A DNA exonuclease induced during meiosis of Schizosaccharomyces pombe.

In meiotic cells of the fission yeast Schizosaccharomyces pombe, a DNA exonuclease activity increased approximately 5-fold after premeiotic S-phase and decreased to the initial level before the meiotic divisions. We have purified this activity, designated exonuclease I, to near homogeneity. The activity co-purified with a polypeptide with an apparent molecular weight of 36,000. With a linear double-stranded DNA substrate, exonuclease I degraded only the 5'-ended strand from each end to produce 3'-single-stranded tails. The enzyme also acted on nicked circular DNA with comparable affinity. The meiotic induction of exonuclease I and its mode of action, similar to that of recombination-promoting exonucleases from bacteria, suggest that exonuclease I is involved in meiotic homologous recombination in S. pombe.

A DNA Exonuclease Induced during Meiosis of Schixosaccharomyces pombe* (Received for publication, August 19,1991) Philippe Szankasi and Gerald R. Smith$ From the Fred Hutchinson Cancer Research Center,Seattle,Washington 98104 In meiotic cells of the fission yeast Schizosaccharornycee pornbe, a DNA exonuclease activity increased approximately &fold after premeiotic S-phase and decreased to the initial level before the meiotic divisions. We have purified this activity, designated exonuclease I, to near homogeneity. The activity co-purified with a polypeptide with an apparent molecular weight of 36,000. With a linear double-stranded DNA substrate, exonuclease I degraded only the B'-ended strand from each end to produce 3"single-stranded tails. The enzyme also acted on nicked circular DNA with comparable affinity. The meiotic induction of exonuclease I and its mode of action, similar to that of recombinationpromoting exonucleases from bacteria, suggest that exonuclease I is involved in meiotic homologous recombination in S. pombe.
Nucleases are believed to play important roles during breakjoin homologous recombination. Both endo-and exonucleases have been proposed to act at various steps in the pathways that lead from parental to recombinant chromosomes. In an early step of recombination, one or more participating DNA strands have to be broken. Such DNA breaks could occur as the result of DNA damage or its repair. Alternatively, specific endonucleases might introduce single or double strand breaks that lead to the initiation of genetic exchange. For example, purified RecBCD enzyme of Escherichia coli introduces a nick at a specific recombination-stimulating octanucleotide, Chi (Ponticelli et al., 1985); the single-stranded DNA product is then believed to initiate joint molecule formation (Smith et al., 1984).
DNA breakage is believed to be followed by the formation of hybrid DNA, the annealing of single strands from different parental chromosomes. This central step accounts for the homology dependence of genetic recombination. E. coli RecA protein (reviewed in Radding, 1982;Cox and Lehman, 1987) and homologous pairing activities from other organisms (fungi, Kolodner et al., 1987;Sugino et aL, 1988;Halbrook and McEntee, 1989;Kmiec and Holloman, 1983; fruit flies, Eisen and Camerini-Otero, 1988;humans, reviewed in Kucherlapati and Moore, 1988) were most commonly detected by their ability to promote the formation of hybrid DNA between two homologous DNA molecules, one being single-stranded.
* This work was supported by National Institutes of Health Grant GM32194 (to G. R. S.) and in part by a postdoctoral fellowship from the Swiss National Science Foundation (to P. S.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. $To whom correspondence and reprint requests should be addressed.
Purified RecA protein, in particular, requires at least partial single-strandedness of one DNA molecule for stable hybrid DNA formation.
The regions of single-stranded DNA that are the substrate for annealing may be produced by DNA unwinding enzymes (Taylor and Smith, 1980) or exonucleases that preferentially degrade only one strand of duplex DNA. Exonuclease VI11 of E. coli and X exonuclease both produce long 3"single-stranded tails (Joseph and Kolodner, 1983;Little, 1967) and are required for the RecE and Red pathways of recombination, respectively (Gillen et al., 1981;Stahl, 1986). Bacteriophage T7 gene 6 exonuclease has properties similar to those of ExoVIII and X exo (Kerr and Sadowski, 1972) and has been implicated in the recombination of phage T7 DNA (Kerr and Sadowski, 1975). Mating type switching (White and Haber, 1990) and the processing of RAD50-dependent double strand breaks (Cao et al., 1990;Sun et al., 1991) in Saccharomyces cereuisiae, as well as plasmid recombination in Xenopus luevis oocytes (Maryon and Carroll, 1991), involve intermediates with long single-stranded tails presumably formed by exonucleases acting on double strand breaks.
In eukaryotes, recombination occurs in numerous cell types throughout the organism's life cycle. During meiosis, however, recombination is induced. For example, in Schizosaccharomyces pombe, intragenic and intergenic recombination are induced up to 200-and lOOO-fold, respectively (Grossenbacher-Grunder, 1985). The degree of meiotic induction is variable from locus to locus. Similar values have been observed in S. cerevisiae (e.g. Menees and Roeder, 1989). This increase in recombination may be accounted for, at least in part, by the induction of meiosis-specific recombination factors. The background levels of recombination in mitotically dividing cells may be due to by-products of DNA damage repair (Resnick, 1979). Supporting the existence of meiosisspecific factors is the observation that of six mutants of S. pombe deficient in meiotic recombination (Ponticelli and Smith, 1989) none tested has a detectable effect on mitotic recombination.' In S. cereuisiae, mutations in the genes HOP1, RED1, MER1, MER2, SPOll, REC102, REC107, and, in certain cases, RAD50 affect recombination only in meiosis (e.g. Engebrecht et al., 1990, and references therein;Malone et al., 1991).
It is important for the biochemical analysis of meiotic recombination to have large quantities of cells undergoing synchronous meiosis. The conditional mutant pat1 -1 14 of S. pombe undergoes meiosis at elevated temperature (Iino and Yamamoto, 1985a) and bypasses the requirements of nitrogen starvation and mating type heterozygosity for the initiation of meiosis (Beach et al., 1985). The patl+ gene encodes a protein kinase homologue (McLeod and Beach, 1986) and negatively regulates both conjugation of haploid cells and the ' K. Larson and N. Hollingsworth, personal communication. 3014 initiation of meiosis (Nielsen and Egel, 1990;Beach et al., 1985). In patl-114-induced meiosis, d e 6 intragenic recombination frequencies are comparable to those observed in standard (patl+) meiotic crosses, and the M26 meiotic hotspot has its full stimulatory effect (Gutz, 1971).* Furthermore, in haploid patl-114 cells at high temperature, recombination between an d e 6 gene on a plasmid and its counterpart on the chromosome is detected in the 1% viable spores at a frequency comparable to that measured for plasmid-by-chromosome recombination in diploid meiosis (Ponticelli and Smith, 1989).3 Others have also observed high frequencies of recombination in patl-114-induced meiosis (Iino and Yamamoto, 1985b;Bahler et aL, 1991). We therefore conclude that the components of the meiotic recombination machinery are expressed during patl-114-induced meiosis, even in haploid cells.
In this paper, we describe the purification of an exonuclease, called ExoI, from patl-114 haploid cells induced to undergo meiosis. Its expression and mechanism of action suggest a role for ExoI during meiotic recombination in S. pombe.

EXPERIMENTAL PROCEDURES
Enzymes and Chemicals-Restriction endonucleases and their buffers, E. coli DNA polymerase I large fragment, T4 polynucleotide kinase, S1 nuclease, and calf intestinal phosphatase were obtained from New England Biolabs, Boehringer Mannheim, or Bethesda Research Laboratories. Unless otherwise indicated, chemicals were obtained from Sigma, J. T. Baker Chemical Co., and U. S. Biochemicals. Yeast RNA was purchased from BDH Biochemicals (Poole, Great Britain).
Cell Growth-S. pombe strain GP535 was grown in YEL (0.5% yeast extract, 3% glucose), supplemented with adenine (80 pg/ml), at 24 'C to full saturation (A686 = 7-11). This preculture was diluted to a final absorbance of 0.1 into 3-liter batches of EMM2* medium (modified EMM2 according to Nurse, 1975, with only 0.5% glucose) in 6-liter flasks. Growth was continued at room temperature until the absorbance reached 0.3 (approximately 5 X lo6 cells/ml; about 15 h), and the flasks were swirled in hot tap water for about 5 min until the temperature increased to 33 "C. The cultures were incubated at 33 "C with slow shaking for the desired time. Samples of 0.8 liter were harvested at each time point for measuring meiotic induction of exonucleases. For large scale enzyme purification, cells were harvested 4.75 h after the temperature shift. The cultures were chilled in their flasks on ice before centrifugation. Cells were washed once in 1/100 volume of 50 mM MOPS/NaOH pH 7.5, 1 m M EDTA, 0.5 M NaCl and resuspended in the same buffer at 1 ml/g of cells (wet weight). The suspension was frozen in liquid N, and stored at -70 'C. 1.2 g (wet weight) were obtained per liter. A portion of the cells for enzyme purification was grown in a 30-liter fermenter (New Brunswick Instruments). Under these conditions, synchrony and timing of meiosis were not fully reproducible, but comparable induction of exonuclease activity was observed after about 6 h at 33 "C. 3H Labeling of S. pombe DNA and DAPI' Staining-From a standard 3-liter culture of GP535, 25 ml were removed immediately after heat induction (see above) to a 100-ml flask, and 75 pCi of [6-3H] uracil (25 Ci/mmol, Du Pont-New England Nuclear) was added. Incubation was continued at 33 "C. At various times, 2-ml samples were removed, mixed with 25 p1 of calf thymus DNA (0.2 mg/ml) and 10 ml of 10% trichloroacetic acid plus uracil (1 mg/ml), and kept on ice for at least 15 min. The precipitate was recovered by centrifugation at 12,000 X g, 0 "C for 10 min, dissolved in 1 ml of 0.6 M NaOH plus uracil (1 mg/ml), and incubated at 65 "C for 2 h to hydrolyze RNA. After cooling on ice, 1 ml of 0.6 M HCl plus 20% trichloroacetic acid was added and the samples were kept on ice for 30 min. The precipitates were collected on glass fiber filters (GF/C, Whatman). Filters were dried and mixed with scintillation fluid (Ready Safe, Beckman). Radioactivity was determined in a Beckman LS3801 scintillation counter.
For staining of nuclei, 10-ml samples of the induced culture were removed at various times and chilled in ice water. The cells were collected by centrifugation, resuspended in 1 ml of 95% ethanol, and kept on ice for 5 min. 1 pl of 4,6-diamidino-2-phenylindole (DAPI, 1 mg/ml) were added, and the cells were kept on ice for 10 min. Cells were pelleted, washed once with 10 ml of H20, resuspended in 5 ml of H20, and stored at 4 "C until they were analyzed by fluorescence microscopy.
Growth was continued until the cells lysed (approximately 1 h and 15 rnin). After addition of chloroform to 0.01%, cell debris was removed by centrifugation at 4500 X g, 4 "C for 10 min in a Beckman JAlO rotor. From the supernatant, the T7 phage particles were harvested by centrifugation at 30,000 X g, 4 "C for 2 h in a Beckman JA14 rotor. The phage pellets were resuspended in a total of 4 ml of 10 mM Tris-HC1, pH 8, and 1.5-ml aliquots were overlayed on gradients consisting of three 1.2-ml steps of a 1:2, 1:1, and 2:l dilution, respectively, of saturated CsCl in 10 mM Tris-HC1, pH 8. The gradients were spun at 35,000 rpm, 18 "C for 30 min in a Beckman SW55 rotor. The phage band in the middle layer was removed, mixed with a 1:l dilution of saturated CsCl solution, and rebanded by equilibrium density gradient centrifugation at 45,000 rpm for 20 h. The phage was then diluted 10-fold in 10 mM Tris-HCI, pH 8 , l mM EDTA (TE) and extracted with phenol (equilibrated with 0.1 M Tris-HC1, pH 8) by gentle inversion for 20 min. After the first extraction, the interphase was collected together with the aqueous phase. The extraction was repeated until no interphase was visible. The DNA was dialyzed against several changes of TE. This protocol yielded up to 1 mg of T7 DNA at 5-8 X lo' cpm/nmol of nucleotides.
3 H -~b e l e d M I 3 DNA-A plaque from phage M13mp18 on E. coli strain V1453 was transferred with a round toothpick into a test tube containing 1 ml of 2 X TY (1% yeast extract, 0.5% NaCl, 1.6% tryptone) supplemented with thymine (50 pg/ml). This phage culture was vigorously shaken at 37 "C for 6 h. In the meantime, E. coli strain V1453 was grown in 1 liter of 2 X TY supplemented with thymine (2 pg/ml) with vigorous shaking at 37 "C. When the 1-liter culture reached an A m of 0.5, the 1-ml phage culture, 200 mg of deoxyadenosine, and 5 mCi of [n~thyl-~HIthymidine (80 Ci/mmol, Du Pont-New England Nuclear) were added, and growth was continued for 4 h.
The cells were harvested by centrifugation, and replicative form M13mp18 DNA was prepared using the alkaline lysis procedure followed by two subsequent steps of CsCl equilibrium density gradient centrifugation (Sambrook et al., 1989). M13mp18 virion DNA was prepared from phage particles as follows. To the culture supernatant, 'A volume of 20% polyethylene glycol 3500, 2.5 M NaCl was added. The solution was stirred for 5 min and kept overnight at 5 "C. The precipitate was harvested by centrifugation at 30,000 X g, 0 "C for 15 min. The pellets were dissolved in TE, and the phage particles were purified by two sequential CsCl density gradient centrifugations. After dialysis against several changes of TE, the phage suspension was extracted with phenol as described above for the purification of ' The abbreviations used are: DAPI, 4,6-diamidino-2-phenylindole; DTT, dithiothreitol; BSA, bovine serum albumin; PMSF, phenylmethanesulfonyl fluoride; TLCK, N"-p-tosyl-L-lysine chloromethyl ketone; MOPS; 4-morpholinepropanesulfonic acid; FPLC, fast protein liquid chromatography. phage T7 DNA, extracted twice with chloroform, and precipitated with ethanol. This protocol yielded approximately 1 mg of viral DNA and 150 pg of replicative form DNA at 5.5 X lo4 cpm/nmol of nucleotides.
Nicked Plasmid DNA-'H-Labeled M13mp18 replicative form DNA (100 pglml), ethidium bromide (32 pg/ml), and the restriction endonuclease HindIII (100 units/ml) in 10 mM Tris-HC1, pH 7.5, 10 mM MgC12, 1 mM DTT were incubated at room temperature for 1 h. The mixture was extracted twice with phenol, once with chloroform, and precipitated with ethanol. These conditions resulted in the formation of approximately 50% nicked circular DNA and less than 5% linear DNA as judged from an ethidium bromide-stained agarose gel. A portion of the nicked DNA was electrophoresed on and eluted from an agarose gel and linearized with the restriction enzyme BglII. On an alkaline agarose gel (Sambrook et al., 1989), the band corresponding to the linear strand was quantitatively converted into a discrete, faster migrating band (not shown), indicating that most of the molecules contained a single nick at the HindIII site.
Internally 32P-Labekd DNA"Ml3mpl8 virion DNA (0.4 pmol of molecules, New England Biolabs) and 0.8 pmol of M13 universal primer (U. S. Biochemicals) in 12 pl of 10 mM Tris-HC1, pH 7.5, 50 mM NaC1,lO mM MgC12, 1 mM DTT were heated to 65 "C and slowly cooled to room temperature. The complementary strand was synthesized with 2.5 units of E. coli DNA polymerase I large fragment for 30 min at room temperature, in the presence of 10 pCi of [CY-'~P] dCTP or [a-"P]TTP (Du Pont-New England Nuclear, diluted to 80 Ci/mmol) and 500 pmol of each of the three unlabeled dNTPs. After enzyme inactivation at 65 "C for 15 min, the DNA was digested with SmuI restriction endonuclease, extracted with phenol and chloroform, and precipitated with ethanol.
Double-stranded DNA was analyzed on 1% agarose gels in TAE buffer (Sambrook et al., 1989) in the presence of ethidium bromide (0.5 pg/ml). Linear and circular single-stranded virion DNA was separated on 1% agarose gels in 0.5 X TBE buffer (Sambrook et al., 1989) and stained afterwards with ethidium bromide (1 pg/ml).
Thin Layer Chromutography-Reaction products were separated on a polyethyleneimine-cellulose plate with fluorescent indicator (J. T. Baker Chemical Co.) according to Randerath and Randerath (1967). Prior to use, the plate was submerged in 1 M acetic acid for 2 min and then dried. Without this step, dCMP migrated as a smear at the front of the chromatogram. 20 nmol of nonradioactive nucleotides and %O of the exonuclease digestion mixtures were spotted 2 cm from the edge of the plate, which was developed at room temperature in 1 M acetic acid for a distance of 2 cm. The solvent was changed to 1 M acetic acid3 M LiCl (9:1), and chromatography was continued, without prior drying of the plate, up to 15 cm. After the plate was dried, the non-radioactive standards were located under UV light. The "Plabeled reaction products were visualized by autoradiography. The d(pGpG) marker was obtained by adding a 5"phosphate with polynucleotide kinase to d(GpG) (Sigma). All standards were diluted in 1 X exonuclease buffer (see below). Exonuclease Assays-For exonuclease assays, the enzyme was diluted in 20 mM MOPS/NaOH, pH 7.5, 0.1 mM EDTA, 10% glycerol, 100 mM NaCl, 1 mM DTT, acetylated BSA (0.1 mg/ml). Unless stated differently, amounts of substrate DNA are given in nanomoles of cleaved ['HIT7 DNA (1.7 pmol of DNA ends), 25 mM MOPS/NaOH, nucleotides. Standard reaction mixtures contained 1 nmol of HaeIII-pH 7.0, 5 mM Mg(OAc)2, 0.5 mM DTT, and acetylated BSA (0.1 mg/ ml) in 50 pl and were incubated at 31 "C for 10 min. The reactions were stopped by adding 25 pl of calf thymus DNA (0.2 mg/ml) and 250 pl of 5% trichloroacetic acid and kept on ice for 10 min. After centrifugation for 5 min in a microcentrifuge, 260 pl of the supernatant fluid was carefully removed and mixed with 3 ml of scintillation fluid. Radioactivity was measured in a scintillation counter. To measure single strand exonuclease activity, the DNA substrate was first boiled for 5 min and chilled in ice water. Alternatively, single-stranded M13mp18 virion DNA was cleaved with the restriction endonuclease HaeIII (Blakesley and Wells, 1975) and boiled for 5 min prior to use in order to minimize secondary structure. 1 enzyme unit was defined as the activity rendering 1 nmol of nucleotides trichloroacetic acidsoluble in 1 min. Exonuclease reactions were carried out in duplicate, and the counts measured were within 10% of each other.
Exonuclease reaction products were digested with S1 nuclease as follows. A 50-p1 Ex01 reaction was stopped by addition of 200 pl of 1.25 X S1 buffer (1 X S1 buffer is 30 mM KOAc, pH 4.6, 250 mM NaCl, 3 mM ZnS04, 5% glycerol) alone or S1 buffer containing S1 nuclease (40 units/ml), and incubation was continued at 31 "C for 30 min. 25 pl of calf thymus DNA (0.2 mg/ml) and 50 pl of 25% trichloroacetic acid were added, and the samples were processed as above.
All steps were carried out at 0-5 "C. 250 g of frozen S. pombe GP535 cells, harvested 4.75 h after induction of meiosis, were thawed in ice water, and DTT, PMSF, TLCK, pepstatin, leupeptin, and aprotinin were added to the same final concentration as in buffer A (see above). A 400-ml stainless steel chamber for the Bead-Beater (Biospec Products) was filled with 200 ml (packed volume) of acidwashed glass beads (0.5-mm diameter), one-third of the thawed cells (for each of three processings), and buffer A to fill the chamber. The chamber was assembled excluding air bubbles, and NaC1-ice water (-5 "C) was used as a coolant. Cells were broken during 16 pulses of 15 s each, in 45-s intervals. The solution was recovered, and the glass beads were rinsed with 40 ml of buffer A. Large cell debris was removed by centrifugation at 10,000 X g for 10 min, and the supernatant was spun at 100,000 X g for 30 min in a Beckman 60Ti rotor. The supernatants, but not the fuzzy layer at the bottom of the tubes, were removed and pooled to give 650 ml of fraction I. Nucleic acids were precipitated by adding 10% polyethyleneimine-HC1, pH 7.5, dropwise, while stirring, to a final concentration of 0.05%. The solution was kept on ice for 20 min, and the precipitate was removed by centrifugation at 10,000 X g for 10 min. To the supernatant (628 ml), 221 g of solid ammonium sulfate was added, while stirring, over a period of 15 min, and stirring was continued for 15 min. After centrifugation at 17,000 X g for 20 min, the supernatant was discarded, and the pellets were dissolved in a total of 120 ml of buffer B7. The solution was dialyzed against 2 liters of buffer B7 for 2 h and then against 2 liters of buffer B7/120 for 9 h. The solution (fraction 11) had a volume of 150 ml.
The conductivity of fraction I1 was adjusted to that of buffer B7/ 120 by adding buffer B7. The solution was loaded at 100 ml/h onto a column of single-stranded DNA cellulose (U. S. Biochemicals, 6 cm2 X 16.7 cm). After washing the column with 300 ml of buffer B7/120, proteins were eluted at the same flow rate with a 420-ml linear gradient of NaCl from 120 mM to 500 mM in buffer B7.14-ml fractions were collected, and 2 p1 of 20-fold dilutions were assayed for exonuclease activity. Fractions containing exonuclease, which peaked at approximately 320 mM NaC1, were pooled to yield 41 ml of fraction 111.
Fraction I11 was dialyzed against 500 ml of buffer B7/140 for 9 h. The conductivity of the solution was adjusted to that of buffer B7/ 140 by adding buffer B7 and loaded at 20 ml/h onto an S-Sepharose (Pharmacia LKB Biotechnology Inc.) column (0.8 cm2 X 7.5 cm). The column was washed with 15 ml of buffer B7/140 and eluted at the same flow rate with a 50-ml linear gradient of NaCl from 140 mM to 500 mM in buffer B7. 1-ml fractions were collected and 2 pl of 50fold dilutions were assayed for exonuclease activity. Fractions with activity, peaking at 255 mM NaCl, were pooled to give 4.9 ml of fraction IV.
Fraction IV was dialyzed for 9 h against buffer B85/50. The solution was adjusted to a conductivity equivalent of B85/50 by adding buffer B85 and applied to a Mono Q column (HR5/5, FPLC system, Pharmacia) at 0.5 ml/min. The column was washed with 10 ml of buffer B85/50 and eluted with a 20-ml linear gradient of NaCl from 50 mM to 500 mM in buffer B85 into 0.5-ml fractions. Enzyme activity, peaking at 125 mM NaCl, was pooled to give 0.96 ml of fraction V.
Fraction V was concentrated to 0.1 ml in a Centricon 10 microcon-centrator (Amicon) and applied to a Superose 12 column (HR10/30, FPLC system, Pharmacia) equilibrated in buffer B75/200. The column was run at 0.25 ml/min, and 0.5-ml fractions were collected. The fraction with the peak of enzyme activity (fraction VI, 0.5 ml) was dialyzed overnight against buffer C, and aliquots were frozen in liquid nitrogen and stored at -70 "C. Small scale preparations of crude extracts from 0.8 liter of culture harvested during a meiotic time course (see above) were done as follows. Cells were thawed and transferred into a 2-ml screw cap tube (Sarstedt) containing 1 ml (packed volume) of glass beads. The tube was filled with buffer A and closed avoiding air bubbles. With the Mini Bead-Beater (Biospec Products), cells were broken during 10 pulses of 30 s at 5 "C with 30-s intervals on ice. The liquid was removed, and the glass beads were rinsed with 1 ml of buffer A. The lysate was spun at 100,000 X g in a Beckman 70.1Ti rotor for 30 min. The supernatant was diluted and assayed for exonuclease activity as described above.
Protein Amlysis-Protein concentrations were determined according to Bradford (1976) using the protein assay dye purchased from Bio-Rad and bovine serum albumin as a standard. Protein samples were separated by NaDodS04-polyacrylamide gel electrophoresis (Laemmli, 1970) on a 0.75-mm gel in a Mighty Small apparatus (Hoefer) and stained with Coomassie Brilliant Blue R-250. Samples in large volumes were concentrated by precipitation in 10% trichloroacetic acid.

RESULTS
An Exonuclease Activity Is Induced during Meiosis of a patl-114 Strain-We used a haploid S. pombe strain (GP535) carrying the temperature-sensitive patl-114 mutation to obtain synchronous meiotic cultures, as described under "EXperimental Procedures." A strong DNA endonuclease activity is present in extracts from both meiotic and mitotic cells of S. pombe. A mutation (endl-458) abolishing this activity has been described (Uemura and Yanagida, 1984); end1 -458 does not affect meiotic recombination.* We used this mutation in our strains in order to more readily detect exonuclease activity, measured as the production of acid-soluble nucleotides from linear duplex DNA (see "Experimental Procedures").
patl-114 cells efficiently enter meiotic differentiation when first grown in minimal medium (EMM2*) to a maximal density of approximately 5 X lo6 cells/ml (Beach et al., 1985). To synchronize large cultures of GP535 prior to the shift to the restrictive temperature and the following onset of meiosis, the cells were first grown to saturation in yeast extract medium (YEL), transferred into minimal medium, and allowed to grow for approximately one generation (see below). This amount of growth in minimal medium was sufficient for the cells to exit from starvation arrest and still allowed reasonable synchrony as judged by monitoring DNA synthesis and meiotic divisions (Fig. 1). To analyze DNA synthesis, a culture of GP535 was allowed to undergo meiosis in the presence of [3H] uracil, as described under "Experimental Procedures." Total 3H incorporated into DNA during premeiotic S-phase was measured at various times in acid-soluble, alkali-resistant material (Fig. 1, squares). To ensure that only DNA and not RNA synthesis was measured, a parallel experiment was carried out with hydroxyurea, an inhibitor of DNA synthesis (Mitchison, 1974) added to 50 mM 30 min after heat induction. In this case, no significant incorporation of 3H was measured during 6 h (not shown). To determine the percentage of cells with four nuclei, resulting from the two sequential meiotic divisions, cells were removed from a heat-induced culture of GP535 at various times, stained with the fluorescent DNA dye DAPI and observed under a fluorescent microscope (Fig.  1, triangles). The bulk of DNA synthesis occurred between 1.5 and 3 h after shift to the restrictive temperature. Because we observed only a single burst of 3H incorporation into DNA soon after the temperature was shifted to 33 "C, we infer that at time zero (Fig. 1) most of the cells were in G1 phase. As observed for starvation-induced meiosis ofpatl+ strains, temperature-induced meiosis ofpatl-114 strains also starts in G1 (Beach et al., 1985). The percentage of cells with four nuclei rapidly increased between 5 and 6 h after shift to the restrictive temperature. Up to 90% of the cells had four nuclei after 7 h (not shown). From samples taken during this meiotic time course, we measured exonuclease activity on uniformly 3H-labeled T7 DNA. We found that maximal activity required a high concentration of DNA ends (see below) and therefore cleaved the t3H]T7 DNA with the restriction endonuclease HaeIII into 69 fragments with an average size of 600 base pairs. Exonuclease activity was induced about &fold, peaking between 4.5 and 5 h (Fig. 1, closed circles). Hence, the activity was maximal between premeiotic S-phase and the meiotic divisions, a time span in which meiotic recombination is believed to occur (Borts et al., 1984;Resnick et al., 1984). To rule out the possibility that the induction of exonuclease activity was a response to heat shock, the same growth protocol was applied to apatl+ strain (458). In this case, no induction of exonuclease activity was observed (Fig. 1, open circles). We conclude that the induction of exonuclease activity is patl-lll-dependent and, therefore, is a true meiotic event. Using a diploid strain homozygous for the patl-114 mutation, we measured the same timing and degree of induction of exonuclease activity (not shown).
Purification of a Meiotic Exonuclease-Large quantities of S. pombe strain GP535 were grown and induced to undergo meiosis for 4.75 h as described under "Experimental Procedures." The purification scheme, summarized in Table I, yielded a polypeptide of about 36 kDa, as estimated by Na-DodS04-polyacrylamide gel electrophoresis (Fig. 2a). The protein was purified approximately 10,000-fold to near homogeneity. The faint additional band around 60 kDa, enriched in fraction VI (Superose 12, Fig. Za), might originate from contamination with keratin during handling of the sample. From the Mono Q column, aliquots of fractions including and surrounding the peak of enzyme activity were analyzed by gel electrophoresis; the amount of enzyme activity coincided with the intensity of the 36-kDa polypeptide (Fig. 2b). The same result was obtained with fractions from the Superose 12 column (not shown). By calibrating the Superose 12 column, the molecular weight of the enzyme was estimated at about 20,000 (not shown). This suggests that the native enzyme is

From 250 g of cells (approximately 10l2 cells) as described under "Experimental Procedures."
Step  (Table I). TCA, trichloroacetic acid. a monomer. The discrepancy in estimates of the molecular weight may be due to aberrant mobility of the protein in NaDodSOl-polyacrylamide gels. It is unclear what caused the immense losses of enzyme on the last column (Superose 12).
We have named this protein exonuclease I (ExoI) of S. pombe. Assuming that the band at 36 kDa in Fig. 2a, Table 11. A divalent cation was essential for activity; M$+ was more effective than Mn2+. As counterion OAc-was slightly favored over C1-. Maximal activity was obtained between 2 and 5 mM Mg(OAc)2, with higher concentrations being slightly inhibitory (50% activity a t 20 mM; not shown). Both Ca'+ and Zn2+ inhibited the activity to 40% at 5 mM and to 2% at 0.1 mM, respectively. The addition of ATP to 1 mM had no detectable influence. In experiments with crude extracts, we did not observe any ATP-dependent nuclease in meiotic or mitotic extracts of S. pombe (not shown). Enzyme activity was slightly salt-sensitive, being reduced to 63% at 100 mM NaCl. The optimal pH was between 7 and 7.5, with approximately 20% as much activity at pH 6 or 8 (not shown). The temperature optimum of ExoI was 36 "C, with 80% as much activity at our standard reaction temperature of 31 "C (not shown). Yeast RNA (1 pg, which provided at least a 10-fold excess of RNA ends over DNA ends as calculated from the average length of the RNA molecules; not shown) did not compete significantly with the substrate DNA. Therefore, ExoI is DNA-specific.
T o compare the activity of ExoI on double-stranded and single-stranded DNA, we carried out reactions with either native or heat-denatured DNA. Fig. 3 shows that ExoI degraded double-stranded DNA 10 times more rapidly than single-stranded DNA. We also determined whether digestion by ExoI is processive or distributive. The experiment in Fig. 4 used reactions to which a 5-fold excess of ends from unlabeled DNA was added to the reactions at different times. The addition of unlabeled DNA before ExoI reduced the production of acid-soluble 3H label (squares) approximately 4-fold compared to the control reaction (closed circles). When added 3 min after ExoI, the unlabeled DNA also reduced the production of acid-soluble label to the same extent (open circles). This implies that ExoI has a low degree of processivity and dissociates from the DNA after hydrolyzing a limited number of phosphodiester bonds.  In this experiment, linearized [3H]M13 DNA was used so that the molecules were not completely hydrolyzed during the time of the reaction (see below for rates).
ExoI has no detectable activity on supercoiled plasmid DNA or circular single-stranded M13 virion DNA as determined by agarose gel electrophoresis (not shown). The enzyme acts only exonucleolytically.
Exonuclease I Degrades Only One Strand from a Doublestranded DNA End-To further characterize the mode of degradation by ExoI, linear blunt-ended DNA labeled at either the 3' or the 5' end with 32P was mixed with the uniformly 3H-labeled substrate DNA. ExoI produced acidsoluble 32P label from the 5' end (Fig. 5b, open squares), but not from the 3' end (Fig. 5a, open circles). Comparable overall activity of ExoI in both experiments was confirmed by measuring the release of the uniform 3H label (Fig. 5, a and b, filled  symbols). This result suggests that only the 5"ended strand is detectably degraded from a given double strand end.
In the following experiment, we determined that ExoI indeed produces single-stranded DNA by sparing the 3' ends. We tested this notion with standard exonuclease reactions followed by digestion with the single strand specific nuclease S1 (Wiegand et al., 1975). Fig. 6a shows that, depending on the extent of digestion with ExoI (hatched bars), a nearly equal amount of DNA was rendered acid-soluble by a subsequent incubation with nuclease S1 (bluck bars). Reactions FIG. 6. Production of single-stranded DNA during digestion by ExoI. a, 1 nmol of HueIII-cleaved L3H]T7 DNA was reacted with ExoI (3.6 X units or 0.25 pl of fraction VI) for the times indicated in triplicate, as standard reactions (hatched burs) or standard reactions followed by incubation with nuclease S1 buffer alone (gray burs) or with S1 nuclease (black bars). b, 1 nmol (0) or 0.5 nmol (0) of HaeIII-cleaved i3H]T7 DNA was reacted with ExoI (3.6 X lo-' units or 2.5 pl of fraction VI) for the times indicated at a ratio of DNA ends to enzyme molecules of 30 and 15, respectively. The change in the production of trichloroacetic acid (TCA)-soluble counts upon addition at 15 min of another aliquot of substrate or at 60 min of another aliquot of enzyme is shown by a dashed line and a dotted line, respectively. For further explanation, see text.

S. pombe Meiotic Exonuclease
continued with low pH, Zn2+-containing S1 buffer but without S1 (gray burs) inhibited further ExoI activity and did not yield additional acid-soluble material.
Since ExoI had a strong preference for double-stranded DNA (Fig. 3), these results suggest that 50% of the input substrate, i.e. the single-stranded product, should be resistant t o extensive digestion. Fig. 6b shows an experiment in which 1 nmol (closed circles) or 0.5 nmol (open circles) of HueIIIcleaved (3H]T7 DNA was digested to the limit by ExoI. 60% of the DNA mass was rendered acid-soluble independent of its amount present in the reaction mixture. That the remaining 40% of the DNA mass were resistant to digestion (e.g. by having been rendered single-stranded) rather than the enzyme loosing its activity was confirmed by two control experiments. First, we added an additional, equal amount of substrate to one sample after 15 min; 50% of this DNA was digested (dashed line). Second, an additional, equal amount of enzyme was added to one sample after 60 min of reaction, and no further digestion was observed (dotted line). In Table I1 (bottom), we show that neither reaction product, singlestranded DNA or deoxyribonucleoside monophosphates, inhibited enzyme activity when they were added at about twice the concentrations produced in the above experiment. This rules out product inhibition in the experiment shown in Fig.  6b. The fact that 60% rather than 50% of the substrate was digested could have several explanations; for example, as Fig.  3 shows, ExoI has a weak activity on single-stranded DNA, and nicks accumulated during substrate preparation might produce short acid-soluble oligonucleotides when rendered single-stranded.
Estimating the concentration of ExoI in fraction VI (see above) and knowing that only one strand is degraded a t a time, we calculated that in the experiment shown in Fig. 3 one ExoI molecule cleaves 750 phosphodiester bonds per min on the average (see also product analysis below).
Exonuclease I Produces 5'-Mononucleotides-We wished to determine the nature of the acid-soluble products of digestion by ExoI. We addressed two questions. (i) Is the acid-soluble material composed of nucleoside monophosphates or short oligonucleotides? (ii) Does cleavage of the DNA backbone by ExoI produce 5'-or 3'-phosphates? The two questions were addressed by the experiment shown in Fig. 7 with the following rationale. A double-stranded DNA substrate containing internal 32P label was constructed by synthesizing the complementary strand of M13 single-stranded DNA with DNA polymerase and the four deoxynucleoside triphosphates, one being "P-labeled at the CY position. If cleavage by ExoI produces 5'-phosphates, then the 32P atoms would be released with the same nucleotide used for the synthesis. If, on the other hand, ExoI produces 3' phosphates, then the 32P atoms would be released with the nucleotides positioned on the 5' sides of the nucleotide used for the synthesis. Fig. 7 shows that the anP label was always released as TMP when TTP was the labeled nucleotide and as dCMP when dCTP was the labeled nucleotide. At least 25% of the released label should have been released as 3'-dGMP (estimated from the published sequence of M13mp18, Yanisch-Perron et al., 1985) if 3"phosphates were produced by ExoI. If dinucleotides were released, those products, of heterogeneous composition, would be expected to migrate slower, as judged from the d(pGpG) marker. The reaction products were quantitated in two ways. First, an aliquot of the reaction was precipitated with trichloroacetic acid (as in standard exonuclease reactions) and radioactivity of the soluble and insoluble fractions was determined by Cerenkov counting. Second, from the thin layer chromatography plate, the spots with the products and the origin were excised, and radioactivity was determined. The two measurements were in agreement and yielded 5% acid-soluble counts and 5% mononucleotides, respectively. This argues against ExoI producing acid-soluble short oligonucleotides that remain at the origin of the chromatogram. We conclude that ExoI releases 5'-nucleoside monophosphates.
Nature of the 5'-Ends Required for Digestion by Exonuclease I-In order to elucidate the role of exonuclease I in uiuo, we measured its activity on different kinds of DNA breaks. First, we determined the K , value for double strand ends, our standard substrate. Reactions were carried out with varying amounts of HaeIII-cleaved or SmaI-cleaved [3H]M13mp18 DNA such that the concentration of 5' ends was between 0.7 and 35 nM. Fig. 8 shows the results displayed in a Lineweaver-Burk plot. The deduced K , values are 8.3 nM 5' ends for the SmaI-cleaved DNA and 10 nM 5' ends for the HaeIII-cleaved DNA (see legend to Fig. 8). Due to the different average substrate length in the two determinations, the ratio of 5' ends to DNA mass differed by a factor of 15. Nevertheless, the K, values are in close agreement, indicating that ExoI interacts preferentially with DNA ends. The deduced VmnX values are 9 pmol and 12 pmol of nucleotides rendered acidsoluble per min per mg of ExoI protein for the HaeIII-and SrnaI-cleaved DNA, respectively.
We then tested the ability of ExoI to act on nicked plasmid DNA. The DNA preparation used contained approximately 50% nicked circular DNA (see "Experimental Procedures"). The residual supercoiled DNA is unlikely to interfere with ExoI since 2 pg of supercoiled DNA do not interfere with  digestion of 0.3 pg of HaeIII-cleaved T7 DNA, and supercoiled DNA is not a substrate (not shown). The small amount of contaminating linear DNA molecules in the preparation of nicked DNA (see "Experimental Procedures") did not contribute significantly to the ExoI activity measured. Fig. 8 shows the results of digestion of nicked circular DNA at three concentrations (squares). The difficulty of preparing this substrate precluded a more extensive analysis. Nevertheless, we estimate from Fig. 8 that the K , value for nicked circular DNA was at least as low as that for double strand ends. We conclude that ExoI acts on both nicks and double strand breaks with similar affinity.
We further tested any specificity of ExoI for the nature of the 5' ends on the substrate DNA. Reactions were carried out with untreated, dephosphorylated, and rephosphorylated linear DNA. Table I11 shows that ExoI acted poorly on DNA with 5'-hydroxyl groups. The residual activity might originate from incomplete dephosphorylation. Activity was restored upon readdition of a 5'-phosphate.

DISCUSSION
We have described the purification of a double-stranded DNA exonuclease that is induced during meiosis of S. pombe.
The enzyme, exonuclease I, degraded only one strand from a given double strand end in the 5' 3 3' direction, producing 3'-single-stranded tails. The enzyme also acted on nicks with comparable affinity.
We looked for such an activity because we are interested in studying the meiotic recombination machinery of S. pombe. To obtain synchronous meiotic cultures, we have utilized the conditional mutation patl-114 from S. pombe. By all criteria analyzed, these cells undergo a true meiosis at the restrictive temperature (see introduction; Iino and Yamamoto, 1985b;Beach et al., 1985).*s3 S. pombe pat1 -1 14 cells were synchronized by starvation in stationary phase prior to heat induction of meiosis. These cultures then underwent synchronous meiosis as we deduce from the timing of premeiotic S-phase and the meiotic divisions. We were interested in determining whether activities proposed to act during recombination were induced in these meiotic cultures. Extracts from cells of such a time course show the sharp induction of a double-stranded DNA exonuclease activity immediately followed by a rapid decrease. We did not investigate whether exonuclease activity before heat induction of meiosis can be attributed to a basal level of the same activity or whether other, mitotic exonucleases are present. Therefore, the induction of ExoI might be greater than 5-fold. We also found a similar induction of an activity that promotes the formation of joint molecules between linear double-stranded DNA and homologous single-stranded DNA.6 Exonuclease activity peaks shortly before the meiotic divisions. The exact time when meiotic recombination takes place in S. pombe has not been determined yet. It would be of great interest, for example, to monitor the appearance of a recombinant restriction fragment as has been done in S. cerevisiae (Borts et al., 1984) and to correlate the result with the time of exonuclease induction.
It is likely that the enzyme designated ExoI is indeed the same activity showing meiotic induction in crude extracts. All column chromatography steps used during purification showed a single peak of double strand exonuclease activity (not shown). In addition, pilot purification experiments showed a 5' + 3' polarity of digestion at earlier steps of purification (not shown). The construction of a mutant lacking ExoI will give the final answer.
Our results show that ExoI acts in the 5' + 3' direction on double strand ends, and we presume that its action on nicked circular DNA proceeds with the same polarity. Such an activity might act either in the formation of 3'4ngle-stranded tails from double strand breaks or in the excision of a strand starting at a nick to form a single strand gap. Both functions could be part of DNA metabolic processes such as DNA damage or mismatch repair and recombination.
We could not detect any endonucleolytic activity of ExoI and therefore postulate the existence of an endonuclease that acts prior to ExoI. Unfractionated extracts from strains used in this study have no detectable endonucleolytic activity on supercoiled plasmid DNA (not shown). Such an activity might be tightly regulated by low abundance, dependence on a certain chromatin configuration, or, perhaps, sequence specificity.
The weak activity of ExoI on single-stranded DNA implies an additional function such as the digestion of single strand loops formed by strand invasion. Such a function would also require the prior action of a single strand endonuclease. Alternatively, ExoI could process 5"single-stranded tails into flush ends thereby creating its own favorable substrate. The origin of such 5"single-stranded tails is, however, unclear.
ExoI requires a 5"phosphate to initiate degradation. This implies that the enzyme acts only on a subset of possible DNA breaks, for example, those introduced by a specific endonuclease. Other breaks, with 5'-hydroxyl groups, possibly caused by some DNA-damaging agent, would not be processed by ExoI and would, therefore, lead into a different pathway of double strand break repair.
The properties of ExoI suggest a role for it during meiotic homologous recombination in S. pombe. Because ExoI acts on both nicks and double strand breaks, we envision two models for its function. Multiple pathways might contribute to meiotic recombination in S. pombe. Indeed, temporally separated pathways of meiotic recombination have been proposed (Carpenter, 1987;Engebrecht et al., 1990); an early pathway that initiates homologue alignment by gene conversion and a later one involving reciprocal exchange to ensure proper disjunction of homologues.
One possibility is that an endonuclease introduces a nick in double-stranded DNA. This break is then processed by ExoI into a single strand gap which would serve as a recipient for an invading complementary single strand from the homologue. Strathern et al. (1991) have demonstrated that the phage fl gene I1 product, when expressed in S. cereuisiae, stimulates mitotic recombination in conjunction with the fl nicking site artificially integrated into the genome. The chromosome containing the fl nicking site acts predominantly as the recipient of genetic information with a bias to the 5' side of the postulated nick.
A second possibility is a double strand break introduced by a specific endonuclease. ExoI would process the ends into 3' overhangs which then serve as a substrate for activities that promote homologous pairing and strand exchange. If the 3' ends were partially degraded by other nucleases, the broken chromosome would act predominantly as a recipient of genetic information in the region of the initial double strand break.
S. pombe has approximately 50 exchanges per meiotic genome (Munz et al., 1989) and a nuclear diameter of roughly 2 microns (Robinow and Hyams, 1989). Assuming initiation of meiotic recombination by double strand breaks, one per exchange, the concentration of double strand ends would be 4 X lo-' M or 5 times the K , for ExoI. We have estimated that there are 1,000 ExoI molecules per cell; if they are all nuclear, the concentration of ExoI would be 4 X M or 10 times the postulated concentration of double strand ends. Thus, both Ex01 and its presumed substrate are present at concentrations above those required for maximal activity in uitro.
Both double strand breaks and single-stranded tails have been reported as recombination intermediates in 5' . cereuisiae and X. lueuis (see introduction; White and Haber, 1990;Cao et al., 1990;Sun et al., 1991;Maryon and Carroll, 1991). We expect that activities similar to ExoI exist in those and other organisms. Indeed, an activity very similar to ExoI has recently been described from mitotic cells of s. cereuisiae (Dolberg et al., 1991). Its expression during meiosis, though, has not been reported yet. The formation of single-stranded DNA as a prerequisite for hybrid DNA formation might be ubiquitous in all recombining creatures. Therefore, enzymes like ExoI might play a crucial role in eukaryotic homologous recombination. Chow and Resnick (1988) and Resnick et al. (1984) have described a nuclease activity from S. cereuisiae that is induced during meiosis and strongly reduced in rad52 mutants. Cells lacking the RAD52 function exhibit increased sensitivity to x-rays and are deficient in various aspects of recombination (reviewed in Malone et al., 1988). In particular, the RAD52 function has been implicated in the repair of double strand breaks. The purified enzyme has endonucleolytic activity on single-stranded DNA and exonucleolytic activity on doublestranded DNA. 32P label is readily released from the 5'-end of double-stranded DNA (Chow and Resnick, 1987). Unfortunately, 3'-end labeled DNA was not analyzed, and it is therefore unclear whether this activity is related to ExoI from S. pombe.
Two exonucleases involved in recombination, X exonuclease from phage X and exonuclease VI11 from a cryptic X-like prophage rac of E. coli, have properties very similar to those of ExoI (see introduction; Little, 1967;Joseph and Kolodner, 1983). Mutations in the gene coding for X exonuclease abolish the recombination system of phage X (Shulman et al., 1970). The E. coli sbcA mutations allow expression of ExoVIII and thereby suppress mutations in recB and recC, to restore recombination proficiency during conjugation and transduction (Barbour et al., 1970;Gillen et al., 1981).
We have purified another exonuclease from S. pombe, ExoII, which is not induced during meiosis. ExoII specifically degrades single-stranded DNA with a 5' + 3' polarity: similar to that of the E. coli recJ gene product (Lovett and Kolodner, 1989). We suggest that ExoII does not act in a ratelimiting step of meiotic recombination or that it functions in other processes, such as mitotic recombination and DNA damage repair.