DNA Polymerase I11 from Saccharomyces cerevisiae 11. INHIBITOR STUDIES AND COMPARISON WITH DNA POLYMERASES I AND II*

The newly identified yeast DNA polymerase I11 was compared to DNA polymerases I and I1 and the mitochondrial DNA polymerase. Inhibition by aphidicolin (I50) of DNA polymerases I, 11, and I11 was 4, 6, and 0.6 pg/ml, respectively. The mitochondrial enzyme was insensitive to the drug. N2-(p-n-butylphenyl)-2’-deox- yguanosine 5”triphosphate strongly inhibited DNA polymerase I (I6o = 0.3 WM), whereas DNA polymerase I11 was less sensitive (Ia0 = 80 FM). Conditions that allowed proteoIysis to proceed during the preparation of extracts converted DNA polymerase XI from a sen- sitive form (Iao = 2.4 p ~ ) to a resistant form (Ia0 = 2 mM). The mitochondrial DNA polymerase is insensitive (I60 > 5 mM). With most other inhibitors tested (N- ethylmaleimide, heparin, salt) only small differences were observed between the three nuclear DNA polymerases. Polyclonal antibodies to DNA polymerase I11 did not inhibit DNA polymerases I and 11, nor were those polymerases recognized by Western blotting. Monoclonal antibodies to DNA polymerase I did not cross- react with DNA polymerases I1 and 111. The results show that DNA polymerase I11 is distinct from DNA polymerases I and 11.

Recent experiments with the cloned DNA polymerase I gene have shown that this enzyme is required during the Sphase of the yeast cell (2, 3). DNA polymerase I has an associated DNA primase activity, encoded by a separate gene, but no associated 3'-5' exonuclease activity (4-7). In contrast, DNA polymerase I1 does have a proofreading exonuclease activity (5,6). The enzyme has not been studied in detail. Its failure to elongate RNA primers in vitro argues against its involvement in Okazaki fragment synthesis in vivo, but no other role has been assigned to this enzyme (5,6).
The most curious observation made during analytical and preparative HPLC separations of extracts prepared under varying conditions was that the major DNA polymerase activity consisted of either DNA polymerase I or DNA polymerase 111 (1). Two extreme examples are given in Fig. 1. Because these results indicated that DNA polymerase I and DNA polymerase 111 might be two different forms of the same enzyme, we carried out inhibitor and immune studies to compare both enzymes with each other and with the minor nuclear DNA polymerase 11, as well as the mitochondrial enzyme. These studies are presented in this paper.

MATERIALS AND METHODS
DNA substrates and general DNA polymerase and exonuclease assays are as described (1). The regular DNA polymerase assay used activated calf thymus DNA as a substrate; the regular exonuclease assay used 3"labeled single-stranded calf thymus DNA as a substrate (1). Monoclonal antibodies to yeast DNA polymerase I were a gift from Dr. L. M. S. Chang (Uniformed Health Services University) (8).
These monoclonals were generated against the native DNA polymerase I, containing predominantly the 140-kDa polypeptide and its proteolytic 110-kDa fragment (7,8)? None of these antibodies inhibit enzyme activity directly, but their affinity for DNA polymerase I was measured by precipitation with secondary antibodies (8): The monoclonal antibodies do not bind to the denatured enzyme and they fail to give a signal on Western blots. Monoclonal antibody y-48 was used by Plevani et al. (25) to immunopurify the DNA polymerase I-primase complex. Of the six independent monoclonal antibodies provided, y-9, y-66, y-98, y-34, y-48, and y-60, the last three bound most strongly to our preparation of DNA polymerase I, whereas the first three were less active. A mixture of equal volumes of y-34 (IgGl), y-48 (IgGSb), and y-60 (IgG1) was used for enzyme binding studies. N"-(p-n-butylphenyI)-2'-deoxyguanosine-5'-triphosphate (BuPhdGTP) and Nz-(p-n-butylphenyl)-2'-deoxyadenosine-5'-triphosphate (BuPhdATP) were generously provided by Dr. George Wright (University of Massachusetts Medical School). Aphidicolin was prepared from the culture supernatant of Cephalosporium aphidicola as described (9). All other chemicals were from Sigma.
DNA Polymerases-DNA polymerase I was partially purified from an extract of PY2 cells made in the absence of protease inhibitors, and purified over phosphocellulose and DEAE-silica gel chromatography (1). DNA polymerase I1 was obtained similarly and, in addition, partially purified through similar chromatographic steps from extracts of PEP4 cells made in the presence of protease inhibitors. DNA polymerase I11 (Fraction VIa) was used for most studies. Fraction IV was generally used for immunological studies (1). Mitochondrial DNA polymerase was partially purified through DEAE-cellulose chromatography as described (10). Measurements of specific activities and comparison with literature values from Refs. 5, 8, and 10 for DNA polymerases I, 11, and the mitochondrial enzyme, respectively, showed that these enzymes were pure for about 3, 10, and 5%, respectively. They were, however, not cross-contaminated.
Immunochemical Techniques-Rabbit anti-DNA polymerase 111 or control serum was incubated with enzyme (30 units of polymerase) with immune precipitation buffer IPB (50 mM KH,PO,, pH 7.0, 1 mM EDTA, 10% glycerol, 1 mg/ml of bovine serum albumin, 150 mM NaCl) added to adjust the total volume to 50 pl. After 2 h at 0 "C, 30 pl of protein A-agarose beads (Sigma; 4.2 mg of extracellular protein A/ml of gel; resuspended in buffer IPB and washed with 10 volumes of IPB) were added and, after agitation at 0 "C for 1.5 h, the beads were spun down and 10-20 p1 of the supernatant used for an assay. Inhibition by anti-DNA polymerase 111 serum was measured as percent activity of that with control serum. The control serum did not inhibit the activity of any of the four DNA polymerases. In fact, addition of large amounts of control serum slightly (10-20%) stimulated enzyme activity.
Culture supernatants to mouse myeloma cell lines secreting monoclonal antibodies to DNA polymerase I (a mixture of y-34, y-48, and y-60, each at about 5 pg/ml) or control supernatants were incubated with enzyme (30 units) with buffer IPB added to adjust the total volume to 20 pl. After 2 h at 0 "C, carrier mouse IgG (75 ng) was added, followed by excess rabbit anti-mouse IgG (1 pg). After another 1 h at 0 "C, the samples were spun for 15 min in a microcentrifuge and 5-10 pl of the supernatant was used for an activity assay. Inhibition of the monoclonal antibodies was measured as percent activity of that with control supernatant. The control supernatant did not inhibit nor stimulate enzyme activity.

RESULTS
Inhibition by Aphidicolin-Both the polymerase and exonuclease activities of the DNA polymerase 111-exonuclease I11 complex were strongly inhibited by the tetracyclic &terpenoid aphidicolin (for a review see Ref. 11). Inhibition of the exonuclease activity by aphidicolin was dependent upon the substrate concentration. At 1.5 pg/ml of single-stranded DNA, the apparent K,, exonuclease activity was inhibited for 50% at 50 pg/ml of aphidicolin. At a saturating DNA concentration of 8 pg/ml, 50% inhibition was at 10 pg/ml of aphidi-Colin (Fig. 2). Similarly, inhibition of the polymerase activity was most potent at saturating template DNA concentrations. Plevani et al. (12) have observed a similar template dependent inhibitory effect for yeast DNA polymerases I and 11. For comparative purposes inhibition of DNA polymerases I and I1 were measured under identical conditions. Both were about an order of magnitude less sensitive to aphidicolin. The mitochondrial enzyme was insensitive to the drug (Fig. 2, Table  I).
Inhibition of BuPhdGTP and BuPhdATP-BuPhdGTP is a potent inhibitor of the mammalian DNA polymerase CY (13). Much higher concentrations of this dGTP analog are required to inhibit DNA polymerase 6 (14). DNA polymerases p and y are insensitive (13). Yeast DNA polymerase I was very sensitive to BuPhdGTP. Under the assay conditions used, 50% inhibition was observed at 0.3 pM BuPhdGTP. In contrast, DNA polymerase III was only weakly inhibited by BuPhdGTP with I,, = 80 p~ (Fig. 3A, Table 1). T o ensure that the low sensitivity of DNA polymerase 111 to BuPhdGTP was not due to some kind of inactivation of the inhibitor (e.g. hydrolysis), loo I + -

FIG. 2. Inhibition by aphidicolin.
The regular DNA polymerase assay was used, except that the dCTP concentration was lowered to 10 p~. The concentration of single-stranded DNA in the exonuclease assay was increased to 8 pg/ml.  * In comparison to the optimum Mg?' concentration.
KC1 or heparin was added to the regular DNA polymerase assay. dGTP was lowered to 10 p~ in the regular assay. e Estimated from extrapolation of the curves (see Figs. 2 and 3). 'dATP was lowered to 10 p~.
Complete assays minus dithiothreitol and template DNA were incubated with N-ethylmaleimide (at 1 mM) for various times at 0 "C. Dithiothreitol and activated calf thymus DNA were then added to 10 mM and 150 pg/ml, respectively, and, after another 5 min at 0 "C, the reactions were incubated at 37 "C for 30 min.
dCTP was lowered to 10 p~ in the regular assay. a mixing experiment was carried out. Equal units of DNA polymerase I and I11 were measured together in BuPhdGTP inhibition studies. The experimental points virtually coincided with the calculated data points indicating that no inactivation of the inhibitor occurred (Fig. 3A). Inhibition of DNA polymerases I and I11 by BuPhdGTP was also independent of the source of the enzyme (ie. from wild-type or protease deficient yeast), or the degree of purity of the enzyme, this is in contrast to DNA polymerase I1 (see below). The mitochondrial DNA polymerase was insensitive to the action of BuPhdGTP (Fig. 3B). Inhibition of DNA polymerase I1 by BuPhdGTP was more complex and depended on the source of this enzyme. DNA polymerase I1 partially purified from a protease deficient strain (PEP4) in the presence of protease inhibitors showed a biphasic response to BuPhdGTP with 60-80% of the enzyme being very sensitive to the inhibitor and 20-40% relatively insensitive (five independent extracts and DEAE-HPLC separations). A similar biphasic response was observed with DNA polymerase I1 partially purified from wild-type cells (PY,) without protease inhibitors. The sensitive component, however, only contributed 5-30% and the insensitive component 70-95% of the total DNA polymerase I1 activity (three independent extracts and DEAE-HPLC separations). An example of two extreme DNA polymerase I1 preparations is given in Fig. 3B were obtained with the dATP analog BuPhdATP (Table I). Except for the inhibition by these inhibitors, DNA polymerase IIa (sensitive) and DNA polymerase IIb (insensitive) were identical within experimental error with regard to all other criteria. This includes sensitivity to aphidicolin (7 pM for IIa, 6 p M for IIb), N-ethylmaleimide (t3,z = 3.1 min for IIa, 2.4 min for IIb), salt (Iso = 0.20 M KC1 for IIa and IIb), high levels of Mg" (Iho = 38 mM for IIa, 40 mM for IIb), and heparin (Iso = 2.0 pg/ml for Ira, 1.5 Fg/ml for IIb). Furthermore, the apparent K , values for dNTPs were virtually identical (5.5 p M for IIa, 4.9 p M for IIb). In addition, neither enzyme IIa nor enzyme IIb were inhibited by antibodies to DNA polymerase I or I11 (see below). The average values of enzymes IIa and IIb for all inhibitors except BuPhdGTP and BuPhdATP are given in Table I. Both enzyme IIa and IIb consistently eluted as sharp peaks at identical salt concentrations from an analytical DEAE-silica gel column (Fig. 1).
Further Characterization of Yeast DNA Polymerases-The K , values for dNTPs were measured for the four yeast polymerases (Table I). No large differences were observed that could be used to differentiate between the yeast DNA polymerases.
The K , value for DNA polymerase I (4.6 pM/nucleotide) is in agreement with those determined by Badaracco et al. (7) (0.66, 1.0, 1.7, and 1.9 p~ for dGTP, dATP, dTTP, and dCTP, respectively) if one takes into consideration that the apparent K , value for a single dNTP with the other three dNTPs at saturation is several fold lower than when all four dNTP concentrations are subsaturating (15).
Other inhibitors than aphidicolin or BuPhdGTP were less informative about the yeast DNA polymerases. Thus, rates of inactivation by N-ethylmaleimide of the three nuclear DNA polymerases were almost identical, whereas the mitochondrial enzyme was much more sensitive. Similarly, inhibition by heparin served to distinguish the mitochondrial enzyme from the three nuclear enzymes but did not differentiate among these (Table I). Salt (K') inhibition paralleled Mg2' inhibition with DNA polymerases I and I11 as the most sensitive enzymes, followed by enzyme 11. The mitochondrial enzyme is most active at 60 mM Mg' and much higher concentrations were needed to inhibit this enzyme by 50% (Table I, 10).
Antigenic Determinants of the Nuclear DNA Polymerases-More than the use of inhibitors antibodies should be useful in defining and distinguishing the three nuclear DNA polymerases. A mixture of three monoclonal antibodies to DNA polymerase I (see "Materials and Methods") was tested for binding to the other two enzymes, Neither enzyme 11, nor enzyme 111, were inhibited by these antibodies (Fig. 4).
The polyclonal serum to DNA polymerase I11 inhibited both DNA polymerase I11 and exonuclease I11 activities to the same extent ( Fig. 5; data not shown for exonuclease 111). DNA polymerases I and I1 were refractory to this antiserum (Fig.  5). In addition, Western blots with the polyclonal serum of fractions from a DEAE HPLC separation of the three nuclear DNA polymerases showed prominent bands in the DNA polymerase 111 region (Fig. 6, lane l ). No cross-reactive bands, except a 70-kDa band, were seen in the DNA polymerase I and I1 regions (lanes 2 and 3 ) . This band is due to a 70-kDa polymerase-unrelated polypeptide which streaks across the entire gradient and is also present in fractions that do not contain any DNA polymerase activity. The 70-kDa protein is probably one of a family of heat shock proteins because the same band is also visualized with antibodies to the major

Quuntitation of DNA Polymerases Z and ZZZ in Various
Extracts-A major consideration in estimating the relative levels of DNA polymerases I and I1 is the possibility that preferential inactivation of one of these enzymes might occur during DEAE-silica gel chromatography. Attempts to quantitate these enzymes prior to DEAE-silica gel chromatography with the use of the mono-and/or polyclonal antibodies failed, presumably because the extracts were too crude to allow reproducible binding of the antibodies. However, because DNA polymerase I is very sensitive to BuPhdGTP and DNA polymerase I11 relatively insensitive, and, in addition, DNA polymerase I1 constitutes only a minor activity (20-30%), this inhibitor can be used to estimate the ratio of DNA polymerase III/DNA polymerase I in crude extracts prior to separation of the enzymes by DEAE-HPLC. At 2 PM BuPhdGTP, DNA polymerase I activity was inhibited by 92% and DNA polymerase I11 activity by 6% (Fig. 3A). Addition of 2 PM Bu-PhdGTP to assays of extracts (Fraction 11) prepared from wild-type cells in the absence of protease inhibitors inhibited total DNA polymerase activity 61-73% (three independent extracts), indicating that DNA polymerase I was the major activity in these extracts. Extracts from protease-deficient cells made in the presence of protease inhibitors gave 14-37% inhibition of total DNA polymerase activity when 2 PM BuPhdGTP was added (five extracts), indicating that now DNA polymerase I11 was the major activity. Interestingly, when extracts (Fraction 11) were prepared with protease inhibitors present, from protease-deficient cells which had been frozen in liquid nitrogen and then thawed prior to breakage, inhibition by 2 PM BuPhdGTP was 34-51% (three extracts). Analytical DEAE-silica gel chromatography of one of such extracts yielded DNA polymerase I11 and DNA polymerase I in the ratio 1.8:1 (results not shown). This is much lower than the ratio obtained (DNA polymerase III/DNA polymerase I is 4-5 in several experiments) when cells were not frozen prior to breakage. Thus, the inhibition data in crude extracts roughly correlate with the actual ratio of DNA polymerase III/DNA polymerase I obtained after DEAE-silica gel chromatography (Fig. 1). These results show that: (i) dialyzed ammonium sulfate preparations (Fraction 11) from proteasedeficient cells contain mostly DNA polymerase 111; (ii) the amount of DNA polymerase I was increased (or DNA polymerase I11 decreased) when these cells were frozen and thawed prior to breakage; (iii) extracts (Fraction 11) from wild-type cells broken in the absence of protease inhibitors contain mostly DNA polymerase I; (iv) because the degree of inhibition observed in crude extracts agrees well with the relative activities of DNA polymerases I11 and I observed after DEAEsilica gel HPLC chromatography, we conclude that none of the activities is preferentially inactivated during chromatography of this column.

DISCUSSION
Since the first report in 1970 on the separation of yeast DNA polymerases I and I1 (17), numerous studies have appeared on these two enzymes (5,6,12,18,19). DNA polymerase I, the major enzyme activity, has been most extensively studied. The enzyme elutes early from a DEAE-cellulose column, has a tightly associated primase activity but no associated 3'-5' exonuclease activitv (4-8. 20). DNA Dolvm-questions: 1) which factors have kept DNA polymerase 111 from being identified in previous work? 2) Why is DNA polymerase 111 only present in extracts from protease deficient cells? 3) Why are DNA polymerase I levels reduced in extracts from protease deficient cells?
The DEAE-HPLC silica gel separations of different extracts are given in Fig. 1. The simplest explanation of these experimental data would define DNA polymerase 111 and DNA polymerase I as two alternative forms of the same enzyme. Thus, under conditions that allow proteolysis to proceed, DNA polymerase I would cochromatograph with DNA primase. Alternatively, when proteolysis is inhibited by use of a protease-deficient strain in combination with a variety of protease inhibitors DNA polymerase I would cochromatograph with the 3'-5' exonuclease. The resolving power of HPLC columns and possibly additional subtle interactions with the silica gel matrix separates these two species.
When two extracts similar to those shown in Fig. 1, A and B were chromatographed over a conventional DEAE-cellulose column, only two peaks of activity were seen in each case. The activity eluting early from the gradient constituted about 80% of the total DNA polymerase activity. The minor DNA polymerase activity (20%) eluted later from the column. The chromatograms of these two DEAE-cellulose separations are almost identical and compare well to such chromatograms obtained by others (results not shown) (5, 6). Thus, DEAEcellulose fails to separate DNA polymerase I from DNA polymerase 111 and only the DEAE-silica gel column allowed separation of the two species. Both the inhibitor studies and the antibody studies, however, indicate that DNA polymerases I and I11 are different enzymes (Figs. 2, 3A, 4-6, Table I). Of the inhibitors tested aphidicolin and BuPhdGTP were most indicative, but caution should be exercised with the use of these inhibitors. The inhibitory effect of aphidicolin can decrease upon extensive purification of the DNA polymerase (7,12). This effect may be related to proteolysis (7). Inhibition of DNA polymerase I11 by aphidicolin, however, did not change upon purification of the enzyme from Fraction I\' to Fraction VII) (1). Our results with inhibition of DNA polymerase I1 by BuPhdGTP or BuPhdATP indicate that the use of these inhibitors can also lead to misleading results (Fig. 3B, Table I). The immunological data, however, support the results with the inhibitors .
Highest levels of DNA polymerase 111 were obtained when cells of a protease-deficient yeast strain were harvested in log phase and processed, without intermediate freezing, in the presence of several protease inhibitors. From these data it is not unreasonable to assume that proteolysis by the yeast proteases during the preparation of extracts inactivates DNA polymerase 111. Because the studies on yeast DNA polymerases cited above did not make use of log phase proteasedeficient cells, processed without intermediate freezing, DNA polymerase 111 would generally have been a minor activity. In addition, since DEAE-cellulose chromatography, used as an early purification step in these studies, fails to separate DNA polymerase 111 from DNA polymerase I, the former would have been present as a contamination during initial purification steps of DNA polymerase I. And due to its lability, DNA polymerase I11 might have gone undetected during further purification of DNA polymerase I (1).
More difficult to understand is why the reduction of proteolysis would lead to a decrease in extractable DNA polymerase I activity. Two possible explanations are: 1) when proteolysis is limited the DNA polymerase I, presumably in a large complex with accessory proteins, is not extracted from yeast cells; 2) unproteolyzed DNA polymerase I is less active on the activated calf thymus DNA substrate used to quantitate DNA polymerase activity.
Re-extraction of yeast cell debris with high salt buffers (containing 1 or 2 M NaC1) in the presence or absence of Triton X-100 (0.2%) did not result in recovery of additional DNA polymerase I activity. Nor could polymerase be extracted from the polymin P pellet using similar high salt buffers (polymin P (0.4%) in the presence of 175 mM ammonium sulfate was used to precipitate nucleic acids from extracts (1)). The second possibility, i.e., that the DNA polymerase I obtained from the protease-deficient strain poorly uses activated calf thymus DNA as a substrate, is an attractive one because it suggests that from these cells a multipolypeptide complex might be obtained which is functional under conditions and on substrates that more closely approach the replication fork. More work is needed to obtain a better understanding of the differential activities of these enzymes in various extracts. Ultimately, the possible relationship between DNA polymerases I and I11 will be clarified by cloning of the gene for DNA polymerase 111. Since the gene for DNA polymerase I has been cloned, a direct comparison of the genes and their expression will be of great aid to the biology of DNA replication in yeast (2).
A comparison of the yeast DNA polymerases with the mammalian enzymes shows that DNA polymerase 111 is most clearly related to the 6 polymerase. Thus, both enzymes have high molecular weight catalytic subunits and possess a 3'-5' exonuclease activity (1, 14,21,22). Like DNA polymerase 111, both the polymerase and exonuclease activities of DNA polymerase 6 are strongly inhibited by aphidicolin (Fig. 2, 23). And the polymerase activities of both enzymes are a few orders of magnitude less sensitive to BuPhdGTP than DNA polymerase I and DNA polymerase a , respectively (Fig. 3A,  14). Thus, DNA polymerase 111 could serve a similar function in DNA replication as proposed for the mammalian DNA polymerase 6, i.e. in synthesis of the leading strand of the replication fork (14, 24). In addition, or alternatively, yeast DNA polymerase 111 in combination with exonuclease IV could serve in Okazaki fragment maturation. DNA polymerase I and DNA polymerase a, respectively, both with an associated DNA primase activity would be responsible for lagging strand DNA replication (14,24). With the gene for yeast DNA polymerase I in hand and the cloning of the gene for DNA polymerase I11 in progress, a genetic approach to these questions will be possible in this organism.