Heat shock protein 70 stimulation of the deoxyribonucleic acid base excision repair enzyme polymerase β

Abstract

Base excision repair (BER) of damaged deoxyribonucleic acid (DNA) is a multistep process during which potentially lethal abasic sites temporarily exist. Repair of these lesions is greatly stimulated by heat shock protein 70 (Hsp70), which enhances strand incision and removal of the abasic sites by human apurinic-apyrimidinic endonuclease (HAP1). The resulting single-strand gaps must then be filled in. Here, we show that Hsp70 and its 48- and 43-kDa N-terminal domains greatly stimulated filling in the single-strand gaps by DNA polymerase β, a novel finding that extends the role of Hsps in DNA repair. Incorporation of deoxyguanosine monophosphate (dGMP) to fill in single-strand gaps in DNA phagemid pBKS by DNA polymerase β was stimulated by Hsp70. Truncated proteins derived from the C-terminus of Hsp70 as well as unrelated proteins were less effective, but proteins derived from the N-terminus of Hsp70 remained efficient stimulators of DNA polymerase β repair of DNA single-strand gaps. In agreement with these results, repair of a gap in a 30-bp oligonucleotide by polymerase β also was strongly stimulated by Hsp70 although not by a truncated protein from the C-terminus of Hsp70. Sealing of the repaired site in the oligonucleotide by human DNA ligase 1 was not specifically stimulated by Hsp-related proteins. Results presented here now implicate and extend the role of Hsp70 as a partner in the enzymatic repair of damaged DNA. The participation of Hsp70 jointly with base excision enzymes improves repair efficiency by mechanisms that are not yet understood.

INTRODUCTION

Heat shock proteins such as Hsp70 associate with certain deoxyribonucleic acid (DNA) base excision repair (BER) enzymes and stimulate their activity (Mendez et al 2000; Kenny et al 2001; Mendez et al 2003). After removal of a damaged base, the resulting potentially lethal apurinic-apyrimidinic (AP) site is transiently present and is then incised on the 5′ side by human apurinic-apyrimidinic endonuclease (HAP1) (also known as REF1), leaving a 3′ OH and a potential single-strand gap. The gap is completed by the deoxyribophosphodiesterase (dRPase) activity of DNA polymerase β (β pol), which first incises on the 3′ site of the AP site, removing the sugar remnant. Then β pol fills in the gap by covalent attachment of a new nucleotide at the 3′ OH site. Ligation is done to seal the replacement base (Matsumoto and Kim 1995). Hsp70 stimulated HAP1 activity 10- to 100-fold (Kenny et al 2001). We questioned whether stimulation of β pol activity would also result from its interaction with Hsp70. Here, we describe experiments that show that Hsp70 and an N-terminal portion of Hsp70-stimulated β pol activity, thereby enhancing the filling in of single-strand gaps that remained after elimination of the potentially lethal abasic site. Mechanisms responsible for Hsp's role in these novel findings are not yet understood.

The structural domains of Hsp70 have been reviewed (Bukau and Horwich 1998). Stimulation of β pol by Hsp70 was greater than with certain other proteins, but the N-terminal 45-kDa (1–385) adenosine triphosphatase (ATPase) domain (potlike) portion of Hsp70 was nearly as potent a stimulator as the intact Hsp70, in contrast to the 25-kDa 386–640 substrate binding and lid domains of Hsp70 or an unrelated protein such as bovine serum albumin (BSA). Buczynski et al (2001) showed that the HSP70 bacterial cytoplasmic homolog, DnaK (1–517), which contains the ATPase domain and the b-sandwich substrate-binding domain but no lid, has a variety of biological activities.

EXPERIMENTAL PROCEDURES

Enzymes and reagents

Phagemid pBKS was from Stratagene (La Jolla, CA, USA). Human β polymerase was from Trevigen, Gaithersburg, MD, USA, [α³²P]deoxyguanosine triphosphate (dGTP) and [α³²P]deoxycytidine triphosphate (dCTP), each at 3000 Ci/mM, were from Amersham Pharmacia Biotech (Piscataway, NJ, USA). A HAP1 expression plasmid was a gift from Dr Ian Hickson (University of Oxford, UK). Purified human uracil–DNA glycosylase (UDG) was a gift from Dr G. Slupphaug (UNIGEN, University of Trondheim, Norway). Recombinant human HAP1 was expressed in Escherichia coli and purified to homogeneity (Mendez et al 2000; Kenny et al 2001). Recombinant human Hsp70 and E coli Hsp70 DnaK were from Stressgen, Victoria, BC, Canada. N-terminal or C-terminal mutant Hsp70 proteins were a kind gift from Dr Jaewhan Song and Dr Richard I. Morimoto (Northwestern University, Evanston, IL, USA) Preincubation of these mutant proteins at 50°C for 10 minutes eliminated certain extraneous nuclease activities. Bovine pancreatic L-1-tosylamido 2-phenylethyl chloromethyl ketone (TPCK)–treated trypsin immobilized on 4% cross-linked beaded agarose was bought from Pierce, Rockford, IL, USA. DNA oligonucleotides were prepared at the Albert Einstein DNA Sequencing and Oligonucleotide Facility. N-terminal 48 kDa and 43 kDa fragments of Hsp70 were prepared by 18 hours of digestion with agarose-bound trypsin. Edman degradation and identification of 8 N-terminal amino acids were determined by standard methods (Laboratory of Macromolecular Analysis and Proteomics, Albert Einstein College of Medicine Bronx, New York, USA). Adenosine triphosphate (ATP) was purchased from Amersham Pharmacia Biotech. T4 DNA ligase (55 kDa) was bought from New England Biolabs Inc., Beverly, MA, USA. Human DNA ligase 1 (100 kDa) was a kind gift from Dr Alan Tomkinson (University of Texas, San Antonio, TX, USA).

Creation of abasic sites

Abasic sites were created by heating the DNA phagemids in 100 mM sodium acetate, pH 5.2, at 70°C for 15 minutes. The DNA was then neutralized with 1/10 volume of 1 M Tris, pH 9. This treatment created ∼2 AP sites per phagemid, ie, 16% surviving supercoiled forms, where the number of breaks per molecule (n) = −log natural logarithm of the fraction of unbroken molecules.

Conversion of abasic sites to single-strand gaps

AP sites were converted to single-strand gaps by endonucleolytic hydrolysis with purified HAP1. To completely convert the AP sites, depurinated phagemid (44 μg) was incubated for 30 minutes at 37°C in 800 μL of 66 mM Tris (pH 7.5), 5 mM MgCl₂, and 1 mM 2-mercaptoethanol with 16 μg of purified HAP1 (Kenny et al 2001). After incubation, the mixture was extracted with phenol to eliminate HAP1, and the DNA was then precipitated with ethanol. To verify conversion of AP sites to gaps, aliquots of phagemid in 0.05% xylene cyanol, 0.05% bromophenol blue, 2 mM ethylenediaminetetraacetic acid (EDTA), and 10% glycerol were heated for 5 minutes at 60°C and resolved by electrophoresis on a 0.8% agarose gel containing ethidium bromide. The residual supercoiled forms and the predominant nicked forms were observed by ultraviolet light and photographed. The photographic negative was scanned using a Molecular Dynamics Image Quant densitometer (Redwood City, CA, USA), and the proportion of nicked to supercoiled forms was calculated.

Gap filling former AP sites in the phagemids

Incorporation of ³²P-labeled guanine nucleotides into the created gapped sites of the phagemids was achieved by repairing the DNA with β pol. The DNA repair enzyme removed 5′ phosphate sugar remnants and incorporated dGMP at the former AP site, which had been incised by HAP1 (Matsumoto and Kim 1995). Purified β pol, and a stimulatory protein (either Hsp70, BSA, or variants of Hsp70), and 2.5 μCi of [α³²P]dGTP were added to gapped phagemid in 50 mM Tris-HCl, pH 8.0, 8 mM MgCl₂, 4 mM dithiothreitol (DTT) (12-μL reaction volume). The total amount of β pol and the final incubation times at 37°C are described in the figure legends. One unit (0.25 μL) of β pol as defined by the manufacturer catalyzes 1 nmol of dGTP to an acid-insoluble form in 1 hour at 37°C. After 30 minutes preincubation of stimulatory proteins, [α³²P]dGTP, and β pol on ice, the gapped phagemid was added, and incorporation was allowed to proceed at 37°C for 30 minutes. Autoradiograms of the gel, made by exposing it to X-ray film (Fuji, Tokyo, Japan), were scanned to determine the intensity of label incorporated into the phagemid DNA.

[α³²P]dGMP incorporation into the nicked phagemids was scored and related to the amounts of β pol or stimulatory proteins. No label was detectable in the supercoiled phagemid forms, although they accounted for 16% of the phagemid placed on the gel.

N-mutant protein of Hsp70 is a ∼44-kDa deletion containing N-terminal amino acids 1–385. C-mutant protein of Hsp70 is a deletion containing C-terminal amino acids 386–640 and GST, an enzyme that contributed to its 56-kDa size. Proteins were determined by the method of Bradford (1976).

Incorporation of ³²P-labeled deoxycytidine monophosphate at an abasic site in an oligonucleotide

A double-stranded 30-bp oligonucleotide was made by annealing 200 pmol of the top strand (shown below), which contained a single uracil 12 nucleotides from the 5′ end, with 300 pmol of the complementary strand.

5′-ATA TAC CGC GGU CGG CCG ATC AAG CTT ATT-3′

3′-TAT ATG GCG CCG GCC GGC TAG TTC GAA TAA-5′

After digestion of the duplex oligonucleotide for 5 minutes in 8 mM MgCl₂, 50 mM Tris, pH 8, 4 mM DTT at 37°C with 31.5 pmol UDG to remove the U and create an abasic site, it was then incised at the abasic site by addition of 35 pmol of HAP1 and further incubation for 5 minutes at 37°C. After phenol extraction and ethanol precipitation, including 18 μg of plasmid pBR322 as a carrier, the oligonucleotide was repaired at the newly created gap by incubation with β pol in buffer containing 5 μCi of [α³²P]dCTP and Hsp70 or BSA, as described in the text.

Proteins were removed from the labeled oligonucleotide by phenol extraction and ethanol precipitation before denaturation with formamide and electrophoresis on a 20% denaturing acrylamide gel containing 7 M urea. The radiolabeled DNA was detected by autoradiography and quantitative densitometry. A small aliquot of each reaction was analyzed on an agarose gel containing ethidium bromide to verify equivalent recovery of the plasmid carrier and, therefore, also of each reaction.

Ligase sealing of filled oligonucleotide gaps

After gap filling by β pol, ligase sealing of oligonucleotide was achieved in 30 minutes or 45 minutes incubation at 20°C in 50 mM Tris, pH 7.5, 10 mM MgCl₂, 10 mM DTT, and 1 mM ATP with T4 DNA ligase or human DNA ligase 1, the predominant human DNA ligase.

RESULTS

Hsp70 stimulates β pol repair of phagemid pBKS DNA

After depurination and depyrimidination of the phagemids by heat and low pH, as described in Experimental Procedures, phagemids were converted to more slowly migrating nicked forms by endonucleolytic cleavage at AP sites with HAP1 (Fig 1A). The DNA contents of aliquots of nicked plasmids in each of the 15 lanes were identical, as determined by ethidium bromide fluorescence measurements. Sixteen percent of the phagemids were left unnicked and had therefore retained their supercoiled forms, indicating an average of 2 AP sites per phagemid molecule, ie, 84% nicked. Although the amounts of nicked DNA in lanes 1–15 were identical, there were marked differences in gap filling measured by incorporation of [α³²P]dGMP by β pol when different stimulator proteins were included in the labeling mixtures (Fig 1B). Supercoiled forms were free of AP sites and therefore resisted HAP1 and repair, as indicated by little or no incorporation of [³²P]dGMP. No label could be detected in unnicked supercoiled forms in contrast to intense labeling of the gapped forms when the phagemid preparations were separated by agarose gel electrophoresis.

Fig 1.

Heat shock protein 70 (Hsp)70 stimulates β pol repair of pBKS deoxyribonucleic acid after strand scission at AP sites by human apurinic-apyrimidinic endonuclease. (A) Gapped pBKS was radiolabeled during 30 minutes of repair (see Experimental Procedures). Each 12-μL incubation mixture contained 0.150 μg of the phagemid, 2.5 μCi of [α³²P]deoxyguanosine triphosphate (dGTP), 1.0 μg of each stimulatory protein, and 1 μL (4 units) of β pol diluted: 1/50 (lanes 5, 10, and 15), 1/100 (lanes 4, 9, and 14), 1/200 (lanes 3, 8, and 13), or 1/400 (lanes 2, 7, and 12). Buffer replaced β pol in treatment groups of lanes 1, 6, and 11. Stimulatory proteins: Hsp70 (lanes 1–5); Hsp70 N-mutant protein (lanes 6–10); Hsp70 C-mutant protein (lanes 11–15). Lane 16 contained 0.100 μg of unlabeled native pBKS. Nicked and covalently closed circular (CCC) forms are shown. CCC signify supercoiled forms. Binding of β pol, [α³²P]dGTP, and stimulatory proteins for 30 minutes on ice was followed by addition of pBKS and incubation at 37°C for 30 minutes followed by gel electrophoresis in 0.8% agarose, stained with ethidium bromide and observed under ultraviolet light, as shown in (A). (B) An aligned autoradiogram of the gel in (A) shows incorporation only into the relaxed forms of pBKS; lane numbers correspond to those in (A). The film was exposed for 40 minutes. (C) Densitometer determinations of the autoradiogram of 1B are plotted; the density determination of the N-mutant protein is omitted; •—•, Hsp70; ▵—▵, C-mutant protein. Data from an otherwise similar experiment, in which incorporation at 37°C was for 15 minutes; •—•, Hsp70; ○—○, bovine serum albumin; •—•, no stimulatory protein. The film was exposed for 60 minutes

Hsp70 stimulated β pol, as seen (Fig 1B) by the greater nucleotide incorporation achieved when Hsp70 was present during 30 minutes of repair at 37°C. In an otherwise identical repair experiment, with labeling for 15 minutes, half as much ³²PdGMP was incorporated by β pol diluted 1/50. BSA was 10- to 100-fold less active in stimulating β pol (Fig 1C). In the absence of added protein, nucleotide incorporation into phagemids by β pol was still detectable, but it was 100-fold less than when Hsp70 was present (Fig 1C).

Hsp70 and the truncated 44-kDa N-terminal fragment of Hsp70 and a 56-kDa C-terminal fragment of Hsp70 that contained an additional GST fragment were compared in an experiment similar to that shown in Fig 1 A,B. The study compared the ability of the proteins to stimulate β pol. Data of Fig 1C clearly show that Hsp70 was more efficient than the C-terminal fragment of Hsp70.

Figure 2a shows that incorporation of [³²P]dGMP into nicked pBKS increased faster when Hsp70 was present; it increased somewhat less rapidly in the presence of BSA and even more slowly without a stimulatory protein. Incorporation of ³²P-labeled guanine nucleotide into the gapped DNA of pBKS would be limited by the amount of phagemid substrate subjected to repair and the initial frequency of AP sites in each pBKS molecule. Therefore, the time course for increase in ³²P label in pBKS would be expected to start from identically low levels of incorporation, increasing to achieve a common plateau when all repair sites were filled. Gapped phagemids were repaired by incubation with [α³²P]dGTP, as described in Experimental Procedures. As expected, the same intensity of label was eventually achieved with Hsp70 and BSA because repair was completed, albeit more slowly, with BSA. Repair of the DNA by β pol shown in the single-sample determinations of Figure 2a was 1.56-, 1.49-, and 1.59- fold greater at 2 minutes, 5 minutes, and 15 minutes, respectively, when Hsp70 was present than with equivalent concentrations of BSA. These results are in general agreement with the results shown in Figure 1C and in other kinetic experiments. For example, Figure 2b again shows more rapid repair with Hsp70 than with BSA; the β pol concentration was reduced to half of that used in the experiment shown in Figure 2a.

Fig 2.

(a) Time course of repair of single-strand gaps in pBKS by β pol. The time course of incorporation of [α³²P]deoxyguanosine monophosphate was determined by measuring incorporation into pBKS, as described in Experimental Procedures and in the experiments of Figure 1. Incubation was stopped at the times shown, and incorporation into relaxed forms was determined by densitometry measurements of a gel autoradiogram, as in Figure 1B. •, 0.25 μg heat shock protein 70; ○, 0.25 μg bovine serum albumin; •, no added stimulatory proteins. β pol (4 units/μL) was diluted 1/50 and then added to the 12-μL reaction volume as described in Experimental Procedures. (b) The time course of repair was measured in an experiment similar to that of (a), with β pol present at half the concentration used in (a)

The amount of [³²P]dGMP incorporated reached 6.6 × 10⁻⁸ μmol for 10 × 10⁻⁸ μmol of gapped phagemid (∼0.150 μg), ie, 0.66 mol of dGMP for 1.0 mol of pBKS. Because 84% of the pBKS was gapped, (ie, 16% unlabeled, supercoiled and 84% labeled, nicked forms), the expected number of gapped sites would be ∼2 per pBKS molecule. Because approximately one-fourth of the pBKS DNA nucleotides would be expected to be GMP, only ∼0.5 of the repaired sites would have been available for repair. In fact, 0.66 molecules per pBKS were found to be in good agreement with this estimate. Depurination at guanines is more frequent than with the other bases. Differences between the stimulatory ability of different proteins are greater with intermediate concentrations of DNA repair enzymes such as β pol (Fig 1C) and HAP1 (Mendez et al 2003). With β pol, as in Figure 2a, Hsp70 was more potent than BSA (Fig 5) or the C mutant (Fig 1C). With the N-terminal fragment, interference from an associated dGTPase prevented comparisons. To circumvent this, a 43-kDa and a 48-kDa N-terminal protein were prepared by 18 hours of digestion of Hsp70 by immobilized trypsin (Fig 3). Comparisons were then possible (Fig 4 A,B).

Fig 5.

Stimulation of repair of an oligonucleotide by heat shock protein 70 (Hsp70). Gapped 30-mer duplexes were incubated with 5 μCi [α³²P]deoxycytidine triphosphate, β pol supplemented with 0.5 μg Hsp70 (lanes 1–3) or with 0.5 μg bovine serum albumin (lanes 4–6), or no stimulatory protein (lanes 7–9). Each reaction (20 μL) included 400 ng of gapped duplex, (20 pmol containing ∼0.3 pmol AP sites), 2 μL of β pol at a 1/50 dilution (lanes 1, 4, and 7), 1/200 dilution (lanes 2, 5, and 8), or buffer only (lanes 3, 6, and 9). Reactions at 1/50 dilution of β pol contained 0.16 units, and reactions at a 1/200 dilution of β pol contained 0.04 units of the enzyme. The denatured samples were analyzed on a 20% acrylamide denaturing gel, as described in Experimental Procedures and in the text. the upper panel represents an autoradiogram from a 3-hour exposure of the gel, and the lower panel represents an 18-hour exposure. Hsp70 (0.5 μg) is 7.0 pmol. BSA (0.5 μg) is ∼7.4 pmol

Fig 3.

Release of 43 kDa and 48 kDa tryptic digest fragments from heat shock protein 70 (Hsp70) at 37°C. After centrifugation to pellet the immobilized trypsin, the supernatants containing digestion products were resolved by electrophoresis on a 12% sodium dodecyl sulfate acrylamide gel that was stained with silver nitrate. The estimated gel inputs of Hsp70 in lanes 1, 2, and 3, before digestion by trypsin, were 800, 400, and 160 ng, respectively. Lanes 4, 5, and 6 represent 250, 125, and 63 ng of undigested Hsp70, respectively. Lane 7 has generic protein molecular weight markers

Fig 4.

Equivalent stimulation of β pol by heat shock protein 70 (Hsp70) and Hsp70 tryptic fragments in a time course of repair. (A) One microliter of a 1/50 dilution of β pol (containing 0.08 units of β pol) was included in 11.4 μL of incubation mixture containing stimulatory proteins (0.20 μg), plasmid (0.15 μg), 32p labeled dGTP, and buffer, as described in Experimental Procedures. •, Hsp70; ▴, tryptic fragments; □, tryptic digest with small peptides eliminated by size filtration through a Bio-Rad Micro-bio P-6 column. Samples taken during the time course of incubation were placed on a 0.8% agarose gel. After electrophoresis, the closed circular nicked plasmids were located by radioautography and ethidium bromide stain. (B) Radiolabel in repaired phagemid deoxyribonucleic acid. The labeled relaxed plasmids were cut out of the gel, and the radiolabel was determined by Cerenkov radiation in a scintillation counter. Symbols are similar to those found in Figure 4A

In a different comparison, we tested DnaK of E coli in the β pol repair assay (with the 30-bp oligonucleotide) because human Hsp70 has amino acid sequences 47% identical to those of the E coli Hsp DnaK (Hunt and Morimoto 1985). Nevertheless, Hsp70 stimulated β pol much more (>5-fold) than did DnaK of E coli, which itself was only slightly more potent than BSA. Interestingly, 2 mM ATP increased the ability of both Hsps to stimulate β pol in the presence of 8 mM Mg⁺² (results not shown).

Stimulation of β pol by Hsp70 tryptic fragment

To compare Hsp70 with the ATP-binding N-terminal portions of Hsp70 produced by trypsinization, 40 μg of Hsp70 was digested for 18 hours at 37°C with immobilized trypsin (Fig 3). Seventy-five percent of Hsp70 had been converted to 48-kDa and 43-kDa fragments, whose N-terminal amino acids were identical to those of Hsp70, as determined by Edman N-terminal amino acid analysis (Mendez et al 2003). Ten percent of the Hsp70 resisted trypsin digestion; smaller peptides accounted for the balance.

In a separate time course of [³²P]dGMP incorporation, the incorporated label was determined by direct counting of the ethidium bromide–stained nicked phagemid forms cut out of the agarose gel. We confirmed the near equivalence of Hsp70 and its 43 kDa and 48 kDa tryptic digests containing the N-terminal sequences of Hsp70 in stimulating β pol. Figure 4A shows that the ATPase domain prepared by tryptic digest when freed of small peptides by gel filtration was nearly equivalent to the unfractionated preparation, despite the presence of only ∼10% of the original full-length Hsp70. Silver staining and protein determinations (Bradford 1976) verified that extraneous proteins could not have contributed to the stimulation achieved with 0.2 μg of Hsp70, trypsin digest, and the Hsps free of small peptides. All 3 preparations stimulated repair of phagemid strand gaps to the same extent, as indicated by achieving similar plateau levels of label incorporation. A small reduction in the stimulation of β pol by the tryptic digests is likely due to mechanical losses in harvesting the Hsps from immobilized trypsin. Undamaged phagemids remained unlabeled as in the previous experiments (not shown). On comparison with radiolabel in the [α³²P]dGTP, we estimated that β pol finally had incorporated 4.8 × 10⁻⁸ μmol of dGMP into 10 × 10⁻⁸ μmol of phagemid, ie, ∼½ of a molecule of dGMP per phagemid. Because each phagemid had ∼2 AP sites that had been converted to single-strand gaps by HAP1, one-quarter of the 2 loci per phagemid had been repaired by β pol. This is consistent with dGMP incorporation, which is restricted to repair sites opposite cytidine. Figure 4B is a confirmation of the study shown in Figure 4A, but repair was determined by direct counting using Cerenkov radiation.

This result means that the ATPase domain causes most of the stimulation, an important conclusion of this article.

Incorporation of ³²P-labeled deoxycytidine monophosphate (dCMP) at the former AP site 12 nucleotides from the terminus of the 30-nucleotide-long gapped oligonucleotide was stimulated by Hsp70

The oligonucleotide was prepared as described in Experimental Procedures. 3600 nanograms of oligonucleotide duplex was divided into 9 aliquots and repaired with β pol supplemented with Hsp70, BSA, or no stimulatory protein addition (Experimental Procedures and the legend to Fig 5). One-tenth of each aliquot was removed to verify recovery of the labeled oligonucleotide after phenol extraction, by measuring the recovery of the pBR322 carrier, after the aliquots were separated by electrophoresis on an agarose gel containing ethidium bromide. Densitometry was done on the plasmid in the photographic negative of the gel. Recovery of the plasmids among the 9 aliquots agreed within 10% (not shown).

However, addition of 1 α³²P-labeled dCMP by β pol at the 3′ OH terminus of the 11-nucleotide-long segment at the gapped site of the duplex was 30-fold greater when Hsp70 was present than when equimolar amounts of BSA were used to stimulate [α³²P]dCMP incorporation. Incorporation of the ³²P-labeled oligonucleotide was quantitated from autoradiograms of the acrylamide gel; the resulting 12 mer was labeled and then dissociated from the unlabeled duplex after denaturation (Fig 5).

When [α³²P]dGTP was used instead of the complementary [α³²P]dCTP, in an otherwise identical experiment, no labeled 12 mers were detected, ie, fidelity of incorporation of dCMP opposite G was greater than 1000-fold compared with dGMP opposite G (not shown).

Ligation of oligonucleotides

Polymerase β and DNA ligase 1 are associated physically and functionally in human DNA BER (Tomkinson 1997). Therefore, we tested the influence of Hsps on the last step in BER ligase sealing. The final step in BER after β pol repair requires ligation of the repaired strand. With progressive ligation, the labeled 12 mers were sealed into less mobile 30 mers (Fig 6 A,B). As expected, labeled 12 mers progressively disappeared. With T4 DNA ligase, Hsp70 and BSA stimulated the reaction very little (Fig 6C). With human DNA ligase 1, Hsp70 and BSA did stimulate the reaction significantly, but these two proteins were nearly equivalent in this, unlike the previous results, in which β pol activity was stimulated more by Hsp70 than by BSA.

Fig 6.

Repair of a 30-bp gapped oligonucleotide by β pol and ligases. A gapped 30-bp oligonucleotide containing a single uridine 12 nucleotides from the 5′ terminus of 1 strand was prepared by sequential treatment with uracil–deoxyribonucleic acid (DNA) glycosylase and human apurinic-apyrimidinic endonuclease. Repair of the single-nucleotide gap was achieved by insertion of ³²P-labeled deoxycytidine monophosphate. Proteins and unincorporated radiolabel were then eliminated by phenol extraction and ethanol precipitation. Each 20-μL reaction included 6.6 pmol of ³²P-labeled oligonucleotide, T4 DNA ligase, and a protein supplement. (A) Aliquots of 3′ ³²P-labeled oligonucleotide were incubated at 20°C for 40 minutes with T4 DNA ligase from E coli to seal the 5′ side of the repair site, followed by boiling for 5 minutes in 50% formamide, quick quenching on ice, and electrophoresis on a 16% denaturing acrylamide gel containing 7 M urea. The amounts of T4 DNA ligase shown were supplemented with 0.5 μg of Hsp70, 0.5 μg of bovine serum albumin (BSA), or buffer. The autoradiogram was developed after 2 minutes exposure. (B) A similar study with human DNA ligase 1 is shown. ³²P-labeled oligonucleotide (3.3 pmol) was placed in each well. The gel autoradiogram was exposed for 17 minutes. (C) Conversion of 3′ ³²P-labeled 12 mers to 30 mers by ligation with T4 DNA ligase and human DNA ligase 1. The autoradiograms of Figure 5 (A and B) were scanned, and densitometer values from labeled 12 mers and repaired 30 mers from each incubation mixture were determined. The proportion of product 30 mers was expressed as a fraction of the total labeled oligonucleotide placed on the gel. With T4 DNA ligase 0.6 pmol was converted; with the human DNA ligase 0.04 pmol of 12 mers was converted during 30 minutes. The amount of T4 DNA ligase or human DNA ligase 1 in each reaction is shown on the abscissa. Hsp70 (0.5 μg) is 7.0 pmol, 1000 ng of T4 DNA ligase is 28 pmol, and 56 ng of human DNA ligase 1 is 5.0 pmol. Human DNA ligase 1, with Hsp70, • human DNA ligase 1 with BSA, ▴; and human DNA ligase 1 with buffer, ▪. T4 DNA ligase with Hsp70, ○; T4 DNA ligase with BSA, ▴; and T4 DNA ligase with buffer, □

Hsp-related proteins, Hsp70, Hsp27, the Hsp70 N-terminal mutant protein, and the Hsp70 C-terminal protein, and other proteins such as BSA and actin monomers were tested in DNA ligase stimulation experiments similar to those shown in Figure 6. All these proteins stimulated human DNA ligase 1 to approximately the same extent, compared with the results with unsupplemented buffer (results not shown). These results are in marked contrast to the stimulation of β pol and the other BER enzymes by Hsp70 and its domains.

DISCUSSION

These results indicate that Hsp70 and some of its derivatives stimulate β pol activity in BER, thereby strengthening the notion that Hsps may be of general importance in DNA repair. The wide distribution of Hsps in nature could well be due to their positive role in DNA repair of oxygen and radiation damage, an indication that Hsps could endow cells with improved survival.

Phagemids with AP sites nicked by HAP1 were the targets of repair by β pol. This approach enabled determination of the stimulatory role of the N-terminal potlike portion of Hsp70 and the lesser roles of other proteins. Precedents for assigning specific biological roles to the N-terminal portion of Hsp70 have been described for human cells and in Hsp70-like chaperone proteins of other species (Buczynski et al 2001; Chaudhuri et al 2001; Yoshida et al 2001; Hundley et al 2002). Most biological activities of Hsp70 have been ascribed to complete eukaryotic Hsp and bacterial (DnaK) Hsps.

Little is known about how Hsps stimulate β pol (first described in this article) and HAP1 (Kenny et al 2001). The crucial structural features of the stimulatory proteins are not known. Other proteins are less potent, and perhaps structural comparisons may reveal structural singularities. Presumably, the ATPase domain is most important. As described here, Hsp70, the compact 43 kDa and 48 kDa N-terminal proteins that were released from Hsp70 by trypsin are more stimulatory than the more extended structures such as BSA and the C-terminal mutant protein of Hsp70. Similar results were obtained in studies on HAP1 stimulation (Mendez et al 2003). We have found that Hsp27, which has a compact structure (Bukau and Horwich 1998), as well as 43-kDa compact actin monomers moderately stimulated UDG and HAP1 (unpublished observation). Nonspecific as well as specific interactions of Hsps with repair enzymes are to be expected (Kenny et al 2001). Actin monomers stimulated transcription of respiratory syncytial virus genome RNA, which was further stimulated by profilin (Burke et al 2000). Presumably, compact barrel-shaped structures such as these are more stimulatory. Structural features of the proteins that correlate with their efficiency of stimulation are now needed to better understand their specificity in DNA repair enzyme activity. Actin monomers did not stimulate β pol (unpublished observation). An example of protein enhancement of human endonuclease III activity has been reported (Marenstein et al 2001); Hsps activate the DNA origin binding protein of Herpes simplex virus, signaling another role involving DNA synthesis (Tanguy Le Gac and Boehmer 2002).

Human β pol makes a mistake in filling a 1-nucleotide gap for every 4500 nucleotides polymerized in certain codons (Osheroff et al 1999). Fidelity of repair is better in filling larger gaps. Our results, using the inappropriate [α³²P]dGTP to attempt to fill the gap in the duplex oligonucleotide substrate, showed less than 1 misincorporation per 1000 nucleotides.

In this article, we have shown that Hsp70 stimulated β pol more efficiently than did BSA or the Hsp70 C-terminal mutant protein. Similar to those results, Hsp70 and a truncated Hsp70 N-terminal portion of Hsp70 stimulated the BER enzyme, HAP1, more than did the unrelated proteins (Kenny et al 2001; Mendez et al 2003). However, with human DNA ligase 1, no stimulatory difference between Hsp70 and BSA could be determined. In our studies with T4 DNA ligase, no accessory proteins were required (Fig 6). Ligation appears to require little chaperone activity, possibly because of the restricted locus of its action, but other explanations are possible.

Hsps have been implicated in the induction of radiation resistance in the adaptive response (Lee et al 2002). Moreover, recent studies support an important role for BER in determining radiosensitivity (Vens et al 2002). The widespread conservation of Hsps in nature may be the result of selection, by protecting the genomes of cells from oxidation and radiation damages by stimulating DNA repair enzymes. In human Hsps, some clues to specific interactions with BER enzymes have been obtained (Mendez et al 2000; Kenny et al 2001; Mendez et al 2003).

Other chaperones such as Hsp90/40/110 might have similar abilities to stimulate β pol, but nothing about that is known.