Exonuclease X of Escherichia coli

DNA exonucleases are critical for DNA replication, repair, and recombination. In the bacteriumEscherichia coli there are 14 DNA exonucleases including exonucleases I-IX (including the two DNA polymerase I exonucleases), RecJ exonuclease, SbcCD exonuclease, RNase T, and the exonuclease domains of DNA polymerase II and III. Here we report the discovery and characterization of a new E. coli exonuclease, exonuclease X. Exonuclease X is a member of a superfamily of proteins that have homology to the 3′-5′ exonuclease proofreading subunit (DnaQ) ofE. coli DNA polymerase III. We have engineered and purified a (His)6-exonuclease X fusion protein and characterized its activity. Exonuclease X is a potent distributive exonuclease, capable of degrading both single-stranded and duplex DNA with 3′-5′ polarity. Its high affinity for single-strand DNA and its rapid catalytic rate are similar to the processive exonucleases RecJ and exonuclease I. Deletion of the exoX gene exacerbated the UV sensitivity of a strain lacking RecJ, exonuclease I, and exonuclease VII. When overexpressed, exonuclease X is capable of substituting for exonuclease I in UV repair. As we have proposed for the other single-strand DNA exonucleases, exonuclease X may facilitate recombinational repair by pre-synaptic and/or post-synaptic DNA degradation.

Examination of multiple sequence alignments by both BLAST (1) and hidden Markov model (2) have helped define a large family of proteins that share sequence homology with the 3Ј-5Ј exodeoxyribonuclease domain of DNA polymerases. The ⑀ proofreading subunit of Escherichia coli DNA polymerase III, encoded by dnaQ, is the archetypal member of this family. Other family members include the bacterial proteins RNase T, RNase D, exonuclease I (ExoI), 1 oligoribonuclease (3), the Saccharomyces cerevisiae PAN2 protein, and the human Werner syndrome protein (WRN) (1,2,4). These proteins share a conserved tripartite set of "Exo" motifs containing negatively charged aspartate and glutamate residues (5). These hallmark residues can be visualized in the crystal structure of the Klenow (proofreading) subunit of E. coli DNA polymerase I to coordinate two divalent cations that catalyze DNA phosphodiester bond cleavage (6 -9). Comparison of the crystal structures of the Klenow fragment and bacteriophage T4 DNA polymerase suggests that the Exo motifs are diagnostic of functional conservation, since both proteins share the same active site structure despite the lack of sequence identity outside of the Exo motifs (2,10). Presumably, other proteins that share these motifs adopt a catalytic site structure and mechanism of action similar to the polymerase exonuclease domain.
It has recently been demonstrated that the Werner syndrome protein (WRN) has a 3Ј-5Ј DNA exonuclease activity (11)(12)(13). Originally identified as a 3Ј-5Ј RecQ-like DNA helicase (14 -16), the Werner syndrome protein also has an N-terminal DnaQ-like nuclease domain (17,18). WRN possesses a weak exonuclease activity with specificity for the 3Ј-ending recessed strand of a partial DNA duplex but is unable to degrade singlestrand DNA alone (12).
We recently reported that RNase T of E. coli, previously described as a 3Ј-5Ј ribonuclease (19 -21), also possessed a potent 3Ј to 5Ј distributive single-strand (ss) DNA-specific exonuclease activity (22). When overexpressed, RNase T was capable of complementing DNA repair defects caused by a deficiency in E. coli Exo I. Unlike exonucleases associated with DNA polymerase, which can degrade from the 3Ј end of doublestrand (ds) DNA molecules, RNase T had no activity on dsDNA substrates (22). Clearly, the architecture of proteins within the DnaQ superfamily allows for different modalities of function, since the same active site configuration can be used to accommodate various substrates (ssDNA, dsDNA, RNA) while retaining the 3Ј-5Ј polarity of degradation.
In addition to the bacterial members of the DnaQ superfamily listed above is an open reading frame of unknown function designated yobC (also known as o220 and b1844) at 41.5 min on the E. coli chromosome. We have cloned, overexpressed, and characterized the protein product of this gene. Overexpression of this gene concomitantly induces high levels of a DNase activity on both ssDNA and dsDNA. We have renamed this open reading frame exoX and the native 25-kDa protein product exonuclease X (ExoX). We have purified a (His) 6 -ExoX fusion protein to homogeneity and characterized its nuclease activity on various DNA substrates. ExoX is an extremely potent 3Ј to 5Ј distributive nuclease capable of degrading 40kilobase bacteriophage T7 ssDNA and dsDNA to completion. Its affinity for ssDNA ends is greater than for dsDNA, and it appears to have no affinity for RNA. When overexpressed, ExoX, like RNase T (23), is capable of substituting for ExoI in vivo, as measured by its ability to increase the UV survival of an ExoI-deficient strain. A mutation in exoX did not by itself cause sensitivity to UV but strongly augmented the UV sensitivity of a strain deficient in ssDNA exonucleases RecJ, ExoI, and ExoVII.

Plasmid Constructions
The exoX gene was amplified by PCR using Pfu polymerase (Stratagene Inc.) from wild type E. coli strain MG1655 genomic DNA using primers 5Ј-CGGAATTCTAAGGAGGGATCCATGTTGCGCATTATC-3Ј and 5Ј-GCTCTAGACTAAGTATTTTCCAG-3Ј and buffer conditions recommended by the manufacturer. Primers were annealed to the genomic DNA at 50°C for 30 s and extended for 2 min at 72°C; 25 cycles of PCR were performed. The PCR product was subsequently digested with restriction endonucleases EcoRI and XbaI and ligated into the compatible sites of pBSSK Ϫ (Stratagene Inc.), producing pExoX. Sequence analysis verified the construct was error-free.
The EcoRI-XbaI fragment from pExoX was cloned into compatible sites within pBSKS Ϫ (Stratagene Inc.) creating pExoXKS Ϫ . Plasmid pExoXfs, a derivative of pExoXKS Ϫ with a frameshift mutation 171 base pairs downstream from the initiation codon of exoX, was constructed by cleavage of pExoXKS Ϫ DNA with restriction endonuclease NcoI, "fill-in" synthesis with Klenow fragment (DNA polymerase I), and blunt end DNA ligation. An exoX (His) 6 -tagged gene fusion was constructed by cloning the 693-base pair BamHI-SacI fragment of pExoX into the same sites within pET28a(ϩ) (Novogen), producing the plasmid pExoX-His.
A 2,919-base pair region of the E. coli chromosome containing the exoX gene was amplified by PCR using primers beginning 1,088 base pairs upstream (5Ј-GGGAATTCGTACCCGTATGCGTGATG-3Ј) and 1,167 base pairs downstream (5Ј-GGTCTAGACGAGGATCATCAATTC-CGG-3Ј) of exoX. The PCR was performed using Turbo Pfu polymerase (Stratagene, Inc.) in buffer conditions recommended by the manufacturer. Primers were annealed to MG1655 E. coli genomic DNA at 60°C for 30 s and extended for 3 min at 72°C; 25 cycles of PCR were performed. The PCR product was digested with XbaI and EcoRI and ligated into compatible sites within the Litmus 29 vector (New England Biolabs Inc.), creating the plasmid pExoXFlank.
A precise deletion of the exoX open reading frame from the plasmid pExoXFlank was performed by PCR. The primers utilized for the PCR flanked exoX and were oriented to replicate the entire plasmid except for the exoX open reading frame. Both primers; 5Ј-GGGGGCGGCCGC-GGCATGCTCCAGGCCG-3Ј and 5Ј-GGGGGCGGCCGCTCCGCAGGC-GTAGCGGG-3Ј contained NotI sites at the primer 5Ј terminus. The PCR was performed as above except that primer annealing was at 45°C, and extensions were performed for 6 min. The resulting 5-kb PCR product was treated with DpnI to remove any original methylated template DNA and then digested with NotI and ligated to a 2.1-kb NotI fragment from plasmid pCK155 (24) containing a Tn5 npt gene, which confers kanamycin resistance. Flanking npt on both sides are 140-base pair resolution sites from the broad host range plasmid RP4 multimer resolution system, which allows for the precise excision of npt (24). The ligation was transformed by electroporation into XL1-Blue (Stratagene Inc.) cells selecting kanamycin-resistance. The resulting plasmid pExoX⌬ had a complete deletion of exoX and a 2.1-kilobase insertion as verified by restriction analysis.

UV Survival Assays
Single colonies were grown in LB or LB ϩ ampicillin (for strains containing plasmids) liquid medium to exponential stage (A 600 ϭ 0.4 -0.5), serially diluted in 56/2 buffer (60 mM Na 2 HPO 4 , 40 mM KH 2 PO 4 , 0.02% , and plated on LB or LB ϩ ampicillin plates. Plates were immediately irradiated with varying doses of UV (254 nm) irradiation and incubated at 37°C in the dark overnight. Total viable cells were determined from serial-diluted unirradiated cells.

Protein Expression and Analysis
ExoX protein expression was induced from the T7 10 promoter of pExoX by 42°C heat shock induction of the T7 RNA polymerase gene on plasmid pTJH30. T7 promoter-mediated expression of the (His) 6 -ExoX fusion protein from plasmid pExoX-His was induced by the addition of isopropyl-1-thio-␤-D-galactopyranoside (1 mM) to STL4337 (pExoX-His transformant of STL2329). Culture growth, [ 35 S]methionine protein labeling in the presence of rifampicin and crude extract preparations for both proteins were performed as described previously (25). Protein concentrations were measured by the method of Bradford (31) with standard reagent (Bio-Rad) and bovine serum albumin (BSA) as standard. All proteins were resolved by 15% SDS-polyacrylamide gel electrophoresis (32) and visualized by Coomassie stain and/or autoradiography.

Purification of His 6 -ExoX
All steps were performed at 4°C using ultrapure reagents. Buffer A contained 50 mM Tris-HCl (pH 7.5), 200 mM NaCl, and 1 mM dithiothreitol. Buffer B contained 10% glycerol, 20 mM Tris-HCl (pH 8.0), 500 FIG. 1. Expression, labeling and purification. Protein expression, 35 S-labeling and SDS-polyacrylamide gel electrophoresis were performed as described under "Experimental Procedures." [ 35 S]Methionine-labeled whole cells extracts are shown (panel A) in the following order: lanes 1 and 2, pBSSK Ϫ ; lanes 3 and 4, pExoX; lanes 5 and 6, pET28a(ϩ); lanes 7 and 8, pExoX-His. Odd-and even-numbered lanes are from uninduced and induced cells, respectively. Labeling experiments were performed in the presence of rifampicin to inhibit expression from E. coli RNA polymerase. Each fraction from the purification of (His) 6 -ExoX was resolved by electrophoresis through a 15% SDS-polyacrylamide electrophoresis gel and stained with Coomassie Blue (panel B). The SDS-polyacrylamide gel shown in panel B contains crude extracts from cells induced for expression of the (His) 6 -ExoX fusion protein from plasmid pExoX-His and pooled protein fractions from the subsequent purification steps on nickel (nitrilotriacetic acid (NTA))agarose and dsDNA cellulose matrices. mM NaCl, 25 mM imidazole, and 1 mM ␤-mercaptoethanol. Buffer C contained 10% glycerol, 20 mM Tris-HCl (pH 7.5), 1 mM EDTA, and 1 mM dithiothreitol (DTT). STL4533 (pExoX-His transformant of BL21) was cultured at 37°C in 34 liters of LB ϩ Km to an A 600 of 0.6, then induced with 1 mM isopropyl-1-thio-␤-D-galactopyranoside for 1 h. Cells were harvested and frozen as described previously (33) in a volume of 700 ml. A crude extract was prepared by lysing cells in 10% sucrose, 50 mM Tris-HCl (pH 8.0), 50 mM NaCl, 5 mM MgCl 2 , 1 mM DTT, 1 mM EDTA, and 1 mg/ml lysozyme for 1 h on ice. Three cycles of freeze/ thawing (10 min 37°C, 10 min 0°C) were performed before a supernatant was obtained by high speed centrifugation at 130,000 ϫ g for 20 min. The crude extract (700 ml, 1.8 g of protein) was adjusted to 200 mM NaCl and 5 mM imidazole before adding 15 ml of Ni 2ϩ -nitrilotriacetic acid-agarose resin (Qiagen) equilibrated in Buffer A. The resin and extract slurry was allowed to mix for 8 h with stirring at 4°C. The resin was washed by repeated low speed centrifugation and resuspension in 175 ml of Buffer A, 85 ml of Buffer B, and finally with 50 ml of Buffer C ϩ 200 mM NaCl. Proteins were eluted from the resin with 35 ml of Buffer C ϩ 500 mM NaCl ϩ 400 mM imidazole. The resulting fraction (6.0 mg of protein) was concentrated to 10 ml using a Centriprep 10 (Amicon) cartridge and then dialyzed against 2 1-liter changes of Buffer C. This fraction was then applied to a 0.5-ml dsDNA cellulose column equilibrated in Buffer C ϩ 25 mM NaCl and washed with 10 ml of Buffer C. Bound proteins were eluted in a single step to Buffer C ϩ 500 mM NaCl. Fractions containing nuclease activity were pooled and dialyzed overnight against 0.5 liters of 60% glycerol, 500 mM NaCl, and 1 mM DTT, and then again against 60% glycerol, 1 mM DTT, 1 mM EDTA. The purified protein (0.75 mg at 0.41 mg/ml) was stored at Ϫ20°C.

DNA Substrates and DNase Assays
Uniformly labeled bacteriophage T7 [ 3 H]DNA with a specific activity of 2.5 ϫ 10 4 cpm/nmol of nucleotide was prepared as described previously (33) using [ 3 H]thymidine (NEN Life Science Products). 3Ј endlabeled substrate was generated by Klenow fragment fill-in synthesis of HindIII-digested pBSSK Ϫ DNA with [ 32 P]dATP. 5Ј end-labeled substrate was generated from HindIII-digested pBSSK Ϫ DNA, treated with shrimp alkaline phosphatase, and phosphorylated with T4 polynucleotide kinase and [␥-32 P]ATP. Both 3Ј and 5Ј end-labeled substrates were purified by G-50 Sephadex quick spin column (Roche Molecular Biochemicals). Unless otherwise stated, enzyme assays employed 0.5 g of T7 [ 3 H]DNA (1.5 nmol) or 0.5 g (0.5 pmol of ends) of 3Ј or 5Ј end-labeled substrate. For ssDNA assays, substrates were incubated for 5 min at 100°C then quenched on ice. Standard reactions contained 10 mM Tris-HCl (pH 8.0), 10 mM MgCl 2 , 1 mM NaCl, and 2 mg/ml BSA in 50 l (for T7 substrates) or 30 l (for end-labeled substrates). Competition assays were performed with 50 ng (50 fmol) of 3Ј end-labeled substrate, varying the amount of either single-strand or double-strand cold competitor DNA. Competition experiments using RNA competitors utilized E. coli tRNA XXI (Sigma) and Torula yeast RNA (molecular mass range 3 ϫ 10 3 -4 ϫ 10 4 g/mol, Sigma). Both RNAs are used by Sigma as ribonuclease assay substrates. Cold RNA stocks were prepared before each competition experiment in cold sterile water. The A 260 of the RNA stocks was determined immediately before use to verify the optical density provided by the manufacturer and to ensure that noticeable degradation had not occurred. When required, protein samples were diluted in a buffer containing 60% glycerol, 10 mM Tris-HCl (pH 8.0), 1 mM DTT, and 1 mM EDTA. Sample incubation, trichloroacetic acid precipitation, and soluble count determination were performed as described previously (34). One unit of DNase activity with T7 DNA substrate corresponds to the release of 1 nmol of acid-soluble product in a 20-min reaction at 37°C.
For agarose gel electrophoresis assays, 0.5 g (0.5 pmol) of either 3Ј or 5Ј 32 P end-labeled ssDNA substrate was incubated with various amounts of (His) 6 -ExoX protein. Reactions were quenched at various time points on ice by the addition of EDTA to 5.0 mM. Standard assay conditions were employed, except that BSA was omitted to allow DNA electrophoresis without further manipulation of the samples after the reactions were quenched. ssDNA substrate samples were boiled for 5 min and cooled on ice before loading onto a 1.0% agarose-NA gel (Amersham Pharmacia Biotech); double-strand substrates were loaded directly onto the gels. Gels were run at 80 V for 45 min in Tris acetate ϩ EDTA buffer (32), dried at 80°C for 30 min, and exposed to film.

Enzymes and Antibiotics
Shrimp alkaline phosphatase was obtained from Amersham Pharmacia Biotech. Lysozyme was obtained from U. S. Biochemical Corp. All other enzymes were obtained from New England Biolabs, Inc. The antibiotics ampicillin, kanamycin, and chloramphenicol were used at 100, 60, and 15 g/ml, respectively.

RESULTS
Overexpression and Purification of a DNase Activity Associated with Exonuclease X-The E. coli open reading frame designated yobC, hereafter known as exoX, was amplified by PCR and cloned directionally into pBSSK Ϫ , placing its expression under T7 promoter control. Induction of exoX expression from the T7 promoter of pExoX led to the production of a single 25-kDa protein, consistent with the expected molecular mass of the protein (Fig. 1A, lane 4). The 25-kDa protein was absent in uninduced extracts (Fig. 1A, lane 3) and from cells carrying vector only (Fig. 1A, lane 2). Crude extracts prepared from these induced cells were tested for DNase activity using uniformly 3 H-labeled T7 DNA substrates. Strains induced for expression of exoX exhibited a 150-fold increase in ssDNase activity and a 280-fold increase in dsDNase activity compared with cells carrying vector alone (Table I). Both ssDNase and dsDNase activities were Mg 2ϩ -dependent, as no increase in activity was noted in the absence of the divalent cation (data not shown).
An N-terminal (His) 6 -tagged ExoX protein fusion was constructed using the pET28a(ϩ) vector (Novogen). (His) 6 -ExoX protein was expressed from the T7 promoter on plasmid pExoX-His (Fig. 1A). Crude extracts prepared from induced cells carrying pExoX-His were tested for DNase activity. Strains induced for expression of (His) 6 -ExoX exhibited a 47fold increase in ssDNase activity and a 170-fold increase in dsDNase activity compared with cells carrying vector alone ( Table I). Comparison of uninduced levels of expression between the pET28a(ϩ) vector and pExoX-His revealed a basal level of expression of the (His) 6 fusion protein without induction of the T7 promoter. Upon induction, increased expression of two [ 35 S]methionine-labeled proteins was noted. The identity of the second band is unknown; however, it did not appear (Fig.  1A, lane 4) when the native protein was expressed in a different strain background.
The (His) 6 -ExoX protein was overexpressed in E. coli strain BL21 and was purified using nickel-agarose chromatography (Fig. 1B). Three high molecular weight contaminants and a lower molecular weight protein were removed by dsDNA cellulose affinity chromatography. The identity of the abundant lower molecular weight band is not known; however, it may be a proteolytic product of the (His) 6 -ExoX protein, since it bound to the Ni 2ϩ -nitrilotriacetic acid resin yet failed to bind dsDNA cellulose affinity column. Furthermore, the protein that bound to the dsDNA cellulose column had significant nuclease activity, whereas the flow-through fraction from the dsDNA cellulose column containing the lower molecular weight protein had nearly none. The purified protein was analyzed by mass spectrometry and was determined to have a molecular mass of 28,552 mass units. The molecular mass determined by mass spectrometry is commensurate with the expected mass of the fusion protein by amino acid sequence.
Properties of the DNase Activity of Exonuclease X-Purified (His) 6 -ExoX was used to determine optimal conditions for ssD-Nase activity. ExoX showed optimal activity at pH 8.0 in the presence of Mg 2ϩ ( Fig. 2A). Similar to other nucleases, the ssDNase activity of ExoX was dependent upon the presence of the divalent cation Mg 2ϩ ; no detectable degradation was seen in its absence or in the presence of Mn 2ϩ (Fig. 2B). BSA enhanced the ssDNase activity of ExoX (Fig. 2C). The ssDNase activity of ExoX was enhanced with low salt concentrations, but above 5 mM salt, the activity decreased with increasing concentration (Fig. 2D). The addition of the sulfhydryl reducing agent, DTT (1-5 mM), did not alter levels of ssDNase activity (data not shown).
In reactions with denatured bacteriophage T7 DNA (40 kb in length), 1 ng (34 fmol) of purified (His) 6 -ExoX linearly degraded 35% of the total DNA (0.5 g, 1.5 nmol in nucleotides) in 20 min. The addition of 10 ng of protein resulted in 100% of the ssDNA substrate being degraded (Fig. 3A). Similarly, in reactions with T7 dsDNA, 5 ng (0.17 pmol) of purified (His) 6 -ExoX linearly degraded 30% of the total DNA in 20 min, and the addition of 40 ng (1.4 pmol) of (His) 6 -ExoX resulted in 100% of the dsDNA substrate being degraded (Fig. 3B). Using data points in the linear range of T7 DNA degradation from Fig. 3, the calculated rate of nucleotide release/protein monomer is 800 nucleotide/min for ssDNA, and 150 nucleotides/min for dsDNA. From its ability to degrade T7 DNA completely, we conclude that ExoX has little or no specificity for DNA sequence or structure. No endonuclease activity was observed in reactions with X174 circular ssDNA, supercoiled dsDNA, or relaxed dsDNA, suggesting that the DNase activity of ExoX is exonucleolytic and not endonucleolytic in nature (data not shown). Electrophoretic analysis of end-labeled reaction products (discussed below) also support this conclusion.
Using uniformly labeled T7 ssDNA as substrate, the extent of ssDNA degradation was determined for a substrate concentration range of 0.2-3.0 nM with 80 ng/ml (2.8 nM monomer) of (His) 6 -ExoX in 5-min reactions using standard buffer conditions. A Lineweaver-Burk plot produced a K m of 1.7 nM and a V max of 50 nmol of nucleotide/min/mg of protein for the degradation of 40 kb of ssDNA molecules (data not shown). The k cat , at high T7 ssDNA substrate concentrations (based on V max ) is 1,400 nucleotides/min/monomer.
The DNase activity of (His) 6 -ExoX was examined on various end-labeled ssDNA and dsDNA substrates (Fig. 4). Purified (His) 6 -ExoX showed efficient release of the terminal nucleotide from 3Ј ssDNA ends. 95% of all 3Ј termini were removed in 20 min with a 1:1,900 ratio (protein monomer to DNA, Fig. 4A). Release of the terminal nucleotide from 5Ј ssDNA ends was less efficient; nearly 500-fold more enzyme was required to achieve comparable extent of release of 5Ј ssDNA ends (Fig. 4B). In reactions with dsDNA, complete release of the terminal nucleotide from 3Ј dsDNA ends (5Ј overhangs) required approximately 10-fold more enzyme (Fig. 4C) as compared with similarly end-labeled ssDNA (Fig. 4A). Nucleolytic activity was barely detected with 5Ј-labeled dsDNA substrate (Fig. 4D). These results imply that ExoX acts as a 3Ј to 5Ј exonuclease on both ssDNA and dsDNA substrates, with ssDNA more efficiently attacked than dsDNA.
A 3Ј to 5Ј polarity of digestion was confirmed by gel electrophoresis of ExoX nuclease reactions with end-labeled ssDNA and dsDNA as substrate. In these reactions BSA was omitted to allow DNA electrophoresis without manipulation of the samples after the reaction; subsequently more enzyme (2-3-fold) was needed for the reactions to go to completion as compared with the values seen in Fig. 2. A 1:700 molar ratio of (His) 6 -ExoX protein to 3Ј end-labeled ssDNA (protein monomers: 3Ј DNA ends) produced a loss of the terminal 3Ј label from nearly all the substrate DNA molecules by 20 min, without detectable shortening of the labeled DNA (Fig. 5A). In contrast, incubation of (His) 6 -ExoX with 5Ј end-labeled ssDNA at a molar ratio of 1:5 (protein monomers: 5Ј DNA ends) resulted in progressive shortening of the labeled DNA with little loss of signal intensity at the earlier time points (Fig. 5B). By 10 min the ssDNA molecules either were degraded completely or had become heterogeneous in size due to asynchronous digestion (Fig. 5B). A 1:100 ratio of (His) 6 -ExoX protein to 3Ј end-labeled dsDNA (protein monomers: 3Ј DNA ends) also produced a loss of the terminal 3Ј label from nearly all the substrate dsDNA molecules by 20 min without detectable shortening of the labeled DNA (Fig. 5C). Reaction of a 3-kb 5Ј end-labeled dsDNA substrate with (His) 6 -ExoX in a 1:2 ratio (protein monomers: 5Ј DNA ends) resulted in progressive shortening of the dsDNA molecules from the 3Ј ends; eventually shortening from both ends resulted in a 1.5-kb ssDNA product, seen at 10 min.
These results are consistent with a 3Ј to 5Ј polarity of DNA degradation by ExoX. Furthermore, the degradation by ExoX must be via a distributive mechanism, because the substrate molecules appear to be degraded uniformly from the 3Ј end when substrate is in excess of enzyme. ExoX must dissociate and rebind to its ssDNA substrate during cycles of degradation, similar to the mechanism found for the ssDNase activity associated with RNase T of E. coli and in contrast to the processive mechanism of DNA degradation exhibited by ssDNA exonucle-ases such as exonuclease I (35) and Re 1 cJ. 2 Competition experiments were performed using 3Ј end-labeled ssDNA as the assay substrate and either ssDNA, dsDNA, tRNA, or yeast RNA as unlabeled competitors. ssDNA proved to be a potent competitor with a 1000-fold excess of cold ssDNA, producing an 80% decrease in released counts. At the same ratio of substrate to competitor, dsDNA showed a modest 6% decrease in counts released, whereas both RNA species failed to compete altogether. These results demonstrate that the affinity of ExoX for ssDNA is considerably greater than for dsDNA and that RNA is not a substrate.
ExoX Is Involved in the Repair of UV-induced DNA Damage-To determine whether the 3Ј-5Ј nuclease activity of ExoX could serve a biological function we asked whether high copy expression of ExoX could substitute for ExoI deficiency in UV repair. Plasmid pExoXKS Ϫ with exoX under lac promoter control was transformed into both RecJ Ϫ ExoI Ϫ ExoVII Ϫ and RecJ Ϫ ExoVII Ϫ strains and assayed for UV survival. A plasmid containing exoX was capable of ameliorating the UV repair defect of a RecJ Ϫ ExoI Ϫ ExoVII Ϫ strain, however it had no effect upon the UV sensitivity of a RecJ Ϫ ExoVII Ϫ strain (Fig.  6). Neither vector (pBSKS Ϫ ) nor plasmid pExoXfs, containing a frameshift early in the exoX coding region, provided any measure of protection. These results demonstrate that ExoX can specifically compensate for loss of ExoI, a 3Ј-5Ј-specific exonuclease, in vivo. In panel A, 0.5 g (0.5 pmol) of 3Ј end-labeled ssDNA was incubated with either 8 pg (0.28 fmol) or 4 pg (0.14 fmol) of (His) 6 -ExoX. In panel B, 0.5 g (0.5 pmol of ends) of 5Ј end-labeled ssDNA was incubated with either 4 ng (0.15 pmol) or 0.8 ng (0.03 pmol) of (His) 6 -ExoX. In panel C, 0.5 g (0.5 pmol) of 3Ј end-labeled dsDNA was incubated with either 80 pg (2.8 fmol) or 40 pg (1.4 fmol) of (His) 6 -ExoX. In panel D, 0.5 g (0.5 pmol) of 5Ј end-labeled dsDNA was incubated with either 40 ng (1.4 pmol) or 4 ng (0.14 pmoles) of (His) 6 -ExoX. All assays were performed under standard assays conditions described under "Experimental Procedures." To confirm a role for ExoX in DNA repair, a null mutant of ExoX was constructed by homologous recombination of plasmid pExoX⌬ into the E. coli chromosome. This null mutant, ⌬exoX1::npt, carries a precise deletion of the exoX open reading frame replaced by a cassette containing the npt gene conferring kanamycin resistance. Using this null allele of exoX, a series of isogenic exonuclease-deficient strains were constructed. The exoX null mutation was added to strains already deficient in one or more of the other known single-strand exonucleases; RecJ, ExoI, and ExoVII. We assayed this set of strains for their ability to survive UV irradiation (Fig. 7). An exoX null mutant alone had no measurable effect upon UV survival compared with a Exo ϩ strain. Similarly, the ⌬exoX1::npt allele added to any other single or double exonuclease mutant had no effect on UV survival compared with their respective progenitor strain. However, the ⌬exoX1::npt null allele added to a RecJ Ϫ ExoI Ϫ ExoVII Ϫ mutant resulted in a strain that was significantly more UV sensitive than the triple mutant alone, demonstrating a synergistic relationship among these nucleases with respect to UV survival. DISCUSSION In E. coli, DNA exonucleases play diverse and important roles in DNA metabolism. The 3Ј-5Ј exonucleases associated with the three polymerases (dnaQ, polA, polB) help maintain genomic fidelity during replication. Exonuclease V, better known as the RecBCD nuclease, is important for conjugal and repair recombination of double-strand breaks (36). Exonuclease VIII, the product of the cryptic Rac prophage, is a component of the RecE pathway of recombination in E. coli (37). There are also three known processive ssDNA-specific exonucleases in E. coli, exonuclease I, exonuclease VII, and RecJ exonuclease (36). These exonucleases catalyze the nucleolytic cleavage of successive phosphodiester bonds on a ssDNA molecule. Exonuclease I degrades ssDNA in a 3Ј to 5Ј direction (38), RecJ exonuclease has 5Ј to 3Ј polarity (33), and exonuclease VII possesses dual polarity acting from either DNA end (39). Mutation of one or several of these genes in combination has pleiotropic effects in DNA repair and recombination in E. coli, including sensitivity to UV irradiation and recombination defects (34,40,41). All three of these exonucleases have been additionally implicated in the process of methyl-directed mismatch repair (42,43). Using in vitro reconstitution experiments with purified proteins, all three exonucleases can mediate the excision step of mismatch repair using synthetic mismatch repair substrates (42). In vivo experiments with strains multiply deficient in all three exonucleases have, however, failed to demonstrate a role for these exonucleases in mismatch repair, suggesting that other, unknown exonucleases exists in E. coli capable of compensating for the loss of ExoI, ExoVII, and RecJ (34,40).
We have demonstrated that the E. coli open reading frame yobC, which has homology to the DnaQ superfamily, encodes a 25-kDa protein that possesses a potent Mg 2ϩ -dependent ssDNA exonuclease activity. Accordingly, we have renamed this open reading frame exoX and its protein product Exonuclease X. The exoX gene appears to be part of a cistron with another unknown open reading frame denoted yobB. The genes are separated by 23 nucleotides. Although exoX does not appear to have any promoter of its own, the region upstream of yobB contains a 70 promoter-like sequence. Bacterial genome data base searches revealed that both exoX and yobB have protein homologs in Salmonella typhi; in addition, yobB has a second FIG. 6. Complementation of ExoI deficiency by ExoX. STL2701 RecJ Ϫ ExoI Ϫ ExoVII Ϫ (filled symbols) or STL2348 RecJ Ϫ ExoVII Ϫ (open symbols) were transformed with either pExoXKSϪ (f, Ⅺ), pExoXfs (q, E), or pBSKS-(OE, ‚). Transformants were assayed for survival after UV irradiation as described previously in Viswanathan et al. (23). 7. ExoX is synergistic with other single-strand exonucleases in UV repair. Shown are UV survival curves for E. coli strains both singly and multiply deficient for exonucleases I, VII, X, and RecJ exonuclease. All strains assayed for UV survival were derived from the BT199 background. Panel A, BT199, Exo ϩ (Ⅺ); STL4534, ExoX Ϫ (f); STL2331, RecJ Ϫ (‚); STL4540, RecJ Ϫ ExoX Ϫ (OE); STL4556, homolog in Pseudomonas aeruginosa. 3 No other strongly homologous proteins were found in the eubacterial data bases, including the proteobacterial genomes of Haemophilus influenzae and Helicobacter pylori. 4 ExoX did, however, have significant homology with the DNA polymerase exonuclease domain of Bacillus subtilis and lesser homology to the DNA polymerase ⑀ subunits of various eubacterial species including Chlamydia trachomatis and Staphloccoccus aureus.
Here we report the purification of a (His) 6 -ExoX fusion protein and the characterization of its nuclease activity on various DNA substrates. ExoX degrades DNA in a 3Ј to 5Ј direction using a distributive mechanism of hydrolysis. Therefore, unlike the processive nuclease ExoI, ExoX undergoes multiple rounds of binding, hydrolysis, and release to degrade long substrate molecules. Its rapid rate of ssDNA degradation (1,400 nucleotides ssDNA/min/monomer) is due in part to the high affinity of the enzyme (K m ϭ 1.7 nM) for ssDNA. Although capable of degrading both ssDNA and dsDNA molecules, ExoX has higher affinity for ssDNA ends as judged by the extent of degradation of various substrates and by competition experiments. The need for high enzyme to DNA stoichiometry for duplex DNA degradation may mean that the duplex DNA binding step is slow relative to ssDNA binding. The mechanism by which ExoX binds and catalyzes phosphodiester bond cleavage of either the single-strand or duplex DNA end may be revealed by more biochemical and structural information.
Previously we demonstrated that RecJ exonuclease, ExoI, and ExoVII are involved in a redundant fashion in the repair of UV-induced lesions (34). We have proposed that these exonucleases act during recombinational repair of lesions that block replication. Here we have demonstrated that ExoX can specifically compensate for UV repair defects associated with the loss of ExoI, itself a potent 3Ј-5Ј exonuclease. This result correlates well with the biochemical characterization of the enzyme. The fact that ExoX could only partially compensate for ExoI even when expressed in high copy may reflect the disparity between a processive nuclease such as ExoI and a distributive nuclease like ExoX.
Alone, an exoX null mutant did not appear to affect the UV survival of E. coli. However, in combination with mutations in RecJ, ExoI, and ExoVII, an exoX mutation greatly attenuated UV survival. These results demonstrate that exonuclease X can play a role in DNA repair and suggests that all four exonucleases likely share a redundant function. Given that the role of ExoX in repair was only evident when three other exonucleases were removed, ExoX clearly plays a minor role in UV repair. This is similar to ExoI (34) and suggests that 3Ј-5Ј exonucleases, ExoI and ExoX, are less important for UV repair than the 5Ј to 3Ј exonuclease activity of RecJ and ExoVII. The UV sensitivity of the quadruple mutant is extreme, approximating that of RecA, in which recombination is completely blocked. We have suggested that ssDNA degradation is required for efficient RecA-dependent recombinational repair to create presynaptic recombinational intermediates and to extend heteroduplex regions post-synaptically (34,44). Although no individual ssDNA exonuclease is indispensable in DNA repair, a redundant function specified by RecJ, ExoI, ExoVII, and ExoX is essential for UV repair and potentially other cellular processes.