Feb 3, 1989 - dAMP:Gte.,a mispairs (1/16 500 000), and below 10-8. (1/100 000 000) for .... Subunit structure and proof of the undegraded state of the human ...
The EMBO Journal vol.8 no.6 pp. 1 833 - 1839, 1989
Exonucleolytic proofreading increases the accuracy of DNA synthesis by human lymphocyte DNA polymerase Oa-DNA primase Gabriele Bialek, Heinz-Peter Nasheuer2, Hilde Goetz' and Frank Grosse Abteilung Chemie and 1Abteilung Immunchemie, Max-Planck-Institut fur experimentelle Medizin, D-3400 Gottingen, FRG 2Present address: Department of Pathology, Laboratory of Experimental Oncology, School of Medicine, Stanford University, Stanford, CA 94305, USA Communicated by F.Eckstein
DNA polymerase-primase complex, isolated with an apparently undegraded a-subunit, was immunoaffinitypurified to near homogeneity from the human lymphoblast line HSC93. The undegraded state of the a-subunit was monitored by Western-blot analysis of crude cellular extracts and all active fractions obtained during purification. The human polymerase- primase consists of four subunits with molecular weights of 195, 68, 55 and 48 kd. The fidelity of the polymerase-primase in copying bacteriophage 4PX174am16 DNA in vitro was determined by measuring the frequency of production of different revertent phages. The overall accuracy was between 4 x 10-6 and 10 x 10-6. This value reflects the spontaneous mutation frequency of 4X174am16 phages in Escherichia coli, and is 10- to 20-fold higher than the accuracy of a conventionally purified enzyme from calf thymus. The frequencies of base pairing mismatches, estimated from pool bias measurements, were 3.5 x 10-7 (1/2 880 000) for dGMP:Ttemp,,te mispairs, between 10-7 and 10-8 for dCMP:Ttempiate (1/35 000 000), dCMP:Atemplate (1/18 200 000) and dAMP:Gte.,a mispairs (1/16 500 000), and below 10-8 (1/100 000 000) for dTM[:Ttnpae, dGMP:Anpate and dGMP:Gtemplate mispairs. In contrast to previous preparations, the intact polymerase-primase possesses a 3'-5' exonuclease activity. This exonuclease removes both matched and mismatched 3'-OH ends, with a preference for mismatched bases. Fidelity was reduced 8-fold by increasing the concentration of the next nucleotide following the incorporated mismatch nucleotide. Upon replacing dGTP by its phosphorothioate analogue at equimolar concentrations of the four nucleoside triphosphates, a 2-fold increase in the number of revertants was observed; biasing dGTPaS 9- and 30-fold over dATP and dCTP, respectively, led to a 50-fold increase in the number of revertants. Taken together, these observations suggest that the 3'-5' exonuclease present in immunoaffinity purified human polymerase-primase proofreads nucleotide misinsertions during DNA synthesis. The exonuclease contributes at least one to two orders of magnitude to the high fidelity characteristics of the intact polymerase- primase
complex. Key words: fidelity/immunoaffinity purification/next nucleotide effect/phosphorothioate nucleotide
(IRL Press
Introduction Cellular DNA replication is a highly accurate process with 10-8 to 10-12 misincorporated nucloeotides per base replicated (Englisch et al., 1985). This ultra-high accuracy is achieved by a series of steps involving (i) nucleotide selection by the replicative DNA polymerases, the accuracy of which was suggested to be limited to 10-4 to 10-5 by the existence of rare tautomers of the nucleotide bases (Watson and Crick, 1953; Topal and Fresco, 1976), (ii) removal of incorrectly paired nucleotides by a 3'-5' proofreading exonuclease (Brutlag and Komberg, 1972; Kunkel, 1988) and (iii) post-replicative mismatch repair systems (Glickman and Radman, 1980; Modrich, 1987). The eukaryotic replicative enzyme, DNA polymerase a -primase complex, had been found previously to have no exonucleolytic activity (Chang and Bollum, 1973; Fisher and Kom, 1977; Banks et al., 1979; Grosse and Krauss, 1980, 1981). Consistent with this observation were the low accuracies between 10-4 and 10-5 determined in vitro (Kunkel and Loeb, 1981; Grosse et al., 1983). Recently, however, purified a-polymerases with an associated exonuclease activity have been reported (Chen et al., 1979; Ottiger and Hubscher, 1984; Skarnes et al., 1986). It was not clear from these studies whether the exonuclease activities were of physiological significance, e.g. as a proofreading activity, or represented a contamination. The recent demonstration of proofreading with the mammalian DNA polymerases -y (Kunkel and Soni, 1988) and 6 (Kunkel et al., 1987), and all finding of a cryptic 3'- 5' exonuclease together with the DNA polymerase a -primase complex of Drosophila melanogaster (Cotterill et al., 1987; Reyland et al., 1988) suggests that exonucleolytic proofreading is not only essential for removing incorrectly inserted nucleotides of prokaryotic organisms but might also be an important contributor to the fidelity of replication in eukaryotes. Here we report the purification of a highly accurate polymerase-primase from human lymphocytes, which contains a 3'-.5' exonuclease activity and present evidence that exonucleolytic proofreading contributes to the high accuracy characteristics of the human polymerase -primase.
Results Immunoaffinity purification of the DNA polymerase primase complex from human cells DNA polymerase - primase was purified from human cells by the same protocol (summarized in Table I) which was used for the purification of polymerase -primase from calf thymus glands (Nasheuer and Grosse, 1987). As described under Materials and methods, human HSC93 cells were disrupted under isotonic conditions. Proteins from the cytoplasmic extract were enriched by batch adsorption to phosphocellulose and further chromatographed on heparinSepharose. The eluate was concentrated by ammonium
a -DNA
1833
G.Bialek et al.
Table I. Immunoaffinity purification of polymerase-primase from human lymphoblastsa Fraction
(1) (2) (3) (4)
Total protein (mg)
Total activity Specific activity Yield
(units)
12000 4200 crude extract 100 18000 phosphocellulose heparin-Sepharose 30 12000 0.05 1300 antibody column
205-
(units/mg)
(%)
3 140 400 21000
(66)b
116-93 ---
100 66 7
68----
aPurification was from 100 g (wet mass) human HSC93 cells. bMeasurements in crude extracts were interfered with by the presence
Subunit structure and proof of the undegraded state of the human lymphocyte DNA polymerase a -DNA prmase complex Analysis of the subunit structure by denaturing polyacrylamide gel electrophoresis revealed four bands with molecular weights of 195, 68, 55 and 48 kd (Figure 1). The level of degradation of the largest subunit was monitored by Western-blot analysis over the course of purification using an antibody against the largest subunit of the immunoaffinitypurified polymerase - primase from calf thymus (Nasheuer and Grosse, 1988). As shown in Figure 2, freshly opened human lymphocytes contain only one band of cross-reacting protein, which migrates at a position corresponding to 195 kd. Only one protein of the same size is found in the finally purified fraction. In the time span between taking samples and performing the electrophoresis procedure (typically 18 h), small amounts of a degraded product with an Mr of 180 kd occurred in the less purified samples of the phosphocellulose and heparin-Sepharose eluates (Figure 2). Enzymatic activities of the intact polymerase - primase Immunopurified polymerase -primase is able to replicate single-stranded 4)X 174 DNA to the double-stranded form (Figure 3). During this process about seven RNA primers, 10-14 nucleotides long, were formed (Bialek, 1988). These data strongly resemble our earlier data on the calf thymus enzyme (Grosse and Krauss, 1984, 1985; Nasheuer and Grosse, 1987, 1988). Accuracy of the intact polymerase - primase We have demonstrated earlier that proteolytic degradation of DNA polymerase a leads to error prone forms of the enzyme (Brosius et al., 1983; Reckmann et al., 1983). Accuracy measurements with an apparently undegraded forn
1834
J
43
195
4
.~ ~68
'p-W
-
of nucleases and dNTPases.
sulphate precipitation and loaded to the immobilized monoclonal antibody SJK287-38 (Tanaka et al., 1982). The affinity material was extensively washed and then eluted in a batch procedure by shifting to pH 13. Immediately after elution the fractions were neutralized by the addition of KH2PO4. The whole procedure was performed within 12 h. From 100 g (wet weight) tissue culture cells, 50 jig of nearly homogeneous polymerase -primase complex was obtained. This represents a 7 500-fold enrichment with a total recovery of 7 %. The specific activity of the Fraction IV enzyme was 21 000 U/mg, DNA primase displayed a specific activity of 2 500 U/mg (Table I).
L lb
-
t-_
55 48
30-
Fig. 1. Polypeptide structure of the immunoaffinity purified DNA polymerase a-primase complex from human HSC93 cells. Right lane: 2 Ag of the immunoaffinity purified human DNA polymerase-primase complex was separated on an 8.5% acrylamide-0.17% methylenebisacrylamide gel in the presence of 0.1% sodium dodecylsulphate as a denaturant (Laemmli, 1970). Left lane: Molecular weight markers (kd) were rabbit muscle myosin (205), E.coli 13-galactosidase (116), rabbit muscle phosphorylase b (93), bovine serum albumin (68), chicken ovalbumin (43) and bovine erythrocyte carbonic anhydrase (30). After electrophoresis, the gel was stained with Coomassie.
Mr X 10-3 ~--1 95
Fig. 2. Immunoblotting of the a-subunit of polymerase-primase containing fractions during the course of purification. Polymerasecontaining fractions at the individual steps of purification were separated on an 8.5% acrylamide-0.17% methylenebisacrylamide gel according to Laemmli (1970). Proteins were transferred to nitrocellulose sheets (BA85, Schleicher & Schiill, Dassel, FRG) for 30 min at 0.8 mA/cm2 gel by using a Semi-Dry electroblotter A apparatus (Ancos, Denmark). Development of immunoreactive bands on the nitrocellulose was carried out as described elsewhere (Nasheuer and Grosse, 1988). Lanes from left to right contained 80 Mg of crude cytosolic extract from human HSC93 cells, 30 Mg each of the phosphocellulose eluate and the heparin-Sepharose eluate and 0.1 Mg of the immunoaffinity column desorbed material, respectively.
Human lymphocyte DNA polymerase
Fig. 3. Replication of single-stranded 4X174am16 DNA by the immunopurified DNA polymerase-primase. FX174am16 (+) strand 100 (60 AM) was replicated with 2.5 U/mI polymerase-primase in I1 of 20 mM Tris acetate, pH 7.2, 75 mM potassium acetate, 5 mM magnesium acetate, 1 mM dithiothreitol, 0.1 mg/ml bovine serum albumin, 0.1 mM each of dATP, dGTP, dCTP and dTTP, 1 mM ATP, and 0.1 mM each of GTP, CTP and UTP. Replication was performed at 37°C, until RFII was generated (4-24 h). Product analysis of an aliquot was on a 1% agarose gel as described (Grosse and Krauss, 1984). Left lane: unreplicated single-stranded 4X174am16 DNA. Middle lane: 900 AM dGTP and 30 AM TTP (30:1 bias). Right lane: 30 AM dGTP and 30 jiM TTP (unbiased case).
of the enzyme might lead to important clues to an understanding of how replication fidelity is achieved in the living cell. Fidelity measurements were performed with the (DX174am16 reversion assay (Fersht and Knill-Jones, 1981, 1983; Grosse et al., 1983). The single-stranded phage DNA was replicated in vitro by purified polymerase -primase. DNA synthesis was initiated by the addition of the four ribonucleoside triphosphates to allow RNA priming. The resulting circular duplex DNA (RFII) was transfected into suppressor tRNA-containing E. coli spheroplasts to allow the production of progeny phages. Progeny-producing spheroplasts were mixed with both the suppressor tRNAcontaining E. coli strain CQ2 to score the total number of phages produced by transfection, and the wild-type strain E. coli C to measure the number of revertants. From the ratio of revertants to total phages, divided by a factor of 0.4, to correct for minus strand expression inside the bacteria (see Material and methods), the total error rate of misincorporations at the aml6 codon was measured to be between 1/100 000 and 1/240 000 for the human polymerase - primase. The frequency of spontaneous mutations of (X174am16 DNA was also between 1/100 000 and 1/240 000, thus in vitro replication of natural DNA by human polymerase-primase is as accurate as the amplification of (X174am16 phages inside the E.coli host. Since
a - DNA
primase
the accuracy measurements were at the natural background level of the assay we carried out extensive controls to assure that (i) replication proceeded over the amber codon and (ii) that the spheroplasts used were able to indicate higher error frequencies of known and more error prone DNA polymerases. Replication products were analysed by agarose gel electrophoresis. Since only samples with >90% of RFH/RFIH were used for transfection (Figure 3) failures to replicate past the amber codon are unlikely. 4X174am16 DNA that was replicated in vitro by the conventionally purified polymerase-primase from calf thymus (Grosse and Krauss, 1981) was transfected to exclude any malfunction of the spheroplasts. DNA replicated in this way gave error frequencies between 1/6 000 to 1/12 000, in agreement with earlier results (Grosse et al., 1983). As errors were not detectable above background of 4'X174aml6 replication, a more stringent nucleotide pool bias assay was used. By shifting the nucleotide pools during in vitro replication, selective pressure can be put to any of the seven evaluable mismatches at the aml6 codon. A subsequent phenotypic characterization of progeny phages by measuring the temperatures at which the resulting phages are able to grow, allows the assignment of progeny phenotypes to the corresponding mismatch that is responsible for the observed phenotype (Fersht and Knill-Jones, 1981, 1983; Grosse et al., 1983). For example, at high concentrations of dGTP and concomitant low concentrations of dATP, a dGMP:Ttempiate misincorporation at the first nucleotide of the aml6 codon (AB5279) is provoked. Such a mismatch leads to revertants that grow at temperatures 244°C (Vwt-phenotype). As Figure 4 shows, there was a significant increase in progenies growing at temperatures >44'C upon applying a dGTP over dATP pool bias. The response of revertants to the bias was linear up to a 30-fold excess of dGTP over dATP, before it leveled off. The frequency for a dGMP:Ttemplate mismatch, extrapolated to unbiased nucleotide concentrations, was calculated from the slope of the linear part of the curve in Figure 4 to be 3.5 x 10-7 (1/2 880 000). In the same manner, pool bias measurements were performed for the other six mismatches. The evaluation of the slope of the dATP over dCTP bias, which forces a dAMP:Gemplate mismatch, yielded an error frequency of 6 x 10- (1/16 500 000) (Table H). The other mispairs did not respond to an up to 30-fold bias of the relevant nucleotides (Table II). Therefore, the base pairing fidelity for mispairs other than dGMP:T and dAMP:G might be much higher than that of the E.coli DNA polymerase III holoenzyme, the enzyme responsible for the replication of 4X174 phages. The frequencies for dCMP:T and dCMP:A mispairs were estimated to be between 10-7 to 10-8 when only the number of revertants from the 30:1 biases were considered (Table H); dTMP:T, dGMP:A and dGMP:G mispairs were not detectable, even under conditions of a 30-fold nucleotide pool bias. From the detection limit of the assay it was estimated that these particular mispairs occur with frequencies of less than 1 X 10-8 (1/100 000 000) (Table II). Intact polymerase - primase contains a 3'- 5' exonuclease Such a high base-substitution fidelity implies the existence of a sophisticated mechanism for error prevention. It was
1835
G.Bialek et al. 10
Table II. Effect of pool bias on the fidelity of intact human polymerase -primase Pool bias Ratio Mispair induced applied
co
0
8
x
G1 Al V 0
6
40
U C.) C
4
Ct Al
0
Tt Al
2
GI TI O
F
,
1
20
,
,
60 40 [dGTPJ/[dATPJ
,
l
80
100
Fig. 4. Dependence of the phage reversion frequency to {wt on the bias of dGTP over dATP during in vitro DNA replication. Replication was performed as described in the legend to Figure 3. Only progeny phages that grew at temperatures >440C (,6wt and wild-type) were scored. The straight line represents a least square fit of the data from a 1:1 over a 30:1 bias. The rate law for a dGTP:T mispair is obtainable from the slope as v = 1.39 x 10-7[dGTP]/[dATP] (R = 0.95). The error rate for a dGTP:T mispair was calculated for equimolar nucleotide concentrations to be 1.39 x 10-7 . 0.4 = 1/2 880 000 (Table II).
therefore tested whether the human polymerase - primase contained a 3'- 5' exonuclease activity to prevent errors. 3'- 5' exonuclease activity was measured as loss of 3'-labelled nucleotides after incubation of polymerase primase with poly(dA).(dT)20[3H]dT (matched primer end) or poly(dA).(dT)20[3H]dA (mismatched primer end), respectively. As shown in Figure 5, immunoaffinity-purified polymerase -primase from human cells does contain significant amounts of 3'-5' exonuclease activity. The exonuclease was 10-fold less active than the proofreading activity of E. coli DNA polymerase I (Klenow fragment) (Figure 5). Human polymerase-primase did not contain detectable amounts of 5'- 3' exonuclease activity (data not shown). -
The 3'- 5' exonuclease of human polymerase primase contributes to the high accuracy of the enzyme complex To determine whether proofreading occurs during DNA synthesis, we looked for next nucleotide and aphosphorothioate effects on base-substitution fidelities. In the first test of fidelity a 30-fold excess of dGTP over dCTP was used to induce a dGMP:G misincorporation during replication of the aml6 codon. A concomitant 30-fold increase of the concentration of the complementary following nucleotide (dTTP) forced the enzyme to the synthesizing pathway and thus permitted less time for the excision of mismatches (Fersht, 1979). Phenotypic characterization of the relevant ts35 phenotype (resulting from the provoked dGMP:Gt,mpjat, misinsertions) revealed 2 8-fold increase as compared to the number of ts35 progenies obtained without increasing the concentration of the next nucleotide (Table -
1836
Cl TI
Phenotype Reversion Estimated evaluated frequencya error rateb (X 10-6)
1:1 G:Ttemplate i/wt 3:1 10:1 30:1 100:1
2.8c 5.3 9.3 13.8 21.3
G:Ttemplate
2.8c 0.8
1:1 30:1
wt
Gt CI
< 1/35 000 000d
1:1 T:Ttemplate tS43 30:1
NDe ND
