-mediated contacts to the nucleosides at the 3Ј-OH nick terminus (see Fig. 2B). Here we probed the roles of residues Gln 72 , Ser 81 , Asn 84 , Thr 135 , Arg 136 and Asn 201 that bind the DNA at and near the nick by replacing them individually with alanine.
The NTase domain also contacts the DNA duplex at sites remote from the nick. Arg 97 , Arg 101 , and Arg 308 make electrostatic interactions with adjacent phosphates of the template DNA strand on the 5Ј-phosphate side of nick (Figs. 1 and 2A). Gln 209 and Arg 218 make direct and water-mediated contacts with two adjacent template strand phosphates on the 3Ј-OH side of the nick (Figs. 1 and 2B). Each of these residues was subjected presently to alanine substitution. We included three additional conserved LigA residues in the alanine scan: Arg 305 , situated near the nick 5Ј AppN terminus ( Fig. 2A); Arg 171 , which donates hydrogen bonds to multiple main chain and side chain atoms within the NTase domain ( Fig.  2A); and Ser 206 , which interacts with Arg 218 near the DNA minor grove (Fig. 2B).
The wild type and mutated E. coli LigA proteins were produced as N-terminal His 10 fusions and purified from soluble bacterial extracts by nickel-agarose chromatography (Fig. 3A). The extent of ligation of singly nicked 3Ј-OH/5ЈPO 4 DNA by wild type LigA, and each mutant was gauged as a function of input enzyme, and the specific activities were normalized to the wild type value (defined as 100%). The results are compiled in Table 1. Our operational definition of a functionally important residue is one at which alanine substitution reduced specific activity in nick joining to Յ10% of wild type LigA. By this criterion, five of the targeted residues were deemed important: Arg 136 , Arg 171 , Ser 81 , Arg 218 , and Arg 308 (Table 1).
By contrast, the alanine mutations at nine other positions reduced activity to between 14 and 39% of wild type LigA. We surmise that the contacts of these residues to DNA or other protein constituents seen in the crystal structure contribute to nick sealing, although not enough to meet our criterion of significance. In some cases, an apparently "nonessential" side chain engages in potentially redundant atomic contacts. For example, the effects of loss of the hydrogen bond from Thr 135 O␥ to a phosphate near the 3Ј-OH end might be softened by the fact that the main chain amide of nearby Gly 137 donates a hydrogen bond to the same nonbridging phosphate oxygen as Thr 135 (Fig. 2A). In the same vein, whereas Arg 101 and Arg 308 FIGURE 1. Schematic summary of E. coli LigA contacts to DNA. The nicked duplex DNA is depicted as a two-dimensional cartoon, with the continuous template strand on the left and the nicked strands on the right. The extrahelical 5Ј-adenylate is shown at right. The DNA contacts of LigA side chains (residue identity in plain text) and main chain amides (residue identity in italics) are indicated by arrows. Amino acids making both main chain and side chain contacts to DNA are in bold font. Water-mediated interactions are shown with waters as red spheres. LigA residues that penetrate the DNA helix and interact with the bases are indicated within the DNA base pair ladder. Residues Lys 115 , Glu 173 , Arg 200 , and Arg 208 that were shown previously to be critical for LigA activity (19) are highlighted in red. make ionic interactions with the same phosphate oxygen atom ( Fig. 2A), the R308A mutation (9% of wild type activity) is more deleterious than R101A (28%). Residues Gln 209 and Arg 218 share a water-mediated phosphate contact to the template strand (Fig. 2B), yet the R218A change (2% of wild type activity) was much more harmful than Q209A (33%). The effects of losing the direct hydrogen bond between Gln 209 and the vicinal phosphate might be masked by the hydrogen bond to the same nonbridging oxygen atom from the main chain amide nitrogen of Leu 210 (Figs. 1 and 2B).
The Arg 97 and Arg 101 side chains project from the same face of an ␣ helix ( 89 EESFLANKRVQDR 101 ) and help dock the ␣ helix to the template DNA strand by ionic contacts to adjacent phosphates ( Fig. 2A), in which case the impact of the R97A change (17% of wild type activity) might be checked by the backup contacts of Arg 101 (a seemingly inessential side chain, insofar as the R101A mutant had 70% of wild type activity). To test this idea, we produced a R97A/R101A double mutant of LigA and found that its specific activity in nick sealing was 5% of the wild type value (Table 1). This result is consistent with func-

Effects of Alanine Mutations on EcoLigA Function in Vivo in
Yeast-A deletion of the gene encoding the essential Saccharomyces cerevisiae ATP-dependent DNA ligase Cdc9 can be complemented by expression of LigA (18). Viability of the yeast cdc9⌬ strain is contingent on maintenance of an extrachromosomal CDC9 gene on a CEN URA3 plasmid. Hence, cdc9⌬ cells cannot grow on medium containing FOA (a drug that selects against the URA3 CDC9 plasmid), but they can grow on FOA if the cells have been transformed with a CEN TRP1 plasmid expressing wild type LigA under the control of the constitutive yeast TPI1 promoter. Here we tested by the plasmid shuffle assay (18) whether the LigA-Ala mutants were functional in yeast. We found that mutations R136A, R171A, and R218A that reduced nicked sealing activity to Յ2% of wild type LigA were all lethal in vivo, i.e. they were unable to support growth of cdc9⌬ on FOA at 25 or 30°C (scored as Ϫ in Table 1). The impaired R308A mutant (9% as active as wild type LigA in vitro) and the R97A-R101A double mutant (5% activity in vitro) were also unable to complement cdc9⌬. By contrast, the nine alanine mutants that retained more than 10% of wild type nick joining activity in vitro were functional in cdc9⌬ complementation in vivo. These results suggest that growth of the cdc9⌬ strain depends on a threshold level of LigA nick sealing activity. The exception to this correlation was S81A, which had 6% of wild type nick joining activity in vitro but still supported yeast growth. All of the viable cdc9⌬ ligA-Ala strains (including S81A) grew as well as the "wild type" cdc9⌬ ligA strain on YPD agar medium at 25 and 30°C, as gauged by colony size (scored as ϩϩϩ in Table 1).
Structure-Activity Relationships at Key Residues of the NTase Domain-We tested the effects of conservative lysine and glutamine substitutions for the four essential arginine residues identified above in the alanine scan (Arg 136 , Arg 171 , Arg 218 , and Arg 308 ) and for Arg 97 , which reduced ligase activity by a factor of six. We also introduced conservative changes at Lys 290 , an important residue that makes a hydrogen bond to the adenine-N1 of NAD ϩ and AppDNA (Figs. 1 and 2C). We showed previously that the K290A mutant of E. coli LigA was lethal in yeast and had 8-fold lower nick sealing activity in vitro than wild type LigA (19). Here we replaced Lys 290 with arginine and glutamine. Twelve conservative LigA mutants were produced in E. coli and purified from soluble bacterial extracts by nickelagarose chromatography (Fig. 3B). The specific activities of the mutants in nick joining were determined. The conservative mutants were also tested for activity in yeast cdc9⌬ complementation. The results are compiled in Table 2.
We found that Arg 136 was strictly essential for LigA activity in vitro and in vivo. The R136K and R136Q mutants were 0.2 and Ͻ0.1% as active as wild type LigA in nick sealing and were lethal in vivo in yeast. We surmise that the multidentate contacts of Arg 136 with either the NAD ϩ ␤ phosphate (Fig. 2C) or the phosphates of the nicked DNA-adenylate (Fig. 2B) are critical for LigA activity, rather than mere positive charge on this side chain. By contrast, positive charge appears to suffice in the case of Arg 308 . Whereas the R308Q mutation (4% as active as wild type LigA and lethal in yeast) mimicked the effects of R308A, the lysine substitution restored nick sealing to 80% of wild type and revived activity in vivo in yeast ( Table 2). We infer that lysine can recapitulate the monodentate contact to the DNA phosphate observed for Arg 308 in the LigA crystal structure ( Fig. 2A).
Glutamine substitutions for Arg 171 and Arg 218 had no salutary effect compared with the alanine mutations, i.e. R171Q and R218Q were lethal in yeast and were 3 and 4% as active in nick sealing as wild type LigA, respectively (Table 2). However, the introduction of lysine raised nick sealing activity above our 10% cut-off criterion (to 15% for R171K and 19% for R218K), which allowed for survival of the cdc9⌬ R171K and cdc9⌬ R218K yeast strains ( Table 2). We surmise that LigA activity relies on a positively charged side chain at positions 171 and 218, although optimal activity is attained with arginine. Note that Arg 171 makes hydrogen bonds from all three guanidinium nitrogens to main chain and side chain atoms of the NTase domain. The terminal guanidinium nitrogens of Arg 218 donate hydrogen bonds to Ser 206 O␥ and a DNA-bound water, respectively (Fig. 2B).
Replacing the adenine-binding Lys 290 residue with glutamine elicited a 30-fold reduction in nick sealing and ablated LigA activity in vivo in yeast ( Table 2). The K290R mutant was only slightly better at nick sealing in vitro (8% of wild type activity). Yeast cdc9⌬ K290R cells were viable but grew poorly ( Table 2). The decrement in nick sealing activity upon the loss of Arg 97 (17% of wild type activity) was partially remedied by glutamine (30% activity) and lysine (63% activity), both of which supported growth of cdc9⌬ (Table 2).
Mutational Effects on Phosphodiester Formation at a Preadenylated Nick-The third step of the ligation pathway entails attack of the 3Ј-OH of the nick on the 5Ј-PO 4 of the DNAadenylate to form a phosphodiester and release AMP. We assayed step 3 in isolation using a preadenylated nicked 36-bp DNA substrate labeled with 32 P at the 5Ј-PO 4 of the 18-mer DNA-adenylate strand; this substrate was incubated with LigA in the presence of magnesium without added NAD ϩ . Under condition of enzyme excess, wild type E. coli LigA sealed the nicked DNA-adenylate with pseudo-first order kinetics (Fig. 4). LigA mutants with alanine and conservative substitutions for Arg 136 , Ser 81 , Arg 171 , Arg 218 , and Arg 308 were tested in parallel with wild type LigA (Fig. 4). A striking finding was that elimination of the motif I Arg 136 side chain, which abolished overall nick sealing, had little effect on the rate of phosphodiester synthesis at a preadenylated nick (Fig. 4A, R136A). This result was surprising to us in light of the extensive set of contacts made by Arg 136 to the phosphates at the nick in the crystal structure ( Fig. 2A), which is taken to mimic the Michaelis complex for the step 3 reaction (9). Moreover, the R136K change, which also abolished overall 3Ј-OH/ 5Ј-PO 4 nick sealing, resulted in a 2-fold increase in the rate of AppDNA sealing compared with wild type LigA (Fig. 4A). The R136Q mutant (severely defective in composite nick sealing) performed the isolated step 3 reaction at about one-third the rate of wild type LigA (Fig. 4A).
Loss of the Ser 81 hydroxyl group had little effect on the rate of sealing at a preadenylylated nick (Fig. 4B), compared with the significant impact of the S81A change on the complete ligation reaction (Table 1). Ser 81 O␥ contacts a phosphate oxygen of the AMP leaving group in the step 3 reaction (Fig. 2A). It is conceivable that this contact is functionally redundant to that made by Arg 136 with the same phosphate oxygen or the contact of the Ser 81 main chain amide with the other nonbridging phosphate oxygen of the adenylate (Fig. 2A). To test this idea, we produced and purified two LigA double mutants: S81A/R136A and S81A/R136K (Fig.  5A). As expected, S81A/R136A was defective in overall nick sealing (Ͻ0.1% of wild type activity) and unable to complement cdc9⌬ ( Table  1). The remarkable finding was that the rate of phosphodiester formation by S81A/R136A was still onethird the rate of wild type LigA (Fig.  5C). This result militates against an essential but functionally redundant role of the contacts made by Ser 81 and Arg 136 to the AppN phosphoanhydride bridge. Rather, we surmise that catalysis of step 3 at a preadenylylated nick does not rely on either of these amino acid side chains. This inference is reinforced by the properties of the S81A/ R136K mutant, which is defective in 3Ј-OH/5Ј-PO 4 nick sealing and cdc9⌬ complementation ( Table 2) yet is still faster than wild type LigA in sealing a preadenylylated nick (Fig. 5C). Thus, the loss of Ser 81 did not diminish the vigorous step 3 activity of the R136K single mutant.
Mutations at Arg 308 had disparate effects on the isolated step 3 reaction (Fig. 4C). The rates of AppDNA sealing by R308A and R308Q were reduced to 25 and 40% of wild type, respectively, whereas R308K was nearly twice as fast as wild type LigA. Similar bipolar effects were seen at Arg 218 , where R218A and R218Q sealed the AppDNA substrate at about one-third the wild type rate, whereas R218K was faster than wild type LigA. These results emphasize the importance of the positive charge at side chains 308 and 218 that interact with the DNA phosphodiester backbone. By contrast, mutations at Arg 171 had a generally suppressive effect (Fig. 4B). The rates of AppDNA sealing by R171A, R171K, and R171Q were ϳ10%, 15 and 6% of wild type LigA, respectively. The R97A-R101A double mutant was one-fourth as fast as wild type LigA in AppDNA sealing (Fig. 5C).  The extents of label transfer to LigA are shown. Each datum is the average of triplicate assays from a single experiment in which wild type LigA was tested in parallel with the mutants specified. The error bars denote the standard deviation. C, sealing a preadenylylated nick. Reaction mixtures were constituted as described under "Experimental Procedures." Each datum is the average of three separate experiments in which wild type LigA was assayed in parallel with the mutants specified. The error bars denote the standard deviation.
Mutational Effects on Ligase Adenylylation-Reaction of LigA with [ 32 P]NAD ϩ and magnesium in the absence of DNA leads to the formation of a covalent LigA-[ 32 P]adenylate adduct. The autoadenylylation activities of the wild type and mutant LigA preparations were assayed in parallel; the results for the alanine mutants are depicted in Fig. 6A and summarized in Table 1. Our criterion for a significant mutational effect on step1 catalysis was a reduction of LigA adenylylation to Յ10% of the wild type value. By this standard, the alanine scan showed that only Arg 136 and Arg 171 were important for the reaction of LigA with NAD ϩ , whereas Gln 72 , Ser 81 , Asn 84 , Arg 97 , Arg 101 , Asn 201 , Ser 206 , Gln 209 , Arg 218 , Arg 305 , and Arg 308 were not. As one might expect, the latter group embraces the many side chains that contact the DNA substrate at points remote from the nick but do not engage in atomic contacts with NAD ϩ in the crystal structure of the LigA-NAD ϩ binary complex (Fig. 2).
Replacing Arg 136 with alanine virtually abolished ligase adenylylation (Ͻ0.1% of wild type activity). Replacing Arg 136 with lysine restored the yield of ligase-adenylate to 15% of the wild type value, whereas glutamine conferred little benefit at step 1 (0.7%). The profound effects of the R136A and R136Q mutations on overall nick sealing correlated nicely with their feeble step 1 activities and contrasted with the lack of deleterious impact of these substitutions on sealing at a preadenylylated nick. The gain of step 1 function seen with R136K attests to the importance of the ionic contacts between this side chain and the pyrophosphate bridge of NAD ϩ during step 1 catalysis. Nonetheless, the gain of step 1 activity in R136K is nowhere reflected in its overall ligation function, which remained at 0.2% of wild type LigA, notwithstanding that R136K step 3 activity exceeds that of wild type LigA. These results implicate Arg 136 as a catalyst of step 2 of the ligation pathway (see "Discussion").
Changing Ser 81 to alanine had relatively little effect on ligaseadenylate formation (44% of wild type), even though Ser 81 directly contacts the pyrophosphate bridge of the NAD ϩ substrate (Fig. 2C). In particular, Ser 81 O␥ donates a hydrogen bond to the same phosphate oxygen of the NMN leaving group that receives a bidentate contact from Arg 136 . This situation raised the prospect that Ser 81 is functionally redundant with Arg 136 . This notion was underscored by our finding that the gain of step 1 function attendant on the R136K change (to 15% of wild type) was erased by the S81A/R136K double mutation (0.1% of wild type) ( Fig. 5B and Table 2).
Mutations of Arg 171 had a generally suppressive effect on adenylyltransferase activity, whereby R171A, R171K, and R171Q yielded 7, 17, and 5% as much radiolabeled ligase-AMP adduct as wild type LigA (Fig. 6). The Arg 171 mutational effects on step 1 activity agreed reasonably well with their impact on the composite nick sealing reaction (Tables 1 and 2). Because Arg 171 makes multidentate contacts that tether secondary structure elements of NTase domain, and it makes no contacts directly to DNA or NAD ϩ , we surmise that Arg 171 plays a predominantly structural role in LigA function, e.g. in stabilizing the proper conformation of the NTase domain, especially establishing the proper position for the nearby Glu 173 side chain. Glu 173 contacts the AMP ribose 2ЈOH and is strictly essential for overall nick sealing and ligase adenylylation (19). An E173A mutation of LigA elicits a 30-fold decrement in the rate of phosphodiester formation at a preadenylylated nick (19).

DISCUSSION
The present structure-guided mutational analysis of the NTase domain of E. coli LigA provides new insights to bacterial ligase function in the following respects: (i) by delineating Arg 308 and Arg 218 as critical constituents of the LigA-DNA interface that bind the phosphodiesters backbone of the template strand at sites remote from the nick; (ii) by highlighting functionally redundant essential contacts of Arg 97 and Arg 101 with the template strand backbone on the 5Ј-PO 4 side of the nick; (iii) by discerning an essential structural role for Arg 171 ; and (iv) by unveiling the essential functions of the active site constituent Arg 136 in the first and second steps of the ligation pathway, albeit not in the third step.
DNA Interface of the NTase Domain-As illustrated in Fig. 1, the NTase domain contacts the template strand backbone at discrete trinucleotide clusters on both sides of the nick. We find here that both regional contacts with the template strand are important for LigA activity. On the 3Ј-OH side of the nick, Arg 218 is essential for overall nick sealing, but Gln 209 is not. The R218A and R218Q changes that suppress nick sealing have no significant impact on ligase adenylylation, consistent with the location of Arg 218 outside the active site for nucleotidyl transfer chemistry. We surmise that Arg 218 contacts to DNA facilitate step 2, which entails initial binding of LigA-AMP to the nicked duplex and transfer of AMP to the 5Ј-PO 4 to form AppDNA. Arg 218 mutations have a milder impact on the sealing of a preadenylylated nick than they do on the composite ligation reaction, presumably because tight binding of the AppDNA adenylate within the nucleoside pocket of the NTase domain is the dominant factor in stabilizing the step 3 Michaelis complex. On the 5Ј-PO 4 side of the nick, we find that Arg 308 is singly important for ligation, as is the combination of Arg 97 plus Arg 101 . Mutations of Arg 97 , Arg 101 , and Arg 308 had little effect on ligase adenylylation, and only a mild effect on AppDNA sealing, again suggesting that NTase-DNA contacts flanking the nick promote step 2 of the ligation pathway.
With the exception of Arg 136 and Lys 290 , the NTase residues studied presently that contact the broken 3Ј-OH and AppDNA strands were found not to be essential per se, according to our cut-off criterion of a 10-fold decrement in nick sealing in vitro and by lack of an overt growth phenotype in the yeast complementation assay. Nonetheless, we see that loss of the Asn 84 side chain reduced sealing by a factor of 5, without affecting ligase adenylylation, implying that Asn 84 DNA contacts contribute modestly to ligation efficacy. Loss of Asn 201 reduced sealing by a factor of 4 and had a similar effect on ligase adenylylation. The reasons for the step 1 defect are not obvious, insofar as Asn 201 does not contact NAD ϩ (7). It is conceivable that the hydrogen bond between Asn 201 with the nearby Asn 198 side chain is pertinent to the modest decline in general activity of N201A. The conserved Asn 198 side chain plays a helix-capping role for the signature LigA ␣-helix ( 198 NPRNAAAGSLRQ) in which several essential side chains are located (9).
Lys 290 donates a hydrogen bond to the adenine-N1 atom of NAD ϩ and the AppDNA strand (Fig. 2). Whereas mutants K290R and K290Q are defective in nick sealing, there is little impact of these conservative changes on the extent of ligaseadenylation in vitro (Table 2). This result, which was unexpected in light of the atomic contacts, hints at a more complex scenario. It is conceivable that altering the adenine environment affects the syn-anti conformational switches of the adenosine nucleotide that accompany each step of the LigA reaction pathway (9).
Distinctive Catalytic Roles of Arg 136 at Different Steps of the LigA Reaction Pathway-The motif Ia residue Arg 136 emerges as the single most important residue identified in this study. Arg 136 is essential for the ligase adenylylation reaction, consistent with its contacts to the phosphoanhydride moiety of NAD ϩ . Yet, the network of phosphate contacts of Arg 136 to the reactive -NpN OH and AppN-nick termini is apparently inessential for phosphodiester bond synthesis. Indeed, sealing at a preadenylylated nick proceeds even when the potentially redundant contacts of Arg 136 and Ser 81 with the AppN moiety are eliminated simultaneously. We can now state that, of the three amino acids that contact the AppN phosphates in the crystal structure (Lys 115 , Arg 136 , and Ser 81 ; Fig. 2), only Lys 115 is essential per se for AppDNA ligation; specifically, the K115Q mutant is inert in the isolated step 3 reaction, whereas K115R is as active as wild type LigA (19).
The distinctive activity profile of R136K versus R136(Q/A) implicates Arg 136 as a catalyst of step 2 of the ligation pathway, insofar as R136K has recovered step 1 adenylyltransferase activity, is fully competent for AppDNA sealing, and yet is defective in sealing a 3Ј-OH/5Ј-PO 4 nick. We surmise that the extensive contacts of Arg 136 seen in the LigA-AppDNA structure ( Fig.  2A) reflect what was required to execute step 2. It appears that lysine cannot fulfill all of the phosphate interactions of arginine with both ends of the nick (involving all three guanidinium nitrogens), thereby accounting for the ultimate failure of nick sealing by R136K. Arg 136 is likely to participate directly in step 2 chemistry by stabilizing the transition state of the lysyl-AMP phosphate (via contact to a nonbridging oxygen), while coordinating one of the phosphate oxygens of the attacking nick 5Ј-PO 4 group ( Fig. 2A).
Conserved Role for Motif Ia Arg/Lys in Step 2 Catalysis-Motif Ia (consisting of a lysine or arginine flanked by serine or threonine) is conserved among NAD ϩ -dependent DNA ligases, ATP-dependent DNA ligases, ATP-dependent RNA ligases, and GTP-dependent mRNA capping enzymes. A comparison of the present findings concerning the role of motif Ia Arg 136 in LigA with the mutational data available for several other members of the covalent nucleotidyltransferase superfamily suggests a general role for the motif Ia arginine/lysine in DNA/ RNA 5Ј nucleotidylation. On the other hand, the essentiality of the motif Ia basic side chain during enzyme nucleotidylation and phosphodiester synthesis varies in a case-specific fashion.
Bacteriophage T4 RNA ligase 2 (Rnl2) exemplifies an ATPdependent RNA ligase clade that repairs nicks in duplex RNAs (21). Changing its motif Ia residue Arg 55 to alanine ablated overall RNA sealing and suppressed each step of the pathway (22). Replacing Arg 55 with lysine revived each step and the composite pRNA sealing reaction, whereas glutamine afforded no benefit. In the crystal structure of Rnl2 bound to adenylylated nick (21), Arg 55 makes a bidentate interaction with the two nonbridging oxygens of the nick 5Ј-PO 4 group, thereby implicating Arg 55 in nick recognition and orientation of the nick 5Ј-PO 4 for nucleophilic attack on the lysyl-AMP (analogous to what we propose here for LigA Arg 136 ).
T4 RNA ligase 1 (Rnl1), the prototypal tRNA repair enzyme, relies differently on its motif Ia residue, Lys 119 . The K119A and K119Q mutations of Rnl1 abolished the overall RNA ligation reaction but did not affect either ligase adenylylation (step 1) or sealing a preformed AppRNA (step 3) (23). Rnl1 Lys 119 is thereby implicated as a specific catalyst of the RNA adenylation reaction (step 2). The conservative K119R substitution restored overall pRNA ligation activity (23). The crystal structure of Rnl1 bound to AMPCPP reveals that Lys 119 coordinates the ␥ phosphate (24); the observed inessentiality of Lys 119 for Rnl1 adenylylation is presumed to reflect functional redundancy with several other ␥ phosphate contacts made by Rnl1 side chains (25). Further elucidation of the role of motif Ia during step 2 catalysis hinges on obtaining a crystal structure of Rnl1 with its broken tRNA substrate.
Deinococcus radiodurans RNA ligase (DraRnl) exemplifies yet another clade of ATP-dependent RNA sealing enzymes (26). Changing its motif Ia residue Lys 186 to alanine severely impaired overall sealing and ligase adenylylation but had little effect on sealing at a preadenylylated nick (analogous to what we find here with LigA R136A).
Chlorella virus DNA ligase (ChVLig) is a minimized eukaryal ATP-dependent DNA ligase that has an intrinsic nick sensing function. Replacing the motif Ia residue Arg 42 with alanine caused a 50-fold reduction in overall nick sealing and suppressed nick-specific DNA binding (27). Arg 42 is proposed to directly coordinate the nick 5Ј-PO 4 and thereby promote step 2 chemistry (27). In the crystal structure of ChVLig bound at a 3Ј-OH/5Ј-PO 4 nick, Arg 42 also engages the terminal phosphodiester -NpN OH (28). This is analogous to the nick-spanning LigA Arg 136 contacts with the 3Ј-OH and AppN strands seen in Fig. 2A.