Purification and cloning of Micrococcus luteus ultraviolet endonuclease, an N-glycosylase/abasic lyase that proceeds via an imino enzyme-DNA intermediate.

Although Micrococcus luteus UV endonuclease has been reported to be an 18-kDa enzyme with possible homology to the 16-kDa endonuclease V from bacteriophage T4 (Gordon, L. K., and Haseltine, W. A.(1980) J. Biol. Chem. 255, 12047-12050; Grafstrom, R. [Abstract] H., Park, L., and Grossman, L.(1982) J. Biol. Chem. 257, 13465-13474), this study describes three independent purification schemes in which M. luteus UV damage-specific or pyrimidine dimer-specific nicking activity was associated with two proteins of apparent molecular masses of 31 and 32 kDa. An 18-kDa contaminant copurified with the doublet through many of the chromatographic steps, but it was determined to be a homolog of Escherichia coli ribosomal protein L6. Edman degradation analyses of the active proteins yielded identical NH2-terminal amino acid sequences. The corresponding gene (pdg, pyrimidine dimer glycosylase) was cloned. The protein bears strong sequence similarities to the E. coli repair proteins endonuclease III and MutY. Nonetheless, traditionally purified M. luteus protein acted exclusively on cis-syn thymine dimers; it was unable to cleave site-specific oligonucleotide substrates containing a trans-syn -I,), or Dewar thymine dimer, a 5,6-dihydrouracil lesion, or an A:G or A:C mismatch. The UV endonuclease incised cis-syn dimer-containing DNA in a dose-dependent manner and exhibited linear kinetics within that dose range. Enzyme activity was inhibited by the presence of NaCN or NaBH4 with NaBH4 additionally being able to trap a covalent enzyme-substrate product. These last findings confirm that the catalytic mechanism of M. luteus UV endonuclease, like those of other glycosylase/AP lyases, involves an imino intermediate.

Purifications of Micrococcus luteus UV damage-specific or pyrimidine dimer-specific nicking activity have resulted in the isolation of UV endonuclease proteins with molecular masses ranging from 11 to 18 kDa (2-6). Haseltine et al. (7) proposed that strand scission at a pyrimidine dimer required two activ-ities, an N-glycosylase and an apurinic/apyrimidinic (AP) 1 endonuclease. The UV endonuclease-associated N-glycosylase activity would cleave the N-glycosylic bond between the 5Ј pyrimidine partner of the dimer and the corresponding deoxyribose moiety. Subsequently, an independent AP endonuclease activity would cleave the sugar-phosphate backbone on the 3Ј side of the apyrimidinic sugar moiety. Grafstrom and co-workers (2) modified this proposal by suggesting that the N-glycosylase and AP endonuclease activities both reside on the same UV endonuclease molecule. Characterization of an 18-kDa protein has shown that the UV endonuclease prefers thyminecontaining dimers over cytosine-containing dimers (under conditions of substrate excess), double-stranded over singlestranded DNA, and apyrimidinic sites at the site of glycosylase action to simple apurinic or apyrimidinic residues (2). The M. luteus UV endonuclease locates pyrimidine dimers, at least in vitro, by a processive sliding mechanism on nontarget DNA (8). The efficiency of this scanning is dependent on both ionic strength and pH. The 3Ј terminus generated by UV endonuclease requires further processing by a class II endonuclease before DNA polymerase I and ligase can seal the gap.
The catalytic mechanism of M. luteus UV endonuclease resembles those of a number of other enzymes that perform the initial incision step of base excision repair; bacteriophage T4 endonuclease V (1), Saccharomyces cerevisiae UV endonuclease (9), Escherichia coli endonuclease III (10 -12), and E. coli Fpg (13,14) also possess both N-glycosylase and AP lyase activities. Of these functionally related enzymes, our laboratory is most familiar with the reaction mechanism of T4 endonuclease V. T4 endonuclease V employs its NH 2 -terminal, ␣-amino group in a nucleophilic attack of the C-1Ј sugar carbon of the 5Ј nucleotide within the dimer (15,16). The existence of the resultant imino or Schiff base intermediate can be verified experimentally by demonstrating both that cyanide can inhibit the enzyme in a substrate-dependent manner (17,18) and that NaBH 4 can reduce the imino intermediate to a covalent enzyme-substrate product (19). The imino intermediate may or may not undergo a subsequent ␤-elimination reaction that follows a syn stereochemical course to generate a trans 3Ј-␣,␤-unsaturated aldehyde and a 5Ј-phosphate product (20 -22). Endonuclease III and Fpg also are believed to utilize primary amino groups to form imino intermediates (23,24), possibly Lys 120 in the case of endonuclease III (25) and a primary amino group at or near the amino terminus in the case of Fpg (24). The AP lyase steps of the M. luteus UV endonuclease (21), endonuclease III (23,26,27), and Fpg enzymes all are known to proceed via ␤-elimina-tion. The ␤-elimination step of endonuclease III, like that of T4 endonuclease V, follows a syn stereochemical course. Thus, a unified catalytic mechanism for the N-glycosylase/AP lyases has emerged; this family of enzymes employs a primary amine nucleophile in its attack on the C-1Ј sugar carbon of the damaged nucleoside and, in doing so, creates an identifiable imino intermediate (28,29). Accordingly, the active site residue of the M. luteus UV endonuclease also is hypothesized to be a primary amine.
Since the M. luteus UV endonuclease protein is established in the literature as an 18-kDa protein, it has been compared most frequently with the 16-kDa T4 endonuclease V. This report will demonstrate, however, that the M. luteus enzyme is actually a 31-or 32-kDa protein; a prominent 18-kDa contaminant copurified with UV endonuclease enzyme throughout many chromatographic steps but was inactive against UVdamaged DNA. Moreover, cloning of the pdg (pyrimidine dimer glycosylase) gene revealed that the UV endonuclease shares extensive sequence homology with the E. coli repair proteins endonuclease III and MutY, not T4 endonuclease V. In addition to describing the isolation and cloning of M. luteus UV endonuclease, or more accurately pyrimidine dimer N-glycosylase/AP lyase, we will discuss characterization of the purified enzyme and a preliminary investigation into its reaction mechanism.

Materials
Lyophilized M. luteus cells (ATCC 4698) and lysozyme were obtained from Sigma. The Sephadex G-100, phenyl-Sepharose CL-4B, SP-Sepharose Fast Flow, Mono S, and Mono P matrices or columns were purchased from Pharmacia Biotech Inc.; the Affi-Gel Blue and heparinagarose matrices were from Bio-Rad. Protein molecular weight markers were purchased individually from Sigma or prestained, low molecular size marker sets from Life Technologies, Inc. were employed. Polyvinylidene difluoride membrane was obtained from Bio-Rad, and microsequencing was carried out in the Protein Sequencing Core Laboratory at Vanderbilt University. [␥-32 P]ATP (3,000 Ci/mmol) was purchased from DuPont NEN. New England Biolabs was the supplier for the T4 polynucleotide kinase and double-stranded M13mp18 and M13mp19 DNAs. The oligonucleotides used in cloning were ordered from Research Genetics (Huntsville, AL) or synthesized in the core facilities of Vanderbilt University or the Sealy Center for Molecular Science. R408 helper phage, E. coli XL1-Blue and LE392 cells, and an M. luteus genomic ZAP II library were purchased from Stratagene. The Sequenase version 2.0 DNA sequencing kit and M13 primers were obtained from the U. S. Biochemical Corp. pBR322 DNA was produced in our laboratory or by the Sealy Center Recombinant DNA Laboratory. All 49-mer, dimer-containing oligonucleotides were provided generously by John-Stephen Taylor and Colin Smith (Washington University, St. Louis, MO). The 37-mer, 5,6-dihydrouracil-containing oligonucleotide and its complement were the gift of Paul Doetsch and L. Augeri (Emory University, Atlanta, GA). Oligonucleotide sizing markers were purchased from Pharmacia. NaBH 4 and piperidine stocks were Sigma products. NaCN was obtained from Fisher Scientific Co.

Purification Scheme 1
Isolation of the 18-kDa Ribosomal Protein-All purification procedures were performed at 4°C. Thirty g of lyophilized M. luteus cells was suspended in 1,500 ml of 20 mM Tris⅐HCl, pH 7.5, 10 mM EDTA, 200 mM KCl (buffer A), allowed to hydrate for several hours, and then lysed overnight by the addition of lysozyme at 600 g/ml into buffer A. The following day, the viscous mixture was sonicated with a Branson Sonifier 450 until its consistency thinned, an indication that the majority of the cells had been disrupted. Microscopic examination of the pre-and postsonication samples also confirmed that essentially complete lysis had been achieved. The cellular debris was pelleted by centrifugation at 13,000 ϫ g for 1 h, and the resultant supernatant was loaded onto a single-stranded DNA-agarose column (20 cm 2 ϫ 75 cm) that had been equilibrated with buffer A containing 10% (v/v) ethylene glycol (30). After the column was loaded and washed thoroughly with the equilibration buffer, bound proteins were eluted with a 2-liter linear gradient of equilibration buffer containing 200 -2,000 mM KCl. Eleven-ml frac-tions were collected and monitored for M. luteus UV endonuclease activity, as in all subsequent purification steps and schemes, by assaying their ability to nick UV-irradiated pBR322 DNA and/or thymine dimer-containing, site-specific oligonucleotides. Although fractions were also examined by SDS-PAGE coupled to either Coomassie or silver staining, they were pooled according to their activity rather than their molecular mass profiles. Single-stranded DNA-agarose fractions 91-141 were combined (1,100 -1,500 mM KCl, 600 ml), concentrated to 200 ml on an Amicon concentrator equipped with a YM10 membrane under 40 p.s.i. of pressure, and loaded in two batches of 100 ml each onto a Sephadex G-100 (20 cm 2 ϫ 100 cm) column that had been equilibrated with 25 mM Na 2 HPO 4 , pH 7.5, 1 mM EDTA, 100 mM KCl, 10% (v/v) ethylene glycol (buffer B). During each run, 13-ml fractions were collected, and the majority of the UV damage-specific nicking activity was found in fractions 41-61. The active fractions from both runs (450 ml) were applied to a heparin-agarose column (3.0 cm 2 ϫ 10 cm) that had been equilibrated with buffer B. After thorough washing of the column with equilibration buffer, elution with a 500-ml linear gradient of 100 -400 mM KCl in buffer B resulted in the enzyme activity peaking between 11-ml fractions 8 and 26 (150 -270 mM KCl, 209 ml). This material, once diluted 2-fold with buffer B minus KCl, was reloaded onto the reequilibrated heparin-agarose column and eluted a second time with a 100 -250 mM KCl, 500-ml linear gradient of buffer B. Twelve-ml fractions 2-15 (110 -150 mM KCl, 168 ml) contained the majority of the UV endonuclease activity and were pooled for further purification on a small single-stranded DNA-agarose column (1.8 cm 2 ϫ 50 cm). The sample was loaded, the column was washed with buffer B, and a 500-ml linear gradient of 100 -1,500 mM KCl in buffer B was run. Fraction 23 (850 mM, 11.5 ml) was prepared for phenyl-Sepharose chromatography by slowly adding (NH 4 ) 2 SO 4 to a final concentration of 1 M. The phenyl-Sepharose column (3.0 cm 2 ϫ 7.0 cm) was equilibrated with Na 2 HPO 4 , pH 7.5, 1 mM EDTA, 1,000 mM (NH 4 ) 2 SO 4 , 10% (v/v) ethylene glycol (buffer C), loaded, washed briefly with its high salt equilibration buffer C, and then eluted of proteins with a 200-ml gradient of 1,000 -0 mM (NH 4 ) 2 SO 4 buffer C. Four-ml fractions 31-41 (380 -180 mM (NH 4 ) 2 SO 4 , 44 ml) were combined, concentrated through Amicon filtration to 10 ml, diluted with 50 ml of 10 mM Tris⅐HCl, pH 7.5, 1 mM EDTA to lower the salt concentration, and again concentrated, this time to 7 ml. Concomitantly, 10 ml of small single-stranded DNAagarose fraction 24 was diluted with 130 ml of 10 mM Tris⅐HCl, pH 7.5, 1 mM EDTA prior to concentration to 10 ml. Each sample was then ready to be loaded onto an FPLC Mono S column that had been equilibrated with 10 mM Tris⅐HCl, pH 7.5, 10 mM NaCl. In the first Mono S run, the phenyl-Sepharose sample was loaded, the column was washed briefly with equilibration buffer, and a linear gradient of 10 mM Tris⅐HCl, pH 7.5, buffer containing 10 -750 mM NaCl was programmed and executed. The majority of the enzyme activity was contained within 1.5-ml fractions 13-16 (6 ml). The second Mono S run (single-stranded DNA-agarose sample) was conducted following the same protocol with UV damage-specific nicking activity eluting primarily in fractions 13-18 (12 ml). Side fractions 8 and 9 from the first Mono S run and 11-13 from the second Mono S run were pooled, at this particular step, according to SDS-PAGE silver stain results; the peaks for the 18-kDa protein and UV damage-specific nicking activity were by now somewhat skewed. The (NH 4 ) 2 SO 4 concentration of the sample was raised to 1 M, and the Mono S material was applied once more to the phenyl-Sepharose column that had been equilibrated in buffer C minus ethylene glycol. After the column had been washed thoroughly, the 18-kDa protein was eluted with a 200-ml linear gradient of equilibration buffer containing 1,000 -0 mM (NH 4 ) 2 SO 4 . Fractions 37 and 38 were combined (6 ml), mixed with an equal volume of 20% (w/v) trichloroacetic acid, set on ice for 30 min to precipitate, and subjected to electrophoresis on a 13% SDS-PAGE gel. The 18-kDa protein (30 -50 pmol) was transferred to polyvinylidene difluoride paper according to the method of Matsudaira (31), excised from the membrane, and microsequenced via Edman degradation.
Cloning of the 18-kDa Ribosomal Protein-Based on the aminoterminal sequence of the 18-kDa protein, sense and antisense 42-mer primers were designed for cloning of the protein: 5Ј-ATGTCCCGCA-TCGGCCGCCTCCCGATCACCATCCCGGCCGGC-3Ј (852) and 5Ј-GC-CGGCCGGGATGGTGATCGGGAGGCGGCCGATGCGGGACAT-3Ј (853). Southern blots performed on restriction digest gels of M. luteus genomic DNA (prepared from the freeze-dried cells) both confirmed that a signal could be detected using either probe and helped to optimize hybridization conditions for the actual cloning procedure. As the initial step in the cloning strategy, a randomly sheared M. luteus genomic ZAP II library was plated on NZY (1% (w/v) NZ amine, 5% (w/v) NaCl, 2% (w/v) MgSO 4 ⅐7H 2 O) ϩ 0.7% agarose-overlaid plates using E. coli LE392 as the host bacterium. Plaques were visible after the bacteriophage had grown overnight at 37°C, so the plates were chilled for 1 h prior to lifting. The nitrocellulose filters were processed successively in 30-s steps with solutions A (1.5 M NaCl, 0.5 M NaOH), B (0.5 M Tris⅐HCl, pH 7.5), and C (3 M NaCl, 300 mM sodium citrate, pH 7.0) to denature the DNA. The filters were baked in a vacuum oven for 2 h at 80°C and then immersed in prehybridization solution (20% formamide, 5 ϫ SSPE where 1 ϫ SSPE equals 50 mM NaH 2 PO 4 ⅐H 2 O, pH 7.4, 750 mM NaCl, 5 mM EDTA), 5 ϫ Denhardt's solution, 100 g/ml fish milt DNA, 0.1% SDS) for 4 h at 48°C. 32 P-Labeled probe 853 was added, and the hybridization was allowed to proceed overnight at 48°C. After a quick rinse, the filters were washed twice for 30 min with 4 ϫ SSPE (200 mM NaH 2 PO 4 ⅐H 2 0, pH 7.4, 600 mM NaCl, 20 mM EDTA), dried, and exposed to XAR-5 film (Kodak) at Ϫ70°C. Eight positive plaques were detected, cored, and eluted into suspension medium (50 mM Tris⅐HCl, pH 7.5, 100 mM NaCl, 8 mM MgSO 4 ⅐7H 2 O, 0.01% (w/v) gelatin). Two different concentrations of each eluate were replated for a second round of screening. Plates for each positive were screened with either probe 852 or 853, and only clones of 1, 2, and 6 survived this second generation hybridization. Again, the positive plaques were cored and eluted into suspension medium with gentle shaking. In addition to being replated for a third round of screening, the positive phage were rescued as phagemid in the following manner. Core eluates were incubated with log phase E. coli XL1-Blue cells and R408 helper phage for 15 min at 37°C. Each cell mixture was then added to 5 ml of 2 ϫ YT (16% (w/v) Bacto-tryptone, 10% (w/v) Bacto-yeast extract, 5% (w/v) NaCl, pH 7.0) medium and allowed to grow for 3 h at 37°C. The cells were lysed by heating the tubes for 20 min at 70°C and centrifuged at 4,000 ϫ g for 5 min. pBluescript phagemid packaged as filamentous phage particles were recovered in the supernatant and plated with log phase E. coli XL1-Blue cells onto 10 Luria Bertani plates supplemented with 50 g/ml ampicillin. Meanwhile, in the third and final round of screening, plaques derived from original clones 1 and 6 hybridized to probe 852; clone 2 proved to have been a false positive. Single colonies were picked from the XL1-Blue plates, cultures grown overnight, and DNA isolated using a STET (10 mM Tris⅐HCl, pH 8.0, 100 mM NaCl, 1 mM EDTA, 5% (v/v) Triton X-100) preparation procedure. A series of restriction enzyme digests/Southern blots suggested that the M. luteus gene could be subcloned into M13mp18 and M13mp19 using a SacI/EcoRI ligation strategy. The ligation mixtures were transformed into E. coli UT481 bacteria, the plaques were probed with 852 and 853, and positives were cored and grown overnight in 10 ml of 2 ϫ YT at 37°C. While replicative form DNA was harvested from the culture pellets, single-stranded M13 DNA was recovered from the supernatants. Sequencing focused on the M13mp18-clone 1 and M13mp19-clone 1 constructs. The M13 universal primer was used in the initial stages of sequencing, but, as sequencing progressed throughout the gene, it became necessary to design additional primers. Homology searching was done using the National Center for Biotechnology Information Blaster program.

Purification Scheme 2
Isolation of the 31/32-kDa M. luteus UV Endonuclease-Many of the experimental details are parallel to those presented in purification scheme 1 and can be inferred from the above protocol. Fifty-four g of M. luteus cells was suspended in 2,500 ml of buffer A and allowed to hydrate overnight. Next, lysozyme was added into the cell slurry to a final concentration of 400 g/ml, and lysis was allowed to proceed until the following day. The concentration of KCl in buffer A was elevated to 400 mM prior to sonication to accelerate mechanical disruption of the cells. Two 30-min centrifugation spins at 2,600 ϫ g were required to obtain a particle-free supernatant that could be loaded onto a singlestranded DNA-agarose column (20 cm 2 ϫ 75 cm). Before loading, however, the cellular extract was diluted with an equal volume (3 liters) of 20 mM Tris⅐HCl, pH 7.5, 10 mM EDTA to return the salt concentration to 200 mM. Thirteen-ml fractions were collected from a 2-liter gradient of buffer A containing 200 -2,000 mM KCl. Fractions 56 -106 (850 -1,400 mM, 700 ml) were pooled and applied to an Affi-Gel Blue column (5.0 cm 2 ϫ 10 cm) that had been equilibrated with buffer A containing 500 mM KCl. The UV endonuclease enzyme bound weakly to the column and emerged in the flow-through, wash, and 8-ml fractions 1-11 (500 -1,500 mM KCl) of a 300-ml, 500 -4,000 mM buffer A gradient. The active effluents were combined (950 ml), diluted 2-fold with 20 mM Tris⅐HCl, pH 7.5, 10 mM EDTA, reloaded onto the Affi-Gel Blue column, and the proteins again eluted with a 300-ml, 500 -4,000 mM buffer A gradient. This time, the flow-through, wash, and fractions 1-21 (500 -2,500 mM KCl) were pooled (2,100 ml) in preparation for loading onto a buffer C-equilibrated phenyl-Sepharose column (3.0 cm 2 ϫ 7.0 cm). Initially, in a trial run, 525 ml of the 2,100-ml Affi-Gel Blue pool was loaded onto the phenyl-Sepharose column, and the remainder was applied in a second run. Before each run, (NH 4 ) 2 SO 4 was slowly mixed into the samples to achieve a final concentration of 1 M, and a 200-ml gradient of 1,000 -0 mM (NH 4 ) 2 SO 4 in buffer B was employed to elute the UV damage-specific nicking activity. Five-ml fractions 25-50 (130 ml) from the first phenyl-Sepharose run and 4-ml fractions 5-29 (100 ml) from the second phenyl-Sepharose run were pooled independently. An additional 25 mM NaH 2 PO 4 , pH 6.8, was added to each pool before it was applied to a Sephadex G-100 column (20 cm 2 ϫ 100 cm) that had been equilibrated in 25 mM NaH 2 PO 4 , pH 6.8, 1 mM EDTA, 100 mM KCl, 10% (v/v) ethylene glycol (buffer D). In each run, as soon as the sample was loaded, buffer D was loaded onto the column, and 12-ml fractions were collected. Fractions 26 -66 from the first Sephadex G-100 run (first phenyl-Sepharose run material) and fractions 41-56 from the second Sephadex G-100 run (second phenyl-Sepharose run material) were combined, and 150 ml of the 600-ml pool was applied to a heparin-agarose column (5.0 cm 2 ϫ 7.0 cm) that had been equilibrated with buffer D. After the column was washed thoroughly with equilibration buffer, elution with a 150-ml linear gradient of 100 -400 mM KCl in buffer D yielded an enzyme activity peak in 5-ml fractions 33-38 (30 ml). This pool, along with the remainder of the Sephadex G-100 pool, was loaded onto a UV-irradiated single-stranded DNA-agarose column (0. Cloning of the 31/32-kDa M. luteus UV Endonuclease-Based on the amino-terminal sequence of the 31-and 32-kDa proteins, the following sense and antisense 38-mer primers were designed for cloning of the UV endonuclease gene: 5Ј-ATGGAGACGGAGTCCACGGGCACGCCGAC-GGGCGAGAC-3Ј (1083) and 5Ј-GTCTCGCCCGTCGGCGTGCCCGTG-GACTCCGTCTCCAT-3Ј (1084). Southern blots were performed on restriction digest gels of M. luteus genomic DNA (prepared from freezedried cells) both to confirm that a signal could be detected using either probe and to optimize hybridization conditions for the actual cloning procedure. The randomly sheared M. luteus genomic ZAP II library was plated and hybridized at 42°C with 32 P-labeled probe 1083. Two of the 10 positive clones from the initial plating of the M. luteus library survived three subsequent rounds of hybridization. These positives were rescued as phagemid into pBluescript, the inserts were subcloned into M13mp18, and the M13mp18-clone 3 construct was sequenced. Unfortunately, although a 4-kilobase fragment of genomic DNA had been isolated, only the first 666 bp of the pyrimidine dimer N-glycosylase/AP lyase coding sequence were present. An additional 1.5 ϫ 10 6 plaques were screened resulting in the isolation of a larger 5.5-kilobase genomic DNA fragment. Using probes spanning the length of the known 666-bp sequence, a series of restriction digestions and Southern blots was performed to determine the position of the gene of interest within the 5.5-kilobase fragment. A 2.0-kilobase MluI fragment thought to incorporate the full-length gene was isolated and its overhang ends blunted so that it could be ligated into the SmaI site of M13mp18.
The pdg gene was sequenced with a battery of sense oligonucleotides and then subcloned into M13mp19 or plasmids for sequencing in the opposite direction with antisense primers. GC-rich stretches of the sequence often were compressed or poorly resolved, thus making the exact sequence difficult to interpret. Those regions that were recalcitrant to standard Sanger dideoxy sequencing were resequenced with 7-deaza-2Ј-deoxy-GTP in the deoxyribonucleoside triphosphate mix or by the Maxam-Gilbert method. Alternatively, sequential and overlapping sections of the gene were amplified by polymerase chain reaction and then sequenced individually.

Purification Scheme 3:
Twenty-five g of M. luteus cells was suspended in 1,250 ml of buffer B minus ethylene glycol, allowed to hydrate overnight, and then lysed for the next 24 h by the addition of 400 g/ml lysozyme into the cell slurry. The KCl concentration of the slurry was raised to 300 mM immediately before sonication. Cellular debris was pelleted by two 30-min rounds of centrifugation at 8,000 ϫ g, and the resultant supernatant was loaded onto a single-stranded DNA-agarose column (20 cm 2 ϫ 75 cm) that had been equilibrated in buffer B. The majority of the UV endonuclease activity eluted in 12-ml fractions 100 -140 (890 -1,300 mM KCl, 500 ml) of a 2-liter, 300 -2,000 mM KCl buffer B gradient. In a trial run, fractions 128 -132 (60 ml) were pooled, dialyzed against 25 mM NaH 2 PO 4 , pH 6.8, 1 mM EDTA (buffer F), and applied to an SP-Sepharose column (1.8 cm 2 ϫ 45 cm) that had been equilibrated in buffer F. A 500-ml linear gradient of 0 -800 mM KCl in buffer F concentrated enzyme activity in 3.5-ml fractions 24 -36 (5-70 mM KCl, 45 ml). Since a limited number of candidate proteins could be detected in these active fractions by silver staining, another SP-Sepharose column (1.8 cm 2 ϫ 100 cm) was packed and equilibrated with buffer F. Singlestranded DNA-agarose fractions 102-127 were pooled (312 ml), dialyzed against 15 mM Na 2 HPO 4 , pH 7.6, 1 mM EDTA (buffer G), and loaded onto the new matrix. After the column was washed, it was eluted of proteins with a 600-ml linear gradient of 0 -400 mM KCl in buffer G. Although UV endonuclease activity was spread out between fractions 60 and 100, fractions 64 -76 (220 -250 mM KCl, 57 ml) were pooled and combined with fractions 24 -36 from the first SP-Sepharose run (102 ml total volume). (NH 4 ) 2 SO 4 was added to a final concentration of 1.8 M before the SP-Sepharose material was loaded onto a phenyl-Sepharose column (1.8 cm 2 ϫ 45 cm) that had been equilibrated with 25 mM NaH 2 PO 4 , pH 6.8, 1.8 mM (NH 4 ) 2 SO 4 , 1 mM EDTA, 100 mM KCl, 10% (v/v) ethylene glycol (buffer H). A 400-ml linear gradient of 1.8 -0 M (NH 4 ) 2 SO 4 in buffer H was run, but no nicking activity was found until 3-ml fractions 180 -290, fractions that were collected during a postgradient wash with buffer H. In fact, UV endonuclease activity continued to trail off in fractions 291-330 as the column was being stripped with 25 mM NaH 2 PO 4 , pH 6.8, 1 mM EDTA, 10% ethylene glycol.

M.luteus UV Endonuclease UV Damage-specific
Nicking Activity on Plasmid DNA In purification schemes 1-3, fractions were monitored routinely for their ability to nick irradiated versus unirradiated plasmid DNA. pBR322 plasmid DNA was irradiated with 254-nm light at 100 microwatts/cm 2 for 245 s to generate 20 -25 pyrimidine dimers/plasmid molecule (32). The DNA was then diluted from approximately 1 g/l to 0.100 g/l in 25 mM Na 2 HPO 4 , pH 7.5, 1 mM EDTA, 100 g/ml bovine serum albumin (purification schemes 1 and 2) or to 0.075 g/l in 25 mM NaH 2 PO 4 , pH 6.8, 25 mM NaCl, 1 mM EDTA, 100 g/ml bovine serum albumin (purification scheme 3). Varying amounts of the column fractions, usually 1-5 l or dilutions thereof, were incubated with 20 l of irradiated or unirradiated pBR322 (2.0 g of DNA in purification schemes 1 and 2, 1.5 g in purification scheme 3) for 30 min at 37°C. Reactions were terminated by the addition of an equal volume of electrophoresis loading buffer (50 mM Tris⅐HCl, pH 8.0, 20 mM EDTA, 40% (w/v) sucrose, 2% (w/v) SDS, 0.02% (w/v) bromphenol blue, 0.02% (w/v) xylene cyanol). Form I (supercoiled), form II (nicked circular), and form III (linear) DNAs were resolved by electrophoresis through a 1% (w/v) agarose gel in 40 mM Tris⅐OAc, pH 8.0, 1 mM EDTA running buffer. The gel was stained in 0.5 g/ml ethidium bromide so that the three topological forms of pBR322 could be visualized on a longwave UV lightbox. To quantitate the data, images of the gels were captured by a digital camera system (The Imager, Appligene) and then analyzed using Bio-Image software (Millipore).

M. luteus UV Endonuclease Damage-and Mismatch-specific Nicking Activities on Site-specific Oligonucleotide Duplexes
In purification scheme 3, fractions were monitored for both their ability to nick irradiated plasmid DNA and their ability to nick a 49-bp oligonucleotide duplex containing a site-specific cis-syn thymine dimer.
In addition, purified M. luteus UV endonuclease (phenyl-Sepharose fraction 315) was tested for its ability to nick 49-mers containing a trans-syn-I, (6 -4), or Dewar thymine dimer (33), a 37-mer containing a 5,6-dihydrouracil lesion, and 50-mers containing an A:G or A:C mismatch. The trans-syn-I, (6 -4), and Dewar thymine dimer-containing duplexes shared the following sequence in the damaged strand: 5Ј-AGCTACCATGCCTGCACGAATTAAGCAATTCGTAATCATGGTCA-TAGCT-3Ј. The cis-syn damaged strand had a slightly altered sequence: 5Ј-AGCTACCATGCCTGCACGTATTATGCAATTCGTAATCATGGT-CATAGCT-3Ј. Each dimer type bridged the two thymines in boldfaced positions 21 and 22 of the 49-mers relative to their 5Ј ends. Control 49-mer incorporated the AATTAA sequence but was left undamaged at the TT site. The damaged strand of the 37-mer duplex had the sequence 5Ј-CTTGGACTGGATGTCGGCACXAGCGGATACAGGAGCA-3Ј with the X in position 21 representing a 5,6-dihydrouracil lesion. The complementary strand incorporated a G opposite the X. In the 50-mer "damaged" strand, 5Ј-AGCTACCATGCCTGCACGAGATAAGCAATT-CGTAATCATGGTCATAGCTA-3Ј, the A at position 21 served as the mismatched base when it was paired with a C or a G in the complement. The control complement incorporated a T opposite the A. Each type of damage-containing strand was 32 P labeled on its 5Ј end with T4 polynucleotide kinase and then annealed to its complement. Depending upon the experiment, varying amounts of phenyl-Sepharose fraction 315 or control enzyme were incubated at 37°C for the indicated length of time with the appropriate duplex 49-mer (0.4 ng ϭ 12.5 fmol in a 10-l volume or 0.8 ng ϭ 25 fmol in a 20-l volume), 37-mer duplex (0.8 ng ϭ 32 fmol in a 20-l volume), or 50-mer duplex (0.5 ng ϭ 16 fmol in a 10-l volume) in 25 mM NaH 2 PO 4 , pH 6.8, 1 mM EDTA, 100 g/ml bovine serum albumin. Reactions were terminated either by placing the reaction tubes in a dry ice-EtOH bath or by treating the reaction mixtures with 1 M piperidine for 30 min at 85°C. When piperidine treatment was carried out, reaction samples were dried down in a Savant SC110 SpeedVac and then resuspended in twice their original reaction volume of loading buffer (95% (v/v) formamide, 20 mM EDTA, 0.02% (w/v) bromphenol blue, 0.02% (w/v) xylene cyanol). When piperidine treatment was not carried out, an equal volume of formamide loading buffer was added to the samples. All samples were heated for 3-5 min at 80°C prior to being loaded onto 15% polyacrylamide gels containing 8 M urea. The oligonucleotide bands were visualized by autoradiography of the wet gels with Hyperfilm-MP film (Amersham Corp.) at Ϫ70°C, typically for several hours with two DuPont Quanta III intensifying screens enclosed in the cassette. When it was necessary to obtain quantitative results, the wet gels were also scanned on a PhosphorImager 450 machine (Molecular Dynamics) and the data analyzed using Image Quant software (Molecular Dynamics).

Purification and Cloning of the 18-kDa Ribosomal Protein-
The original intent of purification scheme 1 was to isolate the 18-kDa M. luteus UV endonuclease that had been reported in the literature (2). Since the M. luteus UV endonuclease was believed to be similar in structure to T4 endonuclease V, the purification scheme incorporated many of the chromatographic steps that are routinely employed by our laboratory to purify endonuclease V: single-stranded DNA-agarose, Sephadex G-100, heparin-agarose, phenyl-Sepharose, and FPLC Mono S (34). Column fractions were assayed for M. luteus UV endonuclease activity by performing nicking assays on UV-irradiated versus nondamaged pBR322 plasmid DNA. The protein makeup of the fractions was assessed via SDS-PAGE coupled to either Coomassie Blue or silver staining. After following the UV damage-specific nicking activity of the M. luteus lysate over five different column types, with some columns being run twice, an 18-kDa protein emerged as the most prominent band. Contaminants were still present even after the FPLC Mono S runs, however, with a 31/32-kDa doublet being particularly noticeable (Fig. 1A). Consequently, the Mono S fractions were pooled according to protein content, and the 18-kDa protein was purified further to Ϸ80% homogeneity on a second phenyl-Sepharose column. Microsequencing of the first 14 amino acids of the 18-kDa protein ( Table I) (35)(36)(37). The amino acid sequence of the 18-kDa protein clearly matched with that of the L6 ribosomal protein (EMBL accession number X17524) and not a repairrelated protein; the NH 2 -terminal amino acid sequences were identical, and our L6 gene sequence differed from the published sequence in only 12 regions of minor sequence discrepancies.
Purification and Cloning of the 31/32-kDa M. luteus UV Endonuclease-After concluding that the 18-kDa protein was not the M. luteus UV endonuclease, the protein and activity profiles collected during purification scheme 1 were reexamined. Reinspection of the data showed that the peaks for the 18-kDa protein and UV damage-specific nicking activity had been slightly offset; faint 31-and 32-kDa protein bands correlated more directly with protein activity. Purification scheme 2 therefore was undertaken with the purpose of isolating this protein doublet. It included the following chromatographic steps: single-stranded DNA-agarose, Affi-Gel Blue, phenyl-Sepharose, Sephadex G-100, heparin-agarose, UV-irradiated single-stranded DNA-agarose, FPLC Mono S, and FPLC Mono P. As was expected from the observations described above, the UV damage-specific nicking activity was retained even after the removal of small molecular weight proteins (M r 11,000 -18,000) from the enzyme preparation. Fig. 1, B and C, illustrates the correspondence between the amount of 31/32-kDa protein and the level of UV damage-specific nicking activity for several FPLC Mono P fractions. Quantitative analysis of  (38), so it followed that fraction 3 had produced approximately six times as many double-stranded breaks per plasmid than fraction 4. Identical NH 2 -terminal amino acid sequences were obtained from microsequencing of the 31-and 32-kDa proteins (Table I), and the common sequence was used to design sense and antisense 38-mer primers.
Cloning identified an 804-bp open reading frame encoding a protein of 268 amino acids with a calculated molecular mass of 29,306 Da ( Fig. 2A). Curiously, a second stop codon was present at bp position 838 which could potentially signal the termination of a longer protein of 279 amino acids with a calculated molecular mass of 30,340 Da. Although we were unsure of how the M. luteus transcription machinery might bypass the first OPA codon, perhaps through an amber suppressor-type mechanism, it seemed plausible that the 268-and 279-residue proteins corresponded to the two forms of purified UV endonuclease protein with apparent molecular masses of 31 and 32 kDa, respectively. Indeed, amino-terminal sequence data obtained for the first 35 residues of the 31-kDa protein and the first 24 residues of the 32-kDa protein matched the NH 2 -terminal sequence predicted by the cloned gene. Comparable to other M. luteus genes that have been sequenced, pdg possessed an overall GC content of 72%, and 94% of its codons contained a G or C in the third position. Also, a purine-rich, Shine-Dalgarno-like  in panel B, lanes 1-3, is presented in panel  C, lanes 1-3. In the assay shown in panel C, enzyme (5 l of each fraction) was incubated with irradiated pBR322 DNA (2.0 g) for 30 min at 37°C. Form I (supercoiled), form II (nicked circular), and form III (linear) DNAs were resolved by electrophoresis through a 1% (w/v) agarose gel. The control lane (C) contained irradiated pBR322 plasmid DNA that had been incubated but without enzyme. To quantitate the silver stain or agarose gel bands, images of the gels were captured by a digital camera system and then analyzed using BioImage software.

M E T E S T G T P T G E T R L A L V R R A R R I
sequence was present 14 to 8 bases upstream of the start codon. A GenBank search revealed that the protein deduced from the pdg open reading frame shared significant homology with two E. coli repair proteins, endonuclease III (31% identity across the length of the M. luteus protein) and MutY (22% identity) (Fig. 2B), and an uncharacterized, pFV1 plasmidencoded protein from Methanobacterium thermoformicicum (the open reading frame 10 product, 16% identity) (25,39). Across all four proteins, the regions surrounding the endonuclease III thymine glycol binding site (Ala-113 through Arg-119) and the putative [4Fe-4S] 2ϩ cluster motif (C-X 6 -C-X 2 -C-X 5 -C) were particularly well conserved. The nucleotide sequence for the M. luteus pdg gene has been submitted to GenBank and assigned accession number U22181.
Purification scheme 3 accomplished two goals: (i) a more streamlined purification procedure was developed, and (ii) enough protein was isolated to characterize the enzyme. The 32-kDa partner of the 31/32-kDa UV endonuclease pair was purified to apparent homogeneity after three chromatographic steps: single-stranded DNA-agarose, SP-Sepharose, and phenyl-Sepharose. Active 31-kDa protein was eluted off the phenyl-Sepharose column prior to its 32-kDa partner, but it was impure (data not shown). Pure 32-kDa protein was eluted off the phenyl-Sepharose column in fractions 300 -325 only after extensive, postgradient washing with low salt buffer (Fig. 3A). Consequently, these fractions were very dilute; each silver stain band represented 400 l of trichloroacetic acid-precipitated material. Each fraction was capable of nicking a 49-mer oligonucleotide duplex containing a site-specific cis-syn cyclobutane thymine dimer. The resultant 20-mers comigrated with the ␤-elimination product generated by T4 endonuclease V (Fig. 3B, lane 18). Fraction 330 lacked the 32-kDa protein and was incapable of nicking the 49-mer substrate.
Characterization of the M. luteus UV Endonuclease-The recent synthesis by Smith and Taylor (33) of a set of deoxyoligonucleotide 49-mers containing defined thymidylyl-(3Ј 3 5Ј)thymidine photoproducts allowed us to investigate the dimer specificity of purified UV endonuclease. The 32-kDa protein (phenyl-Sepharose fraction 315 from purification scheme 3) incised duplex 49-mer containing cis-syn but not trans-syn-I, (6 -4), or Dewar thymine dimers (Fig. 4A). Nor could the UV endonuclease nick a duplex 37-mer containing a 5,6-dihydrouracil lesion, a substrate of endonuclease III (40), or duplex 49-mers containing either an A:G or an A:C mismatch, known substrates of MutY (Fig. 4, B and C) (41). Although it is not evident from the exposure presented in Fig. 4C, MutY was able to cleave the A:C substrate at 4% of the rate that it cleaved the A:G substrate. UV endonuclease product PhosphorImager counts were slightly elevated for the 5,6-dihydrouracil and A:G substrates relative to their undamaged counterparts, but, at most, the counts approached 2% of the positive control counts or Ͻ1% of total counts. Considering that piperidine treatment was not required to detect the cleavage product of the cis-syn thymine dimer-containing substrate and that more UV endonuclease enzyme was introduced into the 5,6-dihydrouracil and mismatch assays than into the pyrimidine dimer assays, its cis-syn pyrimidine dimer-specific nicking capability probably represents the biologically relevant role of M. luteus UV endonuclease. The purified protein nicked cis-syn dimer-containing DNA in a concentration-dependent manner, and piperidine treatment did not enhance the conversion of 49-mer to 20-mer product (Fig. 5). Piperidine treatment did increase the mobility of the 20-mer reaction product, however; removal of the 3Ј sugar fragment through ␦-elimination left a 3Ј phosphate terminus with a faster mobility than the original aldehyde (data not shown). The fact that piperidine treatment failed to convert Regions of strong homology are indicated by upper case as opposed to lower case lettering. Alignments and homology assessments were determined initially using the National Center for Biotechnology Information multiple alignment construction and analysis workbench program (MACAW) and then edited manually. Cysteines within the conserved C-X 6 -C-X 2 -C-X 5 -C coordination motif and the putative catalytic residues for endonuclease III and UV endonuclease (Lys-120 and Lys-135, respectively) are highlighted with boldfaced type. a significant number of AP sites to single-stranded breaks suggested that the AP lyase activity of the M. luteus UV endonuclease was at least as strong as its N-glycosylase activity. Had the UV endonuclease possessed a weak AP lyase, AP sites would have remained in the DNA following the N-glycosylase cut, piperidine treatment would have created DNA breaks, and an increase in 20-mer product would have been observed. Since the dose dependence of the UV endonuclease was not perfectly linear, i.e. 2.50 l of enzyme converted 100% of the 49-mer to 20-mer, whereas 1.25 l of the enzyme converted only 30% of the substrate to product, the kinetics of the UV endonuclease were examined (Fig. 6). As would be expected for a situation of substrate excess or even for this situation of approximately equimolar substrate and enzyme concentrations, initial reaction velocities were proportional to the enzyme concentration: 1.25 l, 1.02% min Ϫ1 (r ϭ 0.991); 2.50 l, 1.87% min Ϫ1 (r ϭ 0.985); and 3.75 l, 3.70% min Ϫ1 (r ϭ 0.975).
Evidence for an Imino Intermediate in the M. luteus UV Endonuclease Reaction-The catalytic mechanism of the M. luteus UV endonuclease was hypothesized to proceed via an imino intermediate like that of other N-glycosylase/AP lyases. Such an intermediate was detected both indirectly by demonstrating that cyanide inhibited the enzyme and directly by trapping the intermediate as (a) covalent enzyme-substrate product(s) with the reducing agent NaBH 4 . UV endonuclease was reacted with cis-syn dimer-containing duplex 49-mer in the presence of NaCl or equimolar concentrations of NaCN (Fig. 7,  A and B). Fifty percent inhibition of the reaction occurred around 6 mM NaCN, an IC 50 very similar to the 3-5 mM range observed for T4 endonuclease V (19). Cyanide, unlike NaBH 4 , reacts with an imino intermediate to form a slowly reversible tetrahedral complex that cannot be isolated by denaturing PAGE. Therefore, no shifted bands representing enzyme-substrate complexes were seen in the Fig. 7A autoradiograph. Stable complexes were formed when NaBH 4 was present in the reaction at Ն10 mM (Fig. 7C). The complexes migrated more slowly than the 49-mer substrate alone and were located just beneath the wells. It is not understood how the upper and lower complexes differed, but they were formed somewhat quantitatively according to the amount of enzyme added to the reactions. The PhosphorImager covalent complex counts in Fig , lanes 1, 4, and 7), phenyl-Sepharose fraction 315 (2.5 l, lanes 2, 5, and 8), or MutY (100 ng, lanes 3, 6, and 9) was incubated with A:T (lanes 1-3), A:G (lanes 4 -6), or A:C (lanes 7-9) 50-mer duplex (0.5 ng ϭ 16 fmol). All reactions were allowed to proceed for 60 min at 37°C before they were terminated by the addition of formamide loading buffer (panel A) or piperidine treatment (panels B and C). Samples were then subjected to electrophoresis on a 15% polyacrylamide gel containing 8 M urea. M lanes contained oligonucleotide sizing markers ranging from 8 to 32 bases. 32 P-Labeled 49-mer substrate and 20-mer product were visualized via autoradiography of the wet gels. To obtain quantitative data, digital images of the autoradiographs were analyzed using BioImage software.  5 l, lanes 1 and 2), or T4 endonuclease V (1.5 l ϭ 450 ng, lanes 17 and 18) was incubated with control (odd lanes) or cis-syn thymine dimer-containing (even lanes) duplex 49-mer (0.4 ng ϭ 12.5 fmol). Reactions were allowed to proceed for 60 min at 37°C before they were terminated by the addition of formamide loading buffer. Samples were then run on a 15% acrylamide gel containing 8 M urea. Lane M contained oligonucleotide sizing markers ranging from 8 to 32 bases. 32 P-Labeled 49-mer substrate and 20-mer product were visualized via autoradiography of the wet gel. enzyme. Fig. 7D illustrates graphically that high salt concentrations whether of NaBH 4 or NaCl dramatically reduced the nicking activity of the M. luteus UV endonuclease. A direct comparison of Fig. 7, B and D, would be misleading; it would lead one to the conclusion that NaBH 4 inhibited the reaction much less efficiently than NaCN. If a NaBH 4 inhibition experiment had been conducted using 1 l rather than 4 or 16 l of UV endonuclease, 50% inhibition probably would have occurred at a more comparable level of NaBH 4 than the ϳ30 mM concentration observed in Fig. 7D. DISCUSSION For a number of years, our laboratory tried unsuccessfully to clone the M. luteus UV endonuclease using strategies designed to exploit the suspected homology between the M. luteus enzyme and T4 endonuclease V (42). As a final strategy, we chose to purify the M. luteus UV endonuclease, sequence its amino terminus, and design best guess oligonucleotide probes to screen a randomly sheared M. luteus genomic ZAP II library. Contrary to what we had expected, the M. luteus UV endonuclease turned out to be a 31/32-kDa protein, not a low molecular weight protein in the range of 11,000 -18,000. Furthermore, the product of the M. luteus pdg gene resembled the E. coli repair proteins, endonuclease III and MutY, not endonuclease V.
The M. luteus UV endonuclease is intriguing in that it constitutes the newest member of an emerging family of [4Fe-4S] 2ϩ cluster DNA repair glycosylases. Moreover, these glycosylases possess widely divergent substrate specificities. The M. luteus UV endonuclease cleaves DNA at pyrimidine dimers, endonuclease III releases thymine glycols and a number of other ring-saturated and ring-fragmented derivatives of thymine (25), and MutY removes undamaged adenines that are mispaired with 8-oxoguanine lesions, 8-oxoadenine lesions, guanines, or cytosines (41,43) The function of the M. thermoformicicum protein has not yet been determined, but it has been speculated to be involved in the repair of G:T mismatches that result from the deamination of 5-methylcytosine by an-other pFV1 plasmid-encoded protein, the GGCC-recognizing methyltransferase (39). Thus, the [4Fe-4S] 2ϩ cluster glycosylases either share a common core structure onto which basespecific recognition motifs have been added or recognize the distortion introduced by DNA damage as opposed to the damage itself. At least in endonuclease III, the [4Fe-4S] 2ϩ cluster appears to contribute to the structural integrity of the protein rather than to play a direct role in catalysis (44,45). One additional piece of evidence supports our hypothesis that the UV endonuclease and endonuclease III are structurally similar: the SP-Sepharose Fast Flow resin served as an invaluable tool in the purifications of both enzymes (46). To eliminate any possibility that we had accidentally cloned the M. luteus homolog of either endonuclease III or MutY, we demonstrated that purified protein was incapable of cleaving a 5,6-dihydrouracilcontaining substrate or mismatch substrates, respectively.
Previous biochemical characterizations of the M. luteus UV endonuclease have examined the enzyme's activity on UVirradiated DNA substrates that undoubtedly contained a mixture of pyrimidine dimer types, cis-syn cyclobutane dimers being the most prevalent followed by pyrimidine-pyrimidone (6 -4) dimers. We examined the dimer specificity of the UV endonuclease on four newly available deoxyoligonucleotide 49mers, each of which contained one the following thymidylyl-(3Ј 3 5Ј)-thymidine photoproducts: a cis-syn, trans-syn-I, (6 -4), or Dewar dimer (33). Although the enzyme nicked only at the cis-syn thymine dimer, literature precedent did exist for this experiment; T4 endonuclease V can cleave a site-specific 49mer duplex containing a trans-syn-I dimer, albeit at 1% of the rate that it cleaves 49-mer duplex containing a cis-syn dimer (33). The major product of the M. luteus UV endonuclease reaction comigrated with the T4 endonuclease V ␤-elimination product; and consistent with the findings of Bailly et al. (21), no ␦-elimination was observed. We did not test whether or not ␦-elimination could be forced under conditions of gross enzyme excess. Occasionally, a secondary product was present between the conventional ␤and ␦-elimination or piperidine product positions. It was never ascertained whether this band was a relevant product or merely an artifact of electrophoresis. The ϫ dilutions thereof) were incubated with cis-syn thymine dimer-containing duplex 49-mer (0.4 ng ϭ 12.5 fmol) for 60 min at 37°C. Reactions were terminated either by placing the reaction tubes in a dry ice-EtOH bath (E) or by heating the mixtures with 1 M piperidine for 30 min at 85°C (q), a treatment that served to convert any remaining AP sites to single-stranded breaks. After the addition of formamide loading buffer, samples were subjected to electrophoresis on a 15% polyacrylamide gel containing 8 M urea. 32 P-Labeled 49-mer substrate and 20mer product were visualized by both autoradiography and PhosphorImaging of the wet gel. The relative amounts of substrate remaining and product generated were determined using Image Quant software. (ϫ) l) were incubated with cis-syn thymine dimer-containing duplex 49-mer (0.4 ng ϭ 12.5 fmol) for increasing lengths of time (0, 5, 10, or 20 min) at 37°C. Reactions were terminated by heating the mixtures with 1 M piperidine for 30 min at 85°C, a treatment that served to convert any remaining AP sites to single-stranded breaks. After the addition of formamide loading buffer, samples were subjected to electrophoresis on a 15% polyacrylamide gel containing 8 M urea. 32 P-Labeled 49-mer substrate and 20-mer product were visualized by both autoradiography and PhosphorImaging of the wet gel. The relative amounts of substrate remaining and product generated were quantitated using Image Quant software. apparent specific activity of the purified UV endonuclease was comparable to that of T4 endonuclease V, but the M. luteus UV endonuclease possessed an AP lyase activity equal to its Nglycosylase activity. Its propensity not to dissociate prior to the ␤-elimination step may explain why the M. luteus enzymesubstrate complex was trapped more readily with NaBH 4 than the T4 endonuclease V imino intermediate. Other investigators have reported that the N-glycosylase activity of the M. luteus UV endonuclease is up to an order of magnitude greater than its apparent AP lyase activity, even in partially purified preparations that may be contaminated by multiple AP endonucleases (2,7,8). These findings are difficult to reconcile with our own unless, as noted by Hamilton and Lloyd (8), the AP lyase activity is more labile over time than is the N-glycosylase activity. Finally, high salt concentrations (100 mM salt on top of 25 mM NaH 2 PO 4 , pH 6.8 buffer) reduced the nicking activity of the UV endonuclease. One of two explanations seems likely: (i) either salt-sensitive processivity facilitated the cleavage of the substrate even though it was only 49 bp long, or (ii) salt inhibited the UV endonuclease via a mechanism that was independent of processivity. The protein was clearly distinct from the M. luteus Ϸ35-kDa class II AP endonucleases A and B in that they require Mg 2ϩ and are sensitive to inactivation by EDTA (47,48).
The catalytic mechanism of the M. luteus UV endonuclease has been shown to involve a Schiff base intermediate. This finding strengthens our theory that all N-glycosylase/AP lyases proceed via an imino intermediate (28,29). Lys-120, a basic residue, was implicated recently in the reaction mechanism of endonuclease III; mutagenesis of Lys-120 to Gln-120 resulted in a 10 4 -fold decrease in the enzyme's activity compared with wild type (25). An analogous Lys-135 residue is present in the M. luteus enzyme. If and when a dependable expression system for the M. luteus UV endonuclease has been developed, we will actively investigate whether the ⑀-amino group of Lys-135 is responsible for the enzyme's catalytic activities. If necessary, the same goal will be pursued using traditional biochemical techniques. Finally, even though the M. luteus UV endonuclease may play only a backup role in vivo to a uvrABC-like nucleotide excision repair system (49 -51), it may play an important role in elucidating the structure/function relationships within the family of [4Fe-4S] 2ϩ cluster DNA repair N-glycosylases.  6, 7, 13, and 14). All reactions were terminated after 60 min at 37°C by the addition of formamide loading buffer, and the samples were subjected to electrophoresis on 15% polyacrylamide gels containing 8 M urea. M lanes contained oligonucleotide sizing markers ranging from 8 to 32 bases. 32 P-Labeled 49-mer substrate, 20-mer product, and enzyme substrate complex(es) were visualized both by autoradiography and PhosphorImaging of the wet gels. Panels A and C show the autoradiographic results of these experiments; panels B and D summarize the PhosphorImager/Image Quant data. The inset in panel C superimposes a 1-week exposure of the gel over a 4-h autoradiograph. Although its placement is essentially accurate, the inset was shifted downward just slightly so as to not obscure the residues marking the wells. The inset arrows point at enzyme-substrate complexes. As is indicated by the slash marks above the wells in panels A and C, the samples spread outward as they ran through the polyurea denaturing gels; the residue that can be seen in each well will not align perfectly with the 49-mer or 20-mer bands.